DRAFT - DO NOT CITE OR QUOTE
www.epa.gov/iris
TOXICOLOGICAL REVIEW
OF
ACRYLAMIDE
(CAS No. 79-06-1)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
August 2009
NOTICE
This information is distributed solely for the purpose of pre-dissemination peer review under
applicable information quality guidelines. It has not been formally disseminated by EPA. It
does not represent and should not be construed to represent any Agency determination or policy.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document is a preliminary review draft for review purposes only. This information
is distributed solely for the purpose of pre-dissemination peer review under applicable
information quality guidelines. It has not been formally disseminated by EPA. It does not
represent and should not be construed to represent any Agency determination or policy. Mention
of trade names or commercial products does not constitute endorsement or recommendation for
use.
08/28/2009 i DRAFT-DO NOT CITE OR QUOTE
-------
CONTENTS —TOXICOLOGICAL REVIEW OF ACRYLAMIDE
(CAS No. 79-06 1)
LIST OF TABLES vii
LIST OF FIGURES xvii
ABBREVIATIONS AND ACRONYMS xviii
FOREWORD xxi
AUTHORS, CONTRIBUTORS, AND REVIEWERS xxii
1. INTRODUCTION 1
2. CHEMICAL AND PHYSICAL INFORMATION RELEVANT TO ASSESSMENTS 3
2.1. CHEMICAL AND PHYSICAL INFORMATION 3
2.2. SOURCES OF EXPOSURE, FATE AND TRANSPORT 4
3. TOXICOKINETICS RELEVANT TO ASSESSMENTS 12
3.1. ABSORPTION 12
3.2. DISTRIBUTION 19
3.3. METABOLISM 22
3.4. ELIMINATION 35
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS.. Error! Bookmark not
defined.
4. HAZARD IDENTIFICATION 59
4.1. STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS 59
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION 86
4.2.1. Oral Exposure 86
4.2.1.1. Subchronic Studies 86
4.2.1.2. Chronic Studies 91
4.2.2. Inhalation Exposure 103
4.2.2.1. Subchronic Studies 104
4.2.2.2. Chronic Studies 104
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION 104
4.3.1. Reproductive Toxicity Studies 104
4.3.2. Developmental Toxicity Studies 123
4.4 HERITABLE GERM CELL STUDIES 134
4.5. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES 141
4.5.1. Neurotoxicity Studies 141
4.5.2. Other Cancer Studies 142
4.6. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION 146
4.6.1. Neurotoxicity Studies Error! Bookmark not defined.
4.6.2. Genotoxicity Studies Error! Bookmark not defined.
4.7. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS Error!
Bookmark not defined.
4.7.1. Oral Error! Bookmark not defined.
4.7.2. Inhalation 160
4.7.3. Mode-of-Action Information 161
4.8. EVALUATION OF CARCINOGENICITY 167
4.8.1. Summary of Overall Weight of Evidence 167
08/28/2009 ii DRAFT-DO NOT CITE OR QUOTE
-------
4.8.2. Synthesis of Human, Animal, and Other Supporting Evidence 169
4.8.3. Mode of Action for Carcinogenicity 173
4.8.3.1. Hypothesized Mode of Action—Mutagenicity 174
4.8.3.2. Alternative Mode of Action—Disruption of Hormone Levels or Signaling 184
4.8.3.3. Conclusion About the Mode of Action 193
4.9. SUSCEPTIBLE POPULATIONS 194
4.9.1. Possible Childhood Susceptibility 194
4.9.2. Possible Gender Differences 196
4.9.3. Other 196
5. DOSE-RESPONSE ASSESSMENTS 198
5.1. ORAL REFERENCE DOSE 198
5.1.1. Choice of Principal Study and Critical Effect—with Rationale and
Justification 198
5.1.2. Methods of Analysis—Including Models (PBTK, BMD, etc.) 203
5.1.3. RfD Derivation—Including Application of Uncertainty Factors 205
5.1.4. Previous RfD Assessment 216
5.2. INHALATION REFERENCE CONCENTRATION (RfC) 216
5.2.1. Choice of Principal Study and Critical Effect—with Rationale and
Justification 216
5.2.2. Methods of Analysis—Including Model (PBTK, BMD, etc.) 218
5.2.3. RfC Derivation—Including Application of Uncertainty Factors 218
5.2.4. Previous RfC Assessment 219
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE AND INHALATION
REFERENCE CONCENTRATION 220
5.3.1 Areas of Uncertainty 221
5.3.2 Uncertainty Factors in Deriving the RfD and RfC..Error! Bookmark not defined.
5.4. CANCER ASSESSMENT 229
5.4.1. Choice of Study/Data—with Rationale and Justification 229
5.4.2. Dose-Response Data 231
5.4.3. Dose Adjustments and Extrapolation Method(s) 232
5.4.4. Human Equivalent Concentration Using the PBTK ModelError! Bookmark not defined.
5.4.5. Oral Slope Factor and Inhalation Unit Risk 241
5.4.5.1. Oral Slope Factor 241
5.4.5.2. Inhalation Unit Risk 242
5.4.6 Application of Age-Dependent Adjustment Factors 244
5.4.7. Uncertainties in Cancer Risk Values 246
5.4.7.1. Areas of Uncertainty 250
5.4.8. Previous Cancer Assessment 254
5.5. QUANTITATING RISK FOR HERITABLE GERM CELL EFFECTS 255
5.5.1. Quantitative Approaches 256
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD 267
AND DOSE RESPONSE 267
6.1. HUMAN HAZARD POTENTIAL 267
6.2. DOSE RESPONSE 270
6.2.1. Noncancer/Oral 270
6.2.2. Noncancer/Inhalation 271
6.2.3. Cancer/Oral 273
6.2.4. Cancer/Inhalation 274
08/28/2009 iii DRAFT-DO NOT CITE OR QUOTE
-------
7. REFERENCES 276
APPENDIX A. Summary of External Peer Review and Public Comments and Disposition 1
APPENDIX B. MUTAGENICITY TEST RESULTS 1
APPENDIX C. DOSE-RESPONSE MODELING FOR DERIVING THE RfD 1
APPENDIX D. DOSE-RESPONSE MODELING FOR CANCER 1
APPENDIX E. KIRMAN ET AL. (2003) PBTK MODEL SUPPORTING DOCUMENTATION
32
APPENDIX F. YOUNG ET AL (2007) PBTK/TD MODEL SUPPORTING
DOCUMENTATION Error! Bookmark not defined.
LIST OF TABLES vii
LIST OF FIGURES xvii
LIST OF ABBREVIATIONS AND ACRONYMS xviii
FOREWORD xxi
AUTHORS, CONTRIBUTORS, AND REVIEWERS xxii
1. INTRODUCTION 1
2. CHEMICAL AND PHYSICAL INFORMATION RELEVANT TO ASSESSMENTS 3
2.1. CHEMICAL AND PHYSICAL INFORMATION 3
2.2. SOURCES OF EXPOSURE, FATE AND TRANSPORT 4
3. TOXICOKINETICS RELEVANT TO ASSESSMENTS 12
3.1. ABSORPTION 12
3.2. DISTRIBUTION 19
3.3. METABOLISM 22
3.4. ELIMINATION 35
3.5. HEMOGLOBIN ADDUCTS AND URINARY METABOLITES AS BIOMARKERS
OF EXPOSURE 41
3.6. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS 52
4. HAZARD IDENTIFICATION 59
4.1. STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS 59
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION 86
4.2.1. Oral Exposure 86
4.2.1.1. Subchronic Studies 86
4.2.1.2. Chronic Studies 91
4.2.2. Inhalation Exposure 103
4.2.2.1. Subchronic Studies 104
4.2.2.2. Chronic Studies 104
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION 104
4.3.1. Reproductive Toxicity Studies 104
4.3.2. Developmental Toxicity Studies 123
4.4. HERITABLE GERM CELL STUDIES 134
4.5. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES 141
4.5.1. Neurotoxicity Studies 141
4.5.2. Other Cancer Studies 142
4.6. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION 146
4.6.1. Studies on the Hypothalamus-Pituitary-Thyroid Axis 146
4.6.2. Genotoxicity Studies 149
08/28/2009 iv DRAFT-DO NOT CITE OR QUOTE
-------
4.7. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS 155
4.7.1. Oral 155
4.7.2. Inhalation 160
4.7.3. Mode-of-Action Information 161
4.7.3.1. NeurotoxicEffects 161
4.7.3.2. Reproductive Effects 166
4.8. EVALUATION OF CARCINOGENICITY 167
4.8.1. Summary of Overall Weight of Evidence 167
4.8.2. Synthesis of Human, Animal, and Other Supporting Evidence 169
4.8.3. Mode of Action for Carcinogenicity 173
4.8.3.1. Hypothesized Mode of Action—Mutagenicity 174
4.8.3.2. Alternative Mode of Action—Disruption of Hormone Levels or Signaling 184
4.8.3.3. Conclusion About the Mode of Action 193
4.9. SUSCEPTIBLE POPULATIONS 194
4.9.1. Possible Childhood Susceptibility 194
4.9.2. Possible Gender Differences 196
4.9.3. Other 196
5. DOSE-RESPONSE ASSESSMENTS 198
5.1. ORAL REFERENCE DOSE 198
5.1.1. Choice of Principal Study and Critical Effect—with Rationale and
Justification 198
5.1.2. Methods of Analysis—Including Models (BMD, equivalent AUCs, in vivo
rate constants, etc.) 203
5.1.3. RfD Derivation—Including Application of Uncertainty Factors 205
5.1.4. Previous RfD Assessment 216
5.2. INHALATION REFERENCE CONCENTRATION (RfC) 216
5.2.1. Choice of Principal Study and Critical Effect—with Rationale and
Justification 216
5.2.2. Methods of Analysis— Including Models (BMD, equivalent AUCs, in
vivo rate constants, etc.) 218
5.2.3. RfC Derivation—Including Application of Uncertainty Factors 218
5.2.5. Previous RfC Assessment 219
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE AND INHALATION
REFERENCE CONCENTRATION 220
5.3.1. Areas of Uncertainty 224
5.4. CANCER ASSESSMENT 229
5.4.1. Choice of Study/Data—with Rationale and Justification 229
5.4.2. Dose-Response Data 231
5.4.3. Dose Adjustments and Extrapolation Method(s) 232
5.4.4. Human Equivalent Concentration (HEC) - based on equivalent areas under
the time-concentration curve (AUC) for serum AA or GA 239
5.4.5. Oral Slope Factor and Inhalation Unit Risk 241
5.4.5.1. Oral Slope Factor 241
5.4.5.2. Inhalation Unit Risk 242
5.4.6. Application of Age-Dependent Adjustment Factors 244
5.4.7. Uncertainties in Cancer Risk Values 246
5.4.7.1. Areas of Uncertainty 250
5.4.8. Previous Cancer Assessment 254
08/28/2009 v DRAFT-DO NOT CITE OR QUOTE
-------
5.5. QUANTITATINGRISK FOR HERITABLE GERM CELL EFFECTS 255
5.5.1. Quantitative Approaches 256
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD 267
AND DOSE RESPONSE 267
6.1. HUMAN HAZARD POTENTIAL 267
6.2. DOSE RESPONSE 270
6.2.1. Noncancer/Oral 270
6.2.2. Noncancer/Inhalation 271
6.2.3. Cancer/Oral 273
6.2.4. Cancer/Inhalation 274
7. REFERENCES 276
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC COMMENTS
AND DISPOSITION 1
APPENDIX B. MUTAGENICITY TEST RESULTS 1
APPENDIX C. DOSE-RESPONSE MODELING FOR DERIVING THE RfD 1
APPENDIX D. DOSE-RESPONSE MODELING FOR CANCER 1
APPENDIX E. DERIVATION OF IN VIVO SECOND ORDER RATE CONSTANTS AND
THE ADDUCT FORMATION SIMULATION MODEL 1
APPENDIX F. ALTERNATE RFC BASED ON HUMAN EPIDEMIOLOGY DATA 1
08/28/2009 vi DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF TABLES
Table 2-1. Summary of acrylamide levels in food (ppb) derived from the FDA data collected
from 2002 through October 1,2003) 7
Table 2-2. Acrylamide levels in food (ppb) as collected by the European Union Joint Research
Center (updated June 2004) 8
Table 2-3. Exposure estimates from 2002-2006 9
Table 2-4. Summary of exposure estimates (|ig/kg-day) by sources and population groups 10
Table 3-1. Second order rate constants for reaction of acrylamide or glycidamide with the N-
terminal valine residue of hemoglobin Error! Bookmark not defined.
Table 3-2. Metabolites detected in urine collected for 24 hours following oral administration of
[l,2,3-13C]-labeled acrylamide (50 mg/kg) to male F344 rats or male B6C3F1 mice27
Table 4-1. Observed deaths and SMRs for selected causes by follow up period for all workers
(compared with the general US population) 64
Table 4-2. Observed deaths and SMRs for selected cancer sites by duration of employment, time
since first employment, and measures of exposure to acrylamide, all U.S. workers,
1950-1994 (compared with the local male populations) 66
Table 4-3. Neurological symptoms self-reported by acrylamide workers and nonexposed
workers 77
Table 4-4. Scoring system for the neurotoxicity index 79
Table 4-5. Group means ± SD of biomarkers in different categories of workers 81
Table 4-6. Correlation coefficients (linear regression) for relationships between biomarkers and
neurotoxicity index 82
Table 4-7. Incidences of symptoms in 210 tunnel workers classified into exposure groups based
on levels of hemoglobin adducts of acrylamide 84
Table 4-8. Light and electron microscopic data for left sciatic nerves from rats exposed to
acrylamide in drinking water for 90 days 90
Table 4-9. Light microscopic data for tibial nerves from F344 rats exposed to acrylamide in
drinking water for 2 years Error! Bookmark not defined.
Table 4-10. Incidences of selected tumors in male and female F344 rats exposed to acrylamide
in drinking water for 2 years 95
Table 4-11. Dosing parameters of groups of rats given acrylamide in drinking water for 106-108
weeks in the carcinogenicity study 96
Table 4-12. Light microscopic data for sciatic nerves from F344 rats exposed to acrylamide in
drinking water for 2 years 98
Table 4-13. Incidences of tumors in male F344 rats exposed to acrylamide in drinking water for
2 years 99
08/28/2009 vii DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-14. Incidences of tumors in female F344 rats exposed to acrylamide in drinking water
for 2 years 100
Table 4-15. Reevaluation and comparison of mesothelial lesions and extent of Leydig cell
neoplasia in male F344 rats exposed to acrylamide in drinking water for 2 years... 103
Table 4-16. Changes in reproductive parameters in F344 rats exposed to acrylamide in drinking
water for two generations 107
Table 4-17. Results of the dominant lethal mutation assay in F344 rats 108
Table 4-18. Results of dominant lethality testing in male Swiss CD-I mice exposed to
acrylamide in the drinking water Ill
Table 4-19. Effects of acrylamide in drinking water on grip strength of mice 112
Table 4-20. Fertility rates and pregnancy outcomes in Long-Evans rats following 72-day oral
exposure of males to acrylamide in the drinking water 114
Table 4-21. Results of sperm analysis (baseline and week 9) and male fertility testing (following
10 weeks of treatment) of Long-Evans rats exposed to acrylamide in the drinking
water 118
Table 4-22. Reproductive effects following exposure of male ddY mice to acrylamide in
drinking water for 4 weeks and subsequent mating with untreated females 121
Table 4-23. Maternal and fetal effects in Sprague-Dawley rats and CD-I mice following oral
(gavage) administration of acrylamide to pregnant dams 125
Table 4-24. Differences in marker enzymes in the small intestine of pups cross-fostered to
acrylamide-treated or control dams during postnatal lactation 133
Table 4-25. Frequency of translocation carriers in offspring derived from males exposed to
acrylamide or glycidamide 135
Table 4-26. Results for specific locus mutations recovered in offspring of male mice exposed i.p
to 50 mg/kg acrylamide on 5 consecutive days 135
Table 4-27. Results for specific locus mutations recovered in offspring of male mice exposed to
acrylamide as a single 100 or 125 mg/kg i.p. dose 136
Table 4-29. Acrylamide initiation of squamous cell carcinomas or papillomas in female
SENCARmice 142
Table 4-30. Acrylamide initiation of skin tumor masses > 1mm in female SENCAR mice.... 143
Table 4-31. Noncancer effects in animals repeatedly exposed to acrylamide by the oral route
Error! Bookmark not defined.
Table 4-32. Neurological effects following exposure to acrylamide in species other than the rat
and mouse 158
Table 4-33. Incidence of tumors with statistically significant increases in both 2-year bioassays
with F344 rats exposed to acrylamide in drinking water 171
08/28/2009 viii DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-34. Circulating thyroid hormone levels in F344 rats following exposure to acrylamide in
drinking water for 14 or 28 days Error! Bookmark not defined.
Table 4-35. Plasma TSH, BrdU incorporation in thyroid, and PCNA expression in thyroid in
male Sprague-Dawley rats exposed to acrylamide by an unspecified route for up to 28
days 193
Table 5-1. Incidence data for degenerative changes detected by light microscopy in nerves of
male and female F344 rats exposed to acrylamide in drinking water for 2 years .... 204
Table 5-2. Predictions (mg/kg-day) from best-fitting models for doses associated with a 10, 5,
and 1% extra risk for nerve degeneration in male and female rats exposed to
acrylamide in drinking water 205
Table 5-3. Predictions (mg/kg-day) from best-fitting models for doses associated with 10, 5, and
1% extra risk for sciatic nerve changes in male and female rats exposed to acrylamide
in drinking water 205
Table 5-4. PBTK model simulation results for HEC based on the rat neurotoxicity BMD . Error!
Bookmark not defined.
Table 5-5. PBTK model simulation results for HEC based on the rat neurotoxicity BMD . Error!
Bookmark not defined.
Table 5-6. Estimated POD (mg/kg-day) from best-fitting models for doses associated with a 5%
extra risk for nerve degeneration in male and female rats exposed to acrylamide in
drinking water 227
Table 5-7. Summary of uncertainty in the acrylamide noncancer risk assessment Error!
Bookmark not defined.
Table 5-8. Incidence of tumors with statistically significant increases in a 2-year bioassay with
F344 rats exposed to acrylamide in drinking water 232
Table 5-9. Points of departure from multistage model fits and rat slope factors derived from
incidences of mammary tumors alone, thyroid tumors alone, or combined incidence
of mammary or thyroid tumors in female rats exposed to acrylamide in drinking water
235
Table 5-10. Predictions from time-to-tumor model for doses associated with 10% extra risk for
TVM alone, thyroid tumors alone, or combined TVM or thyroid tumors in male rats
exposed to acrylamide in drinking water, with associated rat cancer slope factors . 235
Table 5-11. PBTK model simulation results for HEC based on male rat carcinogenicity data
Error! Bookmark not defined.
Table 5-12. PBTK model simulation results for HEC to derive the inhalation unit risk based on
male rat oral exposure cancer data Error! Bookmark not defined.
Table 5-13. Summary of uncertainty in the acrylamide cancer risk assessment Error!
Bookmark not defined.
08/28/2009 ix DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-14. Heritable genetic risk estimates for humans exposed to acrylamide 264
Table C-l. Incidence data for degenerative changes detected by light microscopy in nerves of
male and female F344 rats exposed to acrylamide in drinking water for 2 years 1
Table C-2. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
nerve degeneration in male rats exposed to acrylamide in drinking water 2
Table C-3. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
nerve degeneration in female rats exposed to acrylamide in drinking water 3
Table C-4. Predictions (mg/kg-day) from best-fitting models for doses associated with a 10, 5,
and 1% extra risk for nerve degeneration in male and female rats exposed to
acrylamide in drinking water 4
Table C-5. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
sciatic nerve changes in male rats exposed to acrylamide in drinking water 5
Table C-6. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
sciatic nerve changes in female rats exposed to acrylamide in drinking water 6
Table C-7. Predictions (mg/kg-day) from best-fitting models for doses associated with 10, 5, and
1% extra risk for sciatic nerve changes in male and female rats exposed to acrylamide
in drinking water 7
Table D-l. Incidence of tumors with statistically significant increases in the second 2-year
bioassay with F344 rats exposed to acrylamide in drinking water 1
Table D-2. Risk estimate derived from separate and combined incidence of mammary or thyroid
tumors in female F344 rats exposed to acrylamide in drinking water 4
Table D-3. Risk estimates derived from separate and summed dose-response modeling of
mammary and thyroid tumors in female F344 rats exposed to acrylamide in drinking
water 6
Table D-4. Risk estimates for separate and combined incidence of TVMs or thyroid tumors in
male rats exposed to acrylamide in drinking water 7
Table D-5. Risk estimates derived from modeling separate and summed incidence of TVM and
thyroid tumors in male F344 rats exposed to acrylamide in drinking water 8
Table E-l: Original Model Parameter Values for Rats in the Kirman et al. (2003) PBTK Model.
Source: Kirman et al. (2003) Error! Bookmark not defined.
Table E-2: Data used to recalibrate the Kirman et al. (2003) model parametersError! Bookmark
not defined.
Table E-3: AUC Predictions from the Original Kirman Model versus AUCs Derived from
Hemoglobin AdductData Error! Bookmark not defined.
Table E-4:Recalibrated PBTK Model Parameter Values for the Rat Error! Bookmark not
defined.
08/28/2009 x DRAFT-DO NOT CITE OR QUOTE
-------
Table E-5: Results of the Recalibratedl Kirman et al (2003) Model versus Urinary Metabolite
Data and AUCs Derived from Hemoglobin Adduct Data Error! Bookmark not
defined.
Table E-6: Estimated Internal AUC Acrylamide and Glycidamide Doses Produced by Various
Drinking Water Intakes 32
Table E-7: Available Data for Calibration of the Human PBTK model Error! Bookmark not
defined.
Table E-8. Parameters for the Human (male) Acrylamide PBTK Model 37
Table E-9: Human PBTK Model Predictions versus AUCs and Urinary Metabolites Error!
Bookmark not defined.
Table E-10: Estimated AUCs in Humans for Acrylamide and Glycidamide from a Drinking
Water Exposure Error! Bookmark not defined.
Table E-l 1: Estimated AUCs in Humans for Acrylamide and Glycidamide from An Inhalation
Exposure Error! Bookmark not defined.
Table F-l: Data Generated at NCTR on AA and GA in rats and mice Error! Bookmark not
defined.
Table F-2: Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration
to Rats [Mean ± Standard Deviation (Range)] Error! Bookmark not defined.
Table F-3: Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration
to Micea Error! Bookmark not defined.
Table F-5:. Pharmacokinetic Parameters from AA Administration to Human Volunteers...Error!
Bookmark not defined.
Table 2-1. Summary of acrylamide levels in food (ppb) derived from the FDA data collected
from 2002 through October 1,2003 7
Table 2-2. Acrylamide levels in food (ppb) as collected by the European Union Joint Research
Center (updated June 2004) 8
Table 2-3. Exposure estimates from 2002-2006 9
Table 2-4. Summary of exposure estimates (|ig/kg-day) by sources and population groups 10
Table 3-1. Urinary metabolites collected for 24 hours following oral administration of [1,2,3-
13C]-labeled acrylamide (50 mg/kg) to male F344 rats or male B6C3Fi mice 27
Table 3-2. Comparison of molar percentage of dose excreted in urine of rodents and humans
after oral administration of acrylamide (Source: Hartmann et al., 2009) 38
Table 3-3. The advantages and disadvantages of available biomarkers of exposure for
acrylamide (from Hays and Alyward 2008b) 42
Table 3-4. Estimated human serum AA AUC normalized to administered dose based on
measured Hb adduct levels and in vitro derived second order rate constants (from
Fennell et al., 2005) 43
08/28/2009 xi DRAFT-DO NOT CITE OR QUOTE
-------
Table 3-5. Second order rate constants for reaction of acrylamide or glycidamide with the N-
terminal valine residue of hemoglobin 47
Table 3-6. Measured and estimated AA AUCs normalized to dose in humans and F344 rats.... 48
Table 3-7. Measured and estimated GA AUCs normalized to Dose in Humans and F344 rats.. 49
Table 3-8. Selected published measurements of acrylamide-derived hemoglobin adducts and
urinary metabolites in groups of nonsmokersa (from Doerge et al., 2008) 50
Table 4-1. Observed deaths and SMRs for selected causes by follow up period for all workers
(compared with the general U.S. population) 64
Table 4-2. Observed deaths and SMRs for selected cancer sites by duration of employment, time
since first employment, and measures of exposure to acrylamide, all U.S. workers,
1950-2002 (compared with the local male populations) 66
Table 4-3. Neurological symptoms self-reported by acrylamide workers and nonexposed
workers 77
Table 4-4. Scoring system for the neurotoxicity index 79
Table 4-5. Group means ± SD of biomarkers in different categories of workers 81
Table 4-6. Correlation coefficients (linear regression) for relationships between biomarkers and
neurotoxicity index 82
Table 4-7. Incidences of symptoms in 210 tunnel workers classified into exposure groups based
on levels of hemoglobin adducts of acrylamide 84
Table 4-8. Light and electron microscopic data for left sciatic nerves from rats exposed to
acrylamide in drinking water for 90 days 90
Table 4-9. Light microscopic data for tibial nerves from F344 rats exposed to acrylamide in
drinking water for 2 years 94
Table 4-10. Incidences of selected tumors in male and female F344 rats exposed to acrylamide
in drinking water for 2 years 95
Table 4-11. Dosing parameters of groups of rats given acrylamide in drinking water for 106-108
weeks in the carcinogenicity study 96
Table 4-12. Light microscopic data for sciatic nerves from F344 rats exposed to acrylamide in
drinking water for 2 years 98
Table 4-13. Incidences of tumors in male F344 rats exposed to acrylamide in drinking water for
2 years 99
Table 4-14. Incidences of tumors in female F344 rats exposed to acrylamide in drinking water
for 2 years 100
Table 4-15. Reevaluation and comparison of mesothelial lesions and extent of Leydig cell
neoplasia in male F344 rats exposed to acrylamide in drinking water for 2 years . 103
Table 4-16. Changes in reproductive parameters in F344 rats exposed to acrylamide in drinking
water for two generations 107
08/28/2009 xii DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-17. Results of the dominant lethal mutation assay in F344 rats 108
Table 4-18. Results of dominant lethality testing in male Swiss CD-I mice exposed to
acrylamide in the drinking water Ill
Table 4-19. Effects of acrylamide in drinking water on grip strength of mice 112
Table 4-20. Fertility rates and pregnancy outcomes in Long-Evans rats following 72-day oral
exposure of males to acrylamide in the drinking water 114
Table 4-21. Results of sperm analysis (baseline and week 9) and male fertility testing (following
10 weeks of treatment) of Long-Evans rats exposed to acrylamide in the drinking
water 118
Table 4-22. Reproductive effects following exposure of male ddY mice to acrylamide in
drinking water for 4 weeks and subsequent mating with untreated females 121
Table 4-23. Maternal and fetal effects in Sprague-Dawley rats and CD-I mice following oral
(gavage) administration of acrylamide to pregnant dams 125
Table 4-24. Differences in marker enzymes in the small intestine of pups cross-fostered to
aery 1 ami de-treated or control dams during postnatal lactation 133
Table 4-25. Frequency of translocation carriers in offspring derived from males exposed to
acrylamide or glycidamide 135
Table 4-26. Results for specific locus mutations recovered in offspring of male mice exposed i.p
to 50 mg/kg acrylamide on 5 consecutive days 135
Table 4-27. Results for specific locus mutations recovered in offspring of male mice exposed to
acrylamide as a single 100 or 125 mg/kg i.p. dose 136
Table 4-28. Acrylamide initiation of squamous cell carcinomas or papillomas in female
SENCARmice 142
Table 4-29. Acrylamide initiation of skin tumor masses >lmm in female SENCAR mice 143
Table 4-30. Circulating thyroid hormone levels in F344 rats following exposure to AA in
drinking water for 14 or 28 days 147
Table 4-31. Plasma TSH, BrdU incorporation in thyroid, and PCNA expression in thyroid in
male Sprague-Dawley rats exposed to acrylamide by an unspecified route for up to
28 days 148
Table 4-32. Noncancer effects in animals repeatedly exposed to acrylamide by the oral route 155
Table 4-33. Neurological effects following exposure to acrylamide in species other than the rat
and mouse 158
Table 5-1. Acrylamide oral exposure: selected NOAELs and LOAELs 200
Table 5-2. Incidence data for degenerative changes detected by light microscopy in nerves of
male and female F344 rats exposed to acrylamide in drinking water for 2 years... 204
08/28/2009 xiii DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-3. Predictions (mg/kg-day) from best-fitting models for doses associated with a 10, 5,
and 1% extra risk for nerve degeneration in male and female rats exposed to
acrylamide in drinking water (Johnson et al 1986) 205
Table 5-4. Predictions (mg/kg-day) from best-fitting models for doses associated with 10, 5, and
1% extra risk for sciatic nerve changes in male and female rats exposed to
acrylamide in drinking water (Friedman et al 1995) 205
Table 5-5. Second order rate constants for reaction of acrylamide or glycidamide with the N-
terminal valine residue of hemoglobin 207
Table 5-6. Measured and estimated AA AUCs normalized to dose in humans and F344 rats.. 208
Table 5-7. Measured and estimated GA AUCs normalized to Dose in Humans and F344 rats.209
Table 5-8. Serum AUC from Doerge et al. (2005c) and hemoglobin adduct levels from Tareke et
al (2008) for a 0.1 mg/kg single dose of AAin male and female F344 rats 211
Table 5-9. Summary of uncertainty in the acrylamide noncancer risk assessment 221
Table 5-10. Estimated POD (mg/kg-day) from best-fitting models for doses associated with a
5% extra risk for nerve degeneration in male and female rats exposed to acrylamide
in drinking water 227
Table 5-11. Incidence of tumors with statistically significant increases in 2-year bioassays with
F344 rats exposed to acrylamide in drinking water 232
Table 5-12. Points of departure and oral slope factors derived from Friedman et al. (1995) tumor
incidence data for female rats exposed to acrylamide in drinking water 235
Table 5-13. Points of departure and oral slope factors derived from Friedman et al. (1995)
tumors incidence data for male F344 rats exposed to acrylamide in drinking water.
235
Table 5-14. Points of departure and oral slope factors derived from Johnson et al. (1986) tumor
incidence data for female F344 rats exposed to acrylamide in drinking water 237
Table 5-15. Points of departure and oral slope factors derived from Johnson et al. (1986) tumor
incidence data for for male F344 rats exposed to acrylamide in drinking water.... 237
Table 5-16. Comparison of oral slope factors based on summed risks for tumors at several sites
in two bioassays ofF344 rats exposed to acrylamide in drinking water 238
Table 5-15. Summary of uncertainty in the acrylamide cancer risk assessment 247
Table 5-16. Heritable genetic risk estimates for humans exposed to acrylamide 264
Table B-l. Results of acrylamide mutageni city testing 1
Table C-l. Incidence data for degenerative changes detected by light microscopy in nerves of
male and female F344 rats exposed to acrylamide in drinking water for 2 years 1
Table C-2. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
nerve degeneration in male rats exposed to acrylamide in drinking water 2
08/28/2009 xiv DRAFT-DO NOT CITE OR QUOTE
-------
Table C-3. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
nerve degeneration in female rats exposed to acrylamide in drinking water 3
Table C-4. Predictions (mg/kg-day) from best-fitting models for doses associated with a 10, 5,
and 1% extra risk for nerve degeneration in male and female rats exposed to
acrylamide in drinking water 4
Table C-5. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
sciatic nerve changes in male rats exposed to acrylamide in drinking water 5
Table C-6. Predictions (mg/kg-day) from models for doses associated with a 10% extra risk for
sciatic nerve changes in female rats exposed to acrylamide in drinking water 6
Table C-7. Predictions (mg/kg-day) from best-fitting models for doses associated with 10, 5, and
1% extra risk for sciatic nerve changes in male and female rats exposed to
acrylamide in drinking water 7
Table D-l. Incidence of tumors with statistically significant increases in the Friedman et al.
(1995) bioassay with F344 rats exposed to acrylamide in drinking water 1
Table D-2. Incidences of tumors with statistically significant increases in the Johnson et al.
(1986) bioassay with F344 rats exposed to acrylamide in drinking water 2
Table D-3. Risk estimate derived from separate and combined incidence of mammary or thyroid
tumors in female F344 rats exposed to acrylamide in drinking water 4
Table D-4. Risk estimates derived from separate and summed dose-response modeling of
mammary and thyroid tumors in female F344 rats exposed to acrylamide in drinking
water 6
Table D-5. Risk estimates for separate and combined incidence of TVMs or thyroid tumors in
male rats exposed to acrylamide in drinking water 7
Table D-6. Risk estimates derived from modeling separate and summed incidence of TVM and
thyroid tumors in male F344 rats exposed to acrylamide in drinking water 8
Table D-7. Risk estimates derived from separate incidence of mammary, thyroid, CNS, or oral
cavity tumors in female F344 rats exposed to acrylamide in drinking water 10
Table D-8. Calculation of summed risks for tumors at several sites in female F344 rats exposed
to acrylamide in drinking water in the Johnson et al. (1986) bioassay 11
Table D-9. Risk estimates derived from separate incidence of TVM, thyroid tumors in male
F344 rats exposed to acrylamide in drinking water 12
Table D-10. Calculation of summed risks for tumors at several sites in male F344 rats exposed
to acrylamide in drinking water in the Johnson et al. (1986) bioassay 12
Table E-l. Serum AUC from Doerge et al. (2005c) and hemoglobin adduct levels from Tareke
et al (2008) for a 0.1 mg/kg single dose of AA in male and female F344 rats 2
Table E-2: Fitted AUC's and Rates Based on the Use of Rate Constants Derived from In Vivo
Data 8
08/28/2009 xv DRAFT-DO NOT CITE OR QUOTE
-------
Table F-l. Scoring system for the neurotoxicity index 2
Table F-2. Group means ± SD of biomarkers in different categories of workers 3
08/28/2009 xvi DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
Figure 2-1. Chemical structure of acrylamide (AA) with carbon numbers indicated 3
Figure 3-2. Hemoglobin and DNA adducts of acrylamide and glycidamide 34
Figure 3-3. Schematic of the Kirman et al. PBTK model for acrylamide 53
Figure 3-4. Schematic of the Young et al. PBTK model for acrylamide 54
Figure 5-1. Acrylamide oral exposure: selected NOAELs and LOAELs 199
Figure 5-3. Original parallelogram approaches for estimating risk of heritable germ cell effects.
257
Figure 5-4. Two modifications in the parallelogram approach for estimating risk of heritable
germ cell effects from exposure to AA 258
Figure C-l. Observed and predicted incidences for nerve changes in male rats exposed to
acrylamide in drinking water for 2 years 2
Figure C-2. Observed and predicted incidences for nerve changes in female rats exposed to
acrylamide in drinking water for 2 years 3
Figure C-3. Observed and predicted incidences for nerve changes in male rats exposed to
acrylamide in drinking water for 2 years 5
Figure C-4. Observed and predicted incidences for nerve changes in female rats exposed to
acrylamide in drinking water for 2 years 6
Figure D-l. Observed and predicted incidences for mammary gland tumors in female rats
exposed to acrylamide in drinking water for 2 years 15
Figure D-2: Observed and predicted incidences for thyroid tumors in female rats exposed to
acrylamide in drinking water for 2 years 18
Figure D-3: Observed and predicted incidences for mammary or thyroid tumors in female rats
exposed to acrylamide in drinking water for 2 years 21
Figure F-l. Benchmark Dose Analysis for Calleman et al. (1994) data 3
08/28/2009 xvii DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS
A A aery 1 amide
AAMA N-acetyl-S-(2-carbamoylethyl)-L-cysteine
AAVal acrylamide-hemoglobin-terminal-valine adduct, N-(2-carbamoylethyl)valine
ABT 1-aminobenzotriazole
ADAF age-dependent adjustment factor
AIC Akaike's Information Criterion
ALT alanine aminotransferase
AUC area under the curve
BB Big Blue
BMD benchmark dose
BMDL 95% lower bound on BMD
BMDS benchmark dose software
BMR benchmark response
bw, BW body weight
C-C control dams with control pups
CERHR National Toxicology Program / Center for the Evaluation of Risks to Human
Reproduction
CFR Code of Federal Regulations
CI confidence interval
CIR Cosmetic Industry Review Expert Panel (4) (ref)
CNS central nervous system
C-T control dams with treated pups
dAdo 2'-deoxyadenosine
dCyd 2'-deoxycytidine
dGua 2'-deoxyguanosine
dThd 2'-deoxythymidine
ED effective dose
ENMG electroneuromyographic
EPA Environmental Protection Agency
FAO Food and Agricultural Organization
FDA U.S. Food and Drug Administration
FISH fluorescence in situ hybridization
GA glycidamide
GABA gamma-aminobutyric acid
GAMA N-(R,S)-acetyl-S-(carbamoyl-2-hydroxyethyl)-L-cysteine
GAVal glycidamide-hemoglobin-terminal-valine adduct, N-(2-carbamoyl-2-
hydroxyethyl)valine
GC-MS gas chromatography-mass spectrometry
GD gestational day
GSH glutathione
Hb hemoglobin
HBSS Hanks' balanced salt solution
HEC human equivalent concentration
HID highest ineffective dose/concentration
HSDB Hazardous Substances Data Bank
08/28/2009
xvin
DRAFT-DO NOT CITE OR QUOTE
-------
i.p. intraperitoneal or intraperitoneally
i.v. intravenous or intravenously
IARC International Agency for Research on Cancer
IRB Institute Review Board
IRIS Integrated Risk Information System
IRMM Institute for Reference Materials and Measurements
JECFA Joint FAO/WHO Expert Committee on Food Additives
JIFSAN Joint Institute for Food Safety and Applied Nutrition
LDso median lethal dose
LED 95% lower bound on ED
LFB/PAS luxol fast blue-periodic acid Schiff (59)
LH luteinizing hormone
LOAEL lowest-observed-adverse-effect level
LSD Fisher's Least Significant Difference Test
MF mutant frequency
MLE maximum likelihood estimate
MN micronucleus or micronuclei
MN-RET micronucleated reticulocytes
MOA mode of action
MPDS mortality and population data system; maintained at the University of
Pittsburgh
N3-GA-Ade N3-(2-carbamoyl-2-hydroxyethyl)adenine
NFCS National Food Consumption Survey (Netherlands)
NIOSH National Institute of Occupational Safety and Health
NMA N-methylol aery 1 amide
NOAEL no-observed-adverse-effect level
OR odds ratio
OSHA Occupational Safety and Health Administration
PBTK physiologically based toxicokinetic (as in PBTK model)
PCNA proliferating cell nuclear antigen
PEL permissible exposure limit
PKA protein kinase A
PND postnatal day
POD point of departure
R risk
REL recommended exposure limit
RfC reference concentration
RfD reference dose
SCF Scientific Committee on Food of the European Commission
SEM standard error of the mean
SHE Syrian hamster embryo
SMR standardized mortality ratio
SNFA Swedish National Food Agency
SNT Statens naeringsmiddeltilsy; the Norwegian Food Control Authority
T3 triiodothyronine
T4 thyroxin
T-C treated dams with control pups
TPA 12-O-tetradecanoyl-phorbol-13-acetate
08/28/2009
xix
DRAFT-DO NOT CITE OR QUOTE
-------
TSH thyroid stimulating hormone
T-T treated dams with treated pups
TVM tunica vaginalis mesothelioma
UCL upper confidence limit
UCLE upper confidence limit estimate
UDS unscheduled DNA synthesis
UF uncertainty factor
08/28/2009 xx DRAFT-DO NOT CITE OR QUOTE
-------
FOREWORD
The purpose of this Toxicological Review is to provide scientific support and rationale
for the hazard and dose-response assessment in IRIS pertaining to chronic exposure to
acrylamide. It is not intended to be a comprehensive treatise on the chemical or toxicological
nature of acrylamide.
The intent of Section 6, Major Conclusions in the Characterization of Hazard and Dose
Response, is to present the major conclusions reached in the derivation of the reference dose,
reference concentration and cancer assessment, where applicable, and to characterize the overall
confidence in the quantitative and qualitative aspects of hazard and dose response by addressing
the quality of data and related uncertainties. The discussion is intended to convey the limitations
of the assessment and to aid and guide the risk assessor in the ensuing steps of the risk
assessment process.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
08/28/2009 xxi DRAFT-DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGER
Robert S. DeWoskin, Ph.D., DABT
Office of Research and Development
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
AUTHORS (EPA)
Robert S. DeWoskin, Ph.D., DABT
Office of Research and Development
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Cancer Assessment
Karen Hogan
Office of Research and Development
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
AUTHORS (CONTRACT)
David W. Wohlers, Ph.D.
Peter R. McClure, Ph.D., DABT
Jennifer Rhoades, B.S.
Kelly Salinas, Ph.D.
Environmental Science Center
SRC, Inc.
North Syracuse, NY
Contract Number: GS-OOF-0019L
Hemoglobin Adduct Formation Rate Modeling
Justin G. Teeguarden, PhD, DABT
Battelle, Pacific Northwest Division
902 Battelle Blvd.
Richland, WA, 99352
08/28/2009 xxii DRAFT-DO NOT CITE OR QUOTE
-------
REVIEWERS
This document has been reviewed by EPA scientists, interagency reviewers from other
federal agencies and White House offices, and the public, and peer reviewed by independent
scientists external to EPA. A summary and EPA's disposition of the comments from the
independent external peer reviewers and from the public is included in Appendix A.
INTERNAL EPA REVIEWERS
Ila Cote, Ph.D., DABT
Office of Research and Development
National Center for Environmental Assessment
Kevin Crofton, Ph.D.
Office of Research and Development
National Health and Environmental Effects Laboratory
Sally Darney, Ph.D.
Office of Research and Development
National Health and Environmental Effects Laboratory
Kerry Dearfield, Ph.D.
Office of Research and Development
Office of The Science Advisor
[Currently with the US Department of Agriculture, Food Safety and Inspection Service]
Lynn Flowers, Ph.D., DABT
Office of Research and Development
National Center for Environmental Assessment
Gary Foureman, Ph.D.
Office of Research and Development
National Center for Environmental Assessment
Angela Howard, Ph.D.
Office of Research and Development
National Center for Environmental Assessment
Gene Hsu, Ph.D.
Office of Research and Development
National Center for Environmental Assessment
[Currently with Merck & Co Inc, West Point, PA]
Reeder Sams, Ph.D.
Office of Research and Development
National Center for Environmental Assessment
08/28/2009 xxiii DRAFT-DO NOT CITE OR QUOTE
-------
John Vandenberg, Ph.D.
Office of Research and Development
National Center for Environmental Assessment
EXTERNAL PEER REVIEWERS
CHAIR
Dr. Deborah Cory-Slechta, University of Rochester, Rochester, NY
PANEL MEMBERS
Dr. Alfred Branen, University of Idaho, Coeur d'Alene, ID
Dr. Daniel R. Doerge, National Center for Toxicological Research, Food and Drug Administration,
Jefferson, AR
Dr. James S. Felton, University of California, Lawrence Livermore National Laboratory,
Livermore, CA
Dr Timothy Fennell, RTI International, Research Triangle Park, NC
Dr. Penelope Fenner-Crisp, Independent Consultant, North Garden, VA
Dr. Jeffrey Fisher, University of Georgia, Athens, GA
Mr. Sean Hays, Summit Toxicology, Allenspark, CO
Dr. Steven Heeringa, University of Michigan, Ann Arbor, MI
Dr. Richard M. LoPachin, Albert Einstein College of Medicine, Bronx, NY
Dr. Lorelei Mucci, Harvard Medical School, Channing Laboratory, Boston, MA
Dr. Jerry M. Rice, Georgetown University Medical Center, Washington, DC
Dr. Dale Sickles, Medical College of Georgia, Augusta, GA
08/28/2009 xxiv DRAFT-DO NOT CITE OR QUOTE
-------
Dr. Gina Solomon, Natural Resources Defense Council, San Francisco, CA
Dr. Anne Sweeney, The Commonwealth Medical College, Scranton, PA
Dr. Lauren Zeise, Office of Environmental Health Hazard Assessment, California Environmental
Protection Agency, Oakland, CA
08/28/2009 xxv DRAFT-DO NOT CITE OR QUOTE
-------
1. INTRODUCTION
This document presents background information and justification for the Integrated Risk
Information System (IRIS) Summary of the hazard and dose-response assessment of aery 1 amide.
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 (MOA). The RfD (expressed in units of mg/kg-day) is defined as an estimate
(with uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m3) is
analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for both the respiratory system (portal of entry) and for
effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
acute (<24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
lifetime) exposure durations, all of which are derived based on an assumption of continuous
exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
derived for chronic exposure duration.
The carcinogenicity assessment provides information on the carcinogenic hazard
potential of the substance in question and quantitative estimates of risk from oral and inhalation
exposure may be derived. The information includes a weight-of-evidence judgment of the
likelihood that the agent is a human carcinogen and the conditions under which the carcinogenic
effects may be expressed. Quantitative risk estimates may be derived from the application of a
low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
the estimate of risk per mg/kg-day of oral exposure. Similarly, a plausible inhalation unit risk is
an upper bound on the estimate of risk per ug/m3 air breathed.
Development of these hazard identification and dose-response assessments for
acrylamide has followed the general guidelines for risk assessment as set forth by the National
Research Council (NRC, 1983). U.S. Environmental Protection Agency (U.S. EPA) Guidelines
and Risk Assessment Forum Technical Panel Reports that may have been used in the
development of this assessment include the following: Guidelines for the Health Risk
Assessment of Chemical Mixtures (U.S. EPA, 1986a), Guidelines for Mutagenicity Risk
Assessment (U.S. EPA, 1986b), Recommendations for and Documentation of Biological Values
for Use in Risk Assessment (U.S. EPA, 1988), (proposed) Interim Policy for Particle Size and
Limit Concentration Issues in Inhalation Toxicity (U.S. EPA, 1994a), Methods for Derivation of
DRAFT-DO NOT CITE OR QUOTE
-------
Inhalation Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA,
1994b), Use of the Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995),
Science Policy Council Handbook. Risk Characterization (U.S. EPA, 2000a), Benchmark Dose
Technical Guidance Document (U.S. EPA, 2000b), Supplementary Guidance for Conducting
Health Risk Assessment of Chemical Mixtures (U.S. EPA, 2000c), A Review of the Reference
Dose and Reference Concentration Processes (U.S. EPA, 2002), Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005a), Supplemental Guidance for Assessing Susceptibility from Early-
Life Exposure to Carcinogens (U.S. EPA, 2005b), Science Policy Council Handbook: Peer
Review (U.S. EPA, 2006a), and A Framework for Assessing Health Risks of Environmental
Exposures to Children (U.S. EPA, 2006b).
The literature search strategy employed for this compound was based on the Chemical
Abstracts Service Registry Number (CASRN) and at least one common name. Any pertinent
scientific information submitted by the public to the IRIS Submission Desk was also considered
in the development of this document. The relevant literature was reviewed through April 2009
DRAFT-DO NOT CITE OR QUOTE
-------
2. CHEMICAL AND PHYSICAL INFORMATION RELEVANT TO ASSESSMENTS
2.1. CHEMICAL AND PHYSICAL INFORMATION
Acrylamide (AA) is an odorless, white, crystalline solid. Synonyms include acrylic
amide, acrylic acid amide, ethylenecarboxamide, propenamide, and propenoic acid amide. The
structure of AA is shown below in Figure 2-1 (carbons are numbered).
0
Figure 2-1. Chemical structure of acrylamide (AA) with carbon numbers
indicated.
References for the selected chemical and physical properties of acrylamide listed below
or in the subsequent text include HSDB, 2005; Budavari, 2001; Verschueren, 2001; Lide, 2000;
Lewis, 1997; Hansch et al., 1995; IARC, 1994a; and Petersen et al., 1985.
CAS number:
Molecular weight:
Chemical Formula:
Boiling point:
Melting point:
Vapor pressure:
Density:
Vapor density:
Water solubility:
Other solubilities at 30°C:
Partition coefficient (Kow):
Partition coefficient (Koc):
pH:
Henry's law constant:
Bioconcentration factor:
Stability
Conversion factors:
13,
79-06-1 (Verschueren, 2001)
71.08 (Verschueren, 2001); 74.0 for 1,2,3-"C3 labeled AA
(Fennell et al. 2005)
C3H5NO (Verschueren, 2001)
192.6°C (Verschueren, 2001)
84.5°C (Verschueren, 2001)
0.007 mm Hg at 25°C (HSDB, 2005)
1.12 g/mL at 30°C (Budavari, 2001)
2.46 (air = 1) (Verschueren, 2001)
2.155 g/mL at 30°C (Verschueren, 2001)
Acetone (0.631 g/mL), chloroform (0.027 g/mL), diethyl ether
(0.862 g/mL), ethanol (0.862 g/mL), ethyl acetate (0.126
g/mL), methanol (1.55 g/mL), heptane (0.068 g/mL) (Budavari,
2001; Lide, 2000)
log Kow = -0.67 (octanol/water) (Hansch et al., 1995)
log Koc = 1 (organic carbon/water) (HSDB, 2005)
5.0-6.5 (50% aqueous solution) (HSDB, 2005)
1.7 x 10~9 atm-m3/mol at 25°C (HSDB, 2005)
1 for fingerling trout (Petersen et al., 1985)
Stable at room temperature but may polymerize violently on
melting (HSDB, 2005)
1 mg/m3 = 0.34 ppm, 1 ppm = 2.95 mg/m3 (Verschueren, 2001)
1 gr = 14.07 mmoles
DRAFT-DO NOT CITE OR QUOTE
-------
Acrylamide is a highly water-soluble a,p-unsaturated amide that reacts with nucleophilic
sites in macromolecules in Michael-type additions (Calleman, 1996; Segerback et al., 1995).
Monomeric AA readily participates in radical-initiated polymerization reactions, whose products
form the basis of most of its industrial applications (Calleman, 1996).
2.2. SOURCES OF EXPOSURE, FATE AND TRANSPORT
Acrylamide from industrial sources
Acrylamide was initially produced for commercial purposes by reaction of acrylonitrile
with hydrated sulfuric acid and separation of the product from its sulfate salt. Relatively high
levels of impurities resulted from this process, which was replaced in the 1970s by catalytic
hydration with copper metal or a Raney copper catalyst and lower levels of impurities. With
catalytic hydration, a solution of acrylonitrile in water is passed over a fixed bed of copper
catalyst at 85°C to produce AA. A third production method, developed in 1985, uses
microorganisms to convert acrylonitrile into acrylamide by enzymatic hydration (HSDB, 2005;
IARC, 1994a). Direct uses of acrylamide include photopolymerization systems, adhesives and
grouts, and polymer cross-linking. The primary use of AA is in the production of
polyacrylamides, which are used for enhanced oil recovery in water flooding, in oil well drilling
fluids, in fracturing aids, in sewage treatment flocculants, in soil conditioning and stabilization,
in papermaking aids and thickeners, in adhesion-promoting polymers, in dye acceptors, in textile
additives, and in paint softeners (HSDB, 2005; IARC, 1994a).
Release of AA to the environment may occur during its production and use or in the
production of polyacrylamide. Products and compounds containing polyacrylamide may serve
as sources of exposure to residues of acrylamide. Examples include polyacrylamide compounds
used in oil well drilling operations (well drilling muds), as flocculents in water treatment,
coagulants in food processing, sealing grouts and some coatings, and as foam builders,
lubricants, and emollients in some personal care and grooming products (CFR, 2005; CIR,
1991). Localized contamination may arise from the use of acrylamide in grouting operations
(HSDB, 2005). U.S. EPA (2003) requires drinking water authorities to certify that, for
polyacrylamides used as coagulants or flocculents in drinking water treatment, the level of
acrylamide monomer in the polymer does not exceed 0.05% and the application rate for the
polymer does not exceed 1 mg/L. The National Sanitation Foundation/American National
Standards Institute (NSF/ANSI) Standard 60 for Drinking Water Treatment Chemicals - Health
Effects provides the restrictions for the use of polyacrylamides in well drilling muds and grouts
for potable water wells based on acrylamide monomer levels.
If released to air, the vapor pressure of 0.007 mm Hg at 25°C indicates that AA will exist
solely as a vapor in the ambient atmosphere. Vapor-phase AA will be degraded in the
atmosphere by reaction with photochemically produced hydroxyl radicals; the half-life for this
CRAFT-DO NOT CITE OR QUOTE
-------
reaction in air is estimated to be 1.4 days. The half-life for the reaction of vapor-phase AA with
ozone is estimated to be 6.5 days. Acrylamide is not expected to be susceptible to direct
photolysis in sunlight since it does not absorb light with wavelengths >290 nm (HSDB, 2005).
With a Koc of 10, AA is expected to be highly mobile in soils. Volatilization of AA from
dry or moist soil surfaces is not expected to be an important fate process, based on its vapor
pressure and estimated Henry's law constant of 1.7 x 10~9 atm-mVmol (HSDB, 2005).
Acrylamide is expected to degrade in soil. Degradation in the range of 74-94% within 14 days
and 79-80% in 6 days was reported for AA in several soils that had been moistened to field
capacity (Abdelmagid and Tabatabai, 1982). Half-lives of 18-45 hours were observed for four
central New York soils that had been moistened to 70% field capacity (Lande et al., 1979).
If released to water, AA is not expected to adsorb to suspended solids or sediment, based
on the Koc (HSDB, 2005). In a river die-away test, 90% of AA disappeared in approximately
150 hours (Croll et al., 1974). The hydrolysis half-life of acrylamide has been reported as
>38 years (HSDB, 2005). Volatilization of acrylamide from water surfaces is not expected,
based on the compound's Henry's law constant. An estimated bioconcentration factor of 1 for
fingerling trout (Petersen et al., 1985) suggests that bioconcentration in aquatic organisms is low
(HSDB, 2005). Microbial degradation of acrylamide can occur under light or dark, aerobic or
anaerobic conditions (Brown et al., 1980; Lande et al., 1979; Croll et al., 1974).
Acrylamide was formerly thought to only be present as an industrially manufactured
chemical and not a naturally occurring contaminant (IARC, 1994a). It is now known that
acrylamide is present in cigarette smoke, and can form in certain foods during cooking or
processing.
Acrylamide in cigarette smoke
Acrylamide is a component of cigarette smoke, and AA content in mainstream cigarette
smoke has been estimated at 1.1-2.34 jig per cigarette (Smith et al., 2000). Smoking is a source
of human inhalation exposure, and secondhand smoke could contribute to AA in indoor air,
although no data were found on indoor air levels of acrylamide from environmental tobacco
smoke. Boettcher et al. (2005) measured the AA and AA metabolites in human urine, and
reported median levels in smokers (n = 13) about four times higher than in nonsmokers (n = 16)
indicating that cigarette smoke is clearly an important source of acrylamide exposure.
Acrylamide formation in foods during processing
In early 2002, high concentrations of AA were reported in certain fried, baked, and deep-
fried foods (Swedish National Food Agency, 2002). This discovery dramatically increased the
interest in nonindustrial sources of acrylamide exposure to the general public. Subsequent
research in many European countries and the United States determined that AA is formed
DRAFT-DO NOT CITE OR QUOTE
-------
primarily in carbohydrate-rich foods prepared or cooked at high temperatures (i.e., >120°C)
(Tareke et al., 2002, 2000). The predominant chemistry involves a Maillard reaction, a
nonenzymatic browning reaction that occurs by a condensation of the amino group of the amino
acid, asparagine, and the carbonyl group of reducing sugars (fructose and glucose) during high-
temperature heating (Mottram et al., 2002; Stadler et al., 2002). Thus, browned crispy crusts in
foods like French fries, potato chips, crackers, pretzel-like snacks, cereals, and browned breads
tend to have the highest levels of AA. Acrylamide has been detected in some food products that
are processed at temperatures in the 98-116°C range and in high moisture conditions (e.g.,
canned black olives [not oil cured] and prune juice) [Roach et al., 2003]), so there are other
pathways of formation that do not involve temperatures over 120°C and crispiness, and these are
being further evaluated (JIFSAN, 2004). It is worth noting that, since AA appears to form from
standard cooking methods like baking, frying, and roasting, it has been in the human diet for
many thousands of years.
Dybing et al. (2005) list AA concentrations in various foods in the United States as
determined by the U.S. Food and Drug Administration (U.S. FDA, 2006) in Table 2-1 and, in
Table 2-2, in foods in Europe from data compiled by the Institute for Reference Materials and
Measurements (IRMM, 2004).
Estimates ofacrylamide exposure based on diet and acrylamide content in foods
The FDA has estimated overall daily intake levels ofacrylamide from exposures in the
U.S. diet to be around 0.4 |ig/kg-day with a 90th percentile of 0.95 |ig/kg-day (U.S. FDA, 2006).
Table 2-3 is a compilation by Dybing et al. (2005) of exposure estimates from many different
national organizations. Estimated daily intake in populations around the world are reasonably
similar to FDA's estimate, with the variability assumed to result from cultural differences in food
preferences (i.e., different composition of diet among populations), processing methods (i.e., that
result in different AA levels among local foods), and consumption levels.
A 2004 expert panel review of risk for human reproductive toxicity from exposure to AA
compiled a table of estimates for total exposures, presented here as Table 2-4 (NTP/CERHR,
2004).
©RAFT-DO NOT CITE OR QUOTE
-------
Table 2-1. Summary of acrylamide levels in food (ppb) derived from the
FDA data collected from 2002 through October 1, 2003
Food
commodity
Baby food and
infant formula
French fries and
chips
Protein foods
Breads and
bakery products3
Cereals and
muesli
Crackers and
snack foods
Gravies and
seasonings
Nuts and butters
Chocolate
products
Canned fruits and
vegetables
Coffee, ground
Coffee, brewed
Miscellaneous13
Includes cookies, pies and pastry, bagels.
bHot beverages other than coffee (Postum, caffeine-free coffee substitute), frozen vegetables, dried foods,
dairy, juice, and other miscellaneous.
Data were calculated from the data published by the FDA on the Internet ("Exploratory Data on Acrylamide in
Food," March 2004 [http://www.cfsan.fda.gov/~dms/acrydata.html]). The database contains data collected
from 2002 through October 1, 2003. The categories were used as given by the FDA. For coffee, only data for
roasted coffee were used (total sample number [n] = 439).
Source: Dybing et al. (2005).
n
36
97
21
49
23
32
13
13
14
33
59
20
41
Minimum
0.0
20.0
0.0
0.0
11.0
12.0
0.0
0.0
0.0
0.0
37.0
3.0
0.0
25%
0.0
220.0
0.0
15.0
49.0
92.5
0.0
28.0
2.5
0.0
158.0
6.0
0.0
Median
10.0
318.0
10.0
34.0
77.0
169.0
0.0
89.0
20.5
10.0
205.0
6.5
10.0
75%
31.8
462.0
25.0
96.0
166.0
302.3
0.0
236.0
84.3
70.0
299.0
8.0
43.0
Maximum
130.0
2,762.0
116.0
432.0
1,057.0
1243.0
151.0
457.0
909.0
1,925.0
539.0
13.0
5,399.0
Standard
deviation
36.6
427.9
27.7
107.9
249.1
331.1
43.4
143.0
243.6
411.7
106.3
2.4
1,018.8
DRAFT-DO NOT CITE OR QUOTE
-------
Table 2-2. Acrylamide levels in food (ppb) as collected by the European
Union Joint Research Center (updated June 2004)
Food
commodity
French fries
Chips
Potato fritter3
Fine bakery
ware
Gingerbread
Crispbread
Infant biscuits
Diabetics' cakes
and biscuits
Breakfast
cereals
Coffee, roasted
Coffee,
substitutes
n
741
569
75
485
414
261
63
212
162
102
50
Minimum
5.0
5.0
15.0
5.0
5.0
5.0
5.0
5.0
5.0
79.0
115.6
25%
90.0
378.0
215.0
67.0
152.0
81.0
64.3
92.5
30.0
192.0
439.4
Median
178.0
600.0
492.0
160.0
298.5
251.0
90.0
291.5
60.0
264.0
739.0
75%
326.0
980.0
797.6
366.0
650.7
602.0
275.1
772.3
152.5
337.0
1,321.8
Maximum
2,228.0
3,770.0
2,779.0
3,324.0
7,834.0
2,838.0
910.0
3,044.0
846.0
975.0
2,955.0
aGrated potatoes fried into a pancake.
Note: Data were calculated from the monitoring database on acrylamide levels in food
(http://www.irmm.jrc.be/) maintained by the IRMM, together with the Directorate General for Health and
Consumer Affairs. This database comprises 3,442 samples of acrylamide levels in food products throughout
the EU, including the data collection from the Confederation des Industries Agro-Alimentaires de 1'Union
Europeenne. The categories were used as given in the data collection.
Source: Dybing et al. (2005).
DRAFT-DO NOT CITE OR QUOTE
-------
Table 2-3. Exposure estimates from 2002-2006
Exposure assessment Daily intake jig/kg-day
Mean (age group) Percentilea'b
FAO/WHO (2007) 0.3-0.8
SCF, European Union (2002)
BfR, Germany (2002)
BAG, Switzerland (2002)
AFSSA, France (2002)
FDA, United States (2002)
FDA, United States (2004)
FDA, United States (2006,
2009)
NFCS, Netherlands
0.2-0.4
1.1 (15-18) 3.4a
0.28 (16-57)
SNFA, Sweden (2002)
SNT, Norway (2003)
Source
http ://www .who. int/foodsafety/publ
ications/chem/en/acrylamide_full.
pdf
http://europa.eu.int/comm/food/fs/sc
/scf/out!3 l_en.pdf
http://www.bfr.bund.de/cm/208/Abs
chaetzung_der_Acrylamid_Aufnah
me_durch_
hochbelastete_Nahrungsmittel_in_
Deutschland_Studie.pdf
http ://www .bag.admin. ch/verbrau/a
ktuell/d/DDS%20acrylamide%20pr
eliminary%20communication.pdf
http://www.afssa.fr/ftp/afssa/basedo
c/acrylpoint2sansannex.pdf
http://www.cfsan.fda.gov/~dms/acr
yexpo.html
http://www.cfsan.fda.gov/~dms/acr
yexpo.html
http://www.cfsan.fda.gov/~dms/acr
yexpo.html
Konigs et al. (2003)
Svensson et al. (2003)
Dybing and Sanner (2003)
0.5 (>15) l.T
1.4 (2-14) 2.9a
0.7
0.43 (>2) 0.92b
1.06(2-5) 2.3 lb
0.40 (>2) 0.95a
1.07 (2-5) 2.33b
0.48 (1-97) 0.6a
1.04(1-6) l.la
0.71 (7-18) 0.9a
0.45 (18-74) 1.03
0.49 (males) 1.0lb
0.46 (females) 0.86b
0.36 (9, boys) 0.72b
0.32 (9, girls) 0.61b
0.52 (13, boys) 1.35b
0.49 (13, girls) 1.2b
0.53 (16-30, males)
0.50 (16-30, females)
^percentile.
b90thpercentile.
Source: For all exposures estimates from 2002 to 2004, Dybing et al. (2005) except the FDA estimates; FDA
exposure estimates 2002-2006 (directly from the FDA website:
http://www.cfsan.fda.gov/~dms/acryexpo.html.
DRAFT-DO NOT CITE OR QUOTE
-------
Table 2-4. Summary of exposure estimates (ug/kg-day) by sources and
population groups
Source of exposure Mean or median3 90th percentile or upper boundary3
Diet: general population 0.43 0.92
2-to 5-year-olds 1.06 2.31
Drinking water No data <0.01
Personal care products -0.5 1.1 (female)
Cigarette smoking 0.67 (from cigarette data) 1.3
2.6 (from adduct data)b ~6
Occupational exposures 1.4-18 43 (based on PELe)
Totals (adults)
General population
Nonsmokers 0.98C 2.0
0.85 (from adduct data)
Smokers 1.7 (from cigarette data) 3.2
3.6 (from adduct data)
Occupational exposure"1 45-52
Nonsmokers 2.4-19 45
Smokers 3.1-20 (cigarette data) 46
5-22 (adduct data) 51
3Dose levels in experimental animal studies are expressed as mg/kg-day, human exposures are expressed as
ug/kg-day. To convert figures in table to mg/kg-day, divide by 1,000.
bAcrylamide exposure in smokers based on adduct formation was estimated by taking the value for total exposure in
smokers (3.4 ug/kg-day) and subtracting the value for total exposure in nonsmokers (0.85 ug/kg-day).
Estimated from diet, water, and personal care products. The adduct-derived estimates are considered more
comprehensive.
Occupational exposures include monomer and polymer production and grouting applications.
ePEL = permissible exposure limit. The Occupational Safety and Health Administration (OSHA) permissible
exposure level (PEL) for acrylamide is 0.3 mg/m3. Based on a geometric means of 0.01-0.13 mg/m3 and an upper
bound exposure of 0.3 mg/m3 (PEL), the NTP/CERHR Expert Panel estimated mean and upper bound workplace
acrylamide inhalation exposures at 1.4-18.6 ug/kg bw/day and 43 ug/kg bw/day, respectively.
Source: NTP/CERHR (2004).
Alternate methods for estimating exposure to the general population are based on internal
levels of biomarkers of exposure including levels of hemoglobin adducts or urinary metabolites.
Recent comparisons of biomarker studies from many different studies are being used to estimate
risk (Doerge et al., 2008) or to compare estimates of exposure in the general nonsmoking
population (Hartmann et al., 2008). Hartmann et al. (2008) developed exposure estimates based
on levels of hemoglobin adducts or urinary metabolites as biomarker in a nonsmoking population
of children, adolescents, and adults from the general population in Germany (n=91; 45 males, 46
females; aged 6 -80 years; median age = 36 years). Median daily intakes were estimated at 0.43
(0.21-1.04) jig/kg-day based on hemoglobin adducts levels; and 0.51(
-------
be used as long-term exposure markers, were in both sexes identical with median levels of 30
pmol/g of globin for AAVal and 34 pmol/g of globin for GAVal. The results, however, indicated
that children take up approximately 1.3-1.5 times more AA per kilogram of body weight than
adults. The ratio GAMA/AAMA was also significantly higher in the group of young children (6-
10 years) with a median level of 0.5. The Hartmann et al (2008) results are consistent with the
US population average intake estimates of 0.4 jig AA/kg bw-day, as well as intake estimates
from other Eurpoean countries summarized by FAO/WHO as ranging from 0.2 - 2.0 |ig/kg-day,
with a WHO designated representative average level for the general population of 1 |ig/kg-day
(FAO/WHO, 2005).
Additional information on estimates of exposure based on hemoglobin adducts and
urinary metabolites can be found in the Section 3-5 in the next chapter on the toxicokinetics of
acrylamide, and in Chapter 5 on the derivation of the human equivalent concentration and
reference values.
11 DRAFT-DO NOT CITE OR QUOTE
-------
3. TOXICOKINETICS RELEVANT TO ASSESSMENTS
Much of the information in this section describes interactions of acrylamide (AA) and its
principal and lexicologically significant (epoxide) metabolite, glycidamide (GA) with various
biologically significant targets such as cellular thiols (e.g., glutathione), various proteins and
bases in DNA. The chemical basis for these interactions is strongly associated with the degree
of electrophilicity (electron deficiency) of such agents as AA and GA with nucleophilic centers
(i.e., unshared electrons) that may be present in biological targets. Electrophiles and
nucleophiles are generally characterized as being either "hard" or "soft" corresponding to a
spectral range of high or low charge densities or electronegativity for reactivity (Pearson and
Songstad, 1967). Due to its d,p-unsaturated structure and ready capacity to undergo Michael-
type additions, acrylamide may be classified as a "soft" electrophile. Soft electrophiles like AA
react readily with soft nucleophiles such as the thiol groups of proteins or glutathione. GA, on
the other hand, has a relatively high positive charge density, and acts as a hard electrophile, more
capable of reacting with centers of high electronegativity (i.e., hard nucleophiles) such as the
purine and pyrimidine bases in DNA (Lopachin and DeCaprio, 2005; Dearfield et al, 1995). A
recent evaluation of soft-soft interactions based on frontier molecular orbital characteristics (as
defined by the quantum mechanical parameters for softness [sigma] and chemical potential [mu])
suggest that the thiolate state of cysteine residues is the corresponding adduct target for AA
(Lopachin et al., 2007a). This information is useful in understanding the differences discussed in
this section between the types of adducts formed by AA and GA (e.g., hemoglobin and/or DNA)
and the binding rates.
3.1. ABSORPTION
Hemoglobin adducts as a biomarker of exposure/absorption
Numerous studies, including a recent study by Fennell et al. (2005), support the use of
acrylamide hemoglobin adducts as a biomarker of exposure. (See the Metabolism Section 3.3 for
a detailed discussion of the chemistry of AA and GA hemoglobin adducts, and GA DNA
adducts). Estimates of exposure using hemoglobin adduct levels are based on the assumption
that a measured adduct level represents a steady state level from a continuous exposure to
acrylamide over the previous 120 days, which is the average life span of a red blood cell.
Fennell et al. (2005) calculated AA exposure by using the results of the toxicokinetic study
described above in 24 volunteer adult males. The estimated average daily background exposure
to AA was 1.26 jig/kg-day based on the subject's preexposure background AA-hemoglobin-
terminal-valine adduct levels (AAVal) (averaging about 80 fmol/mg globin). In an occupational
12 DRAFT-DO NOT CITE OR QUOTE
-------
exposure study, Hagmar et al. (2001) reported a background range of 20-70 fmol AAVal/mg
globin in the unexposed reference group. Using the Hagmar et al. (2001) lower range and their
observed average as an upper value (i.e., a range of 20-80 fmol AAVal/mg globin), Fennell et al.
(2005) estimated a daily AA intake of 0.31-1.26 |ig/kg-day. For a 70 kg adult this translates into
a total daily intake of 22-88 jig of AA. As can be seen in Table 2-3, many of the estimates of
daily intakes in adults based on exposure estimates in foods are in the 0.4-0.8 jig/kg-day range,
suggesting that adults with higher adduct levels may be exposed to AA from sources other than
food (e.g., smoking, occupational, or from an as yet unknown source).
Detection of hemoglobin adducts of AA in workers exposed via inhalation and dermal
exposure provides qualitative evidence of absorption by these routes and suggests that dermal
exposure was the predominant route of absorption in these workers (Hagmar et al., 2001;
Bergmark et al., 1993). Hemoglobin adduct levels were measured in 41 Chinese workers who
were exposed to acrylamide for 0.1-8 years (Bergmark et al., 1993). Adducts measured in this
study were those at N-terminal valine residues in hemoglobin. Workers were involved in the
production of acrylamide (via the hydration of acrylonitrile) and polyacrylamide. The adduct
levels in exposed workers ranged from 0.3 to 34 nmol acrylamide/g hemoglobin. Hemoglobin
adducts of AA were not detected in blood samples from 10 control workers from the same city
who had not been exposed to AA (or acrylonitrile). Blood samples from 5 of the 41 exposed
workers were also analyzed for hemoglobin adducts of GA (a principal metabolite of AA in both
humans and animals) (see Section 3.3). There was a statistically significant linear relationship
between levels of hemoglobin adducts of AA and GA in these five workers; the ratio between
GA and AA adducts was approximately 3:10. Average levels of AA in air samples were
1.52 and 0.73 mg/m3 for workplaces involved with polymerization and synthesis processes,
respectively. Workers involved in these processes, however, showed average hemoglobin
adduct levels of acrylamide of 7.3 ±3.4 nmol/g hemoglobin (n = 12, polymerization) and
14.7 ± 10.6 nmol/g hemoglobin (n = 14, synthesis). The study authors calculated the levels of
hemoglobin adducts of AA in these workers that would have resulted from the observed
exposure concentrations, based on an assumption that exposure was only via inhalation (as well
as additional assumptions)1, and derived levels of 0.93 (instead of 7.3) nmol/g hemoglobin for
the polymerization workers and 0.44 (instead of 14.7) nmol/g hemoglobin for synthesis workers.
Thus, Bergmark et al. (1993) state that the observed and predicted adduct levels were
1 The calculation assumed that (1) adducts are stable during the life of erythrocytes; (2) the life span of
human erythrocytes is about 120 days (17 weeks); (3) the second-order reaction rate constant for the reaction of
acrylamide with N-terminal valine residues in human hemoglobin is 4.4 x 1CT6 L/g Hb/hour (based on in vitro
experiments); (4) the human ventilation rate is 0.2 L/min-kg; and 5) inhaled acrylamide is 100% absorbed.
13 DRAFT-DO NOT CITE OR QUOTE
-------
inconsistent with exposure only via inhalation and hypothesize that dermal exposure was the
predominant route of absorption in these workers.
Hagmar et al. (2001) measured hemoglobin adducts in a group of 210 tunnel construction
workers who were occupationally exposed for 2 months without personal protection devices to a
chemical grouting agent containing AA and N-methylolacrylamide. An important caveat in
interpreting the hemoglobin adduct data relative to AA absorption is that both AA and
N-methylolacrylamide form the same N-(2-carbamoylethyl)valine adduct in hemoglobin and
subsequent chemical measures of adduct levels cannot distinguish which parent compound
formed the adduct (Fennell et al., 2003) (see additional discussion in the next section). Blood
samples were drawn within a month after construction work was completed and analyzed for
levels of N-terminal valine adducts. Workers were expected to have experienced dermal
exposure to varying extents, as well as inhalation exposure. Quantitative exposure data were
limited to two personal air samples showing concentrations of 0.27 and 0.34 mg/m3 for the sum
of AA and N-methylolacrylamide; further analysis suggested that the air contained a
50:50 mixture of these compounds. Hemoglobin adduct levels for 18 nonsmoking unexposed
reference subjects varied between 0.02 and 0.07 nmol/g globin. The frequency distribution of
adduct levels in the 210 tunnel workers was as follows: 47 with <0.08 nmol/g globin; 89 with
0.08-0.29 nmol/g globin; 36 with 0.3-1.0 nmol/g globin; and 38 with 1.0-17.7 nmol/g globin.
Adduct levels were determined in blood samples collected at intervals up to 5 months after
cessation of exposure from five workers with initial levels ranging from about 2.2 to 4.4 nmol/g.
Adduct levels decreased to background levels within 120 days, consistent with the approximate
120-day life of red blood cells.
Human oral/dermal exposure
Fennell et al. (2005) evaluated metabolism and hemoglobin adduct formation following
oral and dermal administration of AA to 24 adult male volunteers. The 24 volunteers were all
male Caucasians (with the exception of one Native American), weighing between 71 and 101 kg,
and between 26 and 68 years of age. All volunteers were aspermic (i.e., clinically sterile because
of the potential for adverse effects of AA on sperm), and all volunteers had not used tobacco
products for the past 6 months. The study was conducted in accordance with the Code of
Federal Regulations (CFRs) governing protection of human subjects (21 CFR 50), Institute
Review Board (IRB) (21 CFR 56), and retention of data (21 CFR 312) as applicable and
consistent with the Declaration of Helsinki. The study used radiolabeled [l,2,3-13C]-acrylamide,
and, prior to the conduct of exposures in humans, a low-dose study protocol was evaluated in
rats administered 3 mg/kg [l,2,3-13C]-acrylamide by gavage. The [l,2,3-13C]-acrylamide human
14 DRAFT-DO NOT CITE OR QUOTE
-------
study protocol was reviewed and approved by IRBs both at the researchers' facility (Research
Triangle Institute International), where the sample analysis occurred, and by the clinical research
center conducting the study (Covance Clinical Research Unit [CRU]). The health of the
volunteers, exposed under controlled conditions, was continually monitored.
Acrylamide was administered orally in an aqueous solution (single dose of 0.5, 1.0, or 3.0
mg/kg) or dermally (three daily doses of 3.0 mg/kg) to the male volunteers. Approximately 34%
of the administered dose of AA was recovered in the total urinary metabolites within 24 hours of
administration, representing a lower bound on total absorption from the oral route. No other
estimate of total absorption from an oral exposure was reported.
The results of the dermal exposure in Fennell et al. (2005) indicate much lower levels of
AAVal and GA-hemoglobin-terminal-valine adduct (GAVal) formed than with an equivalent
dose via the oral route. Based on total amount administered, formation of AAVal after dermal
exposure was much lower than after oral administration (4.9 nmol/g globin/mmol AA/kg vs.
74.7 nmol/g globin/mmol AA/kg bw). These numbers can be used to estimate that
approximately 6.6% of the dermally administered dose was absorbed compared to a comparable
orally administered dose, assuming that there was 100% oral absorption. Similarly, dermal
exposure also resulted in much lower formation of GAVal, 9.7% of that formed following oral
exposure. However, approximately 66% of the dermally administered dose of AA was
recovered in the occluding solutions (data not included in the report) and thus was not
systemically absorbed on dermal administration. This suggests that a maximum of 3% of the
dermally applied dose could have been absorbed. An estimate of dermal absorption based on the
formation of AAVal adducts normalized to the absorbed dose yields a value of 17.0% of the
amount formed following oral exposure (12.7 nmol/g globin/mmol AA/kg for dermal vs. 74.7
nmol/g globin/mmol AA/kg for oral). Similarly, GAVal formation following dermal exposure
was 25.3% of that formed on oral administration (7.3 pmol/g globin/mmol AA/kg for dermal vs.
28.9 pmol/g globin/mmol AA/kg for oral). This suggests that as much as 83% of the AA
penetrating the skin was not available systemically. An alternative hypothesis is that AA and
GA clearance is different following dermal exposure, resulting in a lower area under the curve
(AUC) and lower adduct formation on a mg/kg basis. Ongoing study of urinary metabolites in
dermally exposed individuals may help resolve the reason(s) for these differences.
Fuhr et al. (2006) evaluated the toxicokinetics of acrylamide in six young healthy
volunteers after the consumption of a meal containing 0.94 mg of acrylamide. Urine was
collected up to 72 hours thereafter. Unchanged acrylamide, its mercapturic acid metabolite
N-acetyl-S-(2-carbamoylethyl)cysteine (AAMA), its epoxy derivative GA, and the respective
metabolite of GA, N-acetyl-S-(2-hydroxy-2-carbamoylethyl)cysteine (GAMA), were quantified
15 DRAFT-DO NOT CITE OR QUOTE
-------
in the urine by liquid chromatography-mass spectrometry. Toxicokinetic variables were
obtained by noncompartmental methods. Overall, 60.3 ± 11.2% of the dose was recovered in the
urine. Although no GA was found, unchanged acrylamide, AAMA, and GAMA accounted for
urinary excretion of (mean ± SD) 4.4 ± 1.5, 50.0 ± 9.4, and 5.9 ± 1.2% of the dose, respectively.
These results indicate that most of the acrylamide ingested with food is absorbed in humans.
Boettcher et al. (2006b) reported the influence of an AA-free diet on the excretion of
urinary mercapturic acid metabolites derived from AA in three healthy volunteers who fasted for
48 hours. Urinary AA mercapturic acid metabolites were considerably reduced after 48 hours of
fasting, with levels even well below the median level in nonsmokers. These results indicate that
the acrylamide in the diet is the main source of environmental AA exposure in humans, apart
from smoking.
Bjellaas et al. (2007) reported urinary mercapturic acid derivatives of AA and in a
clinical study comprising of 53 subjects. Median intakes (range) of AA were estimated based on
24 hour dietary recall as 21 (13-178) ug for nonsmokers and 26 (12-67) ug for smokers. The
median dietary exposure to acrylamide was estimated to be 0.47 (range 0.17-1.16) ug /kg body
weight per day. The median (range) total excretion of acrylamide in urine during 24 hours was
16 (7-47) ug acrylamide for nonsmokers and 74 (38-106) ug acrylamide for smokers. In a
multiple linear regression analysis, the urinary excretion of acrylamide metabolites correlated
statistically significant with intake of aspartic acid, protein, starch and coffee. Consumption of
citrus fruits correlated negatively with excretion of acrylamide metabolites.
Animal oral exposure
Studies in rats indicate that orally administered AA is rapidly and extensively absorbed
by the gastrointestinal tract (Doerge et al., 2005b; Fennell et al., 2005; Kadry et al., 1999; Dow
Chemical Co., 1984; Dixit et al., 1982; Miller et al., 1982).
Doerge et al. (2005b) compared the toxicokinetics of AA and GA in serum and tissues of
male and female B6C3Fi mice following a single dose by intravenous (i.v.) injection or gavage
of 0.1 mg/kg AA or a comparable dose of 0.1 mg/kg AA from a feeding exposure for 30 minutes.
Study groups also received an equimolar amount of GA from either an i.v. injection or gavage
dose. AA was rapidly absorbed following oral dosing, widely distributed to tissues, and
efficiently converted to GA. Liver levels of GA-DNA adducts were increased at 8 hours
postdosing, which is a time point where AA has been eliminated from the serum. Oral GA
dosing also resulted in rapid absorption, wide distribution to tissues, and liver DNA adduct
levels that were approximately 40% higher than those from an equimolar dose of orally
administered AA. Based on the kinetics of AA following i.v. injection, oral administration from
16 DRAFT-DO NOT CITE OR QUOTE
-------
the diet attenuated AA bioavailability to 23% of the i.v. dose, and aqueous gavage attenuated AA
bioavailability to 32-52%. In contrast, oral exposure resulted in higher relative internal levels of
GA compared with levels following an i.v. exposure, likely due to a first-pass effect but possibly
the result of some other kinetic change.
Fennell et al. (2005) administered 3 mg/kg [1,2,3-13C]-AA by gavage to male F344 rats
(n = 4). The total amount of AA metabolites recovered in urine by 24 hours after dosing was
50%, which is similar to that reported by Miller et al. (1982) and by Kadry et al. (1999).
The time course and extent of urinary elimination of radioactivity from male F344 rats (n
= 3) during a 7-day period following administration of either a single gavage or an i.v. dose of 10
mg/kg [2,3-14C]-acrylamide (in water vehicle) was essentially the same, indicating that 100% of
the oral dose was absorbed (Miller et al., 1982). The time courses of urinary elimination of
radioactivity for groups of rats (n = 3) given single oral doses of 1, 10, or 100 mg/kg
[2,3-14C]-acrylamide were also similar, indicating that the extent of absorption was not affected
by dose level in this experimental range. The rapidity of absorption was demonstrated by
observations that peak plasma levels of radioactivity were attained by 1 hour after administration
and that 53-67% of administered radioactivity was detected in the urine collected within 24
hours of administration (Miller et al., 1982).
Similar results indicating rapid and extensive oral absorption were reported for studies
with male Sprague-Dawley rats (n = 5-7) given single oral doses of 50 mg/kg [1-14C]-
acrylamide (Kadry et al., 1999). Radioactivity was detected in blood 5 minutes after
administration, and peak plasma levels of radioactivity occurred at 38 minutes after
administration. Approximately 51% of administered radioactivity was detected in urine
collected within 24 hours of administration (Kadry et al., 1999).
Animal inhalation exposure
Animal studies indicate that inhaled AA is readily absorbed (Sumner et al., 2003). Male
F344 rats and B6C3Fi mice were exposed to approximately 3 ppm of a mixture of [13C]-labeled
acrylamide and [14C]-labeled acrylamide vapor via nose-only inhalation for 6 hours. Selected
rats and mice were sacrificed immediately following the exposure period for determination of
[14C] content in tissues, an indicator of the extent of absorption of inhaled AA. The remaining
rats and mice were monitored for 24-hour elimination of radiolabeled AA and metabolites via
urine, feces, and expired air. Immediately following the 6-hour exposure period, approximately
18 and 8 jimol of [14C]-equivalents were recovered from tissues and carcasses of the rats and
mice, respectively. At the end of the 24-hour postexposure period, 42% of the total recovered
radioactivity was in urine, feces, and nose-tube and cage washes of rats; less than 3% was in
17 DRAFT-DO NOT CITE OR QUOTE
-------
exhaled air; and 56% remained in the body. In mice, 51% was recovered in urine, feces, and
nose-tube and cage washes; <3% was in exhaled air; and 46% remained in the body. Fractional
absorption could not be determined from the presented data because ventilation rates were
apparently not measured.2
Animal dermal exposure
Studies on dermal absorption in animals indicate that considerable amounts of AA can be
absorbed by the skin within short time frames (Sumner et al., 2003; Frantz et al., 1995; Dow
Chemical Co., 1984).
In male F344 rats, 14-30% (mean 22%) of an occluded dermal dose of [2,3-14C]-labeled
acrylamide (162 mg/kg in distilled water) was absorbed during a 24-hour exposure period
(Sumner et al., 2003). By 24 hours postapplication, approximately 44% of recovered
radioactivity (excluding material from dermal patch and wash of application site at termination
of exposure) was in the urine, feces, and cage washes; 3% was in exhaled air; and 53% remained
in tissues.
Frantz et al. (1995) applied a 0.5% aqueous solution of [14C]-labeled acrylamide to the
skin of male F344 rats at a single dose level of 2 mg/kg. The test material penetrated the skin
and was systemically distributed in male F344 rats within 24 hours; about 31% of the applied
dose penetrated the skin at the dosing site (was not removed by washing) and was considered
available for further absorption.
Peak plasma concentrations of radioactivity occurred at about 2 and 5 hours after dermal
administration of 2 and 50 mg/kg to F344 rats, respectively, indicating rapid absorption by the
skin (Dow Chemical Co., 1984). Aqueous solutions (1%) of [l,3-14C]-labeled acrylamide in a
nonionic detergent were applied at 2 or 50 mg/kg to areas of clipped skin on the backs of groups
of three male F344 rats. Radioactivity was measured in plasma and urine samples collected for
48 hours following administration. The peak concentration following administration of
50 mg/kg was about 20-fold higher than the peak concentration following administration of
2 mg/kg. Following attainment of peak concentrations, plasma concentrations declined with
time, showing slopes that were similar to slopes of curves following i.v. administration of 2 or
2 If reference minute ventilation rates for rats (0.7 cm3/min-gram) or mice (1.5 cm3/min-gram) and
midpoints of the reported ranges of the experimental animal body weights (211 grams, rats, and 30 grams, mice) are
used, the amounts of acrylamide inhaled in the 6-hour exposure period are calculated to be 6.5 and 2 umol
acrylamide/exposure period for rats and mice, respectively. Given that the measured amounts of recovered
acrylamide equivalents were about three- to fourfold higher than these calculated values, it is expected that the
animals had much higher minute ventilation rates during exposure than reference values. Sample calculations:
3 ppm x 71.08/24.45 = 8.7 mg/m3; (8.7mg/m3) x (0.7cm3/min-gram) x (60 min/hour) x (6 hours/exposure) x
(211 grams/rat) x (m3/106cm ) x (mmol/71.08 mg) x (103umol/mmol) = 6.5 umol/rat-exposure period.
18 DRAFT-DO NOT CITE OR QUOTE
-------
50 mg/kg doses of [l,3-14C]-labeled acrylamide. The fraction of dermally applied compound
that was absorbed was not reported.
Results of several in vitro studies describe dermal absorption of acrylamide. Frantz et al.
(1995) applied a 0.5% [14C]-labeled acrylamide in aqueous solution to excised skin discs from
male F344 rats and noted considerable dermal penetration after 24 hours. Approximately 54% of
the radioactivity was recovered in effluents and 13% was retained in washed skin. Diembeck et
al. (1998) applied a 0.5% [14C]-labeled acrylamide in aqueous solution to excised sections of
female pig skin for 24 hours. Approximately 6% of the applied dose was found on the skin
surface; 17.5% in the horny layer, 2% in the epidermis, 52.5% in the dermis, and 22% in the
receptor fluid. Marty and Vincent (1998) applied [14C]-labeled acrylamide (in an aqueous gel of
2% polyacrylamide) to biopsied human abdominal skin for 24 hours at acrylamide
concentrations of 1.28 or 2 ppm. Approximately 28 and 21% of the applied doses, respectively,
were recovered in the receptor fluid. Between 1.6 and 3.4% of applied doses was recovered in
dermis and epidermis. The authors estimated total absorption of acrylamide to be 33.2 and
26.7% at low and high concentration, respectively, based on radioactivity recovered collectively
from the receptor phase, epidermis, and dermis.
3.2. DISTRIBUTION
No human data on distribution of acrylamide were identified. Results from several
animal studies indicate that, following absorption, radioactivity from radiolabeled AA is
distributed among tissues with no specific accumulation in any tissues other than red blood cells
(Barber et al., 2001; Kadry et al., 1999; Crofton et al., 1996; Marlowe et al., 1986; Ikeda et al.,
1985; Dow Chemical Co., 1984; Miller et al., 1982; Edwards, 1975; Hashimoto and Aldridge,
1970) and late-staged spermatids (Sega et al., 1989).
Animal oral exposure
Following 13 daily oral doses of [l,3-14C]-labeled acrylamide (at levels of 0.05 or
30 mg/kg), tissue concentrations of acrylamide in male F344 rats were similar among tissues
with the exception of red blood cells, which showed higher concentrations, presumably due to
the formation of hemoglobin adducts of AA or GA (Dow Chemical Co., 1984). In rats exposed
to 30 mg/kg, mean concentrations (jig equivalents [14C]-acrylamide per gram of tissue) were as
follows: red blood cells, 383.70; liver, 87.74; kidneys, 70.43; epididymides, 70.60; testes, 67.14;
sciatic nerve, 54.00; brain, 53.52; carcass, 47.56; skin, 39.11; and plasma, 16.45. In rats exposed
to 0.05 mg/kg, the mean concentration in red blood cells was 1.26 |ig/g [14C]-acrylamide
equivalents (approximately 61% of the dose that was recovered from all tissues) compared with
19 DRAFT-DO NOT CITE OR QUOTE
-------
a range of 0.07-0.13 |ig/g [14C]-acrylamide equivalents in the other tissues (Dow Chemical Co.,
1984).
In Sprague-Dawley rats given single oral doses of 50 mg/kg [l-14C]-labeled acrylamide,
tissue concentrations of radioactivity, 28 and 144 hours after administration, were indicative of
wide distribution of AA metabolites among tissues with no evidence for accumulation in toxicity
targets, i.e., AA bound, but did not accumulate in erythrocytes or neural tissue (Kadry et al.,
1999). At 28 hours, brain, thyroid, testes, adrenal, pancreas, thymus, liver, kidney, heart, and
spleen showed a narrow range of mean concentrations (based on values for five rats), 0.05-
0.10% of initial dose/g. Higher concentrations were noted in the skin, bone marrow, stomach,
and lung, ranging from 0.15 to 0.18% of initial dose/g, and only the gastric contents showed a
markedly higher concentration, 1.37% of initial dose/g. At 144 hours after administration, tissue
concentrations were uniformly low for tissues including the gastric contents, ranging from
0.01 to 0.05% of initial dose/g, with the exception of skin, bone marrow, and lung, which had
mean concentrations of 0.06, 0.08, and 0.19% of initial dose/g, respectively.
Animal dermal exposure
Following 24-hour dermal exposure of male F344 rats to [14C]-labeled acrylamide
(150 mg/kg), blood cells had the highest concentration of AA equivalents (excluding skin at the
site of exposure), about 1 |imol/g (71 jig equivalents/g), followed by skin at the nondosing site
(-28 |ig/g); liver, spleen, testes, and kidneys (-21 |ig/g); lungs, thymus, brain, and epididymis
(-14 |ig/g); and fat (<4 |ig/g) (Sumner et al., 2003).
Animal inhalation exposure
Immediately following a 6-hour inhalation exposure of male F344 rats to 3 ppm
[14C]/[13C]-labeled acrylamide vapor, blood cells had the highest concentration (-7 |ig/g),
followed by concentrations in testes, skin, liver, and kidneys (-6 |ig/g) and brain, spleen, lung,
and epididymis (-4 |ig/g) (Sumner et al., 2003). Immediately following a 6-hour inhalation
exposure to the same concentration, male B6C3Fi mice showed the following order of
decreasing AA equivalent concentrations: testes (-14 |ig/g), skin and liver (-11 |ig/g), kidney
(-10 |ig/g), epididymis (-8 |ig/g), brain (-7 |ig/g), lung and blood (-6 |ig/g), and fat (-5 |ig/g).
These differences in distribution pattern between rats and mice following inhalation exposure are
unexplained, but more data are needed to support a consistent difference and to determine the
kinetic determinants.
Animal intravenous or intraperitoneal administration
20 DRAFT-DO NOT CITE OR QUOTE
-------
Similar results were reported in male albino Porton rats injected with single i.v. doses of
100 mg/kg [l-14C]-labeled AA (Hashimoto and Aldridge, 1970). Twenty-four hours and 14 days
after dosing, tissue concentrations of radioactivity (jig equivalents/g) were as follows: whole
blood, 90.9 and 54.7; kidney, 36.1 and 6.5; liver, 26.1 and 4.0; brain, 18.6 and 5.1; spinal cord,
12.4 and 5.0; sciatic nerve, 10.6 and 4.0; and plasma, 4.5 and 0.4 (Hashimoto and Aldridge,
1970).
Doerge et al. (2005a) measured DNA adducts following a single intraperitoneal (i.p.)
administration of AA and GA to adult B6C3Fi mice and F344 rats at 50 mg AA/kg or an
equimolar dose of GA (61 mg/kg). GA-derived DNA adducts of adenine and guanine were
formed in all tissues examined for both AA and GA dosing, including both target tissues
identified in rodent carcinogenicity bioassays and nontarget tissues (including liver, brain,
thyroid, leukocytes, mammary gland, and testis in rats), and in liver, lung, kidney, leukocytes,
and testis in mice,; indicating widespread distribution.
Concentrations of radiolabel did not differ in neural tissues (brain, sciatic nerve, spinal
cord) and nonneural tissues (fat, liver, kidney, testes, lung, small intestine, skin, muscle),
following single i.v. injections of 10 mg/kg [2,3-14C]-labeled AA into groups of three male F344
rats sacrificed at time intervals ranging from 15 minutes to 7 days after dosing (Miller et al.,
1982). Radioactivity was rapidly distributed to all tissues and eliminated from most tissues (and
plasma) with biphasic kinetics showing half-lives of elimination of about 5 hours or less for the
first phase and about 8 days or less for the second phase. Peak concentrations of radiolabel were
observed by 1 hour after dose administration in liver, fat, kidney, nervous tissues, and testes.
Red blood cells did not show an elimination of the radioactivity with time up to 70 hours after
dose administration, consistent with the formation of AA and GA adducts with hemoglobin.
Less than 1% of the dose was contained in the brain, spinal cord, or sciatic nerve at any time
point, indicating no special accumulation of AA or metabolites in these targets of AA toxicity
(Miller et al., 1982).
Following i.p. injection of [14C]-labeled acrylamide (125 mg/kg) into male (C3H x
101)F1 mice, peak levels of radioactivity appeared 8-12 days postdosing in sperm heads
recovered from the vasa deferentia and caudal epididymides from a 3-week period of monitoring
(Sega et al., 1989). Essentially all of the covalently bound radioactivity in spermheads was
shown to be alkylated protamine; alkylation of DNA represented generally <0.5% of the sperm-
head alkylation radioactivity. The time course of alkylation of sperm-head protamine paralleled
the time course of AA-induced dominant lethality in mice injected with the same dose
(125 mg/kg) of AA (Sega et al., 1989). In another study using whole-body autoradiography of
Swiss-Webster mice orally exposed to [14C]-labeled acrylamide, (120 mg/kg), radioactivity
21 DRAFT-DO NOT CITE OR QUOTE
-------
moved through the testis and the reproductive tract in a sequence that paralleled the movement
of spermatids (Marlowe et al., 1986).
Further evidence that AA does not accumulate in most tissues is provided by
observations that, 30 minutes after the final i.p. dose in a daily repeated exposure from 10 to
90 days, at dose levels between 3.3 and 30 mg/kg-day, AA concentrations in rat sciatic nerves or
in serum were similar to concentrations in rats exposed to that dose for the first time (Crofton et
al., 1996). The ranges and durations of exposure to groups of three male Long-Evans hooded
rats in this study were 0, 7.5, 15, or 30 mg/kg-day for 10 days of exposure; 0, 5, 10, 15, or 20
mg/kg-day for 30 days; and 0, 3.3, 6.7, or 10 mg/kg-day for 90 days.
Results from studies with pregnant animals indicate that absorbed AA is distributed
across the placenta (Marlowe et al., 1986; Ikeda et al., 1985, 1983). Two hours following i.v.
administration of 5 mg/kg [l-14C]-labeled AA to pregnant beagle dogs (n = 6), concentrations of
radioactivity in blood, brain, heart, and lung were similar in both maternal and fetal tissues
(Ikeda et al., 1985). Average concentrations of radioactivity in maternal tissues were only about
1.1- to 1.2-fold higher than those in fetal tissues. Comparable results were found with pregnant
miniature pigs treated similarly (Ikeda et al., 1985). Whole-body radiographs of pregnant Swiss-
Webster mice, 3 or 24 hours following gavage administration of 120 mg/kg [2,3-14C]-labeled AA
on gestation day (GD) 13 or 17, showed uniform distribution of radioactivity among fetal tissues
that was similar to that seen in maternal tissues, with the exception of increased label in fetal
brain regions at 13 days and in fetal skin regions at 17 days (Marlowe et al., 1986). The
autoradiographic technique used, however, provided only qualitative information.
3.3. METABOLISM
Human metabolism
In the Fennell et al. (2005) study on 24 adult male volunteers previously discussed in the
absorption section, approximately 86% of the urinary metabolites were derived from glutathione
(GSH) conjugation and excreted as N-acetyl-S-(3-amino-3-oxopropyl)cysteine and its S-oxide.
GA, glyceramide (2,3-dihydroxypropionamide), and low levels of N-acetyl-S-(3-amino-
2-hydroxy-3-oxopropyl)cysteine were detected in urine. On oral administration, a linear dose
response was observed for AAVal and GAVal in hemoglobin. The authors reported that the
urinary metabolites of AA in humans showed similarities and differences with data obtained
previously in the rat and mouse. The main pathway of metabolism in humans was via direct
glutathione conjugation, forming N-acetyl-S-(3-amino-3-oxopropyl)cysteine, as observed in the
rat and mouse, and its S-oxide, which has not been reported previously. Epoxidation to GA was
the other important pathway, with glyceramide formed as a major metabolite in humans. GA
22 DRAFT-DO NOT CITE OR QUOTE
-------
was detected in low amounts. The glutathione conjugation of GA, which is a major pathway in
rodents, appeared to occur at very low levels in humans. Metabolism via GA (i.e., derived from
GA and glyceramide) in humans was approximately 12% of the total urinary metabolites. This
is considerably lower than the amount of GA derived metabolites reported for oral
administration of AA in rats (28% at 50 mg/kg, [Sumner et al., 2003]) and in mice (59% at 50
mg/kg [Sumner et al., 1992]).
Boettcher et al. (2005) measured the mercapturic acid of AA and its epoxide GA, i.e.,
N-acetyl-S-(2-carbamoylethyl)-L-cysteine (AAMA) and N-(R,S)-acetyl-S-(carbamoyl-
2-hydroxyethyl)-L-cysteine (GAMA) in human urine as biomarkers of the internal exposure to
acrylamide in the general population. The median levels in smokers (n = 13) were found to be
about four times higher than in nonsmokers (n = 16) with median levels of 127 |ig/L vs. 29 |ig/L
for AAMA and 19 |ig/L vs. 5 |ig/L for GAMA. The level of AAMA in the occupationally
nonexposed collective (n = 29) ranged from 3 to 338 |ig/L, the level of GAMA from below level
of detection to 45 |ig/L. The authors noted that the ratio of GAMA: AAMA varied from 0.03 to
0.53; the median was 0.16, which is in reasonable agreement with results of different studies on
rats. They concluded that the metabolic conversion of AA to its genotoxic epoxide GA seems to
occur to a comparable extent in rats and humans. They also measured the hemoglobin adducts of
AA and GA in the blood of 26 participants. These results were compared with those of the
mercapturic acids to deduce a steady state for AA uptake and demonstrate a higher reactivity of
GA in comparison to AA towards hemoglobin compared to GSH in humans.
Boettcher et al. (2006a) investigated the human metabolism of AA to AAMA and GAMA
in a healthy male volunteer who received a single dose of about 1 mg deuterium-labelled
acrylamide (d(3)-AA), representing 13 |ig/kg body weight, in drinking water. Urine samples
before dosing and within 46 hours after the dose were analyzed for d(3)-AAMA and d(3)-
GAMA by LC-ESI-MS/MS. Total recovery in urine after 24 hours was about 51% as the sum of
AAMA and GAMA and was similar to recoveries in rats (53-66%) given a gavage dose of 0.1
mg/kg bw (Doerge et al., 2007). After 2 days AAMA accounted for 52% of the total AA dose,
and was the major metabolite of AA in humans. GAMA accounted for 5%, and appeared as a
minor metabolite of AA. A urinary ratio of 0.1 was observed for GAMA/AAMA compared to
previously reported values of 0.2 for rats and 0.5 for mice (Doerge et al., 2005a). The authors
conclude that the metabolic fate of AA in humans was more similar to that in rats than in mice as
previously demonstrated in terms of haemoglobin adducts.
Fuhr et al. (2006) evaluated the urinary levels of AA, AAMA, GA, and GAMA in six
young healthy volunteers after the consumption of a meal containing 0.94 mg of acrylamide.
Urine was collected up to 72 hours thereafter. No GA was found. Unchanged acrylamide,
23 DRAFT-DO NOT CITE OR QUOTE
-------
AAMA, and GAMA accounted for urinary excretion of (mean ± SD) 4.4 ± 1.5, 50.0 ± 9.4, and
5.9 ± 1.2% of the dose, respectively. Conjugation with glutathione exceeded the formation of
the reactive metabolite GA. The data suggests an at least twofold and fourfold lower relative
internal exposure for GA from dietary acrylamide in humans compared with rats or mice,
respectively.
Paulsson et al. (2005) evaluated variability in human metabolism of acylamide and
glycidamide due to genetic polymorphic enzymes in the detoxification of acrylamide and its
metabolite glycidamide. Enzymes that enhance conjugation with glutathione (GSH), the
glutathione transferases (GSTs), may influence the detoxification of both acrylamide and
glycidamide, whereas the enzyme epoxide hydrolase (EH) should only catalyse the hydrolysis of
glycidamide. Paulsson et al. estimated the internal doses of acrylamide or glycidamide measured
as specific adducts to hemoglobin (Hb) in blood samples after in vitro incubation with these
compounds. Blood samples from individuals with different genotypes for GSTT1 and GSTM1
were studied. No significant differences in adduct levels depending on genotype were noted. In a
parallel experiment, incubation with ethylene oxide was used as positive control. In this
experiment individuals carrying GSTT1 showed lower adduct level increments from ethylene
oxide than individuals lacking GSTT1. Furthermore, addition of ethacrynic acid or laurylamine,
compounds which inhibit GST and EH, respectively, did not affect the adduct levels. Based on
their results, the authors suggest that neither GSTs nor EH has any significant effect on the blood
dose, measured as Hb-adducts over time, after exposure to acrylamide or glycidamide.
Animal studies
Results from rat and mouse studies also indicate that acrylamide is rapidly metabolized
and excreted predominantly in the urine as metabolites (Twaddle et al., 2004; Sumner et al.,
2003, 1999, 1992; Dow Chemical Co., 1984; Dixit et al., 1982; Miller et al., 1982; Edwards,
1975). Formation of AA and GA hemoglobin adducts in rats was initially reported by Bergmark
et al. (1991) and second rate constants have been subsequently derived by in vitro and in vivo
studies (Bergmark et al. 1993; Fennell et al., 2005; Tareke et al., 2006; Tornqvist et al., 2008).
Bergmark et al. (1991) reported that the hemoglobin binding index of AA to cysteine was found
to be 6,400 pmol/g Hb/|imol AA/kg, higher than for any other substance studied so far in the rat,
and the hemoglobin binding index of GA to cysteine was 1,820 pmol/g Hb/|imol GA/kg. The
difference between AA and GA rates was proposed as being due primarily to a lower reactivity
of GA than AA toward Hb-cysteine and a shorter half-life for GA in blood (based on
determinations of these values in this study). The more recent studies have focused on the AA
24 DRAFT-DO NOT CITE OR QUOTE
-------
and GA binding to the N-terminal valine residue (Bergmark et al. 1993; Fennell et al., 2005;
Tareke et al., 2006; Tornqvist et al., 2008).
A metabolic scheme for acrylamide, based on results from these and other studies, is
illustrated in Figure 3-1. AA reacts readily with glutathione to form a glutathione conjugate,
which is further metabolized to N-acetyl-S-(3-amino-3-oxopropyl)cysteine or S-(3-amino-
3-oxopropyl)cysteine. N-acetyl-S-(3-amino-3-oxopropyl)cysteine has been identified as the
major urinary metabolite of acrylamide in male F344 rats exposed to oral doses of 1-100 mg/kg
[2,3-14C]-labeled acrylamide (Miller et al., 1982) and in male F344 rats and B6C3Fi mice
exposed to oral doses of 50 mg/kg [l,2,3-13C]-labeled acrylamide (Sumner et al., 1992).
25 DRAFT-DO NOT CITE OR QUOTE
-------
acrylamide Hb adducts
Hb
O
H,C=C— C— NH,
2 H 2
acrylamide
GSH
GS-CH2-CH2-CONH2
\
CYP2E1
glycidamide Hb adducts
j i
-Hb
O
\
H,C C—CONH9
2 H 2
glycidamide
GSH
-s
GSH
GS-CH2-CHOH-CONH2
N-AcCys-S-CH2-CH2-CONH2
N-acetyl-S-(3-amino-3-oxypropyl)cysteine
Cys-S-CH2-CH2-CONH2
S-(3-amino-3-oxypropyl)cysteine
CH2OH
GS—C—CONH9
H 2
adducts
HOCH2-CHOH-CONH2
2,3-dihyroxypropionamide
HOCH2-CHOH-COOH
2,3-dihyroxypropionic acid
N-AcCys-S-CH2-CHOH-CONH2
N-acetyl-S-(3-amino-2-hydroxy-3-oxopropyl)cysteine
^ CH2OH
N-AcCys—S—C—CONH,
H 2
N-acetyl-S-(1-carbamoyl-2-hydroxyethyl)cysteine
Note: Processes involving several steps are represented with broken arrows. Abbreviations: Hb, hemoglobin;
GSH, reduced glutathione; N-AcCys, N-acetylcysteine.
Sources: Adapted from Sumner et al. (1999); Calleman (1996); IARC (1994a).
26
DRAFT-DO NOT CITE OR QUOTE
-------
Table 3-1 lists the relative amounts of AA metabolites determined by [13C]-NMR analysis of
urine collected for 24 hours in the latter of these studies. In another study with wild-type
C57BL/6N x Svl29 mice exposed to 50 mg/kg [l,2,3-13C]-labeled acrylamide, N-acetyl-S-
(3-amino-3-oxopropyl)cysteine and S-(3-amino-3-oxopropyl)cysteine accounted for 29 and 20%
of total metabolites excreted within 24 hours in the urine (Sumner et al., 1999).
Table 3-1. Urinary metabolites collected for 24 hours following oral
administration of [l,2,3-13C]-labeled acrylamide (50 mg/kg) to male F344
rats or male B6C3Fi mice
Metabolite21
From AA precursor
N-acetyl-S-(3 -amino-3 -oxopropyl)cy steine
GA
From GA precursor
N-acetyl-S-(3 -amino-2-hydroxy-3 -oxopropyl)cy steine
N-acetyl-S-( 1 -carbamoyl-2-hydroxyethyl)cy steine
2,3 -Dihydroxypropionamide
Percent of total metabolites excreted in urine in 24
hours
(mean ± SD, n = 3)
Rat
67.4 ±3.6
5.5 ±1.0
15.7 ±1.3
9.0 ±1.1
2.4 ±0.7
Mouseb
41.2 ±2.2
16.8 ±2.1
21.3 ±0.6
11.7 ±0.6
5.3 ±1.2
al3C-NMR analysis was used to detect, identify, and quantify metabolites in urine. Urinary metabolites accounted
for about 50% of the administered dose in both species. Unchanged acrylamide was detected in urine but was not
quantified. In other studies with F344 rats exposed to [2,3-14C]-labeled acrylamide, less than 2% of administered
radiolabel was excreted in urine and bile as unchanged acrylamide (Miller et al., 1982).
bln mice, an epoxide degradation product accounted for 4% of the total metabolites excreted.
Source: Sumner etal. (1992).
Another initial step, catalyzed by CYP2E1, involves oxidation of AA to the epoxide
derivative, GA. GA (either at the number 2 or 3 carbon) can react with GSH to form conjugates
that are further metabolized to N-acetyl-S-(3-amino-2-hydroxy-3-oxopropyl)cysteine or N-
acetyl-S-(l-carbamoyl-2-hydroxyethyl)cysteine. GA may also undergo hydrolysis, perhaps
catalyzed by epoxide hydrolases (Sumner et al., 1999, 1992), leading to the formation of 2,3-
dihydroxypropionamide and 2,3-dihydroxypropionic acid. GA and metabolites (or degradation
products) derived from it accounted for about 33 and 59% of the total metabolites excreted in rat
and mouse urine within 24 hours, respectively (Table 3-2), indicating that, under these test
conditions, the rate of transformation from AA to GA is about two-fold greater in mice than in
rats. Similar results were reported in a study of metabolites in urine collected for 24 hours after
6-hour inhalation exposure (nose only) to 3 ppm acrylamide (Sumner et al., 2003). GA and
metabolites derived from it accounted for 36 ± 2.4 and 73 ± 3.7% of total metabolites excreted in
rat and mouse urine within 24 hours, respectively (Sumner et al., 2003).
Doroshyenko et al. (2009) investigated acrylamide toxicokinetics in 16 healthy
volunteers in a four-period change-over trial and evaluated the respective role of cytochrome
27
DRAFT-DO NOT CITE OR QUOTE
-------
P450 2E1 (CYP2E1) and GSTs. Participants ingested potato chips containing acrylamide (1 mg)
without co-medication, after CYP2E1 inhibition (500 mg disulfiram, single dose) or induction
(48 g/d ethanol for 1 week), and were phenotyped for CYP2E1 with chlorzoxazone (250 mg,
single dose). Acrylamide-containing potato chips were prepared by frying 150g batches of self-
prepared potato chips ("Princess" potatoes) at 190°C for 5 minutes. Unchanged acrylamide and
the mercapturic acids N-acetyl- S-(2-carbamoylethyl)-cysteine (AAMA) and N-acetyl- S-(2-
hydroxy-2-carbamoylethyl)-cysteine (GAMA) accounted for urinary excretion [geometric mean
(percent coefficient of variation)] of 2.9% (42), 65% (23), and 1.7% (65) of the acrylamide dose
in the reference period. Hemoglobin adducts clearly increased following the acrylamide test-
meal. The increases in cumulative amounts of acrylamide, AAMA, and GAMA excreted, and in
AA adducts were significant during CYP2E1 blockade [point estimate (90% confidence
interval)] to the 1.34-fold (1.14-1.58), 1.18-fold (1.02-1.36), 0.44-fold (0.31-0.61), and 1.08-fold
(1.02-1.15) of the reference period, respectively, but were not changed significantly during
moderate CYP2E1 induction. Individual baseline CYP2E1 activity, CYP2E1*6, GSTP1
313A>G and 341T>C single nucleotide polymorphisms, and GSTM1- and GSTTl-null
genotypes had no major effect on acrylamide disposition. The changes in acrylamide
toxicokinetics upon CYP2E1 blockade provide evidence that CYP2E1 is a major but not the only
enzyme mediating acrylamide epoxidation in vivo to glycidamide in humans. The authors
reported no obvious genetic risks or protective factors in xenobiotic-metabolizing enzymes could
be determined for exposed subjects.
Age related increases in human CYP2E1 expression have been reported. Johnsrud et al.
(2003) evaluated the content of CYP2E1 in human hepatic microsomes from samples spanning
fetal (n = 73, 8-37 weeks) and postnatal (n = 165, 1 day-18 years) ages. Measurable
immunodetectable CYP2E1 was seen in 18 of 49 second-trimester fetal samples (93-
186 gestational days; median level = 0.35 pmol/mg microsomal protein) and 12 of 15 third-
trimester samples (>186 days, median level = 6.7 pmol/mg microsomal protein). CYP2E1 in
neonatal samples was low and less than that of infants 31-90 days of age, which was less than
that of older infants, children, and young adults (median [range] = 8.8 [0-70]; 23.8 [10-43];
41.4 [18-95] pmol/mg microsomal protein, respectively; each/? < 0.001, analysis of variance,
posthoc). Among those older than 90 days of age, CYP2E1 content was similar. A fourfold or
greater intersubject variation was observed among samples from each age group, with the
greatest variation, 80-fold, seen among neonatal samples. These results suggest that infants less
than 90 days old may have decreased clearance of CYP2E1 substrates such as acrylamide (i.e.,
decreased levels of glycidamide) compared with older infants, children, and adults. However,
actual differences in the total amount metabolized and parent compound cleared (and the
resulting spectrum of adverse effects) would depend upon the delivery rate and substrate
concentration relative to the value of the Michaelis-Menten constant (Km) for CYP2E1
28 DRAFT-DO NOT CITE OR QUOTE
-------
(Lipscomb, 2004; Lipscomb et al., 2003). The higher the substrate concentration relative to Km,
the more marked will be the influence of enzyme level (i.e., maximum activity level) on total
clearance for a saturable enzyme like CYP2E1.
Results from mouse studies indicate that mouse CYP2E1 is the only CYP isozyme that
catalyzes the oxidative formation of GA from AA. Following oral administration of single
50 mg/kg doses of [l,2,3-13C]-labeled AA, no evidence of metabolites formed through GA was
found by [13C]-NMR analysis of urine collected for 24 hours from C57BL/6N x Svl29 mice
devoid of CYP2E1 (CYP2E1 null) or wild-type mice of the same strain treated with the CYP2E1
inhibitor, aminobenzotriazole (ABT) (50 mg/kg i.p. injection 2 hours preexposure) (Sumner et
al., 1999). In contrast, urine collected from wild-type mice contained considerable amounts of
metabolites derived from GA (Sumner et al., 1999). With wild-type mice in this study, 22% of
excreted metabolites were accounted for by metabolites derived from glutathione conjugation
with GA (N-acetyl-S-[3-amino-2-hydroxy-3-oxopropyl]cysteine andN-acetyl-S-[l-carbamoyl-
2-hydroxyethyl]cysteine) and 28% of excreted metabolites were accounted for by GA and its
hydrolysis products (2,3-dihydroxypropionamide and 2,3-dihydroxypropionic acid). The wild-
type and CYP2El-null mice excreted a similar percentage of the administered dose in the urine
within 24 hours (about 30%), suggesting that the CYP2El-null mice compensated for the
CYP2E1 deficiency by metabolizing more of the administered AA via direct conjugation with
GSH.
Figure 3-1 does not include a possible minor pathway hypothesized to result in the
release of CO2 from hydrolysis products of GA. This pathway is not included because of
conflicting results from several studies. Following i.v. administration of 100 mg/kg [1-14C]-
labeled AA to male albino Porton rats, about 6% of the injected dose of radioactivity was
exhaled as CO2 in 8 hours (Hashimoto and Aldridge, 1970), but following administration of [2,3-
14C]-labeled AA to male F344 rats, no radioactivity was detected in exhaled breath (Miller et al.,
1982). Sumner et al. (1992) noted that these results may be consistent with the existence of a
minor pathway involving metabolism of 2,3-dihydroxypropionamide (glyceramide) to glycerate
and hydroxypyruvate with the subsequent release of CC>2 and production of glycolaldehyde, but
they did not detect labeled two-carbon metabolites in urine of mice exposed to [1,2,3-13C]-
labeled AA. In other experiments, no exhaled 14CC>2 was detected following oral administration
of 50 mg/kg [l-14C]-labeled AA to male Sprague-Dawley rats (Kadry et al., 1999), whereas 3-
4% of i.v. injected [1,3-14C]-AA (2 or 100 mg/kg) was detected as 14CO2 in exhaled breath in
male F344 rats (Dow Chemical Co., 1984). During a 24-hour period following a 24-hour dermal
exposure of male F344 rats to 162 mg/kg [2,3-14C]-labeled AA, 14CO2 in exhaled breath
accounted for 1.8 ± 0.2% of radioactivity recovered in exhaled air, urine, feces, and tissues
(Sumner et al., 2003). Similarly, 14CO2 in exhaled breath accounted for 1.7 ± 0.1 and 0.9 ± 0.2%
of radioactivity recovered in exhaled air, urine, feces, and tissues in male B6C3Fi mice and F344
29 DRAFT-DO NOT CITE OR QUOTE
-------
rats, respectively, following nose-only inhalation exposure to 3 ppm of a mixture of
[1,2,3-13C]-AA and [2,3-14C]-AA (Sumner et al., 2003).
Route-to-Route Differences
Results from a rat kinetic study by Sumner et al. (2003) indicate an intraparenteral (i.p.).
or gavage route of exposure had a small effect on the percentage of AA conjugated to GSH vs.
the percentage of AA converted to GA. Following i.p. or gavage administration of 50 mg/kg
[1,2,3-13C]-AA to male F344 rats, 69 ± 0.9% or 71 ± 3.8% of total urinary metabolites,
respectively, were metabolites associated with direct conjugation of AA with glutathione.
Similarly, in the only available animal inhalation kinetic study (i.e, no human inhalation kinetic
studies are available), the metabolites associated with direct conjugation of AA with glutathione
following a 6-hour inhalation (nose only) exposure of male F344 rats to 3 ppm of a mixture of
radiolabeled [1,2,3-13C]- and [2,3-14C]-acrylamide accounted for 64 ± 2.4% of metabolites in
urine collected for 24 hours. The percentages of total urinary metabolites associated with GA
formation were 31 ± 0.9, 28 ± 3.8, and 36 ± 2.4% following i.p., gavage, and inhalation
exposure, respectively.
In this same study, Sumner et al. (2003) report statistically significantly larger
percentages of urinary metabolites associated with GA formation following an inhalation
exposure compared with an i.p. and gavage exposure. GA-Val levels are also higher and AA-
Val levels lower (as indicators of serum AUCs), following the single 6 hr inhalation exposures
versus the single gavage dose in rats, however, statistical significance was not reported for the
adduct level differences, and the numbers are within two fold of each other. Doerge et al.
(2005b, 2005c) report an increased percentage of GA formation observed in mice and F344 rats
from a gavage or dietary exposure compared to an i.v. exposure that, in conjunction with the
Sumner et al. (2003) results, indicate first pass metabolism in the lungs following an inhalation
exposure similar to the first pass metabolism in the liver from an oral exposure, but apparently
the lungs may have a larger percent of oxidative metabolism of AA to GA.
Lehning et al. (1998) report that repeated oral exposures of 26-45 days to AA at
relatively low doses (e.g., 20 mg/kg-day from drinking water concentrations of 20 mM) induces
axonal degeneration, but shorter-term (11 days) exposure to higher i.p. doses (50 mg/kg-day)
does not. Barber et al. (2001) compared AA metabolism and toxicokinetics for these dosing
regimens, but did not find differences that provided a clear explanation for the occurrence of
degeneration with the longer oral dosing regimen. In this study, plasma concentrations of
radioactivity in AA and GA were determined from tail-vein blood samples that were collected
from groups of five to seven Sprague-Dawley rats at nine intervals from 0 to 580 minutes
following a single administration of [2,3-14C]-labeled AA by gavage (24 hours after the last dose
of a drinking water solution of 20 mg/kg-day nonlabeled AA for 34 days) or by a single i.p.
30 DRAFT-DO NOT CITE OR QUOTE
-------
injection (on day 11 of the i.p. administration of 50 mg/kg-day for 11 days). The authors noted
that the toxicokinetics from a single gavage dose had been evaluated in separate experiments,
and in the opinion of the authors, gave a reasonable estimate of the AUCs and half-life of a
drinking water exposure that was simulated with multiple smaller doses (i.e., data not shown).
Barber et al. (2001) also measured the activities of CYP2E1 and epoxide hydrolase in
liver microsomes, as well as concentrations of AA-hemoglobin and GA-hemoglobin adducts
before treatment, after i.p. exposure for 5 or 11 days and after 15, 34, and 47 days of oral
exposure. With both dosing regimens, AA appeared rapidly in plasma and rose to peak
concentrations within 60-90 minutes, followed by peak levels of GA. Respective plasma half-
lives (tVa) were approximately 2 hours and peak plasma levels for each route were directly
related to the magnitude of the respective daily dose (i.e., the i.p. dose and resulting Cmax were
both 2.5 times larger than comparable oral parameters). The only differences found in metabolic
or toxicokinetic parameters for the two dosing regimens involved some, but not all, parameters
that determined GA formation and metabolism. Derived areas under the plasma concentration
vs. time curves (AUCs) indicated that a larger proportion of plasma AA was converted to GA
following a single oral dose of 20 mg/kg (22%) than following a single i.p. dose of 50 mg/kg
(10%). A larger proportion of plasma AA was also converted to GA following the 34 days of
repeat oral dosing (30%) compared with 11-days of i.p. dosing (8%). No correlation was found
to the different enzyme activities involved in GA formation (CYP2E1) or metabolism (epoxide
hydrolase). Concordant with the serum data, concentrations of AA-hemoglobin adducts were
about 36% lower in the oral dosing regimen (8 jimol adduct/g globin at 15 days) compared with
the i.p. regimen (12.5 jimol adduct/g globin at 11 days), and concentrations of GA-hemoglobin
adducts were about two fold higher. Barber et al. (2001) noted that, although it has been
proposed that GA might mediate axonal degeneration, peak concentrations of free GA with the
subchronic oral regime were relatively low and other studies showed that GA is only a weak
neurotoxicant. It was concluded that the mechanism of axonal degeneration did not appear to
involve route- or dose-rate differences in metabolism or disposition of AA.
Doerge et al. (2005b) compared the toxicokinetics of AA and GA in serum and tissues of
male and female B6C3Fi mice following a single dose by i.v. injection or gavage of 0.1 mg/kg
AA, or a comparable dose of 0.1 mg/kg AA from a feeding exposure for 30 minutes. Study
groups also received an equimolar amount of GA from either an i.v. injection or gavage dose.
Oral exposure to AA resulted in higher relative internal levels of GA compared with levels
following an i.v. exposure, due either to a first-pass effect or some other factors that affect the
kinetic disposition from an i.v. dose. Similar results were observed by Doerge et al (2005c) in a
comparable study with F344 rats.
In comparing the results of the Doerge et al (2005b) mouse study with previous studies
from that laboratory at a 500-fold higher concentration (Twaddle et al., 2004), an increase in
31 DRAFT-DO NOT CITE OR QUOTE
-------
relative internal GA levels was observed, suggesting that as dose rate decreases, the conversion
of AA to GA in mice is more efficient.
Differences in mouse and rat metabolism
Twaddle et al. (2004) administered AA at approximately 50 mg/kg via gavage to adult
male and female B6C3Fi mice. Serum concentrations of AA and GA were taken at 0.5, 1, 2, 4,
and 8 hours postdosing. Livers were removed from control and AA-treated mice at all exposure
times, and analyzed for GA derived DNA adducts. The results indicated no systematic sex
differences in AA and GA serum levels at each time point for the species and doses in this study.
Twaddle et al. (2004) estimated an AA half-life of elimination from plasma at 0.73 hours in
these B6C3Fi mice. This value in mice can be compared to an estimate of 2 hours in F344 rats
following a subchronic oral administration of 2.8 mM AA in drinking water for 34 days or
subacute i.p. doses at 50 mg/kg-day for 11 days (Barber et al., 2001). Miller et al. (1982)
estimated a 1.7 hour half-life for AA in rat blood following a 10 mg/kg i.v. dose. For GA,
Twaddle et al. (2004) report that the mice had an elimination half-life for GA of 1.9 hours, which
is identical to that measured by Barber et al. (2001) in rats. Barber et al. (2001) also reported a
GA/AA-AUC ratio of 0.18 for Sprague-Dawley rats treated with 20 mg/kg AA by gavage. This
contrasts to Twaddle et al.'s (2004) observation of equal AUCs for AA and GA in B6C3Fi mice.
Since rats and mice had a comparable GA elimination half-life, this approximately fivefold
difference in internal exposure to GA for mice compared with rats (i.e., a GA/AA-AUC ratio of
1 in mice vs. a GA/AA-AUC ratio of 0.18 in rats) is considered to be the result of an increased
rate of GA formation in the mouse.
Formation of DNA adducts
Doerge et al. (2005a) measured DNA adducts following a single i.p. administration of
AA to adult B6C3Fi mice and F344 rats at 50 mg AA/kg, or an equimolar dose of GA
(61 mg/kg). They report GA-derived DNA adducts of adenine and guanine formed in all tissues
examined, including both target tissues identified in rodent carcinogenicity bioassays and
nontarget tissues, including liver, brain, thyroid, leukocytes, mammary gland, and testis in rats,
and in liver, lung, kidney, leukocytes, and testis in mice. Dosing rats and mice with an
equimolar amount of GA typically produced higher levels of DNA adducts than those observed
from the AA dose.
Doerge et al. (2005a) also measured DNA adduct formation following oral administration
of a single dose of AA (50 mg/kg), and accumulation from repeat dosing at 1 mg/kg-day. The
formation of DNA adducts was consistent with the previously reported mutagenicity of AA and
GA in vitro, which involved reaction of GA with adenine and guanine bases. These results
provide support for a mutagenic mechanism of AA carcinogenicity in rodents.
32 DRAFT-DO NOT CITE OR QUOTE
-------
AA and GA react with nucleophilic sites in macromolecules (including hemoglobin and
DNA [Figure 3-2]) in Michael-type additions (Segerback et al., 1995; Bergmark et al., 1993,
1991; Solomon et al., 1985). Solomon et al. (1985) conducted in vitro studies for the reaction of
AA at pH 7.0 and 37°C for 10 and 40 days with 2'-deoxyadenosine (dAdo), 2'-deoxycytidine
(dCyd), 2'-deoxyguanosine (dGua), and 2'-deoxythymidine (dThd), which resulted in the
formation of 2-formamidoethyl and 2-carboxyethyl adducts via Michael addition. However, AA
reacted extremely weakly with DNA (second order rate constant of 9 x 10~12 L/mg DNA-hour at
pH 7 and 37°C for all adducts), even under in vitro conditions, producing significant levels of
adducts only after incubations of several weeks with high acrylamide concentrations (Solomon et
al., 1985). Based on the second order rate constant derived by Solomon et al. (1985), Segerback
et al. (1995) estimated formation of 25 fmol/mg DNA for all adducts from an in vivo i.p. AA
dose of 50 mg/kg. Only about 14% of these would be adducts to the N-7 atom of guanine. This
amount was considered to be negligible compared with observed levels of N-7-(2-carbamoyl-
2-hydroxyethyl)guanine adducts with GA, which were in the 20-30 pmol/mg DNA range for the
in liver of both mice and rats from a comparable (46-53 mg/kg) i.p. dose (Segerback et al.,
1995). Two additional GA-DNA adducts have been identified in vitro, N3-(2-carbamoyl-
2-hydroxyethyl)adenine (N3-GA-Ade) and Nl-(2-carboxy-2-hydroxyethyl)-2'-deoxyadenosine
(Gamboa da Costa et al., 2003). Using liquid chromatography with tandem mass spectrometry
and isotope dilution, Gamboa da Costa et al. (2003) measured DNA adduct formation in selected
tissues of adult and whole body DNA of 3-day-old neonatal mice treated with AA and GA. In
adult mice, DNA adduct formation was observed in liver, lung, and kidney with levels of
N7-GA-Gua around 2,000 adducts/108 nucleotides and N3-GA-Ade around 20 adducts/108
nucleotides. Adduct levels were modestly higher in adult mice dosed with GA as opposed to
AA; however, treatment of neonatal mice with GA produced five- to sevenfold higher whole
body DNA adduct levels than with AA. The authors suggest that this is due to lower oxidative
enzyme activity in newborn mice. DNA adduct formation from AA treatment in adult mice
showed a supralinear dose-response relationship, consistent with saturation of oxidative
metabolism at higher doses.
33 DRAFT-DO NOT CITE OR QUOTE
-------
o
II
H9C=C—C—NhL
2 H 2
acrylamide
Hb-Val-N-CH2-CH2-CONH2
or
Hb-Cys-S-CH2CH2CONH2
Acylamide Hb Adducts
O
•Hb-Val-NH2
HB-Cys-SH
\
HC - C— CONH
H
glycidamide
GSH
Hb-Val-N-CH2-CH-CONH2
OH
Hb-Cys-S-CH2-CH-CONH2
OH
Glycidamide Hb Adducts
r
DNA-guanine-N7-CH2-CH-CONH2
Glycidamide DNAAdduct
Sources: Dearfield et al. (1995); Bergmark et al. (1993, 1991).
Figure 3-2. Hemoglobin and DNA adducts of acrylamide and glycidamide.
Potential confounders for the hemoglobin adduct biomarker of acrylamide exposure
Other related compounds like acrylonitrile and N-methylolacrylamide (NMA) also form
hemoglobin adducts. NMA is produced by the reaction of formaldehyde with AA and, like AA,
is used in the production of grouting agents. Acrylonitrile can be used as a precursor in one
method to manufacture AA, and is also formed when AA is dehydrogenated.
Studies that use AA hemoglobin adducts as a biomarker for exposure should address the
potential presence of NMA. Acrylonitrile forms an N-(2-cyanoethyl)valine adduct that is
distinguishable from the AA N-(2-carbamoylethyl)valine adduct with gas chromatography/mass
spectrometry (GC-MS) analysis after derivatization with pentafluorophenyl isothiocyanate
(Bergmark et al., 1993). N-methylolacrylamide, however, forms the same adduct as AA, the
N-(2-carbamoylethyl)valine adduct. It is not known whether NMA undergoes loss of the
hydroxymethyl group to form AA, which can then react with globin to form AAVal, or if NMA
reacts directly with globin and then loses the hydroxymethyl group to form AAVal. Both
reactions, involving loss of formaldehyde, could occur on a chemical basis without the
involvement of metabolism (Fennell et al., 2003). There are also differing results on the relative
rate of formation of the N-(2-carbamoylethyl) valine adduct from AA or NMA (Paulsson et al.,
2002; Fennell et al., 2003).
Paulsson et al. (2002) measured hemoglobin adducts (and micronucleus [MN]
frequencies) in mice and rats after AA or NMA treatment. Male CBA mice were treated by i.p.
injection of 0.35, 0.7, and 1.4 mmol/kg for both compounds (i.e., 25, 50, and 100 mg AA/kg, or
35, 71, and 142 mg NMA/kg). The rats were only treated with the highest dose of AA or NMA,
100 mg/kg or 142 mg/kg, respectively. Mice were sacrificed after 48 hours and blood was
collected for hemoglobin adduct measurement. One group of rats was sacrificed after 24 hours
34
DRAFT-DO NOT CITE OR QUOTE
-------
and one group after 48 hours for the hemoglobin adduct analysis. The identical
(N-[2-carbamoylethyl]valine) adduct and the respective epoxide metabolite (N-[2-carbamoyl-
2-hydroxyethyl]valine) adduct were monitored for either the AA or NMA exposure. Per unit of
administered amount, AA gave rise to three to six times higher hemoglobin adduct levels than
NMA in mice and rats. Mice exhibited higher in vivo doses of the epoxy metabolites, compared
with rats, indicating that AA and NMA were more efficiently metabolized in the mice. In mice
the AA and NMA induced dose-dependent increases in both hemoglobin adduct level and MN
frequency in peripheral erythrocytes. Per unit of administered dose, NMA showed only half the
potency for inducing micronuclei compared with AA, although the MN frequency per unit of in
vivo dose of measured epoxy metabolite was three times higher for NMA than for AA. No
increase in MN frequency was observed in rat bone marrow erythrocytes after treatment with
either compound. This is compatible with a lower sensitivity of the rat than of the mouse to the
carcinogenic action of these compounds.
Fennell et al. (2003) also measured levels of N-(2-carbamoylethyl)valine adducts
following gavage exposure of male F344 rats (4/group) to equimolar levels of AA or NMA. The
nominal dose of [1,2,3-13C]-AA was 50 mg/kg, and NMA was administered at a nominal dose of
71 mg/kg. The AA and NMA dose solutions were prepared in distilled water and delivered at
1 mL/kg. In contrast to Paulsson et al. (2002), Fennell et al. (2003) reported that AA exposure
resulted in the formation of 21 ± 1.7 pmol/mg globin (mean ± SD), less than the equimolar dose
of NMA that resulted in 41 ± 4.9 pmol/mg. Since rates of formation of the N-terminal valine
adduct are not comparable (regardless of whether more or less) and both compounds form the
same adduct, caution should be exercised when drawing conclusions about AA exposure based
on N-terminal valine levels if there is also a potential for concurrent exposure to NMA.
3.4. ELIMINATION
Human data
Boettcher et al. (2005) measured the mercapturic acid of AA and its epoxide GA, i.e.,
N-acetyl-S-(2-carbamoylethyl)-L-cysteine (AAMA) and N-(R,S)-acetyl-S-(2-carbamoyl-2-
hydroxyethyl)-L-cysteine (GAMA) in human urine. Median levels in smokers (n = 13) were
found to be about four times higher than in nonsmokers (n = 16) with median levels of 127 |ig/L
versus 29 |ig/L for AAMA and |ig/L versus |ig/L for GAMA indicating that cigarette smoke is
clearly an important source of AA exposure. The level of AAMA in the occupationally
nonexposed collective (n = 29) ranged from 3 to 338 |ig/L, the level of GAMA from
-------
24 hours of administration. A dermal exposure to humans was part of this study but no
elimination data from the dermal exposure were reported.
Boettcher et al. (2006a) investigated the human metabolism of AA to AAMA and GAMA
in a healthy male volunteer who received a single dose of about 1 mg deuterium-labelled AA
(d(3)-AA), representing 13 |ig/kg body weight, in drinking water. Urine samples before dosing
and within 46 hours after the dose were analyzed for d(3)-AAMA and d(3)-GAMA by LC-ESI-
MS/MS. A first phase of increase in urinary concentration was found to last 18 hours with a
broad plateau between 8 and 18 hours for AAMA, and 22 hours for GAMA. Elimination half-
lives of both AAMA and GAMA were estimated to be approximately 3.5 hours for the first
phase and more than 10 hours up to few days for the second phase. Total recovery in urine after
24 hours was about 51% as the sum of AAMA and GAMA and was similar to recoveries in rats
(53-66%) given a gavage dose of 0.1 mg/kg bw (Doerge et al., 2007). After 2 days AAMA
accounted for 52% of the total AA dose, and was the major metabolite of AA in humans.
GAMA accounted for 5%, and appeared as a minor metabolite of AA.
Fuhr et al. (2006) measured AA and metabolite levels in a 72 hour urine collection from
six young healthy volunteers after the consumption of a meal containing 0.94 mg of aery 1 amide.
Overall, 60.3 ± 11.2% of the dose was recovered in the urine. Although no GA was found,
unchanged AA, AAMA, and GAMA accounted for urinary excretion of (mean ± SD) 4.4 ±1.5,
50.0 ± 9.4, and 5.9 ± 1.2% of the dose, respectively. Toxicokinetic variables were obtained by
noncompartmental methods, with apparent terminal elimination half-lives for the unchanged AA,
AAMA, and GAMA of 2.4 ± 0.4, 17.4 ± 3.9, and 25.1 ± 6.4 hours, respectively.
Boettcher et al. (2006b) evaluated urinary mercapturic acid metabolites derived from AA
in three healthy volunteers who fasted for 48 hours. Urinary AA mercapturic acid metabolites
were considerably reduced after 48 hours of fasting, with levels well below the median level in
nonsmokers. These results indicate that, for nonsmokers, AA in the diet is the main source of
environmental AA exposure in humans.
Hartmann et al. (2009) determined the relationship between the oxidative and reductive
metabolic pathways of acrylamide (AA) in the nonsmoking general population, measuring both
blood protein adducts and the urinary metabolites of AA and glycidamide (GA) in an especially
designed study group with even distribution of age and gender. The hemoglobin adducts N-
carbamoylethylvaline (AAVal) andN-(R,S)-2-hydroxy-2-carbamoylethylvaline (GAVal) were
detected by GC-MS/MS in all blood samples with median levels of 30 and 34 pmol/g of globin,
respectively. Concentrations ranged from 15 to 71 pmol/g of globin for AAVal and from 14 to
66 pmol/g of globin for GAVal. The ratio GAVal/AAVal was 0.4-2.7 (median =1.1). The
urinary metabolites were determined by LC-MS/MS. Of all urine samples examined 99% of N-
acetyl-S-(2-carbamoylethyl)-Lcysteine (AAMA) levels and 73% of N-(R/S)-acetyl-S-(2-
carbamoyl-2-hydroxyethyl)-L-cysteine (GAMA) levels were above the LOD (1.5 ug/L).
36 DRAFT-DO NOT CITE OR QUOTE
-------
Concentrations ranged from
-------
Table 3-2. Comparison of molar percentage of dose excreted in urine of rodents and humans after oral administration
of acrylamide (Source: Hartmann et al., 2009)
Species - dose
M - 50 nig/kg"
M - 0.1 mg/kg*
R - 50 mg/kgd
R - 50 mg/kgf
R - 3 mg/kgs
R — 0,1 rug/kg'
R-20ng/kgh
R - 0.1 mg/kgh
H - 3.0 mg/kg*
H-13ng/Kg'
H - 0.5 mg/kgJ
H - 1.0 mg/kg1
H - 3.0 mg/kgj
H - 0.5 tig/kg"
H - 20 re/kg"
AA
NO.
0.5-0.7
NQ
NQ
NQ
2
ND
ND
NQ
ND
4.57 ±1.34
5.02 + 1.65
3J23+0.49
ND
ND
MMA
21.011.10
5-9
34.0 ±1.80
38
29.0 ±4. 50
31
29.7 ±5.13
34.9 + 7.40
22.0 + 5.30
45.1
31.2±6.55
34.4 + 5.21
27.8+7.99
41.4+3.47
37,4+252
AAMA-sulfoxide
ND
ND
ND
ND
ND
ND
ND
ND
420 + 1.10
ND
826±2.39
8.68 + 1.21
725+2.40
7.19 + 1.40
6.33 + 1.77
CA
8.6 + 1.1
16-18
2.8+0.50
3.9
ND
6
ND
ND
0.79+0.24
ND
0.43 ±0.20
0.63+0.33
0.65+0.21
ND
ND
CAMA
17+0.50
9-22
12+0.60
10.5
21 +242
27-29
25.4±6.20
26.7±4.64
ND
2.8
0.82±0.16
0.82+0.11
0.70+0.22
3.83+0.78
3.23+0.69
Gtyceramide
2.70 + 0.60
ND
1.20 + 0.40
O.G
ND
ND
ND
ND
3.30+1.10
ND
ND
ND
ND
ND
ND
£GAJ ,>TAAb
1.3
42
0.47
0.39
0.72
1-1.1
056
0.77
0.16
0,06
0.03
0.03
0.03
0.08
0.07
Total of dose£
50.4
33-48
50.7
53
50.0+ 8.60
64-66
55.1+11.8
61.7 + 10.5
34.0+ 5.70
47.7
45.6±8.50
49.9 + 6.30
39.9+9.90
52.4+3.59
46.9+ 3.70
All information given is referenced to collection periods of 24 1; after administration.
NQ-not quantified. ND-not determined.
a This sum represents GA+GAMA+Clyceramide.
b This sum represents AA+AAMA+AAMA-sulfoxide.
' Total amount excreted within 24 h after exposure calculated as % of dose.
d Sumner et al.. 1992. Gavage male rats; gavagc male mice.
' Doerge ct al., 2007. Cavage male mice: gavagc male rats.
1 Sumner et al., 2003. Cavage male rats.
g Fenncll et al.. 2005. Cavage male rats; oral administration male humans.
h This paper. Cavage male rats.
Boettchcr ct aL. 2005. Oral administration male human. Excretion within 22 h following exposure.
Fennell et al., 2006, Oral administration male humans.
k Hartmann et al., 2009. Oral administration male and female humans. Excretion within 11 h following exposure.
1
Source: Hartmann et al., 2009
38
DRAFT-DO NOT CITE OR QUOTE
-------
Animal data
Results from animal studies indicate that urinary excretion of metabolites is the principal
route of elimination of absorbed AA, with minor amounts of metabolites being excreted via bile
in the feces, and as CC>2 in exhaled breath (Barber et al., 2001; Kadry et al., 1999; Sumner et al.,
1999, 1992; Dow Chemical Co., 1984; Miller et al., 1982; Hashimoto and Aldridge, 1970).
Fennell et al. (2005) administered 3 mg/kg [1,2,3-13C]-AA by gavage to F344 rats. The
low 3 mg/kg dose of AA by gavage to rats resulted in a greater amount of metabolism via GA
(41% of the urinary metabolites) compared with a higher dose of 59 mg/kg (28% of the urinary
metabolites) (Sumner et al., 2003). The fate of GA was primarily conjugation with GSH,
resulting in the excretion of two mercapturic acids. The total amount of AA metabolites
recovered by 24 hours after dosing was 50%, similar to that reported by Kadry et al. (1999) and
Miller et al. (1982).
In male F344 rats given i.v. (10 mg/kg) or oral (1, 10, or 100 mg/kg) doses of
[2,3-14C]-acrylamide, about 60 and 70% of the administered radioactivity was excreted in urine
collected within 24 hours and 7 days, respectively (Miller et al., 1982). Less than 2% of
radioactivity in the urine was accounted for by AA. With either route of administration,
elimination of radioactivity from tissues was described as biphasic, with half-lives of about
5 hours for the first phase and 8 days for the second phase. The elimination time course of
parent AA from tissues followed a single-phase exponential decrease with a half-life of about 2
hours. Calleman (1996) noted that this is a relatively slow elimination half-life for an
electrophilic chemical, citing the elimination half-life of acrylonitrile, a related electrophilic
chemical, at about 10 minutes in rats. Fecal excretion accounted for 4.8 and 6% of administered
radioactivity at 24 hours and 7 days, respectively (Miller et al., 1982). Bile-duct-cannulated rats
given single i.v. doses of 10 mg/kg [2,3-14C]-labeled AA excreted about 15% of the administered
radioactivity in bile as metabolites within about 6 hours; less than 1% of radioactivity in the bile
was in the form of AA. These results are consistent with the existence of enterohepatic
circulation of metabolites.
No radiolabeled CC>2 was captured when two rats given [2,3-14C]-labeled AA were
placed in metabolism cages designed to trap expired air (Miller et al., 1982). In contrast, studies
with radiolabel in the carbon-1 position suggest that exhalation of CC>2 following cleavage of the
amide group is possible but likely represents a minor metabolic and elimination pathway (see
Figures 2-1 and 3-1 for carbon numbering and metabolic pathways, respectively). About 6% of
an injected dose of 100 mg/kg [l-14C]-labeled AA (Hashimoto and Aldridge, 1970) and about
4% of an injected dose of 2 mg/kg [l,3-14C]-labeled AA (Dow Chemical Co., 1984) were
39 DRAFT-DO NOT CITE OR QUOTE
-------
exhaled by rats as CC>2 in 6-8 hours. As noted earlier, however, no exhaled 14CC>2 was detected
following oral administration of 50 mg/kg [l-14C]-labeled AA to male Sprague-Dawley rats
(Kadryetal., 1999).
In studies with male F344 rats given single i.v. doses of [l,3-14C]-labeled AA,
percentages of the administered dose recovered in excreta, carcass, and cage wash after 72 hours
were as follows for four rats exposed to 2 mg/kg: 67% urine; 1.5% feces; 4.2% CQ^ 1.5% skin;
13.1% carcass; and 0.6% cage wash (Dow Chemical Co., 1984). Similar percentages were
reported for four rats injected with 100 mg/kg. Other groups of rats were given single i.v.
injections of 50 mg/kg [l,3-14C]-labeled AA and were killed in groups of 3-4 after 0, 6, 12, 18,
24, or 48 hours for determination of radioactivity in blood plasma, red blood cells, and selected
tissues (testes, epididymis, kidney, and sciatic nerve). The clearance of radioactivity from the
plasma and the tissues was consistent with biphasic elimination with an initial rapid phase,
followed by a slower phase. Plasma elimination half-times were estimated at 2 hours for the
initial phase and 10 hours for the second slower phase. GC/MS analysis indicated that the initial
phase was primarily due to clearance of AA, whereas the second phase was due to clearance of
radiolabeled metabolites from the plasma.
Tong et al. (2004) estimated the second order rate constants for reaction of AA with
human serum albumin and glutathione at 0.0054 and 0.021/mol-second, respectively. These
rates were determined under physiological conditions by following the loss of their thiol groups
in the presence of excess AA. Based on these in vitro values, the authors concluded that the
reactions of AA with these thiols appears to account for most of AA's elimination from the body.
More recently, Doerge et al. (2007) measured 24 hour urinary metabolites, including free
AA and GA and their mercapturic acid conjugates (AAMA and GAMA, respectively), using
LC/MS/MS in F344 rats and B6C3F(1) mice following a dose of 0.1 mg/kg bw given by
intravenous, gavage, and dietary routes of administration. The results were compared with
serum/tissue toxicokinetic and adduct data (DNA and hemoglobin) from previous studies in the
same laboratory using the identical dosing protocols (Doerge et al., 2005a,b, c). The goal was to
investigate relationships between urinary and circulating biomarkers of exposure, toxicokinetic
parameters for AA and GA, and tissue GA-DNA adducts in rodents from single doses of AA.
The molar percentage of the total intravenously delivered dose that was recovered as free AA
and metabolites in a 24 hour urine collection was 57-74 and 54-57% in male and female rats,
respectively; and 62-82 and 60-63% in male and female mice, respectively. Significant linear
correlations were observed between urinary levels of AA with AAMA and GA with GAMA in
the current data sets for rats (AA vs. AAMA, r2 = 0.78,/? < 0.001; GA vs. GAMA, r2 = 0.81,
p < 0.001) and mice (AA vs. AAMA, r2 = 0.86,/? < 0.001; GA vs. GAMA, r2 = 0.57,p < 0.001).
40 DRAFT-DO NOT CITE OR QUOTE
-------
Concentrations of urinary AA or AAMA correlated significantly with average AUC values for
serum AA determined previously in groups of rats (AUC-AA vs. AA, r2 = 0.74,/> < 0.001;
AUC-AA vs. AAMA, r2 = 0.83,^ < 0.001) and mice (AUC-AA vs. AA, r2 = OAl,p< 0.011;
AUC-AA vs. AAMA, r2 = 0.41,/> < 0.01) similarly dosed with AA. Correlation coefficients for
urinary GA and GAMA concentrations versus AUC serum GA and liver GA-DNA adducts were
smaller that for the AA and AAMA, but still significant in rats (AUC-GA vs. GA, r2 = 0.53,
p < 0.001; AUC-GA vs. GAMA, r2 = 0.32, p < 0.02) and mice (AUC-GA vs. GA, r2 = 0.34,
p < 0.022; AUC-GA vs. GAMA, r2 = 0.56, p < 0.0001). Significant linear correlations were also
observed in rats between urinary concentrations of either GA or GAMA with average GA-DNA
adducts (p = 0.001 and 0.2, respectively); data not presented in the publication. In mice, a
significant linear correlation was observed between urinary concentrations of GA (p = 0.03), but
not GAMA (p = 0.2), with average GA-DNA adducts; data not presented in the publication. In
both rats and mice, significant linear correlations were observed between AA or AAMA and
average GA-DNA adduct levels (p = 0.0005 and 0.004, respectively); data not presented in the
publication. Although considerable interindividual variability observed in all urinary
measurements weakened the correlation with either average toxicokinetic or biomarker data
collected from different groups of animals, overall the results indicate that urinary biomarkers do
reflect internal levels of AA and GA, and may be useful (accompanied by appropriate caveats) in
estimating levels of exposure and potential risk for adverse effects.
3.5. HEMOGLOBIN ADDUCTS AND URINARY METABOLITES AS BIOMARKERS
OF EXPOSURE
Hays and Alyward (2008a) report the results of a seminal workshop that conceptualized
and developed methods to generate what were called "Biomonitoring Equivalents" (BEs), which
are estimates of the concentration of a chemical or metabolite in a biological medium that is
consistent with an existing exposure guidance value such as a tolerable daily intake or a
reference dose. BE's address the need for a context to interpret data that are increasingly
becoming available on trace concentrations of chemicals in human biological media. Case
studies for four chemicals (toluene, 2,4-dichlorophenoxyacetic acid, cadmium and acrylamide)
were published to demonstrate the derivation of BEs for various kinds of data. The case study
for acrylamide (Hays and Alyward 2008b) clearly shows the utility of acrylamide hemoglobin
adduct and urinary metabolite concentrations as biomarkers of exposure, as well as the methods
to estimate daily intake levels based upon those concentrations. A table by Hays and Alyward
41 DRAFT-DO NOT CITE OR QUOTE
-------
(2008b), and reproduced below as Table 3-3, summarizes the advantages and disadvantages of
the various biomarkers of exposure for acrylamide. This current toxicological review of
acrylamide utilizes the direct relationship between AA and GA hemoglobin adducts and serum
levels of AA and GA to estimate the dose-response relationship in humans based on the observed
dose-response relationship from animal studies.
Table 3-3. The advantages and disadvantages of available biomarkers of exposure for
acrylamide (from Hays and Alyward 2008b).
Analjrte
AA
GA
AA— as hemoglobin adduct
GA— as hemoglobin adduct
AAMA(mercapturic add metabolite of
AA)
GAMA (mercapturic acid conjugate of
GA) and other urinary metabolites
Medium
Serum
Serum
Blood
Blood
Urine
Urine
Advantages
Relevant to etTect(s) of interest
Relevant to effect(s) of interest
Longer half-life; Relevant to effect(s) of interest
Longer half-life; Relevant to effects) of interest
Non-invasive; slower half-life leads to more stable
profile in urine under chronic exposure conditions
Non-invasive; slower half-life leads to more stable
profile in urine under chronic exposure conditions
Disadvantages
Short half-life: Invasive [requires blood sample)
Short half-life; Invasive (requires blood sample)
Invasive (requires blood sample)
Invasive (requires blood sample)
Not directly related to critical target tissue dose(s);
measure of metabolic deactivation
Not directly related to critical target tissue dose(s);
lower proportion of administered dose than AAMA
Hemoglobin adducts were first proposed as biomarkers of exposure to acrylamide by
WHO (1985), and the initial analytical techniques were developed by Bailey et al. (1986). Early
studies in people who were occupationally exposed or who smoked tobacco evaluated the
relationship between AA and GA hemoglobin adducts and exposure (Bergmark, 1997; Calleman
et al., 1994; Bergmark et al., 1993). AA was reported to form the N-(2-carbamoylethyl) valine
and GA the N-(2-carbamoyl-2-hydroxyethyl)valine and the N-(l-carbamoyl-2- hydroxyethyl)-
valine (Bergmark, 1997; Bergmark et al., 1993; Calleman et al., 1994). The detection of GA
adducts of hemoglobin in AA-exposed workers demonstrated the transformation of acrylamide
to GA in humans (Bergmark et al., 1993). Other related compounds like acrylonitrile and N-
methylolacrylamide (NMA) also form hemoglobin adducts, so these potential confounders
should always be considered in studies that use AA hemoglobin adducts as the basis for
estimating exposure to acrylamide.
Use of Measured Hemoglobin Adducts to Estimate Administered Dose or Serum A UC
The equations used to estimate the area under the time-concentration curve (AUC) of AA
or GA in serum based upon measured hemoglobin adduct levels are straightforward. For a
single dose over a short time frame (i.e., no need to adjust for accumulation or steady state levels
of adducts from multiple doses) the serum AUC is calculated as:
Serum AUC = Hb-adduct concentration / second order rate constant
42 DRAFT-DO NOT CITE OR QUOTE
-------
If the adduct levels are normalized to the administered dose then one can derive an
estimate of the AUC per dose, as exemplified in the following data and discussion from Fennell
etal. 2005.
Fennell et al. (2005) measured the amount of hemoglobin adducts from AA and GA
following administration of a defined dose of AA to adult male volunteers. Both AAVal and
GAVal increased linearly with increasing dose of AA administered orally, suggesting that, over
the range of 0.5-3.0 mg/kg, there is no saturation of metabolism of AA to GA. The ratio of
GAVal: AAVal produced by administration of AA was similar to the ratio of the background
adducts prior to exposure. Compared with the equivalent oral administration in rats (3 mg/kg),
the ratio of [13C]-GAVal: [13C]-AAVal in humans was lower (0.44 ± 0.06) than in rats (0.84 ±
0.07), and the absolute amount (i.e., not scaled to body weight) of [13C]-AAVal formed in
humans was approximately 2.7-fold higher than in the rat. The absolute amount of [13C]-GAVal
was approximately 1.4-fold higher than that formed in the rat.
Table 3-4 shows the data used by Fennell et al. (2005) to estimate the internal serum
concentrations of acrylamide based on adduct levels and second order rate constants that they
measured in vitro by adding AA or GA to extracted human hemoglobin.
Table 3-4. Estimated human serum AA AUC normalized to administered dose based on
measured Hb adduct levels and in vitro derived second order rate constants (from
Fennell et al., 2005)
Hemoglobin adducts/actual Human AA AUC per
administered dose administered dose3
Human Nominal
Dose (route) (mg/kg)
Human Dose
0.5 (oral)
1.0 (oral)
3 (oral)
Combined (oral)
Actual Concentration of 13C3-AAVal
Dose (uM normalized to actual dose
AA/kg bw) (nM/g globin/mM AA/kg)
5.9 ±0.2 86.4±7.5b
12.5 ±0.2 73.4±9.8b
38.7 ±0.5 64.2±17.7b
74.7±14.9b
(13C mM AA/mM-hr
AA/kg bw)
20.2
17.2
15.0
17.5
a Hemoglobin adduct levels divided by second order formation rate constant of 4.27E-06 l/g
globin/hr
The above human serum AA AUC of 17.5 mMoles-hr AA /mMoles of AA /kg bw is
converted to 246 uM-hr of AA per mg AA/kg bw using the following unit conversions (i.e., the
units in Fennel et al. (2005) are divided by the molecular weight of acrylamide (71.08), and
multiplied by 1000 to convert mMoles to uMoles):
43
DRAFT-DO NOT CITE OR QUOTE
-------
U.SmM-hrAA ImM AA 0.227mM-hrAA innn 246uM-hrAA
x = xlOOO =
mMAAIkgbw Jl.QSmgAA mgAAIkgbw mgAAIkgbw
where: the 13C3-AA-Val adduct concentration normalized to administered dose of
74.7 nMoles/g globin/mMoles AM/kg bw is converted to 17.5 mMoles-hr AA /mMoles of AA
/kg bw based on a second order rate constant for AA-Val from in vitro studies of
4.27 x KT6L/gHb/hour.
For a known dosing period Tornqvist et al. (2008) used the following equation
(developed in previous work, Granath et al., 1992) to adjust the total adduct concentration to an
estimate of the daily increase in adduct level.
AA ValDAILY = AA ValTOTAL — — - —
Days oj Treatment
To estimate steady state level of adducts or to estimate the AUC based on an assumed
steady state adduct level, Bergmark et al. (1991) used the following equations:
= AAAUCxkAAx-RBC
GAVal = GAAUC x kOA
2
t
RBC
In the above equations, AAVal is the steady state level of AA-hemoglobin adducts
(umol/g globin), AAAuc is the daily serum AA Auc ( nM h/d), kAA is the rate constant for the
reaction of AA with the N-terminal valine residue of hemoglobin, and IRBC is the life span of a
human red blood cell with often used values ranging from 120 days (Osterman-Golkar et si.
1976) to 126 days in humans (Hartmann et al., 2009).
Fennell et al. (2005) calculated the expected amount of adduct that would accumulate in
adult male humans from continuous exposure based on the amount of adduct formed/day of
exposure, and from the life span of the erythrocyte. Exposure via oral intake to 1 |ig/kg AA
(1.05 fmol AAVal/mg globin/day) for the life span of the erythrocyte (120 days) was estimated
to result in the accumulation of adducts to 63 fmol/mg globin. Daily dermal exposure to 1 |ig/kg
AA (0.18 fmol AAVal/mg globin/day) for the life span of the erythrocyte (120 days) would
44 DRAFT-DO NOT CITE OR QUOTE
-------
result in the accumulation of adducts to 10.8 fmol AAVal/mg globin. With workplace exposure
of 5 days/week, this would decrease to approximately 7.8 fmol AAVal/mg globin.
Hartmann et al (2009) used the following equations to estimate the daily intake of
acrylamide based on measured hemoglobin adduct levels (Schettgen et al., 2003; Calleman et al.,
1996). These equations adjust an assumed steady state adduct level to estimate the internal
serum AUC based, and then apply values for an elimination rate and the volume of distribution
to convert an AUC to an estimated daily intake.
AA[ug/kgbw/day]= g globin} xMW acrylamdexVD
k x erthrocyte lifespan x 1 / 2
where: k is the human Hb adduct formation rate constant for AAVal (4.4 x 10-6 L/ g of globin/
h; Bergmark et al., 1993), the value for the middle erythrocyte lifespan is 63 days, the
elimination rate constant Ek in humans is 0.15 h"1 (Calleman et al., 1996) and the volume of
distribution VD is 0.38 L/kg (Fennell et al., 2005).
An important caveat about these estimates of administered dose or AUC level based on
measured adduct levels is the direct dependence of the results on the value of the second order
rate constant. Currently, there are no available formation rate constants derived from "in vivo"
human data, which would require human studies where data were collected for all three of the
critical variables needed to derive an in vivo human adduct formation rate, namely 1) the
administered dose, 2) the time course serum levels, and 3) the time course adduct levels
(including sufficient post dosing sample times to determine elimination rates). Kopp and Dekant
(2008) only measured human serum data and administered dose, and Fennell et al. (2005) only
measured hemoglobin adduct levels and administered dose. Thus, current derivations of human
serum AUCs or daily intakes that have been reported in the published literature are based on
hemoglobin adduct levels and second order rate constants that were derived from in vitro studies
where AA or GA are added to extracted human hemoglobin (Fennell et al., 2005; Bergmark et
al;., 1993, Tareke et al., 2006). Second order rate constants have also been estimated from in
vitro studies using rat hemoglobin (Tareke et al., 2006; Tornqvist et al. 2009, Fennell et al.,
2005; Bergmark et al;., 1993). Recently, however, rat and mice in vivo data have been
published by Doerge et al (2005 a, b, c) and Tareke et al. (2006) sufficient to allow the
generation of in vivo animal adduct formation rates. EPA used these data, which include in vivo
time course serum data from single doses of acrylamide with different routes of administration
45 DRAFT-DO NOT CITE OR QUOTE
-------
(Doerge et al., 2005c) and the corresponding hemoglobin adduct levels (Tareke et al., 2006) to
derive second order rate constants for AA and GA hemoglobin adducts in rats. Table 3-5 below
is a compilation of these various rat and human adduct formation rate constants. The in vivo rat
adduct formation rates were then used to estimate rat internal AUCs for AA and GA based upon
the time course hemoglobin adduct levels reported by Tareke for rats given acrylamide in
drinking water for 42 days (Tareke et al., 2006; Doerge et al., 2005a). Table 3-6 and Table 3-7
presents these estimated serum acrylamide and glycidamide AUCs normalized to the
administered dose of acrylamide. The values in these tables are used to derive the reference
values in this assessment, and are discussed in greater detail in Section 5.
46 DRAFT-DO NOT CITE OR QUOTE
-------
Table 3-5. Second order rate constants for reaction of acrylamide or glycidamide with the
N-terminal valine residue of hemoglobin.
Second Order Rate Constant for Formation of
Hemoglobin Adducts
(l/g globin/h) x 106
Source
Male
Rat
Female
Rat
Average
or pooled
Male and
Female
Rat
Gender
Not
Specified
Rat
Pooled
Rat and
Mouse Human
AA -Val In Vivo Adduct Formation Rate3
Based on all rat and mice Tareke et al. (2006)
adduct data and measured serum AUCs in
Doerge et al. (2005 b, c) single dose studies.
Based on gender specific rat Tareke et al.
(2006) adduct data and measured serum
AUCs in Doerge et al. (2005c) single dose
studies
Based on all rat Tareke et al. (2006) adduct
data and measured serum AUCs in Doerge et
al. (2005c) single dose studies.
AA -Val In Vitro Rate Adduct Formation
Rate
As reported by Fennell et al. (2005)
As reported by Bergmark et al. (1993)
As reported by Tareke et al. (2006)
As reported by Tb'rnqvist et al. (2008)
7.5
8.9 5.9
7.4
7.5
3.82
2.9
4.6
4.27
4.4
7.4
GA -Val In Vivo Adduct Formation Rate3
Based on all rat and mice Tareke et al. (2006)
adduct data and measured serum AUCs in
Doerge et al. (2005 b, c) single dose studies.
Based on gender specific rat Tareke et al.
(2006) adduct data and measured serum
AUCs in Doerge et al. (2005c) single dose
studies
Based on all rat Tareke et al. (2006) adduct
data and measured serum AUCs in Doerge et
al. (2005c) single dose studies.
GA -Val In Vitro Rate Adduct Formation
Rate
As reported by Fennell et al. (2005)
As reported by Bergmark et al. (1993)b
As reported by Tareke et al. (2006)
As reported by Tb'rnqvist et al. (2008)
32.5
35.3
20.0
27.6
34.0
4.96
12.0
9.5
13.6
6.72
11.0
59.0
a See Appendix E for a complete description of the derivation of the in vivo adduct formation rates.
b Note: Bergmark derived the rat GA-Val residue such that kval = (GA-Val *kcys)/GA cys; the human GA-Val adduct was measured
directly.
47
DRAFT-DO NOT CITE OR QUOTE
-------
Table 3-6. Measured and estimated AA AUCs normalized to dose in humans and F344
rats.
AA ADC in uM-hr per mg AA/kg bw
Average of
Male and Unspecifie
Male Female Female d Gender
Rat Rat Rat Rat Human
AA in Humans
Measured
Kopp and Dekant 2009 - human serum data AA
(single dose of 20 ug/kg, n=3F,3M) 2.83
Estimated using human adduct data and test
animal in vivo rate constants
Fennell et al 2005 - human adduct data and in
vivo rate constants derived from Tareke et al.
(2006) adduct data for all rat and mice in Doerge
et al. (2005 b, c) single dose AUCs. 140.1
Estimated using human adduct data and
human in vitro rate constants
Fennell et al 2005 - human adduct data and
Fennell in vitro rate constants 246.0
Fennell et al 2005 - human adduct data and
Bergmark et al. 1993 in vitro rate constants 238.8
Estimated using human adduct data and rat in
vitro rate constants
Fennell et al 2005 - human adduct data and
Tb'rnqvist et al. 2008 in vitro rate constants 228.5
AAin F344Rats
Measured
Doerge et al. 2005 c - time course data from a
single dietary exposure 18.0 15.0 16.5
Doerge et al. 2005 c - time course data from a
single gavage exposure 24.0 45.0 34.5
Estimated using rat adduct data and rat in vivo
rate constants
Tareke et al (2006) adduct data for the Doerge et
al. (2005a) 42 day drinking water study, and
gender specific in vivo derived rate constants from
Tareke et al. (2006) and Doerge et al. (2005c) 22 48 35
Estimated using rat adduct data and rat in
vitro rate constants
Tb'rnqvist et al. 2008 - adduct data from drinking
water studies and in vitro rate constants 34.0 48.0 41.0
Fennell et al. 2005 - adduct data single dose
gavage studies and in vitro rate constant. 80.2
48 DRAFT-DO NOT CITE OR QUOTE
-------
Table 3-7. Measured and estimated GA AUCs normalized to Dose in Humans and F344
rats.
GA ADC in uM-hr per mg AA/kg bw
Average of Unspecified
Male Female Male and Gender
Rat Rat Female Rat Rat Human
GA in Humans
Estimated using human adduct data and test
animal in vivo rate constants
Fennell et al 2005 - human adduct data and in
vivo rate constants derived from Tareke et al.
(2006) adduct data for all rat and mice in Doerge
et al. (2005 b, c) single dose AUCs. 12.5
Estimated using human adduct data and in
vitro rate constants
Fennell et al 2005 - human adduct data and
Fennell in vitro rate constants 60.4
Fennell et al 2005 - human adduct data and
Bergmark et al. 1993 in vitro rate constants 37.0
Estimated using human adduct data and rat in
vitro rate constants
Fennell et al 2005 - human adduct data and
Tb'rnqvist et al. 2008 in vitro rate constants 29.9
GAin F344Rats
Measured
Doerge et al. 2005 c - time course data from a
single dietary exposure 19.0 15.0 17.0
Doerge et al. 2005 c - time course data from a
single gavage exposure 13.0 44.0 28.5
Estimated using rat adduct data and rat in
vivo rate constants
Tareke et al (2006) adduct data for the Doerge et
al. (2005a) 42 day drinking water study, and
gender specific in vivo derived rate constants
from Tareke et al. (2006) and Doerge et al.
(2005c) 15.0 48.0 31.5
Estimated using rat adduct data and rat in
vitro rate constants
Tb'rnqvist et al. 2008 - adduct data from drinking
water studies and in vitro rate constants 18.0 34.0 26.0
Fennell et al. 2005 - adduct data single dose
gavage studies and in vitro rate constant. 52.1
49 DRAFT-DO NOT CITE OR QUOTE
-------
Urinary Metabolites as Biomarkers of Exposure
Urinary metabolites (along with hemoglobin adducts) have been measured in a number of
studies to estimate daily intake levels in the general population. Doerge et al. (2008) compiled a
summary of selected studies reproduced below as Table 3-8.
Table 3-8. Selected published measurements of acrylamide-derived hemoglobin
adducts and urinary metabolites in groups of nonsmokersa (from Doerge et al.,
2008)
study
Paulsson et al. (54)
Boettcher et al, (49)
Bjellaas et al. (55)
Bjellaas et al. (22)
Urban et al. (56)
Vesper et al, (57)
Fennell et al. (55)
Keller! et al, (59)
Chevolleau et al. (60)
Vesper et al. (61)
group
size (n)
5
16
65
44
60
6
24
13
52
61
AA.MA
("9"-)
—t>
29
39
—
73
—
NDC
26
—
"
GAMA
(wg/L)
—
5
31
—
16
—
ND
3
—
"
AA-Val
(fmol/mg)
27
19
—
38
28
43
76
—
27
51
GA-Val
(fmol/mg)
26
17
—
20
3
26
29
—
22
34
s Urinary concentrations of acrylamide- and glycidamide-derived mercapturic
acids in urine (AAMA and GAMA, respectively) and N-terminal valine adducts of
hemoglobin with acrylamide and glycidamide (AA-Val and GA-Val, respectively)
are reported from the respective studies of nonsmoking humans.b -, not measured.
c Not detected,
[Paulsson et al., 2003b; Bjellaas et al., 2005 [55], 2007b [22]; Urban et al., 2006; Vesper
et al., 2005 [57], 2007 [61]; Fennell et al., 2005; Kellert et al., 2006; Chevolleau et al., 2007;
Boettcher et al., 2005]
Hays and Alyward discuss the methods used to estimate the external dose of acrylamide
based upon the many studies that have measured urinary concentration and correlated the levels
to daily intake. As an example for the glutathione metabolite of AA, acetyl-S-(2-
carbamoylethyl)-L-cysteine (AAMA), under steady-state exposure conditions consistent with
chronic exposure, the daily elimination of AAMA on a molar basis should be equal to
approximately 50% of the daily intake (Fuhr et al., 2006; Boettcher et al., 2006). The daily mass
of AAMA excreted in urine as a function of the daily intake of AA can be estimated as follows:
50 DRAFT-DO NOT CITE OR QUOTE
-------
AAMA =AAx
unne
MWAA
xO.5
where: AAMAurine is the mass of AAMA excreted in urine per day (mg); AA is the total daily
dose of acrylamide (mg); M\¥AAMA and MWAA are the molecular weights of AAMA and AA
(234.1 and 71.08), respectively.
These estimates of daily intakle based upon urinary concentrations requires some
additional assumptions (with the accompanying uncertainties). The best estimates would be
based on a 24-h urine specimen, and the subject's age, gender, lean body mass (a function of
height and weight), dietary patterns, and other factors including kidney function status would all
be known. Hays and Alward (2008b) note that, in practice, collection of 24 hour samples is
difficult and impractical for large biomonitoring studies such as the NHANES/CDC effort. As a
result, urinary concentrations are generally reported based on spot urine sample collection. The
absolute concentration of compounds in such samples can vary substantially due simply to
differences in hydration rates and to other factors. Thus, in addition to reporting absolute urinary
concentrations of such chemicals (for example, in units of ug/L), CDC and other researchers
generally also report levels adjusted to creatinine levels (e.g., ug chemical/g creatinine). While
hydration status introduces variability into interpretation of urinary concentrations on a volume
basis, creatinine adjustment also introduces variability into the analysis. Because the total intake
is also a function of weight (these values are generally specified in terms of mg of intake per kg
bodyweight per day), estimates of the creatinine-adjusted concentration in urine associated with
any daily intake can also vary substantially among individuals. The reader is referred to the
Hays and Alyward (2008b) acrylamide case study for additional discussion of these uncertainties
and two approaches to estimating daily intake based on urinary metabolite levels that adjust for
creatinine ( ug chemical/g creatinine) and for urinary volume ( ug chemical/liter of urine).
As an example of a daily itake estimate, Bjellaas et al. (2007) reported urinary
mercapturic acid derivatives of AA in a clinical study of 53 subjects. Urinary metabolite levels
were determined using solid-phase extraction and liquid chromatography with positive
electrospray MS/MS detection. The median (range) total excretion of AA in urine during 24
hours was 16 (7-47) ug AA for nonsmokers and 74 (38-106) ug AA for smokers. Median
intakes (range) of AA were estimated based on 24 hour dietary recalls as 21 (13-178) ug for
nonsmokers and 26 (12-67) ug for smokers. The median dietary exposure to AA was estimated
to be 0.47 (range 0.17-1.16) ug /kg body weight per day. In a multiple linear regression
analysis, the urinary excretion of AA metabolites correlated statistically significant with intake
of aspartic acid, protein, starch and coffee. Consumption of citrus fruits correlated negatively
with excretion of AA metabolites.
51 DRAFT-DO NOT CITE OR QUOTE
-------
In the Hartmann et al., 2009 analysis of aery 1 amide (AA) exposure in the nonsmoking
general population, hemoglobin adduct levels in blood and mercapturic acid excretion in urine
were used to calculate daily AA intake, and gave practically identical values. The median daily
intakes were 0.43 (0.21-1.04) ug/kg of body weight(bw)/day using Hb adducts and 0.51 (
-------
Aery lam ide
Glycidamide
c
0
5
O
?
5
3
«*
>
fc
F
k
'
4
* Lung
,
II Tissue**;
Liver
L
r
_>
_^ —
•*-
0
o
CO
t/J
o
**"
o
>
Lung
cc
—
Tissues
O '
Liver
,
T
Source: Kirman et al. (2003).
Figure 3-3. Schematic of the Kirman et al. PBTK model for acrylamide.
Young et al. (2007) also developed a PBTK/TD (toxicodynamic) model (see Figure 3-4)
that simulates AA and GA kinetics in mice, rats, and humans, and adds representation of GA-
DNA adduct formation (considered a toxicodynamic event in the pathway leading to
mutagenicity). The Young et al. model parameter values were based on rat and mouse kinetic
data generated at the US FDA's National Center for Toxicological Research (NCTR) (Doerge et
al., 2005a, b, c) and from the literature (Sumner et al., 2003, 1992; Barber et al., 2001; Raymer et
al., 1993); on published human urinary excretion data (Fuhr et al., 2006; Fennel et al., 2005) and
on human hemoglobin adduct data from a dietary exposure (Boettcher et al., 2005). Young et al.
use the PBTK model to fit individual animal PK data, and then evaluate the resulting differences
in parameter values (and distributions). The Young et al. (2007) model simulated liver DNA-
adduct levels based upon data from Doerge et al. (2005a), and was subsequently used to
integrate the findings of rodent neurotoxicity and cancer into estimates of risks from human AA
exposure through the diet (Doerge et al., 2008). The approach taken in Young et al. (2007) was
to adjust the model parameter values to fit individual data sets, rather than develop a single set of
parameters that best fit all of the data. For the Young et al. (2007) model to be used for specific
EPA application in deriving a toxicity value, additional work is needed to determine which
individual parameter values would be the most appropriate to use for each derivation, or
preferably, what set of parameters could be developed that would best fit all of the data.
53
DRAFT-DO NOT CITE OR QUOTE
-------
J npiil
l'BPK-1
Aery lam ide
(AA)
! oixkr
metabolism
i'BPK-2
G lye id amide
-------
The following discussion provides a general description of the Kirman et al. (2003),
Walker et al. (2007), and Young et al. (2007) PBPK models. The reader is referred to the
published articles for additional detailed information on model parameters and simulation
results.
Kirman et al. (2003) PBTK model
A diagram of the Kirman et al. (2003) model is presented in Figure 3-3. This model
simulates the distribution of AA and GA within five compartments—arterial blood, venous
blood, liver, lung, and all other tissues lumped together. The arterial and venous blood
compartments are further divided into serum and blood cell subcompartments to model specific
data sets (e.g., chemical bound to hemoglobin in red blood cells). Different routes of exposure to
AA are represented in the Kirman et al. model including intravenous (i.v.), intraperitoneal (i.p.),
gavage, oral drinking water, and inhalation. Metabolism of AA and GA are represented only in
the liver. Hepatic metabolism of AA proceeds via two pathways: (1) saturable epoxidation by
cytochrome P-450 to produce GA; and (2) first-order conjugation with glutathione (GSH) via
glutathione S-transferase (GST) to ultimately yield N-acetyl-S-(3-amino-3-oxopropyl)cysteine.
Hepatic metabolism of GA proceeds either with: (1) a first-order conjugation with GSH to yield
N-acetyl-S-(3-amino-2-hydroxy-3-oxopropyl)cysteine andN-acetyl-S-(carbamoyl-2-hydroxy-
ethyl)cysteine; or (2) with further saturable metabolism by epoxide hydrolase to yield 2,3-di-
hydroxypropionamide. Based on the reactivity of AA and GA with GSH, and the potential for
depletion of hepatic GSH with sufficiently high doses of AA, GSH depletion and resynthesis are
also represented in the model structure. Free GA enters into the GA portion of the model from
the oxidative metabolism of AA in the liver compartment. The model also represents binding of
AA and GA to hemoglobin, or to liver, tissue, or blood macromolecules. The model was
originally developed in ACSL, version 11.8.4 (Aegis Technologies Group, Huntsville, AL), and
has subsequently been revised in acslXtreme version 2.3.014, as well as inplemented in Excel.
The model parameters values and sources include measured or calculated values for rat
physiological parameters from the literature (tissue volumes, blood flows), estimates for the
tissue partition coefficients for AA based on a published algorithm or specific chemical
properties (e.g., solubility in water and octanol, vapor pressure), estimates for GA tissue partition
coefficients from values for AA using a proportionality constant of 3.2 derived from the ratio of
structural analogs (acrylonitrile and its epoxide metabolite, cyanoethylene oxide), and estimates
of metabolism and tissue binding rates optimized to fit tissue levels of administered
[14C]-radiolabeled AA (Ramsey et al., 1984; Miller et al., 1982), or to urinary metabolite levels
(Raymer et al. 1993 Sumner et al., 1992; Miller et al., 1982). Once the initial metabolism
parameters were defined, these values were held fixed, and the model terms for tissue binding
were adjusted to match the tissue-binding data sets, which include the radiolabel time-course
55 DRAFT-DO NOT CITE OR QUOTE
-------
data of Miller et al. (1982) and Ramsey et al. (1984). The model terms for metabolism were
fine-tuned by refitting simulations of the reparameterized model to the metabolism data sets.
Similarly, the model terms for tissue binding were fine-tuned by refitting simulations of the
reparameterized model to the tissue binding data sets. This process was repeated until an
adequate visual fit was achieved for all data sets using a single set of parameter values.
Walker et al. (2007) PBTK model
The original Kirman et al. (2003) model was not parameterized for humans, and the data
used to calibrate the model were limited (i.e., urinary metabolite data and AA radiolabel).
Additional kinetic and hemoglobin binding data in rats and humans (Boettcher et al., 2005;
Fennell et al., 2003, 2005; Sumner et al., 2003; Bergmark et al., 1991) were used by Walker et
al. (2007) to recalibrate the Kirman et al. (2003).
Walker et al. (2007) recalibrated the Kirman et al PBTK model based on hemoglobin
adduct data as a surrogate for serum levels of AA and GA because the formation of hemoglobin
adducts occurs as a direct function of the blood concentration of the reactive agents and the time
that red cells are exposed in vivo. The use of urinary data as a surrogate for serum levels is
based on the assumption that urinary metabolites (and ratios of urinary metabolites) are an
accurate reflection of specific metabolic pathways and actual levels in the blood or tissues from
those pathways. Uncertainties in this assumption arise if not all of the metabolic pathways that
could have a significant effect on disposition are known, and if there are other clearances that
may be influencing the levels of urinary metabolites or their ratios. The relative levels of
"unrecovered" metabolites are also a source of uncertainty, since fractional recoveries in urine
(i.e., the total amount of parent and metabolite recovered in urine compared to the dose) are
typically far less than 100%. Hemoglobin adduct levels, however, provide a direct measure of
the total amount of parent AA and GA metabolite in the blood over a given time period, which is
quantified as the area under the curve (AUC in amount-unit time/volume). AUC is the integral
of "concentration" (e.g., mg or mmol/L) x "time" (e.g., minutes or hours). Under the reasonable
assumption that the amount of parent or reactive toxicant in blood indicates the amount available
to bind to tissue macromolecules or DNA, hemoglobin adducts provide a more relevant internal
metric to use to calibrate a PBTK model for use in estimating the risk of AA-induced toxicity.
A caveat in the use of the model developed Walker et al. (2007) or any updated model
based on the currently available studies, is that the estimated serum AUCs are directly related to
the value of the second order adduct formation rate, and at present there are only three estimates
of this rate, all derived from in vitro studies (Bergmark et al., 1993; Fennell et al., 2005; Tareke
et al, 2006); and only one human serum study (Kopp and Dekant, 2009), with an estimated AUC
normalized to administered mg/kg bw n the Kopp and Dekant (2009) study that is not consistent
with the normalized AUCs reported in Fennell et al. (2005) based on the in vitro rates. There is
56 DRAFT-DO NOT CITE OR QUOTE
-------
a clear data need for accurate human in vivo second order rate constants for AA and GA
hemoglobin adduct formation and elimination. Specifically studies are needed that measure all
three of the critical variables needed to resolve these rate constants - administered dose, time
course serum levels, and time course adduct levels.
Young et al. (2007) PBTK/TDModel
Young et al. (2007) published a PBTK model developed by the US FDA's National
Center for Toxicological Research (NCTR) to simulate AA and GA kinetics in mice, rats, and
humans, and to add representation of GA-DNA adduct formation. The model was developed in
a general purpose PBTK/TD modeling software program called PostNatal (developed at NCTR).
PostNatal is a Windows based program that controls up to four PBTK models under one shell
with multiple input and output options for various routes (or combinations of routes) of
exposure. Each PBTK unit is comprised of 28 organ/tissue/fluids compartments, and each unit
can be maintained as an independent unit or be connected through metabolic pathways to
simulate complex exposure regimens or to evaluate drug metabolism and disposition in adult
mice, rats, dogs, or humans. For the PBTK model for AA, Young et al. represented the kinetics
of AA, GA, AA bound to glutathione, and GA bound to glutathione in separate models coupled
by input and output terms with urinary excretion represented in each model (see Figure 3-4).
AA or GA dosing is represented by the input terms in the AA and GA model, respectively.
Physiological parameter values in the Young et al. model (organ/tissue weights, blood
flows) are assigned with values within the PostNatal program based on animal species, gender,
and total body weight (specific values and literature sources not specified). The data used to
calibrate the Young et al. model for rats and mice include AA serum levels in rats from an i.p.
acute exposure (Raymer et al., 1993), plasma AA and GA levels, and AA and GA hemoglobin
adduct levels following relatively high (50 mg/kg bw) repeat i.p. dosing in rats for 11 days or 2.8
mM of AA in drinking water for 47 days (Barber et al., 2001), urinary excretion profile and AA
and GA hemoglobin adduct levels following dosing via i.p. (50 mg/kg bw), gavage (50 mg/kg
bw) dermal (150 mg/kg bw) or inhalation (3 ppm for 6 hours) (Sumner et al., 2003); and serum
and tissue (liver, lung, muscle, brain) levels of AA and GA, and liver GA-DNA adduct data in
rats and mice following relatively low dose dosing via i.v. (AA and GA at 0.1-0.12 mg/kg bw),
gavage (AA and GA at 0.12 and 50 mg/kg bw), diet (~0.1 mg/kg bw over 30 minutes), and in
drinking water (~1 mg/kg bw AA over 42 days) (Doerge et al., 2005a, b, c). The single and
multiple oral data from Barber et al. (2001) were combined with the urinary elimination data of
Sumner et al. (2003, 1992) and simulated with the model. The Raymer et al. (1993) data were
also combined with the urinary elimination data of Sumner et al. (2003, 1992) and simulated in a
similar manner. The NCTR tissue data (Doerge et al., 2005a, b, c) were used to develop
partition coefficients. Only those tissues specifically analyzed for AA or GA were partitioned
57 DRAFT-DO NOT CITE OR QUOTE
-------
differently from the blood compartment, i.e., assigned a partition coefficient other than 1 (see
Table 3-3). Values for the human parameters were calibrated against urinary excretion data
(Fuhr et al., 2006; Fennel et al., 2005) and hemoglobin adduct data from a dietary exposure
(Boettcher et al., 2005).
Values for the metabolism and elimination of AA or GA, for AA or GA binding to
hemoglobin, and for GA-DNA adduct formation were derived by optimizing the fit of the
simulation results to individual animal data (i.e., by minimizing the weighted sum of squares of
the difference between each data point and its simulated value). All rate constants for the
metabolic and elimination processes, the binding and decay of AA or GA to hemoglobin, and the
binding of GA to liver macromolecule are represented as first order (i.e., rate constants of min"1).
Although Young et al. calibrated their model parameter values in a logical sequence against the
data identified in the paper, a number of sensitive parameters were allowed to vary when fitting
the individual animal data so as to optimize the model fit to each set of data. The authors
evaluate the resulting differences among the model parameter values relative to gender and study
conditions for insights into the toxicokinetics of AA and GA, and to assess the uncertainty in the
model parameter values. Although in some cases there are statistically significant differences in
the fitted model parameter values for basic physiological functions such as excretion of AA-GSH
conjugates in urine (which varies as much as four to sixfold for model fits to different studies),
the authors argue that the ranges of values are not exceedingly wide considering that different
routes of administration for different chemicals are all being compared, and that there is very
little difference for each metabolic rate constant when comparing across gender, dose, and route.
For use in the derivation of a toxicity value, a PBTK model is generally developed with
the aim of resolving a single set of parameter values that either fits all of the available data best
(i.e., provides the broadest predictive capability) or fits the most relevant data for a specific
application (e.g., oral and inhalation data for a route-to-route extrapolation). Evaluating the
importance of uncertainty in a parameter value or combination of values also depends upon the
choice of the dose metric used in a risk assessment, and how sensitive that metric is to the
parameter(s) of interest. For the Young et al. (2007) model to be applicable for use in the
development of toxicity values for AA, some additional work will therefore be needed to identify
a single set of parameters, and to evaluate the sensitivity of various dose metrics to the
parameters that are the most uncertain.
58 DRAFT-DO NOT CITE OR QUOTE
-------
4. HAZARD IDENTIFICATION
The importance of assessing the potential health effects from exposure to AA in food has
resulted in a unique international collaboration as reflected in international meetings (JIFSAN,
2004, 2002), research programs (U.S. FDA, 2009), special journal issues (Mutation Research
vol. 580, issues 1-2, 2005), hazard and exposure assessments (JECFA, 2005; NTP/CERHR,
2004), and internet sites (U.S. FDA, 2009; FAO/WHO, 2009) solely dedicated to providing the
research and regulatory community (as well as the private and public sectors) access to the latest
information. The discussion here identifies key studies that were used to derive EPA's
noncancer and cancer toxicity values and that provide scientific support to the cancer descriptor
and the characterization of the noncancer and cancer modes of action.
4.1. STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS
Numerous case reports of occupational exposure to AA involving both inhalation and
dermal exposure report neurological impairment in humans from exposure to AA, but levels of
exposure are generally not measured (Gjerl0ff et al., 2001; Mulloy, 1996; Dumitru, 1989;
Donovan and Pearson, 1987; Kesson et al., 1977; Mapp et al., 1977; Davenport et al., 1976;
Igisu et al., 1975; Takahashi et al., 1971; Fullerton, 1969; Auld and Bedwell, 1967; Garland and
Patterson, 1967). Substances like AA that are highly reactive with short half-lives in the blood
are more challenging to monitor for estimates of exposure. AA, however, forms adducts with
hemoglobin that persist throughout the life of the adducted red blood cell (estimated at around
120 days), and hemoglobin adducts have been used as biomarker of exposure. There are two
cross-sectional health surveillance studies of AA-exposed workers that correlate AA-hemoglobin
adduct levels and measures of neurological impairment in AA workers (Hagmar et al., 2001;
Calleman et al., 1994).
A quantitative human study on the toxicokinetics of AA was conducted by Fennell et al.
(2005) to evaluate metabolism and hemoglobin adduct formation following oral and dermal
administration of AA to 24 adult male volunteers. The 24 volunteers were all male Caucasians
(with the exception of one Native American), weighing between 71 and 101 kg, and between
26 and 68 years of age. All volunteers were aspermic (i.e., clinically sterile because of the
potential for adverse effects of AA on sperm), and had not used tobacco products for the past
6 months. The study was conducted in accordance with the Code of Federal Regulations (CFRs)
governing protection of human subjects (21 CFR 50), IRB (21 CFR 56), and retention of data
(21 CFR 312) as applicable and consistent with the Declaration of Helsinki. The study used
[1,2,3-13C]-AA, and, prior to the conduct of exposures in humans, a low-dose study protocol was
evaluated in rats administered 3 mg/kg [1,2,3-13C]-AA by gavage. Subjects were administered a
59 DRAFT-DO NOT CITE OR QUOTE
-------
single oral dose of 0.5, 1.0, or 3.0 mg/kg or a daily dermal dose of 3.0 mg/kg for 3 consecutive
days. A comprehensive physical exam was conducted on each individual upon check-in to the
clinic, at 24 hours after compound administration, and 7 days after checkout. This exam
included medical history, demographic data, neurological examination, 12-lead ECG, vital signs
(including oral temperature, respiratory rate, and automated seated pulse and blood pressure),
clinical laboratory evaluation (including clinical chemistry, hematology, and complete
urinalysis). Each individual also had screens for HIV, hepatitis, and selected drugs of abuse and
provided a semen sample to confirm aspermia. Additional ECG, neurological evaluation,
abbreviated physical examination, and subjective evaluation were conducted at 4 hours after
each AA administration.
No adverse events were reported in the oral phase of the Fennell et al. (2005) study.
With the dermal administration, one individual was observed to have a mild contact dermatitis,
which is a known response to AA and was part of the informed consent. This individual was
seen by a dermatologist who performed a skin biopsy that was consistent with a delayed
hypersensitivity reaction. The skin reaction resolved 39 days after the first application of AA
and 23 days after the reaction was manifested. An increase in the liver enzyme alanine
aminotransferase (ALT) was observed above the upper limit of the reference range (normal) in
four of the five individuals who received AA by dermal application, one of whom had a
preexisting elevation of this enzyme prior to receiving the dose (data and time of observation not
reported). One individual who received dermal AA also had an elevation in serum aspartate
transaminase (data and time of observation not reported). The elevated liver function tests
returned to within or near the reference range at subsequent determinations and were judged to
be not clinically significant by the study physician. When administered to the skin, AA may
cause a moderate increase in ALT levels. Serum prolactin, testosterone, and luteinizing hormone
did not differ between subjects who received AA at these levels and those who received placebo
(data not reported). All blood parameters and hormone levels were within the normal range.
There were no neurological or cardiovascular findings in the study participants at either 24 hours
or 7 days postexposure.
The recent discovery of AA in foods has prompted a number of studies to evaluate a
potential association between dietary AA intake and cancer. Available epidemiology studies on
increased risk of cancer from AA in food include a number of case-control studies (Wilson et al.,
2009a; Pelucchi et al., 2007, 2006; Michels et al., 2006; Mucci et al., 2005, 2004, 2003) and
numerous reports from several ongoing prospective studies (Larsson et al., 2009a, b, c, d; Wilson
et al., 2009b; Hogervorst et al., 2008a, b, 2007; Mucci et al., 2006). These studies evaluated
Swedish, Danish, Dutch, or Italian populations; available assessment of a U.S. population is
restricted to the prospective study of Wilson et al. (2009b). Some of the tumor sites observed in
60 DRAFT-DO NOT CITE OR QUOTE
-------
animal studies (thyroid, testis, central nervous system [CNS]) have also not been evaluated, and
there are limitations in some of the study methods and cohort sizes.
In addition two case-control studies have examined possible associations between AA-
Hb adduct levels in red blood cells and risks for breast cancer (Olesen et al., 2008) and prostate
cancer (Wilson et al., 2009a).
Two cohort mortality studies (Collins et al., 1989; Sobel et al., 1986) with follow-up
analyses (Marsh et al., 2007, 1999; Swaen et al., 2007) evaluated increased risk for cancer in AA
workers.
No human studies were identified that assessed the potential for adverse reproductive or
developmental effects from exposure to AA.
An important factor in evaluating epidemiology studies that relate dietary intake to
effects concerns the characterization of the variability in AA internal dose relative to differences
in diet composition and consumption rates. Hagmar et al. (2005) observed relatively narrow
interindividual variation in AA adduct levels, and suggests that estimates of individual dietary
AA intake will need to be very precise to be useful in cancer epidemiology. Hagmar et al.
(2005) evaluated variation in dietary exposure to AA relative to measurement of AA hemoglobin
adduct levels (as a biomarker of exposure) in blood samples from the Malmo Diet and Cancer
Cohort (n = 28,098). The blood donors were well characterized with regard to their food habits,
and 142 individuals were selected to obtain the highest possible variation in the adduct levels
from AA (i.e., none, random, or high intake of coffee, fried potatoes, crispbreads, and snacks,
food items estimated to have high levels of AA). The median hemoglobin adduct level in the
randomly selected group of nonsmokers was compatible with earlier studies (0.031 nmol/g). The
variation in the average internal dose, measured as hemoglobin adducts, was somewhat smaller
than estimated for daily intake by food consumption questionnaires in other studies. Among
70 nonsmokers, the AA adduct levels varied by a factor of 5 (range: 0.02-0.1 nmol/g), with
considerable overlap in AA-adduct levels among the different dietary groups. There was a
significant difference between men with high dietary exposure to AA compared to men with low
dietary exposure (p = 0.04). No such difference was found for women. As expected, smokers
had a higher level (range: 0.03-0.43 nmol/g) of AA adducts. Smoking women with high dietary
exposure to AA had significantly higher AA adduct levels compared to smoking women with
low dietary exposure (p = 0.01), however, no significant difference was found in smoking men.
Cohort mortality studies
Collins et al. (1989) conducted a cohort mortality study of all male workers (8,854, of
which 2,293 were exposed to AA) who had been hired between January 1, 1925 and January 31,
1973 at four American Cyanamid factories, three in the United States (Fortier, LA
[1295 workers]; Warners, NJ [7,153 workers]; and Kalamazoo, MI [60 workers]) and one in the
61 DRAFT-DO NOT CITE OR QUOTE
-------
Netherlands (Botlek [346 workers]). Estimations of AA exposure were based on available
monitoring data and worker knowledge of past jobs and processes. Industrial hygiene
monitoring was in place at all four plants in 1977. AA levels monitored at that time were
typically considered to be representative of levels during the entire period of plant operation.
Workers were classified as unexposed when cumulative AA exposure was less than
0.001 mg/m3-years. Exposure groups were divided into three categories of cumulative exposure:
0.001 to 0.030, 0.030 to 0.30, and greater than 0.30 mg/m3-years. Smoking history records were
available for approximately 35% of the total cohort, 76% of whom were smokers. Smoking
status of the other workers was unknown. Mortality rates among the factory workers were
compared with the expected number of deaths among men of the United States from 1925 to
1980 or the Netherlands from 1950 to 1982 to derive standardized mortality ratios (SMRs) as a
measure of relative risk for each cohort. No statistically significantly elevated all cause or
cause-specific SMRs were found among AA-exposed workers (including cancer of the digestive
or respiratory systems, bone, skin, reproductive organs, bladder, kidney, eye, CNS, thyroid, or
lymphatic system). All causes of both exposed and nonexposed workers were significantly (p <
0.05) lower than expected (SMRs = 0.81 and 0.91, respectively; 95% confidence intervals [CI]
were not reported). Trend tests showed no increased risk of mortality due to cancer at several
sites (digestive tract, respiratory system, prostate, CNS, or lymphopoietic system) with
increasing level of exposure to AA.
The most recent update report (Marsh et al., 2007) of the cohort of Collins et al. (1989)
includes study periods of 1925-2002 for the 8,508 workers in the three facilities in the United
States, and 1965-2004 for the 344 workers at the Botlek plant in the Netherlands (the original
cohort of 346 included 2 females who were excluded in the follow-up). In the Dutch cohort,
deficits in deaths were reported for all sites ofapriori interest (Marsh et al., 2007). Among the
workers at the three facilities in the United States (during which 4,650 deaths occurred among
the 8,508 workers in the period of 1925-2002), excess and deficit overall mortality risks were
observed for cancer sites implicated in experimental animal studies: brain and other CNS (SMR
0.67, 95% CI 0.40-1.05), thyroid gland (SMR 1.38, 95% CI 0.28-4.02), and testis and other
male genital organs (SMR 0.64, 95% CI 0.08-2.30); and for sites selected in the original report
(Collins et al., 1989) of this cohort: respiratory system cancer (SMR 1.17, 95% CI 1.06-1.27),
esophagus (SMR 1.20, 95% CI 0.86-1.63), rectum (SMR 1.25, 95% CI 0.84-1.78), pancreas
(SMR 0.94, 95% CI 0.70-1.22), and kidney (SMR 1.01, 95% CI 0.66-1.46). None of the
mortality excesses were statistically significant, except for respiratory system cancer, which
Collins et al. (1989) attributed to muriatic acid exposure. Table 4-1 lists all of the observed
deaths and SMRs for selected causes among the U.S. workers who died between 1950 and 2002.
Table 4-2 lists the SMRs from observed deaths for selected cancer sites (rectum, pancreas, and
kidney) for all U.S. workers who died between 1950 and 2002, according to the following
62 DRAFT-DO NOT CITE OR QUOTE
-------
exposure parameters and categories: duration of employment (<1, 1-, and 15+ years), time since
first employment (<20, 20-, and 30+ years), duration of exposure (unexposed, 0.001-, 5-, and
20+ years), cumulative exposure (<0.001, 0.001-, 0.03-, and 0.30+ mg/m3-years), and estimated
mean exposure concentrations (unexposed, 0.001-, 0.02-, and 0.3+ mg/m3). In these exploratory
exposure-response analyses of rectal, pancreatic, and kidney cancers, no statistically
significantly elevated SMRs were found.
63 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-1. Observed deaths and SMRs for selected causes by follow up period for all workers (compared with the
general U.S. population)
Cause of death (ICDA-8)3
All causes (000-999):
All malignant neoplasms (140-209)
Buccal cavity and pharynx (140-149)
Digestive organs and peritoneum (150-159)
Esophagus (150)
Stomach (151)
Large intestine (153)
Rectum (154)
Liver (155, 156)
Pancreas (157)
Respiratory system (160-163)
Larynx (161)
Lung (162, 163)
Bone (170)
Skin (172, 173)
Prostate (185)
Testis and other male genital organs (186, 187)
Bladder (188)
Kidney (189)
Brain and other CNS (191, 192)
Thyroid gland (193)
All lymphopoietic tissue (200-209)
Lymphosarcoma and reticulosarcoma (200)
Hodgkin's disease (201)
Leukemia and aleukemia (204-207)
Other lymphatic tissue (202, 203, 208)
Benign neoplasms (210-239)
Diabetes mellitus (250)
Diseases of the circulatory system (390^-58)
Nonmalignant respiratory disease (460-519)
Cirrhosis of the liver (571)
All external causes of death (800-998)
Unknown causes (999.9)
People (n)
Person-years
"Monson life table program ICD-8 categories, labels and codes for U.S.
Pittsburgh.
bp<0.01.
Obs
3,557
913
24
240
32
48
72
26
13
45
369
15
354
2
10
73
1
29
23
15
3
62
6
9
23
23
10
47
1,569
196
83
251
202
1925-1994
SMR
0.93b
1.06
0.99
1.08
1.27
1.24
0.97
1.31
0.72
1.04
1.19b
1.24
1.21b
0.69
0.67
0.98
0.36
1.30
1.16
0.69
2.10
0.80
0.62
1.33
0.75
0.76
1.01
0.76
0.91b
0.76b
0.96
0.72b
1995-2002
95% CI
0.90-0.96
0.99-1.13
0.63-1.47
0.95-1.22
0.87-1.79
0.91-1.64
0.76-1.22
0.86-1.93
0.38-1.24
0.76-1.39
1.07-1.32
0.70-2.05
1.08-1.34
0.08-2.50
0.32-1.23
0.77-1.24
0.01-1.99
0.87-1.87
0.73-1.74
0.39-1.15
0.43-6.14
0.61-1.03
0.23-1.35
0.61-2.52
0.48-1.12
0.48-1.14
0.49-1.86
0.56-1.02
0.86-0.95
0.66-0.87
0.76-1.19
0.63-0.81
Obs
1,093
291
8
68
9
8
33
4
4
9
110
1
109
1
4
38
1
10
4
3
0
22
0
1
9
12
2
37
383
99
9
20
108
SMR
0.95
0.97
1.67
0.92
1.01
0.99
1.29
0.94
0.45
0.62
1.08
0.32
1.12
2.30
0.74
0.93
2.96
1.09
0.57
0.55
-
0.74
-
2.13
0.79
0.68
1.11
1.18
0.78b
0.77b
0.79
0.77
95% CI
0.89-1.00
0.86-1.08
0.72-3.28
0.75-1.23
0.46-1.91
0.43-1.95
0.89-1.81
0.26-2.40
0.12-1.16
0.28-1.18
0.89-1.31
0.01-1.79
0.92-1.35
0.06-12.83
0.20-1.89
0.66-1.28
0.07-16.51
0.52-2.00
0.16-1.47
0.11-1.62
0.00^1.89
0.47-1.12
0.00-14.14
0.05-11.85
0.36-1.51
0.35-1.20
0.14^1.02
0.83-1.63
0.70-0.86
0.62-0.93
0.36-1.50
0.47-1.19
Obs
4,650
1,204
32
308
41
56
105
30
17
54
479
16
463
3
14
111
2
39
27
18
3
84
6
10
32
35
12
84
1,952
295
92
271
310
1925-2002
SMR
0.93b
1.04
1.10
1.05
1.20
1.19
1.05
1.25
0.63
0.94
1.17b
1.05
1.18b
0.90
0.69
0.97
0.64
1.24
1.01
0.67
1.38
0.78C
0.60
1.38
0.76
0.73
1.03
0.91
0.88b
0.76b
0.94
0.72b
95% CI
0.90-0.96
0.98-1.10
0.75-1.56
0.94-1.18
0.86-1.63
0.90-1.55
0.86-1.27
0.84-1.78
0.37-1.02
0.70-1.22
1.06-1.27
0.60-1.71
1.08-1.30
0.19-2.63
0.38-1.15
0.79-1.16
0.08-2.30
0.88-1.70
0.66-1.46
0.40-1.05
0.28^1.02
0.63-0.97
0.22-1.31
0.66-2.53
0.52-1.08
0.51-1.02
0.53-1.79
0.72-1.12
0.84-0.92
0.68-0.85
0.76-1.15
0.64-0.81
8,508
288,126
plants for 1925-1989; corresponding rates for 1990-
4,565
32,219
2001 from the mortality and population data system (MPDS) maintained at the University of
8,508
320,345
64
DRAFT-DO NOT CITE OR QUOTE
-------
cp < 0.05.
Source: Marsh et al. (2007).
65 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-2. Observed deaths and SMRs for selected cancer sites by duration
of employment, time since first employment, and measures of exposure to
acrylamide, all U.S. workers, 1950-2002 (compared with the local male
populations)
Duration of employment
(years)
<1
1-
15+
Time since first
employment (years)
<20
20-
30+
Duration of exposure
(years)
Unexposed
0.001-
5-
20+
Cumulative exposure
(m^/m3 -years)
0.001
0.001-
0.03-
0.30+
Mean intensity of
exposure (mg/m3)
Unexposed
0.001-
0.02-
0.30+
Rectum
Obs
8
13
7
o
J
5
20
21
o
5
i
2
21
1
4
2
21
4
0
3
SMR
0.63
1.25
1.05
0.71
0.79
1.04
0.85
1.32
1.15
1.96
0.85
1.43
2.44
0.75
0.85
2.96
-
2.08
95% CI
0.27-1.24
0.66-2.13
0.42-2.12
0.15-2.06
0.26-1.83
0.64-1.61
0.52-1.29
0.27-3.86
0.14-4.14
0.24-7.07
0.52-1.29
0.04-7.98
0.67-6.25
0.09-2.71
0.52-1.29
0.81-7.58
0.00-1.64
0.43-6.09
Pancreas
Obs
22
17
15
4
11
39
38
6
6
4
38
3
4
9
38
5
5
6
SMR
0.82
0.84
1.17
0.66
0.96
0.92
0.78
1.12
1.55
1.81
0.78
1.65
0.94
1.71
0.78
1.34
1.11
1.85
95% CI
0.51-1.24
0.49-1.35
0.66-1.93
0.18-1.68
0.48-1.72
0.65-1.26
0.55-1.08
0.41-2.43
0.57-3.38
0.49-4.63
0.55-1.08
0.34-4.83
0.26-2.40
0.78-3.25
0.55-1.08
0.44-3.14
0.36-2.60
0.68-4.03
Kidney
Obs
10
9
8
2
4
21
19
4
3
1
19
1
4
3
19
2
3
3
SMR
0.73
0.87
1.20
0.58
0.65
1.00
0.78
1.31
1.37
0.88
0.78
0.88
1.56
1.15
0.78
0.86
1.27
1.77
95% CI
0.35-1.34
0.40-1.64
0.52-2.37
0.07-2.09
0.18-1.66
0.62-1.52
0.48-1.22
0.36-3.36
0.28-4.00
0.02-4.88
0.47-1.22
0.02-4.92
0.42-4.00
0.24-3.36
0.47-1.22
0.10-3.10
0.26-3.71
0.37-5.18
Source: Marsh et al. (2007).
66
DRAFT-DO NOT CITE OR QUOTE
-------
Although an earlier update analysis (Marsh et al., 1999) of the Collins et al. (1989) cohort
reported a significant 2.26-fold risk (95% CI 1.03-4.29) for pancreatic cancer among workers
with cumulative exposure to AA >0.30 mg/m3-years, the excess in the most recent update
(Marsh et al., 2007) was not statistically significant (SMR 1.71, 95% CI 0.78-3.25). Marsh et al.
(2007) concluded that exposure to AA at the levels reported in their study sites "was not
associated with elevated cancer mortality risks". Limitations of the study are the large
proportion of short-term workers in the cohort, incomplete smoking data, and somewhat limited
follow-up duration (about 54% of the cohort had died through 2002). Strengths of the study
include the relatively large size of the cohort and the quantitative measures of exposure that were
made; with continued follow-up, additional important information will be gathered.
Sobel et al. (1986) conducted a mortality study on a cohort of 371 workers assigned to
AA and polymerization operations at a Dow Chemical facility in the United States. The cohort
was identified from annual and monthly census lists generated between 1955 and 1979. Analysis
and review of air monitoring data and job classifications resulted in estimates of personal 8-hour
time-weighted average AA concentrations of 0.1-1.0 mg/m3 before 1957, 0.1-0.6 mg/m3 from
1957 to 1970, and 0.1 mg/m3 thereafter. Fourteen of the 371 workers had been exposed to
organic dyes in another area of the facility for 5 or more years but moved to the AA areas when
organic dye processes were discontinued. SMRs, calculated for categories in which at least two
deaths were observed, were based on mortality of white males in the United States.
A total of 29 deaths from all causes was observed among the cohort up until 1982,
compared to 38 expected. Incidences of tumors of the CNS, thyroid gland, and endocrine
organs, as well as mesotheliomas, were of particular interest within the cohort in view of a report
of increased tumor incidences at these sites in AA-exposed rats (Johnson et al., 1986); however,
no statistically significantly increased incidences of cancer-related deaths were observed.
Mortality from cancer among the entire cohort was slightly elevated (11 vs. 7.9 expected) but
was lower than expected when the workers with previous exposure to the organic dyes were
excluded (4 deaths vs. 6.5 expected). This study is limited by small cohort size, exposure to
other chemicals (e.g., acrylonitrile), relatively short duration of employment for many of the
workers (276 were employed for 4 years or less, 167 of whom had less than 1 year of
employment at the facility), limited follow-up duration, and the inability to detect small
increases in risk among site-specific cancers.
Swaen et al. (2007) provide an update of the Sobel study cohort (of 371 AA workers) and
expand the cohort to include employees hired since 1979. A total of 696 AA workers were
followed from 1955 through 2001 to ascertain the long-term health effects of occupational
exposure to AA among production and polymerization workers and the cause of death.
Exposure to AA was retrospectively assessed based on personal samples from the 1970s onwards
and area samples over the whole study period. The study reports fewer of the AA workers died
67 DRAFT-DO NOT CITE OR QUOTE
-------
(n = 141) compared to an expected number of 172.1 (SMR 81.9, 95% CI 69.0-96.6). No cause-
specific SMR for any of the investigated types of cancer was exposure related. The authors
report more total pancreatic cancer deaths (n = 5) than expected (n = 2.3) (SMR 222.2, 95% CI
72.1-518.5), however, 3 of the 5 were in the low dose group, with no apparent dose-response
relationship with AA exposure, and thus questionable support for an AA related carcinogenicity.
Although these studies provide no good evidence of a cancer risk from occupational exposure to
AA at production facilities, additional studies are needed to further evaluate the potential
carcinogenicity in humans from exposure to AA.
Case-control studies
No statistically significant associations were found between high consumption of foods
with high (300-1,200 |ig/kg) or moderate (30-299 |ig/kg) AA concentrations and an increased
risk of large bowel, kidney, or bladder cancer in a reanalysis (Mucci et al., 2003) of an existing
population-based case-control study (Augustsson et al., 1999). Augustsson et al. (1999)
identified the existing population to study the relation between heterocyclic amines in fried foods
and cancer of the large bowel and urinary tract. Individuals in this study were born in Sweden
between 1918 and 1942 and resided in Stockholm for at least 1 month between November 1992
and December 1994. Cases were identified from a national cancer registry. Controls were
selected from a national population registry and matched by age and gender to cases.
Questionnaires concerning dietary habits in the 5 years previous to the study were mailed to
692 controls and 875, 391, and 186 cases of cancer of the large bowel, bladder, and kidney,
respectively. Based on completed questionnaires, the final sample size was 538 controls,
591 large bowel cancer cases, 263 bladder cancer cases, and 133 kidney cancer cases. In an
unconditional logistic regression analysis, odds ratios (ORs) were calculated for frequency and
amounts consumed of 14 food types with high (e.g., potato crisps, French fried potatoes) or
moderate (e.g., various types of breakfast cereals and breads) levels of AA vs. each type of
cancer. No statistically significantly elevated ORs were found for frequent consumption of any
of these food types and risks for large bowel, bladder, or kidney cancer. A summary measure of
dietary AA intake was estimated for each individual, based on the results of the questionnaire
and median concentrations of AA in foods determined by the Swedish National Food
Administration. Quartiles of the summary dietary AA measure were based on distribution in the
control group and were modeled as categorical variables with the lowest quartile as the referent
group. Tests for trend were calculated using likelihood ratio tests, where the categorical medians
of each quartile were modeled as covariates. In regression analyses that adjusted for age and
gender or several additional potential confounding variables (e.g., smoking, alcohol intake, and
fruit and vegetable intake), no statistically significant trends for increasing ORs with increasing
AA exposure measure were found for the three types of cancers. Strengths of this study include
68 DRAFT-DO NOT CITE OR QUOTE
-------
the population basis of the design, the moderately high participation rate, the large number of
cases, and the estimation of individual dietary exposures to AA. Limitations of the study to
detect increased cancer risks include the relatively low dietary intake of the study population
compared with the intake of AA in rat bioassays demonstrating cancer and the restriction of the
cases to large bowel, kidney, and bladder cancers. Other limitations include the relevance a
5-year recall questionnaire would have to a lifetime exposure estimate for individuals born
between 1918 and 1942. There may also have been considerable changes in food processing and
the types of food in the diet over that time period, e.g., potato crisp and French fry intake may
have been considerably different pre-World War II, and breads and cereal products have changed
considerably over time.
In the renal cancer cell study, Mucci et al. (2004) reanalyzed data from a large
population-based Swedish case-control study of renal cell cancer. Again, food frequency data
were linked with national food databases on AA content, and daily AA intake was estimated for
participants. The risk of renal cell cancer was evaluated for intake of food items with elevated
AA levels and for total daily AA dose. Adjusting for potential confounders, there was no
evidence that food items with elevated AA, including coffee (OR [highest vs. lowest quartile] =
0.7; 95% CI = 0.4-1.1), crispbreads (OR [highest vs. lowest quartile] = 1.0; 95% CI = 0.6-1.6),
and fried potatoes (OR [highest vs. lowest quartile] = 1.1; 95% CI = 0.7-1.7), were associated
with a higher risk of renal cell cancer risk. There was also no association between estimated
daily AA intake through diet and cancer risk (OR [highest vs. lowest quartile] = 1.1; 95%
CI = 0.7-1.8;/> = 0.8 for trend). The authors state that the results of this study were in line with
the previous studies examining dietary AA, suggesting that there is no association between
dietary AA and risk of renal cell cancer.
In the breast cancer evaluation, Mucci et al. (2005) assessed AA intake of more than
43,000 women, including 667 breast cancer cases, who were enrolled in the Swedish Women's
Lifestyle and Health Cohort. AA intake was determined from food frequency questionnaires
(FFQs) reported by the women in 1991, and the women's health status was tracked via national
health registers until the end of 2002. The average daily AA intake among the participants was
estimated at 25.9 jig/day, with less than 1.5% of the women consuming more than 1 |ig/kg-day
of AA. The foods that contributed the most to AA intake were coffee (54% of AA dose), fried
potatoes (12% of dose), and crispbreads (9% of dose). Mucci et al. (2005) compared women in
the study who had the lowest daily AA intake with women whose intake was higher and reported
no significant increased risk of breast cancer in the higher intake group.
A different research group reported similar findings for a broad spectrum of cancers.
Pelucchi et al. (2006) evaluated data from an integrated network of Italian and Swiss hospital-
based case-control studies to investigate the relation between dietary AA intake and cancers of
the oral cavity and pharynx (749 cases, 1,772 controls), esophagus (395 cases, 1,066 controls),
69 DRAFT-DO NOT CITE OR QUOTE
-------
large bowel (1,394 cases of colon cancer, 886 cases of rectal cancer, 4,765 controls), larynx
(527 cases, 1,297 controls), breast (2,900 cases, 3,122 controls), ovary (1,031 cases,
2,411 controls), and prostate (1,294 cases, 1,451 controls). All the studies included incident,
histologically confirmed cancer cases and controls admitted to the same network of hospitals for
acute nonneoplastic conditions. Odds ratios were derived from multivariate logistic regression
models, adjusted for energy intake and other major covariates of interest. The ORs for the
highest vs. the lowest quintile of AA intake were 1.12 (95% CI = 0.76-1.66) for cancer of the
oral cavity/pharynx, 1.10 (95% CI = 0.65-1.86) for esophageal, 0.97 (95% CI = 0.80-1.18) for
colorectal, 1.23 (95% CI = 0.80-1.90) for laryngeal, 1.06 (95% CI = 0.88-1.28) for breast,
0.97 (95% CI = 0.73-1.31) for ovarian, and 0.92 (95% CI = 0.69-1.23) for prostate. None of the
risk trends were significant. The authors concluded that this uniquely large and comprehensive
data set did not show any consistent association between intake of AA and the risk of breast and
several other common cancers.
Pelucchi et al. (2007) subsequently reported the results of a case-control study to
investigate the relation between dietary AA intake and renal cell cancer that was conducted in
four areas of Italy between 1992 and 2004. The study design was similar to that of Pelucchi et
al. (2006). Incident, histologically confirmed renal cell cancer cases were 767 patients (494
men, 273 women). Controls consisted of 1,534 subjects (988 men, 546 women) matched with
cases by study center, sex, and age; controls were admitted to hospitals for acute nonneoplastic
conditions, which were not related to known or potential risk factors for renal cell cancer or
long-term dietary modifications. ORs for increasing quartiles of total AA intake (20.4-31.2,
31.2-44.1, and>44.1 ug/day) were 1.21 (95% CI 0.94-1.57), 1.14 (95% CI 0.86-1.51), and 1.20
(95% CI 0.88-1.63), respectively, compared to the lowest quartile (<20.4 ug/day ); there was no
trend in risk (p = 0.35). The study authors stated that with respect to estimated total AA intake,
risk of renal cell cancer was consistent across strata of sex and age. Estimated average AA
intake was 37 ug/day. With respect to consumption of selected foods containing AA and their
relative contribution to estimated total AA intake (fried/baked potatoes, 29.6%; white bread,
28.6%; sweet biscuits, 15.0%; coffee, 12.4%; crackers, 6.5%), only white bread exhibited
statistically significantly elevated ORs (1.49, 95% CI 1.18-1.87; 1.70, 95% CI 1.25-2.30) for
weekly portions of 7-<21 and >21, respectively. The study authors indicated that the
relationship between white bread consumption and renal cell cancer might be explained by a
high glycemic content and consequent effect on levels of insulin-like growth factors. It was
concluded that the study results confirm the results of Mucci et al. (2004, 2003) in which there
was no significant association between food items containing elevated levels of AA and risk of
kidney or renal cell cancer.
Wilson et al. (2009a) conducted a case-control study to assess possible associations
between AA and prostate cancer risk using two measures of AA exposure: intake from FFQs
70 DRAFT-DO NOT CITE OR QUOTE
-------
and AA-Hb adduct levels in blood samples. Dietary data were available for 1,499 prostate
cancer cases and 1,118 controls from a Cancer of the Prostate in Sweden (CAPS) population-
based case-control study. AA-Hb adduct levels were measured in blood samples from a subset
of 170 prostate cancer cases and 161 controls. Incident cases of prostate cancer were
pathologically or cytologically verified; clinical data were available for 95% of the cases in the
study. Controls were randomly selected from the Swedish Population Registry and were
frequency matched to cases by five-year age groups and region of residence. No significant
association was found between AA exposure (as measured by FFQ or AA-Hb adduct levels) and
risk of prostate cancer. The FFQ OR for the highest versus the lowest quintile was 0.97 (95%
CI: 0.75-1.27), with adjustments made for age, smoking, body mass index, zinc intake and
energy intake. The AA-Hb adduct OR for the highest versus the lowest quintile was 0.93 (95%
CI: 0.47-1.85), with adjustment for age, region, body mass index, laboratory batch, and smoking.
Michels et al. (2006) conducted a case-control study to evaluate whether diet during
preschool age affected a woman's risk of breast cancer later in life. The case-control study is a
nested study that included 582 women with breast cancer and 1,569 controls free of breast
cancer, selected from participants in two prospective cohort studies, the Nurses' Health Study
and the Nurses' Health Study II. The cohorts in the two prospective studies consisted of
121,700 and 116,678 female registered nurses, respectively, born between 1921 and 1965. For
both cohorts, biennial self-administered questionnaires provided updated information on
demographic, anthropometric, and lifestyle factors and on newly diagnosed diseases, including
breast cancer. Pathology reports confirmed a breast cancer diagnosis, and the current study was
restricted to cases of invasive breast cancer. Information concerning childhood diet of the nurses
at ages 3-5 years was obtained from the mothers of the participants with a 30-item food-
frequency self-administered questionnaire. The median year of birth of the mothers was 1914
for case mothers and 1913 for control mothers. The median year of birth for the cases is not
reported but is calculated from the data in the report to be around 1939. The date of the
questionnaire is not stated in the report, but 1993 is when the cases were identified.
Frequencies of intake of the individual foods were converted into servings/day (e.g.,
number of glasses of milk per day) or servings/week depending on the food, and used as
continuous variables. For 718 nurses, complete data on the frequencies of food intake were
available, but for 1,433 participants data were missing or the mother did not remember the
frequency of intake of one or more food items. On average mothers marked the "don't
remember" option for 8.5% of the food items and left 3.8% of food items blank. Overall, the
proportion of missingness (blanks and don't remembers) ranged from 4.5% for milk to 21% for
cheese. Odds ratios were obtained using unconditional logistic regression models. The
association between food consumption and breast cancer was estimated for each individual food
item, combinations of foods, and nutrients. Of the 582 breast cancer cases and 1,569 controls,
71 DRAFT-DO NOT CITE OR QUOTE
-------
63% were premenopausal, 27% were postmenopausal, and 10% were of uncertain menopausal
status.
The results indicated an increased risk of breast cancer among woman who had
frequently consumed French fries at preschool age. For one additional serving of French fries
per week, the OR for breast cancer adjusted for adult life breast cancer risk factors was 1.27
(95% CI = 1.12-1.44). Consumption of whole milk was associated with a slightly decreased risk
of breast cancer (covariate-adjusted OR for every additional glass of milk per day = 0.90; 95%
CI = 0.82-0.99). Intake of none of the nutrients calculated was related to the breast cancer risk
in this study. The authors noted that they did not observe a similar association of breast cancer
with frequent consumption of hot dogs or ground beef, suggesting that French fry consumption
was not a marker of "fast food" habits. A caveat here is the time frame of the 3- to 5-year-olds,
which for at least half of the cases would be in the early 1940s, when restaurants and diets were
considerably different from today.
The study results suggest a possible association between diet before puberty and the
subsequent risk of breast cancer, but the conclusions and the study are of limited use. No
information is available on cooking methods or AA content in the foods being evaluated, and the
ability of mothers to accurately recall preschool diets from 30 to 50 years ago is questionable.
The researchers do attempt to assess the validity of the diet questionnaire protocol by
administering a questionnaire to mothers of participants in a similar longitudinal study
population (the Fels Longitudinal Study) for whom 7-day diet records were kept by the mothers
when the participants were 3-6 years old. These participants were born between 1929 and 1950,
and the questionnaire was administered in 1997. The mothers in this validation study ranged in
age from 60 to 93 years old. The sample size of completed questionnaires was small (n = 29).
Spearman correlations of mean daily consumption of foods reported by the mothers on the 7-day
diet records and on the recall questionnaire were 0.46 (p = 0.2) for whole milk, 0.37 (p = 0.07)
for broccoli, and 0.36 (p = 0.07) for French fries. Since these mothers took records during the
years of interest for the Fels cohort (in contrast to the mothers in the Nurses' Health Study
cohort), the above correlations can be considered an upper bound, suggesting high uncertainty in
the accuracy of the recall results.
Olesen et al. (2008) conducted a nested case-control study to examine associations
between breast cancer and AA exposure using AA-Hb and GA-Hb adduct levels in red blood
cells as biomarkers. The study design included separate analyses for endocrine receptor positive
and negative (ER+ and ER-) breast cancer cases. The study included 374 breast cancer cases
and 374 age-matched controls selected from a cohort of 24,697 postmenopausal women
participating in the Danish Diet, Cancer, and Health prospective cohort study. Information on
cancer occurrence was obtained from the Danish Cancer Registry. ER status was obtained from
the Danish Breast Cancer Co-operative Group.
72 DRAFT-DO NOT CITE OR QUOTE
-------
The median age of the cases and controls at entry into the cohort was 57 years; the
median length of follow-up was 4.2 years. ER status was obtained for 348 (93%) of the breast
cancer cases; 269 were reported as ER+. Mean AA-Hb and GA-HB adduct levels were 47 and
26 pmol/g globin in the 374 breast cancer cases, 47 and 28 pmol/g globin in the 374 matched
controls, 48 and 27 pmol/g globin in the 269 ER+ breast cancer cases, and 40 and 23 pmol/g
globin in the 79 ER- breast cancer cases. No significant association was found between AA-Hb
or GA-Hb adduct levels and total breast cancer either with or without adjusting for smoking
status. However, the study authors reported a statistically significant positive association
between AA-Hb adduct level and ER+ breast cancer (estimated incidence rate ratios of 4.9, 95%
CI 1.2-20) per 10-fold increase in AA-Hb adduct level in smokers and 2.7 (95% CI 1.1-6.6) per
10-fold increase in AA-Hb level after adjustment for smoking.
Prospective studies for cancer
Mucci et al. (2006) conducted a prospective study to evaluate an association between AA
in food and risk of colon and rectal cancers using prospective data from the Swedish
Mammography Cohort. The cohort comprised 61,467 women at baseline between 1987 and
1990. Through 2003, the cohort contributed 823,072 person-years, and 504 cases of colon and
237 of rectal cancer occurred. Mean intake of AA through diet was 24.6 |ig/day (Q25-
70 = 18.7-29.9). Coffee (44%), fried potato products (16%), crispbreads (15%), and other
breads (12%) were the greatest contributors. After adjusting for potential confounders, the
authors report no association between estimated AA intake and colorectal cancer. Comparing
extreme quintiles, the adjusted relative risks (95% CI;/? for trend) were for colorectal cancer
0.9 (0.7-1.3;p = 0.80), colon cancer 0.9 (0.6-1.4;/? = 0.83), and rectal cancer 1.0 (0.6-1.8;
p = 0.77). Intake of specific food items with elevated AA (e.g., coffee, crispbreads, and fried
potato products) was not associated with cancer risk.
Wilson et al. (2009b) evaluated possible associations between AA in food and risks of
breast cancer in a cohort of 90,628 registered nurses from the Nurses' Health Study II (United
States) who were premenopausal, had baseline diet information, were without a diagnosis of
cancer before baseline in 1991, and had plausible energy intake. FFQs with more than 130 food
items (including major AA-containing foods) were completed in 1991, 1995, 1999, and 2003.
Newly diagnosed cases of breast cancer were identified in biennial follow-up questionnaires.
Pathology reports confirmed 98% of the self-reported breast cancers. Information on
ER/progesterone receptor (PR) status, available for 916 of the breast cancer cases, indicated that
597 were ER+/PR+ and 196 were ER-/PR-.
During 14 years (945,764 person-years) of follow-up, 1,179 cases of breast cancer were
identified in the cohort of 90,628 premenoposal women. The ages at breast cancer diagnosis
ranged from 26 to 56 years. Mean AA intakes in the lowest and highest quintiles were 10.8 and
73 DRAFT-DO NOT CITE OR QUOTE
-------
37.8 ug/day, respectively. The major contributors to AA intake were French fries (23%), coffee
(15%), cold breakfast cereal (12%), potato chips (9%), and other potatoes (5%). Women in the
highest quintile of AA consumption tended to be current smokers and were less likely to exercise
than women in the lowest quintile. After adjusting for potential confounders, the authors
reported no association between estimated AA intake and risk of breast cancer. The relative risk
(95% Cl',p for trend) of premenopausal breast cancer was 0.92 (0.76-1.II; p = 0.61) for the
highest quintile versus the lowest quintile. Results were similar regardless of ER or PR status of
the tumors, smoking status, and specific AA-containing food type.
Larsson and coworkers conducted a series of prospective studies to evaluate associations
between exposure to AA in food and risks of breast cancer (Larsson et al., 2009d), endometrial
cancer (Larsson et al., 2009a), and epithelial ovarian cancer (Larsson et al., 2009b) in cohorts of
Swedish women (n > 61,000 in each cohort); and colorectal cancer in a cohort of 45,306
Swedish men (Larsson et al., 2009c). The cohorts were cancer free at enrollment in 1987-1990,
completed FFQs at baseline and again in 1997, and were followed for averages of 17.4-17.7
years. The mean daily intake of AA at baseline was 24.6 ug (±7.6, SD) in the female cohorts
and 36.1 ug (±9.6, SD) in the male cohort. After adjusting for potential confounders, the authors
reported no association between estimated AA intake and risk of breast, endometrial, or
epithelial ovarian cancer in the female cohorts, and colorectal cancer in the male cohort.
Comparing extreme quartiles, the adjusted relative risk (95% Cl;p for trend) in the female
cohorts were 1.17 (0.84-1.64; p = 0.76) for breast cancer, 0.96 (0.76-1.21; p = 0.72) for
endometrial cancer, and 0.86 (0.63-1.16; p = 0.39) for total ovarian cancer. In the male cohort,
the adjusted relative risks (95% CI;p for trend) were 0.95 (0.74-1.20;/? = 0.69) for colorectal
cancer, 0.97 (0.71-1.31;/? = 0.78) for colon cancer, and 0.91 (0.62-1.34;/? = 0.78) for rectal
cancer.
Hogervorst and coworkers selected the Netherlands Cohort Study on diet and cancer to
evaluate associations between exposure to AA in food and risks of endometrial, ovarian, and
breast cancer (Hogervorst et al. 2007); renal cell, bladder, and prostate cancer (Hogervorst et al.,
2008a); and gastrointestinal cancer (Hogervorst et al., 2008b). At baseline (1986), the
participants completed a self-administered questionnaire on diet and other cancer risk factors. A
case-cohort approach was used in which cases were enumerated for the entire cohort (consisting
of 58,279 men and 62,573 presumed menopausal women) to provide the numerator information
for estimating incidence rates and randomly-sampled subcohorts (2,589 women in the study of
endometrial, ovarian, and breast cancer; 5,000 cohort members for the studies of other cancer
sites) from the entire cohort at baseline to provide the denominator information for estimating
incidence rates.
Hogervorst et al. (2007), identified 327, 300, and 1,835 cases of endometrial, ovarian,
and breast cancer, respectively during 11.3 years of follow-up. The estimated mean daily intake
74 DRAFT-DO NOT CITE OR QUOTE
-------
of AA was 21 ± 11.9 ug/day in the subcohort. The investigators found no increased risk of
breast cancer, but reported increased risks of postmenopausal endometrial and ovarian cancer
with increasing dietary AA intake, particularly among never-smokers. Comparing the lowest
quintile of AA intake (mean intake 8.9 ug/day) with the highest quintile (mean intake 40.2
ug/day), adjusted hazard rate ratios (HRs) (95% CI; p for trend) were 0.93 (0.73-1.19; p = 0.79)
for breast cancer, 1.29 (0.81-2.07; p = 0.18) for endometrial cancer, and 1.78 (1.10-2.88;/? =
0.02) for ovarian cancer. Among never-smokers, HRs were 1.10 (0.80-1.52; p = 0.55) for breast
cancer, 1.99 (1.12-3.52;p = 0.03) for endometrial cancer, and 2.22 (1.20-4.08; p = 0.01) for
ovarian cancer.
Hogervorst et al. (2008a) identified 339, 1,210, and 2,246 cases of renal cell, bladder, and
prostate cancer, respectively, during 13.3 years of follow-up. The estimated mean daily intake of
AA was 21.8 ± 12.0 ug/day in the subcohort. The investigators found no increased risk of
bladder or prostate cancer, but reported increased risk of renal cell cancer with increasing dietary
AA intake. Comparing the lowest quintile of AA intake (mean intake 9.5 ug/day) with the
highest quintile (mean intake 40.8 ug/day), adjusted HRs were 0.91 (0.73-1.15; p = 0.60) for
renal cell cancer, 1.06 (0.87-1.30;/? = 0.69) for prostate cancer, and 1.59 (1.09-2.30;p = 0.04)
for ovarian cancer.
Hogervorst et al. (2008b) identified 2,190, 563, 349, and 216 cases of colorectal, gastric,
pancreatic, and esophageal cancer, respectively, during 13.3 years of follow-up. The estimated
mean daily intake of AA was 21.7 ± 12.1 ug/day in the subcohort. This study found no
significant association between AA intake and risk of colorectal, gastric, pancreatic, or
esophageal cancer. Comparing the lowest quintile of AA intake with the highest quintile,
adjusted HRs were 1.00 (0.84-1.20;/? = 0.94) for colorectal cancer, 1.06 (0.78-1.45;/? = 0.77)
for gastric cancer, 0.98 (0.68-1.40; p = 0.75) for pancreatic cancer, and 0.83 (0.54-1.30;
p = 0.68) for esophageal cancer.
Cross-sectional neurological evaluations
He et al. (1989) studied 71 workers (45 males and 26 females) between 17 and 41 years
of age who were exposed to AA 8 hours/day, 6 days/week for 1 to 18 months at a factory in
China. A referent group consisted of 33 male and 18 female unexposed workers (17 to 35 years
of age) from the same town. Production of AA was initiated in May 1984, and subjects were
tested in October 1985. Atmospheric concentrations of AA reached 5.56-9.02 mg/m3 between
March and June 1985 during an exceptional increase in production, and decreased to an average
of 0.0324 mg/m3 after July 1985. The workers were evaluated in October 1985. An AA level of
410 mg/L was measured in the water in which three of the workers washed their hands. Clinical
and laboratory examinations included personal interviews to obtain information on demographic
factors, occupational history, symptoms, past illnesses, and family history. Physical and
75 DRAFT-DO NOT CITE OR QUOTE
-------
neurological examinations, visual acuity, and visual field testing, skin temperature
measurements, electrocardiography, and electroencephalography were performed. Laboratory
analysis included routine blood and urine tests, liver function (serum glutamate pyruvate
transaminase and the thymol turbidity test for increased globulin components in sera), serum
hepatitis B surface antigen, serum p-glucuronidase, and immunoglobulins. Sixty-nine of the
exposed workers and 48 of the referent workers were subjected to electroneuromyographic
examinations that included measurements of electrical activity in abductor pollicis brevis and
abductor digiti minimi muscles of the hand, maximal motor nerve conduction velocity in the
lower arm and leg, maximal sensory nerve conduction velocity in the lower arm, and the H-
reflex and Achilles tendon reflex. Statistical methods employed included the %2 test to analyze
symptoms and clinical signs and the Student's t-test to assess electroneuromyographic
parameters. The level of statistical significance wasp < 0.05.
The prevalence of a variety of symptoms reported by the exposed and referent groups is
shown in Table 4-3. Compared to the referent group, significantly greater percentages of the
AA-exposed group reported skin peeling from the hands, anorexia, numbness and coldness in
hands and feet, lassitude, sleepiness, muscle weakness, clumsiness of the hands, unsteady gait,
difficulty in grasping, and stumbling and falling. The authors stated that initial symptoms of
skin peeling were the result of dermal exposure to aqueous AA and that other symptoms
appeared following 3 to 10 months of occupational exposure. Additional statistically significant
signs included greater percentages of exposed workers exhibiting erythema of the hands, sensory
impairments (vibration, pain, and touch sensation), diminished reflexes in biceps, knee, and
ankle, loss of reflexes in the knee and ankle, and intention tremor. Results from visual acuity
and visual field testing were normal.
76 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-3. Neurological symptoms self-reported by acrylamide workers and
nonexposed workers
Symptoms
Skin peeling from the hands
Numbness in the hands and feet
Lassitude
Sleepiness
Muscle weakness
Clumsiness of the hands
Anorexia
Unsteady gait
Coldness of the hands and feet
Difficulty in grasping
Stumbling and falling
Sweating
Dizziness
Cramping pain
Acrylamide group (n = 71)
Number
38
15
14
12
11
8
8
6
6
5
5
27
7
6
Percent
53.5a
21. lb
19.7b
16.9b
15.4b
11. 2a
11. 2a
8.4a
8.4a
7.0a
7.0a
38.0
9.8
8.4
Reference group (n = 51)
Number
2
2
1
0
0
0
1
0
0
0
0
14
2
5
Percent
3.9
3.9
1.9
0
0
0
1.9
0
0
0
0
27.4
3.9
9.8
zp < 0.05.
V < 0.01(X2 test).
Source: He etal. (1989).
Electrical activity, monitored in both the abductor pollicis brevis and abductor digiti
minimi muscles of the hand of 69 exposed workers, revealed denervation potentials
(3/69 exposed workers), prolonged duration of motor units (40/69), increased polyphasic
potentials (29/69), and discrete pattern of recruitment (9/69). These abnormalities were not seen
in the group of 48 referent workers, with the exception of prolonged duration of motor units
(4/48 referents). Significantly increased mean duration and mean amplitude of motor unit
potentials were seen in both the abductor pollicis brevis and abductor digiti minimi muscles of
the exposed group. Twenty-seven of the 69 exposed subjects had neuropathologic signs (e.g.,
impairment of distal sensation or reflexes). When these 27 were excluded from the exposed
group, the remaining 42 subjects (i.e., with no observed neuropathologic signs) still
demonstrated a statistically significant effect of AA exposure on motor unit potentials (with the
exception of mean amplitude in the abductor pollicis brevis muscle). The H-reflex was
nonresponsive in 18 of the 27 exposed subjects with neuropathologic signs and was significantly
longer in mean latency among the 9 subjects in which a reflex was detected. Seventeen of the 27
exposed subjects with neuropathologic signs, and 4 of the 42 exposed subjects without
neuropathologic signs were nonresponsive to the Achilles tendon reflex test. Among the
remaining exposed subjects with (n = 10) or without (n = 38) neuropathologic signs, considered
separately or combined (n = 48), observed Achilles reflexes were significantly longer in mean
latency compared with referent values. Sensory action potentials in the wrist (both median and
ulnar nerves) and sural nerve of the 27 exposed subjects with neuropathologic signs, as well as
the entire group of 69 exposed subjects, were significantly lower in mean amplitude than those
77
DRAFT-DO NOT CITE OR QUOTE
-------
of the referents. Similar measurements in the elbow revealed a significantly lower mean
amplitude in the 27 exposed subjects with neuropathologic signs. Assessment of nerve
conduction velocity, electrocardiography, electroencephalography, and laboratory test results
revealed no statistically significant exposure-related effects.
This study associated abnormalities in nervous activity with occupational exposure to
AA. The results suggest that some measures of abnormal electrical activity may be used to
identify early stages of AA-induced neurotoxicity. However, exposure scenarios were poorly
characterized. Dermal exposure was likely a major source of exposure for at least some of the
exposed workers, as evidenced by numerous reports of peeling of the skin and excessive
sweating of the hands. But inhalation exposure was also likely, based on measurable
concentrations of airborne AA. The study does not include information concerning dose-
response relationships or hemoglobin adduct levels in the group of exposed workers. Nor were
adjustments made for confounding factors such as smoking and exposure to other chemicals.
Calleman et al. (1994) performed a cross-sectional analysis of hemoglobin adduct
formation and neurological effects in a group of 41 factory workers (34 males and 7 females,
aged 18 to 42 years) who were exposed to AA (and acrylonitrile, from which AA is formed) for
1 month to 11.5 years (mean 3 years) during the production of AA in a factory in China. Other
reports on this population include those by Bergmark et al. (1993) who detected GA adducts of
hemoglobin in AA-exposed workers indicating that the transformation of AA to GA occurs in
humans, and by Deng et al. (1993). AA mean exposure concentrations, measured during the
summer of 1991, were 1.07 and 3.27 mg/m3 in the synthesis and polymerization rooms,
respectively. Exposure concentrations measured during the time of collection of biomarker data
(September 1991) were lower, averaging 0.61 and 0.58 mg/m3 in the synthesis and
polymerization rooms, respectively. The exposed group included 13 synthesis workers,
12 polymerization workers, 5 packaging workers, and 6 ambulatory workers, classified
according to their primary work location. The remaining four workers were either exposed for
less than 6 months (two subjects) or had not been exposed to AA during the 4 months preceding
the study. Blood sampling and medical and neurological examinations were performed
approximately 1 hour after a work shift. The beginning of a work shift marked the beginning of
24-hour urine sampling. For vibration sensitivity testing, a referent group consisted of 105
unexposed healthy adults (51 males and 54 females aged 20-60 years). A historical control of
80 persons was used as referent for electroneuromyography tests. A group of 10 nonexposed
male workers from the same city as the exposed group was used as a referent group for
biomarkers of exposure and signs and symptoms of neurotoxicity.
Information regarding demographic factors, smoking and drinking habits, height and
weight, occupational history, past illnesses, current symptoms, and reproductive history were
collected by questionnaire. Vibration sensitivity thresholds were measured in fingers and toes
78 DRAFT-DO NOT CITE OR QUOTE
-------
using the Vibratron II instrument (Deng et al., 1993). Physical and neurological examinations
and electroneuromyographic (ENMG) testing were similar to those described by He et al.
(1989). A neurotoxicity index, with a maximal score of 50, was used to express severity of
peripheral neuropathy (Table 4-4); the information used to derive the score was collected by
questionnaire. The prevalence of specific symptoms was also assessed individually. Biomarkers
of exposure to AA that were reported in the study included free AA in plasma, mercapturic acids
in urine, and the hemoglobin adduct formed by the reaction of AA with the N-terminal valine of
hemoglobin (AAVal).
Table 4-4. Scoring system for the neurotoxicity index
Endpoint
Numbness of extremities
Cramping pain
Loss of position sensation
Loss of pain sensation
Loss of touch sensation
Loss of vibration sensation0
According to tuning fork
Vibration threshold in big toe
Vibration threshold in index finger
Clumsiness of hands
Difficulty grasping
Unsteady gait
Decrease or loss of ankle reflexes
Muscular atrophy
Electroneuromyographic abnormalities'1
Maximum total score
Points"
1
1
2
0, 1,2, or3b
0, I,2,or3b
1
0,1, or 2
0,1, or 2
4
4
4
3 or 5
6
0.5 per abnormality (maximum 6)
50
aPoints were intended to reflect weight given to these observations by a clinical physician diagnosing a peripheral
neuropathy.
bWorkers who had lost their pain or touch sensation were assigned 1 to 3 points depending on the extent of loss:
fingers, hands, or forearms.
°The ratio between the vibration threshold of an individual and that of the corresponding control group with regard
to age was used for scoring vibration sensitivity using the Vibratron instrument. One point was given if this ratio
was 1.5-2.5 for fingers or 1.5-4.0 for toes and 2 points if it was 2.5-5.0 for fingers or 4.0-8.0 for toes.
Abnormalities consisted of measured alterations in electrical activity of selected muscles and nerves.
Source: Calleman et al. (1994).
Statistical analyses included the %2 test to analyze symptoms and clinical signs and the
Student's t-test to assess ENMG parameters. Variance analysis and the <2-test were used in the
comparison of vibration thresholds between the reference group and the exposed group.
Univariate and multivariate linear regression analysis was used to estimate correlation
coefficients and levels of statistical significance for biomarkers of exposure. The level of
statistical significance wasp < 0.05.
Significant differences in vibration threshold were observed among three age subgroups
of referents (<31, 31-40, and >40 years of age). Comparisons of vibration threshold between
AA-exposed workers and referents within these age groupings showed a significant increase in
79 DRAFT-DO NOT CITE OR QUOTE
-------
the exposed workers. Comparison of the results of ENMG measurements between the exposed
workers and the referent group revealed a 10-20% decrease in conduction velocity in the
peroneal and sural nerves and 25-36% increase in latency in median, ulnar, and peroneal nerves
within the exposed group.
The prevalence of symptoms and signs of adverse health effects in the AA-exposed
workers (n = 41) that were not reported in the referent group (n = 10) included statistically
significant incidences of numbness (71%), fatigue (71%), sweating of hands and feet (68%), skin
peeling (59%), loss of pain sensation (54%), loss of touch sensation (46%), dizziness (44%),
anorexia (41%), loss of vibration sensation (41%), and nausea (39%). Other signs and
symptoms that were observed only in the exposed group but were not found to be statistically
different from referents included loss of ankle reflexes (29%), headache (27%), unsteady gait
(22%), loss of knee jerk (20%), unsteady Romberg sign (20%), and loss of triceps and biceps
reflexes (10%).
Group mean biomarker levels and neurotoxicity indices are presented in Table 4-5 for
controls and the work locations of packaging, polymerization, ambulatory, and synthesis. The
average neurotoxicity index scores, as well as the averages of the hemoglobin adduct levels of
AA, decreased with physical distance from the synthesis room where the monomer itself was
handled. This relationship was not reflected by measured free plasma AA, urinary mercapturic
acid, or hemoglobin adduct levels of acrylonitrile or by results of hand or foot vibration
sensitivity measurements or estimates of accumulated in vivo doses of AA. Statistically
significant correlations were reported between each of the biomarkers of exposure and the
calculated neurotoxicity indices, with the exception of free plasma AA concentrations.
80 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-5. Group means ± SD of biomarkers in different categories of
workers
Controls
Packaging
Polymerization
Ambulatory
Synthesis
Free AAa
(umol/L)
0.92
2.2
1.3
2.0
1.8 ±0.8
Merc. ac.b
(umol/24
hours)
3± 1.8
93 ±72
58 ±75
53 ±35
64 ±46
AAValc
(nmol/g)
0.0 ±0.0
3. 9 ±2.5
7.7 ±3.4
9.5 ±7.3
13.4 ±9.8
ANVald
(nmol/g)
0.23 ±0.18
19.1 ±5.7
19.1 ±12.9
16.3 ±3.7
19.5 ±7.6
AccD,4/
(mM/hour)
0.0 ±0.0
8.1 ±6.6
27.0 ±23. 9
37.6 ±21.9
68.3 ±64.2
NInf
0.0 ±0.0
8.9 ±9.1
10.0 ±5. 8
11.3 ±9.8
19.2 ±10.6
Tree plasma AA.
bUrinary mercapturic acid.
'Hemoglobin adduct between N-terminal valine and AA.
dHemoglobin adduct between N-terminal valine and acrylonitrile.
Predicted cumulative in vivo AA dose (based on rates of AA-hemoglobin adduct formation in human globin
hydrolysates and mean AA exposure concentrations measured in areas of polymerization and synthesis by station
sampling) (see Section 3.1 and Bergmark et al. [1993] for additional information).
fNeurotoxicity index.
Source: Calleman et al. (1994).
A principal finding of the study of Calleman et al. (1994) was the strong correlation
between hemoglobin adduct levels of AA and neurological impairment (Table 4-6), as assessed
by a combined index of self-reported symptoms and clinically assessed effects. No significant
correlation was found between free plasma AA levels and neurotoxicity index, but significant
correlations were found between neurotoxicity index and the other markers of exposure indicated
in Table 4-5. The data provide a description of the relationship between an internal measure of
dose (hemoglobin adducts) from repeated exposure to AA (1 month-11.5 years; mean = 3 years)
and an index of neurological impairment. Quantitative assessment of contributions of dermal
and inhalation exposure were not made, although in the synthesis area of the factory where
neurological symptoms were most severe, dermal exposure was considered to have been the
major exposure route.
81
DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-6. Correlation coefficients (linear regression) for relationships
between biomarkers and neurotoxicity index
X variable
Free AAa
Merc. ac.b
AAVaf
ANVald
AccDAAe
Y variable
NInf
NIn
NIn
NIn
NIn
Correlation coefficient
0.15
0.42
0.67
0.69
0.60
p-Value
0.31
0.01
O.001
O.001
0.001
"Free plasma AA.
bUrinary mercapturic acid.
'Hemoglobin adduct between N-terminal valine and AA.
dHemoglobin adduct between N-terminal valine and acrylonitrile.
Predicted cumulative in vivo AA dose (based on rates of AA-hemoglobin adduct formation in human globin
hydrolysates and mean AA exposure concentrations measured in areas of polymerization and synthesis by station
sampling) (see Section 3.1 and Bergmark et al. [1993] for additional information).
fNeurotoxicity index.
Source: Calleman et al. (1994).
Hagmar et al. (2001) performed a health examination on a group of 210 tunnel
construction workers who had been occupationally exposed for 2 months to a chemical grouting
agent containing AA and N-methylolacrylamide. Workers were expected to have experienced
dermal as well as inhalation exposure. The workers were exposed to the grouting agent for
55 days (August 4 through September 30, 1997), after which exposure was stopped due to the
development of neurological symptoms in cows that drank water from a creek that contained
leakage water from the tunnel. One week after grouting stopped, 210 workers (of 242 total
workers) agreed to participate in the study. Venous blood samples were drawn and
questionnaires and physical examinations were administered 1-5 weeks after exposure was
stopped. Quantitative exposure data were limited to two personal air samples showing
concentrations of 0.27 and 0.34 mg/m3 for the sum of AA and NMA; further analysis suggested
that the air contained a 50:50 mixture of these compounds. Workers were classified by exposure
level. The levels were designated as "high" (103 subjects who had injected the grouting agent),
"some" (89 subjects), or "none" (18 subjects without obvious exposure), based on self-reported
exposure. The health examination included an extensive questionnaire and a physical
examination that included unspecified tests of peripheral nerve function. Blood samples for the
analysis of adducts of AA with N-terminal valines in hemoglobin were drawn within a month
after construction work was completed. A group of 50 subjects who claimed recently developed
or deteriorated peripheral nervous function at the initial physical examination was subjected to
more detailed neurophysiologic examinations and 6-month follow-up clinical (n = 29) and
neurophysiological (n = 26) examinations. Those with remaining symptoms were examined for
up to 18 months postexposure.
An important caveat in interpreting the hemoglobin adduct data relative to neurotoxic
responses to AA in the Hagmar et al. (2001) study is that both AA and NMA form the same
82 DRAFT-DO NOT CITE OR QUOTE
-------
N-(2-carbamoylethyl)valine adduct in hemoglobin. Fennell et al. (2003) measured levels of this
adduct following separate exposure to equimolar doses of AA and NMA to rats and reported
formation of 21 ±1.7 pmol/mg globin from AA and 41 ± 4.9 pmol/mg from NMA (mean ± SD,
n = 4). Since the levels of adduct formation were not comparable and there is no way to
distinguish whether the N-(2-carbamoylethyl)valine arose from reaction of hemoglobin with AA
or with NMA, conclusions about AA exposure (with adducts as the surrogate for internal
exposure) vs. responses are confounded by not being able to reliably distinguish the AA internal
dose from the NMA internal dose in humans.
Hemoglobin adduct levels for 18 nonsmoking unexposed reference subjects varied
between 0.02 and 0.07 nmol/g globin. Adduct levels in 47 of the 210 tunnel workers did not
exceed the highest level of the referents. The remaining workers were divided into three
categories according to adduct levels as follows: 89 with 0.08-0.29 nmol/g globin, 36 with 0.3-
1.0 nmol/g globin, and 38 with 1.0-17.7 nmol/g globin. The study authors noted a significant
(p < 0.05) association between self-reported exposure categories and adduct levels.
Clear relationships (statistically significant trend tests) were found between increasing
levels of hemoglobin adducts and increased incidences of self-reported symptoms of peripheral
neurological impairment and irritation of the eyes. Statistically significant positive correlations
(p < 0.05) between prevalence of peripheral nervous symptoms, irritant symptoms, and
symptoms of general discomfort with adduct levels were found. For example, in the groups with
adduct levels <0.08 nmol/g globin, 0.08-0.29 nmol/g globin, 0.3-1.0 nmol/g globin, and
>1.0 nmol/g globin, incidences of reported numbness or tingling in the feet or legs were
2/47 (4%), 10/89 (11%), 9/36 (25%), and 14/38 (37%), respectively. This symptom is consistent
with peripheral nervous impairment and was noted with the highest frequency among the
reported symptoms in this study. Irritant symptoms and symptoms of general discomfort
typically disappeared following the end of a workday, whereas peripheral nervous symptoms
persisted. Follow-up examinations revealed that 58% of the subjects with early signs of
impaired peripheral nervous function improved, while only 4% showed signs of deterioration.
Table 4-7 summarizes the symptoms showing the greatest increases in incidences with
increasing hemoglobin adduct levels.
83 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-7. Incidences of symptoms in 210 tunnel workers classified into
exposure groups based on levels of hemoglobin adducts of acrylamide
Symptoms with trend test p-value <0.001
Numbness/tingling in feet or legs
Leg cramps
Eye irritation
Nose irritation
Throat irritation
Coughing
Headache
Hemoglobin adducts of acrylamide (nmol/g globin)a
<0.08
2/47 (4)
3/47 (6)
6/47 (14)
6/47 (14)
4/47 (10)
4/47 (10)
6/47 (14)
0.08-0.29
10/89(11)
6/89 (7)
19/87 (23)
17/89 (21)
19/89 (23)
9/89(11)
27/89 (33)
0.30-1.00
9/36 (25)
2/36 (6)
17/36 (47)
13/36 (36)
17/36 (47)
11/36(31)
11/36(31)
1.00-17.7
14/38 (37)
10/38 (26)
29/38 (76)
20/38 (53)
28/38 (47)
19/38 (50)
24/38 (63)
"Percentages of workers reporting symptoms are noted in parentheses.
Source: Hagmaretal. (2001).
The principal findings of the study of Hagmar et al. (2001) are the positive correlations
between measures of exposure (hemoglobin adducts) and self-reported symptoms of
neurological impairment. Pairwise comparisons (Fisher's Exact test performed by Syracuse
Research Corporation) between the group of subjects with adduct levels <0.08 nmol/g globin and
each of the three groups with higher adduct levels (0.08-0.29, 0.30-1.00, and >1.00 nmol/g
globin) show statistically significantly (p < 0.05) increased prevalence of numbness or tingling
in the feet or legs for the two higher exposure groups, but not in the group with lower adduct
levels (0.08-0.29 nmol/g globin). This analysis indicates that an adduct level in the range of
0.08-0.29 nmol/g globin was the NOAEL, and 0.30-1.00 nmol/g globin was the LOAEL, for
self-reported symptoms of AA-induced peripheral neuropathy. Limitations of this study, with
respect to describing dose-response relationships for chronic exposure to AA, are the relatively
short period (2 months) of occupational exposure to AA, the possible confounding contribution
of NMA to the noted effects, and the fact that both AA and NMA form the same N-terminal
valine hemoglobin adduct (Fennell et al., 2003) that was used as an internal measure of dose.
Myers and Macun (1991) investigated peripheral neuropathy in a cohort of 66 workers in
a South African factory that produced polyacrylamide. The investigation followed clinical
diagnosis of peripheral neuropathy in five workers at the factory. The workforce was divided
into a number of exposure categories, based on environmental sampling and discussions with
workers. Exposure levels for the various tasks ranged from 0.07 to 2.5 times the National
Institute of Occupational Safety and Health (NIOSH) recommended exposure limit (REL) of
0.3 mg/m3. Workers were then classified as being exposed to airborne AA when exposure levels
exceeded the REL (n = 22), and unexposed when exposure levels were below the REL (n = 41).
Workers completed a questionnaire that was designed to capture social, medical, and
occupational history. A standard blind neurological examination was also performed.
The mean age of the subjects was 30 years and the mean length of service 24 months; no
significant differences were seen for these variables between exposed and unexposed groups.
84
DRAFT-DO NOT CITE OR QUOTE
-------
The exposed group showed higher prevalences of abnormalities for all symptoms (weakness,
sensation, balance, fatigue, visual, loss of weight, urogenital, and fingertip skin), most signs
(fingertip effects, light touch, tactile discrimination, pain), and reflexes, coordination, motor
weakness, gait, and Rombergism. Statistically significant differences between exposed and
unexposed groups for individual effects were seen only for abnormal sensation symptoms and
signs in fingertip skin (including color, peeling, and sweating). The overall prevalence of
AA-related abnormalities (inclusive) among the exposed was 66.7%, which was statistically
significantly higher (p < 0.05) than that of the unexposed group (prevalence of 14.3%). The
authors stated that most workers observed to have abnormalities (number not reported) were
employed in areas where exposures were highest (1.6-2.5 times the REL).
Bachmann et al. (1992) performed a follow-up investigation in July 1990 at the same
South African factory that had been examined in 1986 by Myers and Macun (1991). The study
design was similar to that of Myers and Macun (1991) but included measurements of vibration
sensation threshold with a Vibratron II vibration sensation tester that was not available in the
earlier investigation. Among 82 workers employed at follow-up, increased prevalences of
symptoms of tingling and numbness in hands and feet, weakness and pain in arms and legs,
peeling hand skin, and sweating hands were reported by exposed workers, compared with those
classified as being unexposed. The symptoms of numbness, limb pain, and peeling and sweating
of hands were statistically significantly increased in exposed workers. Results of clinical
examinations provided supporting evidence for the reported increased symptoms of peeling and
sweating of the hands. No gross neurological abnormalities were found. Mean vibration
sensation thresholds were similar among unexposed and exposed groups, even when adjusting
for age, and no association was found between vibration thresholds and any symptoms.
The studies of Myers and Macun (1991) and Bachmann et al. (1992) show an association
between occupational exposure to AA above the NIOSH REL of 0.3 mg/m3 and signs and
symptoms of mild neuropathy. However, in the absence of more reliable measures of exposure
(e.g., hemoglobin adduct levels), meaningful effect levels were not established.
Case reports
Numerous case reports have been published in which exposure to AA, predominantly in
occupational settings, has been associated with observed cutaneous and neurological effects
ranging from dermal effects, such as peeling of skin in fingertips, to numerous signs of impaired
neurological performance in peripheral nervous system and CNS (Gjerl0ff et al., 2001; Mulloy,
1996; Dumitru, 1989; Donovan and Pearson, 1987; Kesson et al., 1977; Mapp et al., 1977;
Davenport et al., 1976; Igisu et al., 1975; Takahashi et al., 1971; Fullerton, 1969; Auld and
Bedwell, 1967; Garland and Patterson, 1967). Although these reports provide supportive
85 DRAFT-DO NOT CITE OR QUOTE
-------
evidence of AA-induced neurotoxicity, they lack information regarding primary exposure routes
and exposure-response relationships.
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION
4.2.1. Oral Exposure
The standard bioassay database for subchronic and chronic oral exposures to AA consists
of one 90-day drinking water study in F344 rats (Burek et al., 1980) that demonstrated
neurotoxicity and two 2-year drinking water studies in F344 rats with the main effects being
neurotoxicity and cancer (Friedman et al., 1995; Johnson et al., 1986, 1984).
4.2.1.1. Subchronic Studies
Neurotoxic effects
Burek et al. (1980) administered AA to groups of 6-week-old male (23-29/group) and
female (10/group) F344 rats in the drinking water for up to 93 days at concentrations designed to
result in AA intakes of 0, 0.05, 0.2, 1, 5, or 20 mg/kg-day. Ten rats/sex/group were assigned to
the basic 90-day study and were observed for body weight and water consumption (recorded
weekly) throughout the treatment period. Following 7 and 33 days of treatment, three control
and three high-dose male rats were sacrificed for interim electron microscopic examination of
the sciatic nerve. Ten male (nine in the high-dose group, due to one death prior to treatment
termination) and all female rats from each treatment group were subjected to gross and
histopathologic examination of all major organs and tissues at the end of the treatment period, at
which time three other male rats from each group were processed for electron microscopic
examination of the sciatic nerve. The remaining rats (all males) in each group were observed for
signs of recovery from treatment-related effects for up to 144 days following cessation of
treatment. Three rats/group were subjected to microscopic examination of the sciatic nerve on
days 25 and 111 posttreatment. Body weights were recorded for two rats/dose level prior to
sacrifice on recovery day 111. At the end of the 144-day recovery period, the remaining four
rats of each dose level were weighed and sacrificed for gross and histopathologic examination of
all major organs and tissues. Three of these rats were processed for electron microscopic
examination of the sciatic nerve.
All rats were observed daily (during the 5 day workweek) for general health and clinical
signs. Hindlimb foot splay was measured weekly in four control and four high-dose (20 mg/kg-
day) male and female rats until the onset of neuropathy was detected, after which neuropathy in
the high-dose group was monitored by clinical signs. After neuropathy was detected in high-
dose rats, male and female rats in the 5 mg/kg-day dose groups were also subjected to weekly
testing of foot splay (rats in the lower treatment groups were not tested due to the lack of
86 DRAFT-DO NOT CITE OR QUOTE
-------
response at 5 mg/kg-day). Blood samples collected from seven rats/sex in the control and high-
dose groups on treatment day 76 and from all rats alive on day 60 of the recovery period were
examined for packed cell volume, total erythrocyte count, total and differential leukocyte counts,
and hemoglobin concentration. The study design included urinary sampling from 10 control and
10 high-dose rats per sex on treatment day 76 and at the end of the treatment period. Blood
serum was collected from the 10 rats/sex/dose that were sacrificed at the end of treatment and
from the 4 male rats/group that were maintained throughout the 144-day recovery period. Blood
urea nitrogen, alkaline phosphatase, serum glutamic pyruvic transaminase, and serum
cholinesterase activity were determined.
Light microscopic examinations were performed on brain, spinal cord, and peripheral
nerves (including brachial plexus, sciatic, and femoral nerves) that had been fixed in
glutaraldehyde-paraformaldehyde and stained with hematoxylin eosin. Additional sections of
brain, spinal cord, and peripheral nerves were subjected to the luxol fast blue-periodic acid
Schiff (LFB/PAS) reaction for myelin staining and to Bodian's stain to elucidate more subtle
axonal changes. Myelin and axonal degeneration was classified as severe (degeneration in
approximately 50% of the observed fibers), moderate (degeneration in 20-50% of observed
fibers), slight (degeneration in less than 20% of observed fibers), very slight (effects restricted to
focal or multifocal changes in individual nerves), or equivocal (nerves could not be graded as
clearly normal). Only the sciatic nerve was examined by electron microscopy. Three blocks of
sciatic nerve fibers, two longitudinal and one transverse, were selected per rat for thin sectioning
and ultrastructural analysis. Ultrastructural alterations were counted by examining a maximum
of 50 fields per block, a field defined as a section through any Schwann cell. This resulted in an
examined maximum of 150 fields/rat or 450 fields/treatment group of three rats.
Hematology, urinary and clinical chemistry parameters, body weights, organ-to-body
weight ratio data, foot spread results, and water consumption were statistically analyzed by one-
way analysis of variance followed by Dunnett's test. The level of significance chosen was
p < 0.05. The study report did not, however, include individual or averaged incidences or extent
of changes in these parameters, so an independent analysis of the results of body and organ
weights, water consumption, foot splay, hematology, urinalysis, or serum chemistry was not
possible.
Significantly lower body weights were reported in male and female rats of the 20 mg/kg-
day group relative to controls: 8% lower (males and females) on treatment days 13 and 20, and
21 and 24% lower (males and females, respectively) on treatment day 91. No significant body
weight effect was seen in rats of lower dose groups. At the 20 mg/kg-day dose level, treatment-
related effects on organ weights included significantly decreased absolute liver, kidney, and
thymus weights in males (also testicular) and females, significantly decreased absolute brain and
heart weights in females (trend for decreased weights in males), increased relative brain, heart,
87 DRAFT-DO NOT CITE OR QUOTE
-------
liver, and kidney weights in males and females, and decreased relative thymus (females only)
and testicular weight in males. Absolute and relative liver weight was increased in 5 mg/kg-day
males. Marginally statistically significant increases in relative heart weight in 0.05 and
0.2 mg/kg-day females were not considered to be of toxicological significance due to the lack of
a dose response. Female rats of the 20 mg/kg-day dose level exhibited significantly decreased
water consumption (15-39% decreased) between treatment days 20 and 90. Although decreased
water consumption was noted in high-dose males, the decrease reached the level of statistical
significance in only 4 of the 13 intervals recorded. The few instances of significantly increased
water consumption in low-dose rats did not follow a consistent pattern or trend, and may be of
no toxicological significance. By day 144 of the posttreatment recovery period, the high-dose
group had recovered with higher (but not statistically significant) body weights than controls,
significantly higher absolute liver and kidney weights, as well as significantly higher relative
brain and liver weights.
Significantly increased instances of hindlimb foot splay were observed in 20 mg/kg-day
male and female rats on treatment day 22 (incidences were not reported), which became more
pronounced on treatment day 29. Foot splay testing was terminated with this treatment group (to
prevent injury), but clinical signs of neuropathy (including curling of the toes, rear limb splay,
incoordination, and posterior weakness) progressed in severity throughout the remainder of the
treatment period. Beginning on treatment day 29, rats of the 5 mg/kg-day dose level were tested,
but foot splay was not detected at this treatment level in either males or females. No other
treatment-related clinical effects were observed in the 5 mg/kg-day males or females or any of
the lower dose groups. By day 7 of the posttreatment recovery period, the 20 mg/kg-day groups
showed cleared signs of improvements continuing to day 111 with only slight posterior
weakness and curling of the toes. By day 144, these high dose treated rats appeared clinically
similar to the controls.
At the end of the treatment period, serum cholinesterase activity was increased and
alkaline phosphatase activity was statistically significantly increased in 20 mg/kg-day females.
Significant decreases in packed cell volume, total erythrocyte count, and hemoglobin
concentrations in 20 mg/kg-day males and females and 5 mg/kg-day females were noted.
Results of urinalysis did not reveal any AA-induced abnormalities. By day 144 posttreatment,
the 20 mg/kg-day group (sex not specified) had statistically significant decreased serum
cholinesterase levels and no significant differences in other clinical chemistry parameters.
Upon necropsy, gross observations of rats following the 92- or 93-day treatment period
revealed treatment-related alterations only in the 20 mg/kg-day treatment group, including
perineal soiling, decreased adipose tissue, decreased liver size, darkened kidneys, foci or mottled
appearance of lungs, decreased size or flaccid testicles, decreased size of male accessory
genitalia, decreased uterus size, altered appearance of peripheral nerves, atrophy of skeletal
88 DRAFT-DO NOT CITE OR QUOTE
-------
muscle in the posterior portion of the body, bladder distention, and diffuse mural thickening of
the stomach. The authors did not include incidence data regarding gross examination data,
however. Histopathologic examination at the 20 mg/kg-day treatment level revealed effects such
as atrophy of skeletal muscle (2/10 males, 8/10 females), slightly increased hematogenous
pigment in the spleen (4/9 males), ulcerative gastritis or hyperkeratosis in the nonglandular
stomach (4/10 males), atrophy of mesenteric fat (8/10 females), vacuolization of the smooth
muscle in the bladder wall (1/10 males, 2/9 females), inflammation in the lungs (3/10 males,
5/10 females), and testicular effects that included atrophy (10/10), mineralization in seminiferous
tubules (5/10), and increased cellular debris and/or decreased spermatogenic segments in the
tubular lamina of the epididymides (9/10). The statistical significance of these findings could
not be assessed because incidence data for controls were not reported. By day 144
posttreatment, only the high dose rats had persistent gross pathological effects, primarily dark
testicles and slightly distended bladders. The testicular histological lesions consisted of focal or
multifocal atrophy to individual seminiferous tubules, some with mineral and cellular debris, and
indication of partial reversibility of the testicular atrophy.
Results of sciatic nerve examinations using light and electron microscopy are
summarized in Table 4-8. Light microscopic examination of the sciatic nerve sections (stained
with hematoxylin and eosin) revealed severe degeneration in the 20 mg/kg-day group that was
characterized by demyelinization (LFB/PAS-treated sections) and axonal degeneration
(Bodian's-treated sections) in 10/10 females and similar but less severe effects in males
(degeneration moderate in 5/10 and severe in the other 5). These lesions were also seen in other
peripheral nerve sections (brachial plexus and femoral nerve) but varied in severity from
equivocal to severe (incidences not reported). The authors noted equivocal to very slight
degenerative changes in peripheral nerves of 5 mg/kg-day males (9/10) and females (6/10) but
found no light microscopic evidence of peripheral nerve lesions in 0.05, 0.2, or 1 mg/kg-day
treatment groups. Very slight to slight degenerative changes (demyelinization, swollen
astrocytes and axons) were seen in spinal cord sections of 20 mg/kg-day male (5/10) and female
(9/10) rats. No treatment-related lesions were observed at any dose level within brain sections
examined by light microscopy. After 144 days of posttreatment recovery no nerve tissue
alterations were observed in any of the 5 mg/kg-day or lower dose groups. In the high dose
group, alterations ranged from very slight to slight in the sciatic nerve and no alteration in
sections of the brachial nerve. The authors stated that if the recovery period had been extended
beyond 144 days, the remaining tissue changes would likely have completely reversed.
89 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-8. Light and electron microscopic data for left sciatic nerves from
rats exposed to acrylamide in drinking water for 90 days
Endpoint
Electron microscopy
Number of rats (only males were examined)
Total fields examined
Axolemma invaginations
Axolemma invaginations with cell
organelles and/or dense bodies
Schwann cells without axons and/or with
degenerating myelin
Incidence of fields with any alteration
Light microscopy
(10 rats/ sex/ dose were examined)
Moderate to severe degeneration
Female
Male
Equivocal to very slight degeneration
Female
Male
Dose (mg/kg-day)
0
3
450
36
32
0
68/450
0/10
0/10
0/10
0/10
0.05
3
450
24
15
0
39/450
0/10
0/10
0/10
0/10
0.2
3
350
27
17
0
44/350
0/10
0/10
0/10
0/10
1
3
453
30
78
0
108/453
0/10
0/10
0/10
0/10
5
3
443
33
109
7
149/443
0/10
0/10
6/10
9/10
20
3
435
8
48
183
239/435
10/10
10/10
0/10
0/10
Source: Bureketal. (1980).
Electron microscopic examinations of sciatic nerve preparations from three male
rats/group included the examination of fields (defined as a section through any Schwann cell) for
signs of axolemma invaginations, axonal invaginations with cell organelles and/or dense bodies,
and Schwann cells without axons and/or with degenerating myelin. After 7 days of treatment, no
significant differences were seen between control and 20 mg/kg-day rats (other treatment groups
were not subjected to 7-day interim sacrifice). After 33 days of treatment, 20 mg/kg-day male
rats exhibited increased prevalence of fields showing axolemma invaginations with cell
organelles and/or dense bodies and fields exhibiting Schwann cells without axons and/or with
degenerating myelin (other groups were not subjected to 33-day interim sacrifice). Following
90 days of treatment, severe axonal degeneration and axonal loss were seen at the 20 mg/kg-day
dose level. Approximately 55% of the fields examined exhibited alterations in myelinated
nerves or Schwann cells (compared with 12 and 21% after treatment days 7 and 33,
respectively). Similar, but less severe, ultrastructural alterations in approximately 34% of the
fields examined were seen in the 5 mg/kg-day dose group. At the 1 mg/kg-day dose level,
approximately 24% of the fields examined showed axolemma invaginations with or without cell
organelles and/or dense bodies, but not more severe signs of ultrastructural alterations. The
alterations in the sciatic nerve fields examined in the control, 0.05, and 0.2 mg/kg-day groups
were roughly comparable (15, 9, and 12%, respectively), suggesting that there were no adverse
effects at the 0.05 and 0.2 mg/kg-day doses. Importantly, the increase in lesions observed via
electron microscopy in the 1 and 5 mg/kg-day groups appeared to have completely reversed by
90
DRAFT-DO NOT CITE OR QUOTE
-------
days 25 and 111 posttreatment, respectively. The observed lesions in the 20 mg/kg-day group
were partially or completely reversed by day 144 posttreatment.
In summary, the 90-day toxicity study of F344 rats exposed to AA in the drinking water
(Burek et al., 1980) identified a NOAEL of 0.2 mg/kg-day and a LOAEL of 1 mg/kg-day, based
on ultrastructural degeneration (axolemma invaginations with or without cell organelles and/or
dense bodies) in the sciatic nerve of male rats (as detected by electron microscopic examinations,
which were limited to males). The increased frequency was characterized by the study authors
as "slight" for the LOAEL at 1 mg/kg-day, and the lesions were reversible (back to control
levels) by day 25 posttreatment in all 1 mg/kg-day treated rats. At the resolution of the light
microscope, the 5 mg/kg-day dose was the lowest dose resulting in degenerative effects in the
sciatic nerve of male and female rats.
4.2.1.2. Chronic Studies
Johnson et al. (1986, 1984) study
Johnson et al. (1986, 1984) conducted a chronic toxicity and carcinogenicity study in
which groups of F344 rats (90/sex/treatment group) were administered AA in the drinking water
at concentrations calculated to provide AA doses of 0, 0.01, 0.1, 0.5, or 2.0 mg/kg-day for up to
2 years. Ten rats/sex/treatment group were randomly selected for interim sacrifices after 6, 12,
or 18 months of treatment. Rats were observed twice daily on workdays for clinical signs and
examined monthly for palpable masses. Individual body weights were recorded monthly and
fasting body weights were measured at scheduled necropsy. Based on body weight and water
consumption data from a subgroup of 20 rats/treatment group, recorded weekly for the first
3 months and monthly (water consumption measured for 1 week each month) thereafter,
concentrations of AA in the drinking water were adjusted to maintain target doses for the
remaining rats of each treatment group. During the final 6 months of treatment, mean group
weights of all rats, rather than those of the subgroup, were used in calculating the concentrations
of AA required to maintain target treatment levels.
Blood and urine were collected randomly from 10 rats/sex/group at 3 months and just
prior to 6-, 12-, 18-, and 24-month scheduled necropsies. Hematological parameters investigated
included packed cell volume, hemoglobin, total erythrocytes, leukocyte count, platelet count, and
red cell indices. Stained blood smear examinations and differential leukocyte counts were
conducted. Urine was analyzed for specific gravity, pH, protein, glucose, blood, ketones,
bilirubin, and urobilinogen. During necropsy, blood serum was collected and analyzed for
concentrations of glutamic-pyruvate transaminase, alkaline phosphatase, blood urea nitrogen,
total protein, albumin, glucose, and cholinesterase.
Complete postmortem gross pathologic examinations were performed on all rats in the
study. Organ-to-body weight ratios were calculated for brain, heart, liver, kidneys, and testes.
91 DRAFT-DO NOT CITE OR QUOTE
-------
Representative sections from all major organs and tissues were stained with hematoxylin and
eosin and subjected to histopathologic examination. Light microscopic examinations were
performed on sections of three separate peripheral nerves (tibial nerve and two unspecified
nerves), three locations of the spinal cord, and six sections through the brain and olfactory bulbs
that had been stained with hematoxylin and eosin.
Cumulative mortality data were analyzed by the Gehan-Wilcoxon test. Analysis of
variance and Dunnett's t test were used to analyze body weight data, clinical chemistry,
hematology, urine specific gravity, and organ weight. Cumulative incidence of microscopic
pathologic findings was analyzed by Fisher's Exact probability test. For observations with a
control incidence of at least 6%, a Bonferonni correction for multiple treatment-control
comparisons was applied. In the absence of a positive Fisher's Exact test for a microscopic
lesion, the Cochran-Armitage test for linear trend was performed. Supplemental mortality-
adjusted tests of Peto, and the analogous extension of the Cochran-Armitage test, were
performed when deemed appropriate. The level of significance chosen for all tests wasp < 0.05.
Additional groups of rats (18/sex/group) were added to the study for independent
assessment of neurohistopathologic effects (the results of this portion of the study were reported
by Johnson et al., 1985). Three rats/sex were sacrificed at each scheduled interim examination
(3, 6, 12, and 18 months) and at terminal sacrifice (24 months). An additional three rats/sex/dose
were placed on study to provide for adequate number of rats at the 24-month sacrifice. All
survivors were sacrificed at 24 months. Both light and electron microscopic examinations were
performed on nerve tissue samples taken from the same regions as those described above. As in
the Burek et al. (1980) study, preparations for light microscopy included the use of LFB/PAS
reaction for myelin staining and Bodian's stain to elucidate more subtle axonal changes.
Nonneoplastic results—primarily neurotoxicity
Incidence data were presented only for mortality and tibial nerve degeneration (at
terminal necropsy). Other nonneoplastic results were typically described according to statistical
comparison with controls, but the report did not include incidence or mean data.
Based on water consumption data, AA doses varied from 94 to 105% of target levels.
Cumulative mortality data showed no apparent dose-related effect before 21 months of
treatment, after which the 2.0 mg/kg-day group (especially females) exhibited increasing
mortality that was significantly higher than controls after 24 months of treatment (approximately
32% in 2.0 mg/kg-day females vs. 20% in control females and 41% in 2.0 mg/kg-day males vs.
26% in control males). Beginning on treatment day 89, mean body weight of 2.0 mg/kg-day
males was significantly lower (about 2%) than controls. By the end of the study, the difference
had increased to approximately 4%. No consistent significant treatment-related body weight
effects were seen in 2.0 mg/kg-day females or rats of either sex from lower dose groups. There
92 DRAFT-DO NOT CITE OR QUOTE
-------
were no treatment-related effects on food or water consumption. Clinical observations,
hematology, clinical chemistry, and urinalysis did not reveal any indications of treatment-related
effects in any treatment group. On study day 210, some male and female rats from all dose
groups exhibited excessive lacrimation and enlarged salivary glands consistent with
sialodacryoadenitis virus infection. Both males and females appeared to be equally affected, and
the symptoms resolved within about 10 days.
Light microscopic examination of peripheral nerve section revealed degenerative changes
that consisted of focal swelling of individual nerve fibers with fragmentation of the myelin and
axon and formation of vacuoles containing small round eosinophilic globules and macrophages.
The study authors graded nerve degeneration as very slight, slight, moderate, or severe but did
not further characterize the grading scheme. "Minimal" tibial nerve degeneration was observed
in control and all treated groups beginning at the 12-month necropsy. Although the report
indicated that 12-month assessment revealed increases in both incidence and degree of
degeneration in the 2.0 mg/kg-day group, particularly the males, the actual data were not
presented, precluding an independent analysis of the findings. Incidences of nerve degeneration
increased in controls and treated groups alike throughout the remainder of the treatment period.
Table 4-9 summarizes the light microscopic findings in tibial nerve sections of the groups of rats
from the main study that were treated for 2 years. There were no indications of significant
effects on incidence of very slight or slight degeneration in control or treated males or females.
There was a statistically significant trend towards increased moderate and severe degeneration in
tibial nerves of male rats up to the 2.0 mg/kg-day dose level, although the increase for the pooled
moderate-to-severe data at the high dose was not statistically different from controls. There was
a statistically significant increase in pooled incidence of slight-to-moderate degeneration in tibial
nerves for female rats at 2.0 mg/kg-day.
93 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-9. Light microscopic data for tibial nerves from F344 rats exposed
to acrylamide in drinking water for 2 years
Endpoint
Males
Number of rats examined
Within normal limits
Degeneration
Very slight
Slight
Moderate
Severe
Moderate + severe
Females
Number of rats examined
Within normal limits
Degeneration
Very slight
Slight
Moderate
Slight + moderate
Dose (mg/kg-day)
0
60
2
30
19
8
1
9
60
12
45
o
6
0
o
J
0.01
60
o
J
29
22
5
1
6
60
10
43
7
0
7
0.1
60
4
23
21
12
0
12
60
10
45
5
0
5
0.5
60
o
J
25
19
13
0
13
60
11
42
7
0
7
2
60
4
19
21
12
4
16a'b
61
8
37
13
3
16c'd
aThe data for moderate and severe degeneration were pooled due to low incidence.
blndicates a linear trend by the Mantel-Haenszel extension of the Cochran-Armitage test (p < 0.05) for
pooled moderate and severe degeneration. Note no statistical significance for the high dose group.
°The data for slight and moderate degeneration were pooled due to low incidence.
dStatistically different from control group, mortality adjusted via Mantel-Haenszel procedures (p < 0.05).
Source: Johnson etal. (1986).
Electron microscopic examinations of peripheral nerve sections from rats in the groups
destined for independent neuropathologic assessment revealed slightly increased incidences of
axolemma invaginations in 2 mg/kg-day male (but not female) rats, relative to controls, at 3- and
6-month interim sacrifices. There were no indications of treatment-related degenerative effects
at lower treatment levels. At 12-month interim examination, degenerative myelin and axonal
changes were observed in controls as well as all treatment groups and were considered to be the
result of aging. High background incidences of degenerative changes at 18 and 24 months
precluded the usefulness of electron microscopic analysis to detect differences between control
and exposed groups.
In summary, the most significant noncancer chronic effects observed in F344 rats
exposed to AA in the drinking water for 2 years (Johnson et al., 1986, 1985, 1984) were
increased incidences of axolemma invaginations (observed by electron microscopy) in the tibial
branch of the sciatic nerve of male rats following 3 and 6 months of treatment and increased
prevalence of "moderate" to "severe" degeneration (observed by light microscopy) in both males
and females following 2 years of treatment. A NOAEL for these neurological effects was
identified at 0.5 mg/kg-day, and a LOAEL was identified at the 2.0 mg/kg-day dose level.
94
DRAFT-DO NOT CITE OR QUOTE
-------
Neoplastic results—tumors at multiple sites
Until the last few months of treatment, observations of palpable masses were infrequent.
The authors noted that rats dosed at 2.0 mg/kg-day appeared to have slightly increased
incidences of palpable masses during the last 4 months of treatment, most of which were
subsequently identified as tumors originating from the skin or subcutaneous tissues and glands,
particularly the mammary gland. Study results provide evidence of carcinogen!city from chronic
high dose exposure to AA, as presented in Table 4-10. Upon histopathological examination at
the end of 2 years, male F344 rats exposed to 2.0 mg/kg-day of AA in water had developed
statistically significantly increased incidences of thyroid (follicular cell) adenomas (no
carcinomas), mesotheliomas of the tunica vaginalis testis (i.e., scrotal sac), and benign adrenal
pheochromocytoma. Female F344 rats exposed to 2.0 mg/kg-day for 2 years developed
statistically significantly increased incidences of mammary gland benign tumors (adenoma,
fibroadenoma, or fibroma), CNS tumors of glial origin, thyroid (follicular cell) adenomas or
adenocarcinomas, squamous papillomas of the oral cavity, uterine adenocarcinomas, benign
clitoral gland adenomas, and pituitary gland adenomas. Statistically significant increases in
tunica vaginalis testicular mesotheliomas were also observed in male rats exposed to 0.5 mg/kg-
day of AA in water. No other significant increases were observed at other sites for males or
females at AA doses less than or equal to 0.5 mg/kg-day.
Table 4-10. Incidences of selected tumors in
exposed to acrylamide in drinking water for
male and female F344 rats
2 years
Tumor type
Males
CNS tumors or glial proliferation suggestive of early tumor
Thyroid (follicular cell) adenoma (no carcinomas found)
Tunica vaginalis testis mesothelioma
Squamous cell carcinoma or papilloma, oral cavity
Pheochromocytomas, benign (adrenal)
Females
Mammary gland adenocarcinoma
Mammary gland benign tumors (adenoma, fibroadenoma, or
fibroma)
CNS tumors of glial origin
Thyroid (follicular cell) adenoma or adenocarcinoma
Squamous cell carcinoma, oral cavity
Squamous papilloma, oral cavity
Uterus adenocarcinoma
Clitoral adenoma, benign
Pituitary gland adenoma
Dose (mg/kg-day)
0
5/60
1/60
3/60
6/60
3/60
2/60
10/60
1/60
1/58
0/60
0/60
1/60
0/2
25/59
0.01
2/60
0/58
0/60
7/60
7/59
1/60
11/60
2/59
0/59
0/60
3/60
2/60
1/3
30/60
0.1
0/60
2/59
7/60
1/60
7/60
1/60
9/60
1/60
1/59
0/60
2/60
1/60
3/4
32/60
0.5
3/60
1/59
11/603
5/60
5/60
2/58
19/58
1/60
1/58
2/60
1/60
0/59
2/4
27/60
2.0
8/60
7/59a
10/603
6/60
10/603
6/61
23/6 la
9/6 la
5/60a
1/61
7/6 la
5/60a
5/5a
32/60a
"Significantly different from control, p < 0.05, after Mantel-Haenszel mortality adjustment.
Source: Johnson etal. (1986).
95
DRAFT-DO NOT CITE OR QUOTE
-------
In summary, chronic exposure of male and female F344 rats to the highest dose of
2.0 mg/kg-day of AA in water (Johnson et al., 1986) resulted in increased incidences for tumors
at multiple sites in both sexes. Chronic exposure to the next lowest dose of 0.5 mg/kg-day
resulted in a significant increase only in male testicular sac mesotheliomas. No significant
increases over controls were observed in female tumors at the 0.5 mg/kg-day dose or in male or
female tumors at doses lower than 0.5 mg/kg-day.
Friedman et al. (1995) study
A second cancer bioassay in F344 rats exposed to AA in drinking water (Friedman et al.,
1995; Tegeris Laboratories, 1989) included 204 male rats in the 0.1 mg/kg-day group to increase
the statistical power sufficient to detect a 5% increase in incidence of scrotal sac mesotheliomas
over an expected background incidence of this tumor for F344 rats of about 1%. The study also
had different dose group spacing for female rats to improve the characterization of the dose-
response relationships (see Table 4-11). Ambiguities in the Johnson et al. (1986) study (e.g.,
abnormally high background for CNS and oral cavity tumors in the control males and possible
confounding from a sialodacryoadenitis virus infection) also prompted the design and conduct of
this second study.
Table 4-11. Dosing parameters of groups of rats given acrylamide in
drinking water for 106-108 weeks in the carcinogenicity study
Group
1
2
3
4
5
Males
Number of rats
102
102
204
102
75
Dose (mg/kg-day)
0
0
0.1
0.5
2.0
Females
Number of rats
50
50
-
100
100
Dose (mg/kg-day)
0
0
-
1.0
3.0
Sources: Friedman et al. (1995); Tegeris Laboratories (1989).
Water consumption was measured weekly throughout the study. Body weight and food
consumption were recorded for each animal prior to the start of treatment, weekly for the initial
16 weeks of treatment, and every 4 weeks thereafter. All animals were observed twice daily for
mortality, morbidity, and obvious clinical signs of toxicity. Physical examinations were
performed weekly for the first 16 weeks, every 4 weeks for the ensuing 24 weeks, and biweekly
for the remainder of the study. Examinations for palpable masses were initiated in study
month 6 but the frequency of these examinations was not included in the study report.
Complete postmortem gross pathologic examinations were performed on all rats in the
study. Brain, liver, kidneys, and testes were excised and weighed. Group mean organ weights
and organ-to-body weight ratios were calculated. Representative sections from all major organs
and tissues (including the sciatic nerve) were stained with hematoxylin and eosin for
96
DRAFT-DO NOT CITE OR QUOTE
-------
histopathologic examination. Initially, microscopic examination was completed only on high-
dose and control rats. Based on histopathologic results in these groups, examinations were
performed on specific tissues harvested from rats of lower dose groups. Histopathologic
examination was performed on thyroid, brain (three levels, females only), mammary glands
(females), and testes (males) in all rats. In addition, spinal cord (three levels), uterus, and gross
lesions were evaluated in all control and high dose females, and in low dose female rats found
dead or sacrificed moribund. Brain (three levels), spinal cord (three levels), and gross lesions
were examined in all control and high-dose males and in low- and mid-dose male rats found
dead or sacrificed moribund. No special staining methods were used to enhance light
microscopic detection of degenerative changes in nervous tissues.
Body weight, food consumption, and water consumption were analyzed by one-way
analysis of variance; Dunnett's t-test was used to determine if means of treated groups were
significantly different from controls. Statistical evaluations included comparisons of all groups
relative to each control group, as well as to pooled controls. Pairwise t-tests were used to
compare the mean absolute organ weights (and mean percentage relative organ weights) between
the pooled control groups and each treated group by sex and organ. Two-sided trend tests were
performed to determine whether the mean weights increased or decreased with increasing dose.
Statistical analysis of survival included the Kaplan-Meier method, the log rank test, and a test for
dose-related trend in survival. Tumor incidence data were also analyzed using lifetime tumor
rates that were not time adjusted, utilizing the Cochran-Armitage trend test. Tarone's method of
analysis was used to assess the lethality of mesotheliomas of the tunica vaginalis testis. For all
tumor types, the interval-based method of Peto and the logistic score test were used. Results of
statistical tests were generally considered significant at thep < 0.05 level.
Nonneoplastic results—primarily neurotoxicity
Cumulative mortality data were depicted graphically, and statistical significance was not
reported. There were only minor dose-related increases in cumulative mortality observed among
the male rat groups during the first 60 weeks of treatment, after which mortality increased in
high dose males compared with all other groups, increasing by the end of the study to 75% vs. 53
and 44% in control groups 1 and 2, respectively. Differences in mortality among the male
control groups were greater than differences among either control groups and the low- or mid-
dose-treated males at study end. There were only minor differences in female rat mortality
within the first 23 months; however, by study end, mortality rates in controls 1 and 2 and the 1.0
and 3.0 mg/kg-day treatment groups were 40, 28, 35, and 49%, respectively.
Group mean body weights for control and treated groups were depicted graphically. No
significant differences were seen among experimental groups regarding food or water
consumption. Mean body weights of 2.0 mg/kg-day male rats were consistently decreased from
97 DRAFT-DO NOT CITE OR QUOTE
-------
those of control groups starting at week 8 and were significantly decreased from week 40 (398 g
vs. 408 g in controls, approximately 2.5% lower) to study end (375 g vs. 412 g in controls,
approximately 9% lower). Body weights of 0.1 and 0.5 mg/kg-day males did not differ
significantly from controls at any time during the study. Mean body weights of 3.0 mg/kg-day
females were significantly lower than controls from week 3 to study end, although the data in the
graphical depiction indicated that the difference was greatest near study end and did not exceed
8%. Slight but significantly lower mean body weight was observed in 1 mg/kg-day females
from weeks 8 to 32. However, this treatment group did not exhibit significant differences in
mean body weight at other time points. The study authors did not provide data concerning organ
weights but stated that slight differences (significant in some cases) between group mean organ
weights generally reflected group differences in mean final body weight.
. Table 4-12 summarizes the light microscopic findings in sciatic nerve sections of
selected rats of each sex and treatment level. Sciatic nerve degeneration was characterized by
vacuolated nerve fibers of minimal-to-mild severity. The authors did not include results of
statistical analysis of increased incidences of sciatic nerve degeneration among high-dose male
and female rats, relative to controls. However, application of Fisher's Exact test shows
significantly increased incidences of sciatic nerve degeneration among both male and female
high-dose rats.
Table 4-12. Light microscopic data for sciatic nerves from F344 rats exposed
to acrylamide in drinking water for 2 years
Endpoint
Males
Number examined
Degeneration3
Females
Number examined
Degeneration3
Dose (mg/kg-day)
0
83
30
(36%)
37
7
(19%)
0
88
29
(33%)
43
12
(28%)
0.1
65
21
(32%)
-
0.5
38
13
(34%)
-
1
—
20
2
(10%)
2
49
26
(53%)b
-
3
—
86
38
(44%)b
"Number of sciatic nerves (% of examined nerves) that exhibited light microscopic evidence of degeneration.
bStatistically different from control groups according to Fisher's Exact test (p < 0.05) performed by Syracuse
Research Corporation.
Sources: Friedman et al. (1995); Tegeris Laboratories (1989).
The authors stated that palpable masses in male rats, located primarily in the inguinal
area and most likely associated with inflammation of the preputial gland, were observed
beginning in the first 12 months of the study. The incidences of these masses were similar in all
dose groups during the second year of treatment. Although no dose-related differences were
seen in the percentage of rats with masses at individual locations, the total percentage of rats
98
DRAFT-DO NOT CITE OR QUOTE
-------
with palpable masses was increased in the high-dose group, compared with either control group
or the pooled controls (specific data not presented).
To summarize, the noncancer effects, the Friedman et al. (1995) study observed
peripheral nerve degeneration based on light microscopic examination (electron microscopy was
not conducted) in F344 rats exposed to AA in drinking water for 2 years. A NOAEL of 1
mg/kg-day was identified in female rats (0.5 mg/kg-day in male rats) with a LOAEL of 2 mg/kg-
day for male rats.
Neoplastic results—tumors at multiple sites
Incidences of selected neoplastic lesions in male and female rats are presented in Tables
4-13 and 4-14, respectively. Histopathologic examination revealed significantly increased
incidences of male thyroid gland (follicular cell) adenoma (and adenoma or carcinoma
combined) and tunica vaginalis mesothelioma in the 2.0 mg/kg-day group. Females exposed to
1.0 and 3.0 mg/kg-day developed a significantly increased incidence of mammary gland
fibroadenomas or combined fibroadenomas and carcinomas. Only the high-dose (3.0 mg/kg-
day) females exhibited a significantly increased incidence of thyroid gland follicular cell
neoplasms (adenomas or carcinomas combined).
Table 4-13. Incidences of tumors in male F344 rats exposed to acrylamide in
drinking water for 2 years
Number of animals/group
Tissue/lesion
Brain (glial origin)3
Astrocytoma
Oligodendroglioma
Spinal cord (glial origin)
Astrocytoma
Reproductive organs and accessory tissues
Tunica vaginalis testis mesothelioma
Thyroid gland (follicular cell)
Adenoma
Carcinoma
Adenoma or carcinoma (combined)
Dose (mg/kg-day)
0
102
0
102
0.1
204
0.5
102
2.0
75
1/102
0/102
0/82
4/102
2/100
1/100
3/100
0/102
1/102
0/90
4/102
0/102C
2/102
2/102c
0/98
1/98
1/68
9/204
9/203
3/203
12/203
0/50
1/50
0/37
8/102
5/101
0/101
5/101
2/75
0/75
1/51
13/75b
15/75b'd
3/75
17/75b'e
99
DRAFT-DO NOT CITE OR QUOTE
-------
aDoes not include two rats with "malignant reticulosis" of the brain, one dosed male and one control male. The
male 0.1 mg/kg-day group had only 98/204 brains and 68/204 spinal cords examined. The male 0.5 mg/kg-day had
only 50/102 brains and 37/102 spinal cords examined. All male brains of high-dose rats and all male control brains
(both subgroups) were examined, but only 82/102 and 90/102 control spinal cords and 51/75 high dose spinal cords
were examined. (Footnote from Rice, 2005).
bSignificantly different from control, p < 0.05.
°The data reported in Table 4 in Friedman et al. (1995) list one follicular cell adenoma in the second control group;
however, the raw data obtained in the Tegeris Laboratories (1989) report (and used in the time-to-tumor analysis)
list no follicular cell adenomas in this group. The corrected number for adenomas (0) and the total number of
combined adenomas and carcinomas (2) in the second control group are used in this table and this assessment.
dTwelve rats had a single follicular cell adenoma and three rats had multiple follicular cell adenomas.
eA single rat had both an adenoma and a carcinoma.
Source: Friedman etal. (1995).
Table 4-14. Incidences of tumors in female F344 rats exposed to acrylamide
in drinking water for 2 years
Number of animals/group
Tissue/lesion
Brain (glial origin)3
Astrocytoma
Oligodendroglioma
Spinal cord (glial origin)
Astrocytoma
Mammary gland
Fibroadenoma
Adenocarcinoma
Adenoma or carcinoma (combined)
Thyroid gland (follicular cell)
Adenoma
Carcinoma
Adenoma or carcinoma (combined)
Dose (mg/kg-day)
0
50
0/50
0/50
0/45
5/46
2/46
7/46
0/50
1/50
1/50
0
50
0/50
0/50
0/44
4/50
0/50
4/50
0/50
1/50
1/50
1.0
100
2/100
0/100
0/21
20/94b
2/94
21/94b
7/100
3/100
10/100
3.0
100
2/100
0/100
1/90
26/95b
4/95
30/95b
16/100C
7/100
23/100b
aDoes not include five dosed female rats with "malignant reticulosis" of the brain. All female brains were
examined, but only 45/50, 44/50, 21/100, and 90/100 spinal cords in control 1, control 2, low-, and high-dose
females, respectively, were examined. (Footnote from Rice, 2005).
bSignificantly different from control, p < 0.001 as reported by Friedman et al. (1995).
Statistically different from control groups according to Fisher's Exact test (p < 0.05) performed by Syracuse
Research Corporation.
Source: Friedman etal. (1995).
These findings confirm the results of the earlier Johnson et al. (1986) drinking water
bioassay with F344 rats; i.e., significantly increased incidences of thyroid follicular cell tumors
in males and females, tunica vaginalis testis mesotheliomas in males, and mammary gland
tumors in females. Results of the study of Johnson et al. (1986) that were not reported as being
replicated in the study of Friedman et al. (1995) include the statistically significantly increased
incidences of adrenal pheochromocytomas in males, CNS tumors of glial origin in females, oral
cavity tumors in females, and clitoral or uterus tumors in females.
100 DRAFT-DO NOT CITE OR QUOTE
-------
In a review of the Friedman et al. (1995) study data, Rice (2005) noted that, although
glial tumors of brain and spinal cord were reported not to be increased, not all of the brains and
spinal cords in the test animals were examined, and seven cases of a morphologically distinctive
category of primary brain tumor described as "malignant reticulosis" were reported but were
excluded from the authors' analysis. Rice (2005) comments that it is unusual to exclude brain
tumors of this kind from the results of a bioassay. The neoplasms diagnosed as "malignant
reticulosis" are of uncertain origin but have some features in common with anaplastic
astrocytomas. Both astrocytomas and neoplasms consistent with a descriptive designation of
"malignant reticulosis" are also induced in rats by the structurally closely related compound,
acrylonitrile (IARC, 1994b) and by the simple epoxide carcinogen, ethylene oxide (IARC,
1999). Rice (2005) concluded that the primary brain tumors were underreported in the Friedman
et al (1995) study and provided the following details from his review of the study records (also
see footnotes in Tables 4-13 and 4-14):
. . . tabulated data in the study report does not include seven rats with "malignant
reticulosis" of the brain, including five dosed females, one dosed male and one
control male. The male 0.1 mg/kg-day group had only 98/204 brains and 68/204
spinal cords examined. The male 0.5 mg/kg-day had only 50/102 brains and
37/102 spinal cords examined. All male brains of high-dose rats and all male
control brains (both subgroups) were examined, but only 82/102 and 90/102
control spinal cords and 51-75 high dose spinal cords were examined. All female
brains were examined, but only 45/50, 44/50, 21/100 and 90/100 spinal cords in
control, control, low- and high-dose females, respectively were examined.
EPA agrees that the brain tumor incidence rates and analyses should have been more
fully documented in the Friedman et al. (1995) report tables and discussion, and concurs with the
Rice (2005) conclusion that the CNS tumors be considered one of the tumor types replicated in
the Friedman et al. (1995) study, even though the incomplete brain and spinal cord tumor data
set precludes a quantitative analysis of CNS tumor incidence in the characterization of dose-
response.
latropoulos et al. (1998) reevaluated reproductive tissue from the 38 male rats originally
diagnosed with tunica vaginalis mesotheliomas and arrived at a different diagnosis than the
original analysis (which considered all of the mesotheliomas to be malignant as reported in
Friedman et al. [1995] and Tegeris Laboratories [1989]). Using criteria specified by McConnell
et al. (1992), tissue blocks and slides were reevaluated and reclassified into one of three types of
mesothelial lesions: (1) focal mesothelial hyperplasia, (2) benign mesothelioma, and
(3) malignant mesothelioma. Proliferating cells from the mesothelial lesions were stained for
proliferating cell nuclear antigen to assess the fraction of cells that were replicating. In addition,
for each rat, the extent of Ley dig cell neoplastic proliferation was assessed as occupying 25, 50,
75, or 100% of the testes. The evaluations were reported to have been conducted in a blinded
101 DRAFT-DO NOT CITE OR QUOTE
-------
fashion. The reevaluation assessed that not all of the previously diagnosed mesotheliomas were
malignant (see Table 4-15). All rats reevaluated as having malignant mesotheliomas were
assessed as having 75 or 100% of the testes occupied by Ley dig cell neoplasia. In contrast, rats
reevaluated as having focal mesothelial hyperplasia or benign mesothelioma showed either no
Ley dig cell neoplasia or 25 or 50% of the testes occupied by Ley dig cell neoplasia. The
comparison suggests that the extent of Ley dig cell neoplasia and the development of malignant
mesotheliomas may have been linked.
102 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-15. Reevaluation and comparison of mesothelial lesions and extent
of Leydig cell neoplasia in male F344 rats exposed to acrylamide in drinking
water for 2 years
Dose
(mg/kg-day)
Control
Group 1
Control
Group 2
0.1
0.5
2.0
Rat
no.
126
134
170
179
257
301
335
353
432
457
473
484
514
564
594
603
606
729
732
756
758
762
767
780
783
810
813
814
816
821
824
832
841
844
847
850
868
878
Diagnosis3
No mesothelial tissue was present
Benign mesothelioma, focal
Malignant mesothelioma
Benign mesothelioma, focal
Malignant mesothelioma
Focal mesothelial hyperplasia
Focal mesothelial hyperplasia
Malignant mesothelioma
No mesothelial change
Malignant mesothelioma
Malignant mesothelioma
Malignant mesothelioma
Focal mesothelial hyperplasia
Malignant mesothelioma
Focal mesothelial hyperplasia
Malignant mesothelioma
Focal mesothelial hyperplasia
Malignant mesothelioma
Malignant mesothelioma
Benign mesothelioma, focal
Benign mesothelioma, focal
Malignant mesothelioma
Focal mesothelial hyperplasia
Benign mesothelioma, focal
Benign mesothelioma, focal
Benign mesothelioma, focal
Malignant mesothelioma
Benign mesothelioma, focal
Malignant mesothelioma
Focal mesothelial hyperplasia
Focal mesothelial hyperplasia
Malignant mesothelioma
Benign mesothelioma, focal
Malignant mesothelioma
Benign mesothelioma, focal
Benign mesothelioma, focal
Malignant mesothelioma
Benign mesothelioma
Evidence of metastasis or invasion
Metastasis to mesentery
Metastasis to seminal vesicles
Metastasis to peritoneal cavity
Invasion through the serosa
Metastasis to neighboring skeletal muscle
Metastasis to mesentery
Invasion through the serosa
Metastasis to mesentery
Metastasis to hepatic serosa
Metastasis to mesentery, splenic serosa
Metastasis to splenic serosa
Metastasis to neighboring skeletal muscle,
splenic serosa
Metastasis to urinary bladder
Invasion through the serosa
Metastasis to seminal vesicles, epididymis
Metastasis to neighboring skeletal muscle,
mesentery
Metastasis to mesentery
Leydig cell
neoplasiab
L+++
L+
L+++
L++
L++++
L+
L+
L+++
—
L++++
L+++
L++++
L+
L+++
L+
L+++
L+
L+++
L+++
L+
-
L++++
-
L++
L++
L+
L+++
L+
L+++
-
L+
L+++
L++
L+++
L++
L++
L+++
L++
"Rats previously diagnosed as having mesothelioma of the tunica vaginalis testis (Friedman et al., 1995).
bLeydig cell neoplasms occupying 25% (+), 50% (++), 75% (+++), or 100% (++++) of testes; - denotes no
neoplasm.
Source: latropoulos et al. (1998).
4.2.2. Inhalation Exposure
Information on the response to subchoronic or chronic exposure to inhaled AA in animals
is limited to three subchronic studies in cats, dogs, and rats from the mid-1950s (Hazleton
103 DRAFT-DO NOT CITE OR QUOTE
-------
Laboratories, 1954, 1953) with demonstration of neurotoxicity dependent on dose and species
tested. No chronic animal inhalation studies for exposure to AA were identified.
4.2.2.1. Subchronic Studies
Exposure of four cats to AA vapors at a mean analytical concentration of 1.65 ppm
(4.8 mg/m3), 6 hours/day, 5 days/week for 3 months, resulted in no apparent clinical signs or
adverse effects on body weight (Hazleton Laboratories, 1954). Results of periodic blood studies
(hematocrit, hemoglobin, sedimentation rates, and white blood counts) and plasma
pseudocholinesterase activity levels were within normal limits.
Exposure of dogs and rats to an aerosol of AA dust at a concentration of 15.6 mg/m3,
6 hours/day, 5 days/week for up to 12 exposures, resulted in progressive signs of neurotoxicity
and death (Hazleton Laboratories, 1953). Simultaneously exposed guinea pigs showed no
neurotoxic signs.
4.2.2.2. Chronic Studies
No chronic inhalation animal studies were identified.
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION
There is a large database for reproductive effects from oral exposure to AA, and the
reproductive section begins with a discussion of the recent expert panel review of the database
(NTP/CERHR, 2004).
There were no inhalation studies found in the literature that measured reproductive or
developmental in animals exposed to AA.
4.3.1. Reproductive Toxicity Studies
An NTP-sponsored expert panel (NTP/CERHR, 2004) conducted a comprehensive
review of reproductive and developmental toxicity studies for a variety of exposure routes: by
drinking water in rats or mice (NTP, 1993; Smith et al., 1986; Zenick et al., 1986), by gavage in
rats (Sublet et al., 1989; Working et al., 1987b), by i.p. injection in mice (Holland et al., 1999;
Nagao, 1994; Ehling and Neuhauser-Klaus, 1992; Dobrzynska et al., 1990; Shelby et al., 1987,
1986), and by dermal application in mice (Gutierrez-Espeleta, 1992). The NTP/CERHR (2004)
report summarized that the lowest effective doses of AA reported were 30 ppm in drinking water
in rats (a cumulative dose of about 200 mg/kg by the time of mating) (Smith et al., 1986),
6.78 mg/kg-day in drinking water in mice (a cumulative dose of 949 mg/kg over the 20-week
exposure period) (NTP, 1993), 15 mg/kg-day for 5 days by gavage in rats (Sublet et al., 1989),
75 mg/kg i.p. in mice (single dose) (Ehling and Neuhauser-Klaus, 1992), and 25 mg/kg-day for
5 days applied dermally to mice (Gutierrez-Espeleta, 1992). The panel concluded that the
104 DRAFT-DO NOT CITE OR QUOTE
-------
dominant lethal data provide firm in vivo postmetabolic evidence of genotoxicity in mammals
and that AA was effective via all routes in all species at comparable doses. The report notes that
the stage effect was consistent but that the dominant lethal test does not effectively assess
damage in spermatogonial stem cells. The panel cautioned against assigning stage-specific
effects in these studies based on the kinetics of spermatogenesis, given that some chemical
agents (including, perhaps, AA) may alter the kinetics of spermatogenesis. In the case of AA,
the dominant lethal studies most likely indicate an effect on the ability of epididymal
spermatozoa and spermatids to fertilize an oocyte, along with potential pre- and postimplantation
genetic effects. Although the antifertilization effect may be due to nongenetic actions, the doses
needed to elicit the antifertilization effects were generally higher than that needed to elicit the
postimplantation genetic effects, and thus the antifertilization effects are of limited utility for
predicting human risk.
The following discussion presents details of the oral studies, including two-
generation/dominant lethal studies (Tyl et al., 2000a; Chapin et al., 1995) and dominant lethal
studies (Tyl et al., 2000b; Sublet et al., 1989; Working et al., 1987a, b; Smith et al., 1986; Zenick
et al., 1986). The results for other reproductive function endpoints are also discussed (Sakamoto
et al., 1988; Sakamoto and Hashimoto, 1986; Zenick et al., 1986).
Tyl et al. (2000a) two-generation/dominant lethal study
Tyl et al. (2000a) performed a two-generation reproduction and dominant lethal study of
AA in F344 rats. Groups of FO weanlings (30/sex/group) were exposed to AA in the drinking
water at concentrations that would provide dose levels of 0, 0.5, 2.0, or 5.0 mg/kg-day during a
prebreeding period of 10 weeks. The breeding period consisted of 14 days of cohabitation,
during which males and females were paired one-to-one. During mating, gestation, and the first
week of lactation, female rats of each treatment group were given the same concentration of AA
in the drinking water as that to which they had been exposed during the final week prior to
mating; during the cohabitation mating period, males were exposed to AA based on the body
weights of the corresponding females during mating to avoid overexposure of the females. As
soon as each successful mating was confirmed, each pair was separated. Mated females were
weighed on GDs 0, 6, 13, and 20. Dams and litters were weighed on postnatal days (PNDs) 1, 4,
7, 14, 21, and 28. Pups were weaned on PND 28. Following mating, FO males were maintained
on their respective AA doses until 2 days prior to being mated with naive unexposed females in
the dominant lethal portion of the study, after which impregnated females were separated from
the males and sacrificed on GD 14. Gross examinations were performed and number of ovarian
corpora lutea and number and distribution of total uterine implantation sites, resorption sites, and
live and dead implants were determined.
105 DRAFT-DO NOT CITE OR QUOTE
-------
Thirty Fl male and 30 Fl female rats of each dose group were selected to be continued
on AA (in the same manner as their parents) to produce F2 pups. The prebreeding treatment
period for Fl rats was 11 weeks. All FO and Fl parental rats in all treatment groups were
subjected to gross necropsy. In addition, 30 male and 30 female Fl parental rats each from
control and high-dose groups were subjected to histologic examination of major reproductive
tissues and representative target neurological tissues (peripheral nerves, brain, and spinal cord).
Sciatic and tibial nerve sections from six high-dose male and three control male Fl adults and
spinal cord sections from three high-dose and two control female Fl adults were stained with
Bodian's method for additional histologic examination. Selected Fl and F2 weanling rats were
subjected to the same histologic examinations as were the Fl parental rats. The study report
does not indicate that tissues from FO rats were histologically examined.
Results for quantitative continuous variables were analyzed using Levene's test for equal
variances, ANOVA, and t-tests. Nonparametric data were statistically evaluated by using the
Kruskal-Wallis test, followed by the Mann-Whitney U-test for pairwise comparisons. Fisher's
Exact test was used to compare frequency data. For all statistical tests, the level of significance
wasp < 0.05.
FO males in all three treatment groups showed statistically significantly reduced mean
body weight compared with controls (-4-6% decreased), starting after 4-6 weeks and continuing
through 13 weeks when exposure ceased. Body weights in 2.0 and 5.0 mg/kg-day Fl males
showed similar depressions of body weight throughout their 13 weeks of exposure. Body
weights in FO females were statistically significantly lower than controls during the latter
4 weeks of the prebreeding period in the 2.0 and 5.0 mg/kg-day groups (-4-6% decreased), at
the end of gestation in the 5.0 mg/kg-day group (-9% decreased), and most of the lactation
period in the 5.0 mg/kg-day group (-4-6% decreased). Body weights in Fl females were
statistically significantly lower than controls during the latter 8 weeks of prebreeding in the 2.0
and 5.0 mg/kg-day groups (-5% decreased), at the end of gestation in the 2.0 (-3% decreased)
and 5.0 mg/kg-day groups (-12% decreased), and during the middle 3 weeks of lactation in the
5.0 mg/kg-day group (-4-6% decreased). In F2 offspring, statistically significant changes in
body weight were restricted to the 5.0 mg/kg-day group at PND 14 (-7% decreased). The
depressions in body weight, although not large, provide evidence of mild systemic toxicity, most
consistently in 2.0 and 5.0 mg/kg-day FO and Fl adult males.
Increased incidences of rats with foot splay occurred in FO exposure groups relative to
controls. Incidences for foot splay were 3/30, 10/30, 7/30, and 10/30 for control through
5.0 mg/kg-day FO males and 1/30, 2/30, 6/30, and 6/30 for FO females. Fisher's Exact test
(performed by Syracuse Research Corporation) indicated that incidences were statistically
significantly (p < 0.05) elevated in the low- and high-dose male groups; incidences in the mid-
and high-dose female groups were marginally (p = 0.51) elevated compared with controls. No
106 DRAFT-DO NOT CITE OR QUOTE
-------
foot splay was observed in Fl males or Fl females in any groups. Head tilt was displayed by
some FO and Fl males and Fl females, but the incidences of this sign of neurotoxicity were not
statistically significantly different from controls, except for a marginally significant (p = 0.056)
elevation in the 5.0 mg/kg-day Fl males (0/30, 0/30, 0/30, and 4/30).
Gross examinations of all FO rats, all Fl pups that died during lactation, and selected Fl
weanlings yielded no treatment-related findings. Histopathologic examination of reproductive
and nervous system tissues of the Fl weanlings revealed no signs of treatment-related adverse
effects. Histopathology of selected nervous system tissues from control and 5.0 mg/kg-day Fl
adults and all necropsied F2 weanlings showed no exposure-related lesions with conventional
staining (hematoxylin and eosin). However, when peripheral nerve sections (from sciatic and
tibial nerves) were examined with Bodian's stain, minimal to mild axonal fragmentation and/or
swelling was observed in 6/6 Fl 5.0 mg/kg-day males compared with 0/3 control Fl males
(female tissues were not examined). Spinal cord sections from three high-dose females and two
control females, stained by the same method, showed no lesions (male tissues were not
examined). Tissues from FO rats and Fl rats in lower exposure groups were not examined
histologically.
AA treatment did not significantly affect FO or Fl reproductive parameters involving
success of mating and impregnation or gestation length, but 5.0 mg/kg-day induced statistically
significantly decreased numbers of implantations/dam and live pups/litter on PND 0, and
increased postimplantation loss in the FO and Fl generations (Table 4-16). Fl and F2 pup
survival between PNDs 0 to 4 was unaffected by treatment, with the exception that, in the
5.0 mg/kg-day group, three one-pup Fl litters and three one-pup F2 litters did not survive.
Table 4-16. Changes in reproductive parameters in F344 rats exposed to
acrylamide in drinking water for two generations
Parameters
FO parents/Fl mating (30 pairs/group)
No. males impregnating
No. females pregnant
No. implantations/dam
No. live pups/litter (PND 0)
Postimplantation loss (%)
Fl parents/F2 mating (30 pairs/group)
No. males impregnating
No. females pregnant
No. implantations/dam
No. live pups/litter (PND 0)
Postimplantation loss (%)
Dose (m
0
17
20
10.4 ±2.5
9.8±3.1
7.9 ± 18.5
23
23
11. 3 ±1.5
10.8 ± 1.5
4.4 ±7.6
0.5
24
24
10.0 ±3.6
9.8 ±3.5
2.1 ±4.7
25
25
10.0 ±3.4
10.0 ±2.9
3.3 ±7.9
g/kg-day)
2.0
22
26
10.2 ±2.2
9.7 ±2.4
5.7 ±9.1
25
27
10.5 ±2.1
9.6 ±2.4
9.1 ±14.4
5.0
21
18
6.8±3.1a
4.5±2.6a
34.4±25.9a
27
23
6.8±3.3a
5.1±3.2a
23.1 ±28.2
aSignificantly (p < 0.05) different from control value. Values are group means ± SD.
Source: Tyl et al. (2000a).
107 DRAFT-DO NOT CITE OR QUOTE
-------
No effects on Fl pup body weights were seen on PNDs 1, 4, or 7. However,
measurements made on PNDs 14, 21, and 28 (when rat pups had begun to drink and feed
themselves) revealed significantly reduced pup weight (8-11% lower than controls) in
5.0 mg/kg-day males. Significantly reduced mean F2 pup body weight (approximately 8%) was
seen only in 5.0 mg/kg-day pups and only on PND 14.
Dominant lethal results
In the dominant lethal mutation protocol in which exposed male rats were mated with
nonexposed female rats, exposure did not adversely affect fertility or mating indices or the
number of corpora lutea (Table 4-17). However, the total number of implants/litter and the
percentages of pre- and postimplantation loss were statistically significantly different from
controls in nonexposed females mated to treated 5.0 mg/kg-day FO males.
Table 4-17. Results of the dominant lethal mutation assay in F344 rats
Parameter
No. males paired
No. females paired
No. fecund males3
No. fertile males'3
No. plug- or sperm-positive females
No. pregnant females
Mating index0
No. corpora lutea/dam
No. total implants/litter
Percent preimplantation loss
Live implants/litter
Nonlive implants/litter
Percent postimplantation loss
Acrylamide dose (mg/kg-day) in the drinking water
0.0
30
60
29 (96.7%)
28 (93.3%)
57 (95.0%)
52 (91.2%)
52/60 (86.7%)
11.8±2.1d
10.0 ±2.3
14.3 ±19.6
9.4 ±2.2
0.6 ±0.7
6.2 ±7.0
0.5
30
60
30 (100.0%)
29 (96.7%)
56 (93.3%)
50 (89.3%)
50/60 (93.3%)
11.5±1.1
9.9 ±2.5
14.3 ±21.2
9.5 ±2.5
0.4±0.7e
3.7±6.8e
2.0
30
60
30 (100.0%)
29 (96.7%)
59 (98.3%)
57 (96.6%)
57/60 (95.0%)
11.8±1.1
10.2 ±2.2
13.5 ± 18.4
9.6 ±2.3
0.6 ±0.7
6.1 ±6.9
5.0
30
60
30 (100.0%)
29 (96.7%)
57 (95.0%)
52 (91.2%)
52/60 (86.7%)
11.4 ±1.2
8.6±2.7f
24.9±22.7f
7.5±2.6g
l.l±1.0e
14.2±17.1f
aNumber of males that produced at least one plug- or sperm-positive female.
bNumber of males that produced at least one pregnant female.
°Ratio of pregnant females to paired females.
dMean ± SD.
> < 0.05.
fp<0.0l.
sp< 0.001.
Source: Tyl et al. (2000a).
Chapin et al. (1995) two-generation/dominant lethal/grip strength study
Chapin et al. (1995) conducted a two-generation continuous breeding reproductive
toxicity study in CD-I mice that included an assessment of grip strength in FO and Fl adult mice.
Male and female CD-I mice (20/sex/treatment group) were individually housed and
administered AA in the drinking water at concentrations of 3, 10, or 30 ppm for 7 days, followed
by continuous dosing during 14 weeks of cohabitation as mating pairs. At test concentrations of
108 DRAFT-DO NOT CITE OR QUOTE
-------
3, 10, and 30 ppm, the authors estimated AA doses of 0.81, 3.19, and 7.22 mg/kg-day for both
male and female FO mice, based on water consumption data of FO females. A control group
consisted of 40 mating pairs. Mice were monitored for clinical signs, but the frequency of
observations was not specified. Body weights of FO mice were recorded following the delivery
of each litter produced during the cohabitation period, at necropsy, and at other unspecified time
points. Pups from each litter were counted, sexed, weighed, and killed. Reproductive indices
measured included fertility (number of pairs delivering at least one litter), number of litters/pair,
and number of live pups/litter, sex ratio, day of delivery, and pup birth weight. Parental food
and water consumption were measured for 1 week both immediately prior to (study week 1) and
following (study week 16) the cohabitation period (study weeks 2-15). Forelimb and hindlimb
grip strength were assessed in 10 male and 10 female FO mice/group during study weeks 0, 3, 6,
9, 12, and 17.
At the end of the 14-week cohabitation period, the FO pairs were separated and dosed for
an additional 6 weeks, during which time pregnant dams were allowed to deliver and wean Fl
litters. The Fl pups were culled to two/sex/litter and maintained on the same dosing regimen as
their parents. Upon reaching 74 days of age, Fl females were mated to nonsibling males of the
same treatment group for up to 1 week then separated and continued on their respective AA
treatment levels until delivery of the F2 generation. Reproductive variables evaluated for the Fl
parental mice were the same as those for the FO generation. Grip strength was measured in Fl
parental mice at weeks 3, 5, 7, 10, and 16 (necropsy week) of treatment. At necropsy, body and
selected organ weights were recorded. Microscopic examinations were performed on sural and
gastrocnemius nerves of both sexes of Fl mice, testes and epididymides of Fl males, and visible
gross lesions.
During the 6-week separation period following 98 days of FO cohabitation, selected
control and exposed FO males were cohabited with three untreated females for up to 4 days in
order to evaluate dominant lethal effects in the males. Pregnant females were subjected to
necropsy on GD 16. Uteri were examined for number of live, dead, and resorbed implants.
Following the 6-week separation period, crossover mating tests of control and high-dose
male and female FO mice were performed, which resulted in pairings of control males with
control females, control males with high-dose females, and high-dose males with control
females. The pairs were allowed to mate for 1 week, during which time AA treatment was
suspended. Treatment then continued throughout gestation and delivery. Reproductive indices
measured included fertility (number of pairs delivering at least one litter), number of litters/pair,
and number of live pups/litter, sex ratio, day of delivery, and pup birth weight. Estrous cyclicity
in parental females was assessed for 12 days following delivery. At necropsy, body and selected
organ weights were determined for all FO mice. Sperm quality was assessed in male FO mice.
109 DRAFT-DO NOT CITE OR QUOTE
-------
Grip strength measurements were performed by testing forelimb first, then hindlimb.
The results of three such trials were averaged for each animal tested. Grip strength values were
compared by ANOVA. In the dominant lethal tests, all data from females mated to a given male
were pooled. Differences in results between treated and control groups were considered
significant at the level ofp < 0.05.
AA treatment did not affect body weight or food consumption in FO males or FO females,
but water consumption was erratic in males. In Fl mice selected for mating, exposure-related
effects on body weight were not found, except for an 8% decrease in body weight, compared
with controls, in 30-ppm females. The authors estimated AA doses to be approximately 0.86,
2.9, and 7.7 mg/kg-day, based on water consumption during the week following mating. To
compare with other AA toxicity studies, approximate average doses for the groups in this study
are taken to be 0, 0.8, 3.1, and 7.5 mg/kg-day.
No treatment-related effects were observed regarding proportion of FO fertile pairs,
percentage of cohabiting FO pairs with litters, average number of Fl litters/pair, proportion of
live Fl pups born, sex ratio, or mean live Fl pup weight. A slight, but statistically significant,
decrease in aggregate mean number of live Fl pups was observed at 30 ppm (12.2 ± 0.5, n = 18,
vs. 13.6 ± 0.5, n = 39, for controls). This 10% change was due to significantly reduced numbers
of live pups in the second and third litters of high-dose mice but not in the first, fourth, or fifth
litters.
AA treatment had no adverse effect on postnatal survival or body weight gain prior to
weaning in Fl mice selected for mating. No treatment-related effects were seen regarding the
numbers of impregnated F1 females or percentage of Fl females that delivered offspring. The
mean number of live F2 pups was significantly decreased in the 30-ppm group (7.9 ±1.0 live
pups/litter vs. 14.8 ± 0.5 in controls) in the absence of a significant treatment-related alteration in
live pup birth weight. Postpartum dam body weight was significantly lower (11%) in 30-ppm Fl
dams (34.1 ± 0.9 g vs. 37.7 ± 0.9 g in controls).
Dominant lethal results
When exposed FO male mice were mated with nonexposed females, dominant lethal
effects were observed at the 30-ppm exposure level. Significantly increased early resorptions,
total postimplantation loss, and decreased number of live fetuses were observed in the 30-ppm
group (see Table 4-18). Percentages of impregnated females were 83, 83, 81, and 77 for the
control through 30-ppm groups, respectively, indicating no effects on male fertility.
110 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-18. Results of dominant lethality testing in male Swiss CD-I mice
exposed to acrylamide in the drinking water
Number of males tested
Early resorptions
Dead fetuses
Total implantation loss
Live fetuses
Acrylamide concentration (ppm)
0
20
0.86 ±0.1a
0.03 ±0.02
0.98 ±0.12
12.5 ±0.3
3
20
0.78 ± 0.26
0.06 ± 0.03
0.99 ±0.28
12.5 ±0.2
10
19
1.04 ±0.17
0.04 ±0.02
1.14±0.16
12.5 ±0.4
30
20
1.74±0.17b'c
0.09 ± 0.06
1.95±0.17b'c
11.5±0.4b
aMean ± standard error of the mean (SEM); number/litter/male.
bSignificantly different from controls (p < 0.05).
"Dose-related trend (p < 0.05).
Source: Chapinetal. (1995).
The crossover mating tests of control males with control females, control males with
high-dose females, and high-dose males with control females resulted in averages of 11.4, 11.5,
and 9.4 pups/litter, respectively. The study authors found no statistically significant differences
in litter sizes among the different groups but suggested that the smaller average litter size in the
group of high-dose males mated with control females (9.4 pups/litter, compared with 11.4 and
11.5 pups/litter in the other two groups) indicated that the dominant lethal effect was related to
toxicity in males rather than females. However, the study report did not include additional
details of the results (incidence data or variation from mean values).
Necropsy results of all FO mice did not reveal any signs of treatment-related adverse
effects on body weight or absolute or relative weights of liver, kidneys/adrenals, right ovary,
right testis or cauda epididymis, prostate, or seminal vesicles. Sperm analysis revealed no
treatment-related effects on epididymal sperm concentration, motility, frequency of abnormal
forms, or total spermatid heads/testis. However, the mean number of spermatids/gram testis was
statistically significantly (p < 0.05) lower in the 10- and 30-ppm FO males (11.1 ± 0.4, 10.6 ±
0.4, 9.8 ± 0.8, and 10.0 ± 0.5 spermatids/gram testis in controls through 30 ppm). No AA-related
effect on estrous cyclicity was seen in females (data were not shown).
Gross necropsy of Fl parental mice did not reveal treatment-related effects on male
terminal body weight or weight of liver, kidneys/adrenals, right testis, epididymis, or seminal
vesicles. AA treatment did not adversely affect female terminal body weight, absolute or
relative liver weight, or right ovary weight. Absolute kidney and adrenal weight (combined) of
10- and 30-ppm females was significantly lower than controls (550.4 ± 8.5 mg, 540 ± 12.2 mg,
503.7 ± 11.7 mg, and 519 ± 22.9 mg for controls, low-, mid-, and high-dose groups,
respectively). Relative liver weight was significantly increased (12 and 6%) in mid- and high-
dose females, respectively. The authors reported a dose-related decrease in absolute mean
prostate weight that was statistically significant in the 30-ppm male Fl group (controls
34.6 ±1.9 mg; high dose 29.7 ±1.7 mg), but mean weights of other treatment groups were not
111 DRAFT-DO NOT CITE OR QUOTE
-------
specified. Relative prostate weights were not significantly different from controls. No
significant effects were seen regarding sperm quality or estrous cycle length. Upon
histopathologic examination, testicular degeneration was noted in 1/10 mid- and high-dose males
but was not observed in males of low-dose or control groups. AA treatment did not increase the
incidence of grossly visible lesions or histopathologic findings in examined nerve tissues of male
or female Fl parental mice.
Grip strength results
Absolute grip strength increased over time in control and exposed FO groups during
17 weeks of exposure, and was reported to not be adversely affected by exposure. However,
30-ppm male and female FO mice showed statistically significantly smaller increases over time,
relative to controls (see Table 4-19). Statistically significantly reduced forelimb absolute grip
strength was observed in 10- and 30-ppm Fl males (compared with controls) following 10 weeks
of AA treatment. However, the biological significance of this finding is uncertain since the
authors found no treatment-related effects on grip strength in Fl males or females following 3, 5,
7, or 16 weeks of treatment.
Table 4-19. Effects of acrylamide in drinking water on grip strength of mice
FO relative grip strength increase (%)a
Males
Forelimb
Hindlimb
Females
Forelimb
Hindlimb
Fl absolute grip strength (g)b
Males
Forelimb
Hindlimb
Females
Forelimb
Hindlimb
Acrylamide concentration (ppm)
0
43.4 ±18.3
108.9 ±12.2
37.3 ±13. 8
112.4 ±28.6
96.4 ±4.1
118.2 ±4.0
79.6 ±2.7
103.1 ±3.6
3
39.6 ±10.4
66.4 ±14.1
44.3 ± 12.6
126.0 ±14.8
94.8 ±4.4
123.5 ±5. 5
74.7 ±5.0
126.0 ±14.8
10
2.4 ±11.7
89.8 ±11. 8
3.2±5.4C
94.8 ±15. 8
81.4±4.8C
122.8 ±5. 9
76.7 ±4.8
102.7 ±6.3
30
6.9 ± 5.5c'd
67.6 ± 9.2c'd
1.4±7.3c'd
72.6 ±12.1
84.5 ± 2.6c'd
115.6 ±2.2
80.0 ±4.3
102.2 ±4.1
"Percentage increase in grip strength during growth after 17 weeks of treatment (mean ± SEM, n = 10).
bGrip strength measured at Fl parental treatment week 10 (mean ± SEM, n = 10).
0Significantly different from controls (p < 0.05).
dDose-related trend (p < 0.05).
Source: Chapinetal. (1995).
In summary, the results presented by Chapin et al. (1995) identified 30 ppm AA in
drinking water (7.5 mg/kg-day) as a LOAEL and 10 ppm (3.1 mg/kg-day) as aNOAEL for
reproductive toxicity effects (e.g., increased early resorptions, total postimplantation loss;
decreased number of live fetuses, decreased number of live Fl and F2 pups/litter) that appear to
112 DRAFT-DO NOT CITE OR QUOTE
-------
be male-mediated in Swiss CD-I mice. No clear and consistent exposure-related effects on
fertility, gross necropsy, organ or body weights, or histology of testicular or nervous system
tissues were found. Mild changes in grip strength were noted in FO and Fl male and female
mice of the 30-ppm exposure groups and in FO female and Fl male mice of the 10-ppm exposure
groups.
Additional oral exposure dominant lethal studies
In a study designed to assess dominant lethal effects of AA, groups of male Long-Evans
rats (10-1 I/group) were administered AA in the drinking water at concentrations of 0, 15, 30, or
60 ppm for a total of 80 days (Smith et al., 1986). Based on twice weekly recording of body
weights and water consumption, the authors calculated the AA doses in the 15-, 30-, and 60-ppm
exposure groups to be 1.5, 2.8, and 5.8 mg/kg-day, respectively. During the final 8 days of
treatment, each male rat was paired nightly with two virgin untreated females until each male
had impregnated two females or until the end of the treatment period. Sperm-positive female
rats were sacrificed on GD 14 and examined for numbers of corpora lutea and for living and
dead fetal implants. Fertility rates and percentages of pre- and postimplantation losses were
calculated. Following the completion of the mating period, six males of each group were
sacrificed for histologic analysis of sperm. Segments of sacral, sciatic, and tibial nerves were
excised, fixed, and stained with hematoxylin and eosin or toluidine blue for histopathologic
examination. The remaining treated males were sacrificed 12 weeks after the end of treatment
for assessment of reciprocal translocations in spermatocytes. Data on fertility rates were
analyzed using %2 statistics. Effects on pre- and postimplantation loss were analyzed using
Kruskal-Wallis ANOVA with Mann-Whitney U-test for posthoc comparisons.
There were no statistically significant differences among controls and treated rats
regarding body weights or water consumption. As shown in Table 4-20, fertility rates did not
differ significantly among the groups. A significant elevation in preimplantation loss occurred
only in females that had been mated with high-dose males. Postimplantation loss was
statistically significantly higher in females mated with mid- or high-dose males relative to low-
dose or control males. At the high dose, the percentage was more than 6 times higher than that
of controls. None of the treated males exhibited hindlimb splaying, a characteristic sign of
AA-induced neurotoxicity. No significant pathological lesions were seen in preparations of the
sciatic nerve. The NOAEL in this study is 15 ppm (1.5 mg/kg-day) and the LOAEL is 30 ppm
(2.8 mg/kg-day) for male-mediated reproductive effects (increased postimplantation loss). No
histological changes were found in sacral, sciatic, and tibial nerves, and no evidence of hindlimb
splaying was found in rats exposed to AA concentrations as high as 60 ppm (5.8 mg/kg-day).
113 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-20. Fertility rates and pregnancy outcomes in Long-Evans rats
following 72-day oral exposure of males to acrylamide in the drinking water
Number of
males/group
9
9
10
11
Exposure
level (ppm)
0
15
30
60
Dose
(mg/kg-day)
0
1.5
2.8
5.8
Fertility (%)a
87
76
95
80
Preimplantation
loss (%)b
10.4 ±1.8
9.3 ±2.3
12.2 ±1.4
25.1±4.0d
Postimplantation
loss (%)c
5.7 ±1.6
7.2 ± 1.6
13.3±2.1e
36.7±5.6d
a(Number pregnant/number mated) x 100.
b([(Number corpora lutea - number implants]/[number corpora lutea]) x 100.
°([Number implants - number fetuses]/[number implants]) x 100.
dSignificantly different from control, low-, and mid-dose groups, p < 0.01.
Significantly different from control, low-, and high-dose groups, p < 0.01.
Source: Smith etal. (1986).
Several additional studies have demonstrated reversible dominant lethal effects and
reversible effects on male fertility in animals orally exposed to AA for short time periods.
Working et al. (1987a, b) observed reversible male-mediated reproductive effects (dominant
lethal effects: increased implantation losses) in F344 rats exposed to 30 mg/kg-day for 5 days.
Sublet et al. (1989) observed dominant lethal effects (increased implantation losses) and effects
on male impregnation success in Long-Evans male rats exposed to oral doses as low as 15
mg/kg-day for 5 days. In this study, males were gavaged with 0, 5, 15, 30, 45, or 60 mg/kg AA
for 5 days prior to mating. Reduced fertility and increased preimplantation loss were found in all
dose groups except 5 mg/kg at week 1 posttreatment. Increased postimplantation loss was seen
at weeks 2 and 3 in the 15, 30, 45, and 60 mg/kg groups. In sperm samples collected from the
45 mg/kg group, the percentage of motile sperm was modestly decreased to a statistically
significant degree (58% vs. 73% in controls) at week 3 but not at weeks 2 or 4. Sublet et al.
(1989) concluded that altered motility of sperm may have contributed to, "but can not completely
account for, the poorer reproductive performance of these males." Similarly, Tyl et al. (2000b)
observed significantly decreased fertility and increased postimplantation losses following mating
of untreated female rats with males that had been administered AA at gavage doses of 15, 30, 45,
or 60 mg/kg-day for 5 days prior to mating. No statistically significant effects were seen
regarding motility or concentration of epididymal sperm from AA-treated males, although sperm
beat cross frequency (in cycles/second), a measure of sperm motion and swimming pattern, was
significantly increased in the 60 mg/kg-day group. Clinical signs of neurotoxicity, including
unsteady movement and lethargy, were observed at the 45 and 60 mg/kg-day dose levels. High-
dose males exhibited significantly lower hindlimb grip strength than controls, in the absence of
microscopic evidence of sciatic nerve lesions.
GA as the putative toxin for dominant lethal effects
114 DRAFT-DO NOT CITE OR QUOTE
-------
To determine the relative potencies between AA and GA for dominant lethal effects,
Adler et al. (2000) administered 1-aminobenzotriazole (ABT), an inhibitor of CYP450
metabolism, to reduce the levels of the epoxide GA. Male mice were pretreated with ABT (i.p.
at 3 x 50 mg/kg) on 3 consecutive days followed by AA treatment (i.p. at 125 mg/kg) on day 4.
Parallel groups of animals were treated with AA (i.p. at 125 mg/kg), ABT (i.p. at 3 x 50 mg/kg)
or with the solvent double-distilled water. The experiment was repeated once with slightly
varied mating parameters. The authors state that results of both experiments showed that ABT
inhibited or significantly reduced the AA-induced dominant lethal effects supporting the
hypothesis that the AA metabolite GA is the ultimate clastogen in mouse spermatids. In the
NTP/CERHR (2004) review, however, the panel noted that the dominant lethals were decreased
2 weeks after treatment, but that, during the first week after treatment ABT did not decrease the
dominant lethal effect of AA, suggesting either that AA itself has dominant lethal effects or that
ABT requires more than 1 week to completely prevent metabolism to GA. A lack of a good
explanation for the delay before effect and other weaknesses in the results/argument (including a
decrease in the rate of dominant lethals in their study compared with other studies in mice, lack
of direct confirmatory evidence that ABT actually affected AA metabolism, and evidence that
ABT was also spermatotoxic and did not effectively antagonize the spermatotoxic effect of AA
treatment) prompted the panel to conclude that this study alone does not provide compelling
evidence for the effect of ABT treatment in support of the hypothesis that GA is the ultimate
clastogen in mouse spermatids.
More definitive support for GA as the primary toxin for dominant lethal effects comes
from a recent study by Ghanayem et al. (2005a), who compared germ-cell mutagenicity in male
CYP2El-null and wild-type mice treated with AA. CYP2El-null and wild-type male mice were
treated by i.p. injection with 0, 12.5, 25, or 50 mg AA in 5 mL saline/kg-day for 5 consecutive
days. At defined times after exposure, males were mated to untreated B6C3Fi females. Females
were killed in late gestation, and uterine contents were examined. Dose-related increases in
resorption moles (chromosomally aberrant embryos) and decreases in the numbers of pregnant
females and the proportion of living fetuses were seen in females mated to AA-treated wild-type
mice. No changes in any fertility parameters were seen in females mated to AA-treated
CYP2El-null mice. The authors state that their results constitute the first unequivocal
demonstration that AA-induced germ cell mutations in male mice require CYP2E1-mediated
epoxidation of AA. A further study by Ghanayem et al. (2005b) demonstrated the absence of
AA-induced genotoxicity in somatic cells in CYP2El-null mice compared with wild-type mice
treated with AA. These results support further evaluation of CYP2E1 polymorphisms in human
populations as a major determinant of variability in, and susceptibility to, AA genotoxicity in the
human population. The results also provide insight into results from previous investigations of
AA's germ cell activity in mice where stronger effects were observed after repeated
115 DRAFT-DO NOT CITE OR QUOTE
-------
administration of low doses compared with a single high dose. The differences may be due to
nonlinearities in AA metabolism (and thus internal levels and distribution of GA) for different
dose rates and durations.
Other reproductive function studies
Zenick et al. (1986) reproductive function study
Zenick et al. (1986) examined the potential effects of AA on male and female
reproductive function in Long-Evans rats. Male reproductive function was assessed in rats that
were given 0, 50, 100, or 200 ppm of AA in the drinking water (average AA intakes of 0, 4.6,
7.9, and 11.9 mg/kg-day)3 for 10 weeks. During a 3-week pretreatment period, males were
allowed to mate several times with ovariectomized, hormonally primed females. Body weights
of males were recorded at least once per week, and water consumption was monitored daily
throughout the study. During the treatment period, males were observed for clinical signs of
toxicity (frequency of observations was not reported) and mated with untreated primed females
on a weekly basis. Copulatory behavior (mount frequency, number of mounts and intromissions,
and ejaculation latency) with primed females was recorded during the mating session in which a
baseline was established (1 week prior to the start of AA treatment) and on alternating weeks
during treatment. At baseline and at treatment week 9, mated females were sacrificed and
ejaculate was removed from the genital tract for measurements of total sperm count, percent
motility, sperm morphology, and seminal plug weight. During treatment week 10, each control
and mid-dose (100 ppm) male was housed with an untreated estrous female overnight in order to
assess the reproductive success of AA-treated males. Following the sacrifice of dams on GD 17,
the number of fetuses and implantation sites were recorded. All treated males that survived the
treatment period were sacrificed during the following week and assessed for selected organ
weights (liver, brain, kidney, adrenals, spleen, heart, and reproductive organs). Histologic
examinations were performed on one testis and one epididymis per rat; the other testis and
epididymis were used for spermatid and sperm counting. The level of significance wasp < 0.05
for results of statistical analyses.
During treatment week 5, one 200-ppm male was found dead and two others were
sacrificed moribund. All other 200 ppm high-dose males were sacrificed during week 6 (i.e., this
dose group was terminated due to high mortality). No mortality was observed in any other
treatment groups. Throughout treatment, until death or sacrifice at week 6, the high-dose group
exhibited significantly lower mean body weight and water consumption than controls. Body
weight and water consumption in the mid-dose group were consistently, but not statistically
significantly, lower than controls. There were no statistically significant treatment-related
3 Calculated from graphically presented data on body weight and water consumption.
116 DRAFT-DO NOT CITE OR QUOTE
-------
effects on body or organ weights or sperm parameters in 50- or 100-ppm males following 10
weeks of treatment.
Hindlimb splaying was observed in the 200-ppm males by treatment week 4 and less
severely in 100-ppm males at week 8. Clinical signs of neurotoxicity were not seen in the
50-ppm group. Prior to the appearance of clinical signs of neurotoxicity, biweekly assessments
of copulatory behavior (data plotted graphically as square root or logarithmic transformations)
revealed statistically significantly increased numbers of mounts in the 100- and 200-ppm groups
relative to controls. At week 9, a nonsignificant increase in number of mounts was noted in low-
dose males. At treatment weeks 4 and 9, high- and mid-dose males, respectively, exhibited
statistically significant increases in the number of intromissions compared with controls. No
statistically significant treatment-related changes were seen in mount or ejaculation latency,
although the authors noted that only 4/12 200-ppm and 11/15 100-ppm males ejaculated within a
30-minute period on the final weeks of assessment (weeks 6 and 9, respectively).
Results of sperm analysis through week 9 of treatment and male fertility testing following
10 weeks of treatment are shown in Table 4-21. Mean sperm count was statistically significantly
lower in mid-dose males compared with controls, but the authors indicated that vaginal leakage
may have influenced total sample recovery, particularly in light of the fact that no adverse effects
on sperm parameters were seen in low- and mid-dose males examined histologically after
10 weeks of treatment. Sperm motility and morphology evaluations could not be performed in
the mid-dose group because sperm was recovered from the uterus of only 1 of the 11 females in
which ejaculation had been observed. Low-dose treatment had no statistically significant effect
on sperm parameters assessed. Statistically significant findings of fertility testing (performed
only on controls and mid-dose males) included a decreased number of pregnant females and
increased postimplantation loss in the mid-dose males.
117 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-21. Results of sperm analysis (baseline and week 9) and male
fertility testing (following 10 weeks of treatment) of Long-Evans rats exposed
to acrylamide in the drinking water
Parameter
Sperm count (X 10)
Baseline
Week 9
Sperm motility (%)
Baseline
Week 9
Sperm morphology (% normal)
Baseline
Week 9
Seminal plug weight (mg)
Baseline
Week 9
Females sperm positive/females mated
Females pregnant/females mated (%)
Postimplantation loss (%)e
Acrylamide concentration (ppm)
0
(n = 15)
46 ± 12b
56 ±18
43 ±9.1
41 ±11. 3
96 ±2.7
94 ±3.6
115 ±20
118 ±42
14/14
11/14(79%)
8.0 ±1.1
50
(n = 15)
45 ±19
36 ±23
39 ±9.2
46 ±11.2
96 ±2.3
96 ± 2.0
100 ±38
117 ±27
—
100
(n = ll)a
43 ± 14
14 ± 20C
41 ±6.3
a
95 ± 1.8
a
111±20
146 ± 49
15/15
5/15 (33%)f
31.7±3.8f
"Four males failed to ejaculate in a 30-minute trial.
bMean±SD.
'Significantly different from control, p < 0.05.
dSperm recovered from the uterus of only 1 female.
ePostimplantation loss = ([number of implants - number of fetuses]/[number of implants]) x 100
ip<0.01.
Note: The 200 ppm male dose group was terminated at week 6 due to high mortality.
Source: Zenicketal. (1986).
In a female reproduction assessment phase, Zenick et al. (1986) exposed regular-estrous
female Long-Evans rats (15/group) to AA in the drinking water at concentrations of 0, 25, 50, or
100 ppm for 2 weeks prior to mating and throughout gestation and lactation. The study authors
did not specify the intake levels of AA for the various exposure groups; however, dam body
weights were recorded at least once per week and water consumption was monitored daily
throughout the study. Based on graphically presented weekly mean body weight and water
consumption data, time-weighted average AA doses were approximately 3.4, 5.6, and
11.1 mg/kg-day during the 2-week prebreeding period; 5.3, 9.5, and 17.2 mg/kg-day during
3 weeks of gestation; and 6.5, 11.3, and 15.4 mg/kg-day during 3 weeks of lactation for the 25-,
50-, and 100-ppm treatment groups, respectively. Overall average doses for females were
calculated to be 5.1, 8.8, and 14.6 mg/kg-day.
During treatment week 3, untreated males were placed with the females at night for up to
7 nights. Presence of sperm in the vagina or a copulatory plug marked day 1 of gestation. Dams
were observed for clinical signs of toxicity, but the frequency of clinical observations was not
reported. Rat pups were sexed and weighed at birth (weighed weekly thereafter). Litters were
118 DRAFT-DO NOT CITE OR QUOTE
-------
culled to four/sex on lactation day 4 and to two/sex at weaning. Terminal sacrifice was
performed on PND 42.
High-dose dams exhibited hindlimb splaying as early as gestation week 2. The mean
body weight of this treatment group was statistically significantly lower than that of controls by
the end of the prebreeding treatment period and was more than 10 and 20% lower than controls
at some time points during gestation and lactation, respectively. Slightly, but significantly lower
mean body weight (approximately 6% lower) was seen in mid-dose dams but only during
lactation. The body weight effects were at least partially reflected in decreased water
consumption.
No statistically significant effects were seen regarding mating efficiency, live litter size,
or 4- or 21-day pup survival in any treatment group. Comparisons of body weights between
pups of treated dams and pups of control dams revealed slightly (but statistically significantly)
lower mean pup birth weights in male and female pups of high-dose dams. Significantly
depressed mean body weights were seen in male and female pups of mid- and high-dose dams
during lactation and postweaning periods (approximately 30-35 and 10% lower, respectively).
The study authors stated that statistical analysis revealed an association between cumulative AA
dose to dams and effects on pup body weight, but no significant associations between pup body
weights and dam body weights or water consumption.
In summary, the Zenick et al. (1986) study supports a LOAEL of 100 ppm of AA in
drinking water (7.9 mg/kg-day) for 10 weeks, based on male-mediated reproductive effects
(decreased percentage impregnation of nonexposed females and increased postimplantation loss)
in Long-Evans rats. No NOAEL was identified, as reproductive performance was not assessed
in the 50-ppm exposure group. Increased numbers of mounts and incidence of hindlimb splaying
were observed in the 100- and 200-ppm (7.9 and 11.9 mg/kg-day) exposure groups. Effects on
female reproductive performance were only observed as depressed body weights in offspring of
50- and 100-ppm dams, accompanied by decreased dam body weight. No effects on mating
efficiency, liver litter size, or pup survival were observed. For female-mediated reproductive
effects (decreased pup body weight), this study supports a LOAEL of 50 ppm (8.8 mg/kg-day)
and a NOAEL of 25 ppm (5.1 mg/kg-day).
Sakamoto and Hashimoto (1986) reproductive function study
Sakamoto and Hashimoto (1986) conducted a crossover study in ddY mice. In the
assessment of male reproductive effects, groups of males (14 controls and 14 at the high dose,
9/group at the other dose levels) were administered AA at levels of 0, 0.3, 0.6, 0.9, or 1.2 mM in
the drinking water for 4 weeks, resulting in doses of approximately 0, 3.3, 9.0, 13.3, and
16.3 mg/kg-day, respectively, based on body weight and water consumption data provided by the
authors. Half of the treated males were allowed to mate with untreated females (one male per
119 DRAFT-DO NOT CITE OR QUOTE
-------
three females) for a period of 8 days. All of the dams in each group (only half of the high-dose
group) were sacrificed on GD 13 and examined for numbers of implantations and resorptions.
After the remaining dams of the high-dose pairings were allowed to deliver, the number and
body weights of offspring were recorded. Offspring were observed for 4 weeks for any signs of
abnormal behavior and body weight gain. The remaining treated males were sacrificed
immediately following the dosing period, after which weights of liver, testis, and seminal vesicle
were recorded. Sperm counts and sperm morphology were assessed from epididymal samples.
The high-dose males exhibited slight signs of hindlimb weakness during or following
exposure. As shown in Table 4-22, results of examinations after 13 days of gestation revealed
significantly decreased fertility at the highest exposure level, significantly reduced numbers of
fetuses/dam, and increased numbers of resorptions at the two highest exposure levels relative to
controls. Significant decreases in both fertility and number of offspring were seen among dams
allowed to deliver. There were no significant treatment-related effects regarding pup body
weights or selected organ weights. Sperm analysis revealed significantly reduced numbers of
sperm and increased percentages of abnormal sperm in high-dose males.
120 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-22. Reproductive effects following exposure of male ddY mice to
acrylamide in drinking water for 4 weeks and subsequent mating with
untreated females
Effects observed following 13 days of gestation
Treatment
(mM)
0
0.3
0.6
0.9
1.2
Calculated dose
(mg/kg-day)
0
3.3
9.0
13.3
16.3
Fertility rate3
8/9
9/12
11/12
10/12
2/9c
Number of fetuses/dam
11.3 ± 1.4b
11.2±2.5
10.4 ±3.9
7.8±3.7d
2.5±1.5d
Number of
resorptions/dam
0.3 ±0.4
0.7 ±0.7
1.3 ±2.9
2.9 ±3.4
3.0±0.0d
Effects observed on the day of delivery
Treatment
(mM)
0
1.2
Calculated dose
(mg/kg-day)
0
16.3
Fertility rate
12/15
3/15c
Number of
offspring/dam
11.1±1.2
3.7 ± 1.2d
Offspring body weight
(g)
1.75 ±0.12
1.81 ±0.16
Effects on sperm count and morphology
Treatment
(mM)
0
0.3
0.6
0.9
1.2
Calculated dose
(mg/kg-day)
0
3.3
9.0
13.3
16.3
Sperm count
(xl05/mg epididymis)
35. 8 ±4.3
43.7 ±6.3
47.7±4.2d
49.9±7.1d
23.1±2.8d
Percentage abnormal
sperm
3.65 ±0.73
4.37 ±2.54
4.22 ±0.88
4.21 ±2.80
8.12±2.32d
"Number of fertile females/number of mated females.
bMean±SD.
°p < 0.05 vs. control by Fisher's Exact test.
dp < 0.05 by one-way ANOVA followed by Duncan's multiple-comparison procedure.
Source: Sakamoto and Hashimoto (1986).
The results identify 0.6 mM AA (9.0 mg/kg-day for 4 weeks) as a NOAEL and 0.9 mM
(13.3 mg/kg-day) as a LOAEL for male-mediated reproductive effects (decreased number of
fetuses/dam) in ddY mice (Sakamoto and Hashimoto, 1986). At a higher exposure level, 1.2
mM (16.3 mg/kg-day), more severe effects were observed, including decreased fertility,
increased resorptions, and sperm alterations. In female mice exposed to 1.2 mM (18.7 mg/kg-
day) for 4 weeks and mated with nonexposed mice, no clearly adverse reproductive effects were
observed.
Sakamoto et al. (1988) histology of testicular lesions
Sakamoto et al. (1988) administered AA (95% purity) to ddY mice as a single oral dose
(presumably gavage) of 100 or 150 mg/kg at age 30 days (prepubertal) or 60 days (adult).
Animals were killed 1, 2, 3, 5, 7, or 10 days after dosing. Testes were fixed in Bouin's fluid for
1 hour, cut, and then further fixed in formalin. Sections were stained with periodic acid-Schiff
stain and hematoxylin and eosin. Four animals were used for each treatment condition and
evaluation time point. The 150 mg/kg dose was lethal to 50% of the 30-day-old mice and 65%
121 DRAFT-DO NOT CITE OR QUOTE
-------
of the 60-day-old mice. In the prepubertal mice, body weight was significantly decreased at 1
and 5 days after dosing with 150 mg/kg AA. The numeric values for mean body weight at 2 and
3 days after dosing were similar to the 1- and 5-day values, but the larger standard deviation
prevented identification of statistical significance. In the adult mice, body weight was
significantly reduced 1, 2, and 3 days after dosing with 150 mg/kg AA. There were no
significant alterations in testicular weight at either dose of AA. There were no deaths and no
significant effects on body weight at 100 mg/kg AA in either age group. Histologic
abnormalities in the testes of prepubertal animals treated with 150 mg/kg AA appeared in
spermatids, particularly round spermatids (Golgi and cap phase) 1 day after treatment. Nuclear
vacuolization and swelling were the most common lesions in the spermatids. Degeneration of
spermatocytes and spermatogonia was also noted. By the second day after treatment, spermatid
degeneration was more prominent. On day 3, multinucleated giant cells were frequent. By days
7-10, clearing of the histologic abnormalities was evident. The description of the pattern of
histologic alteration was similar after treatment with 100 mg/kg and in adult animals. Overall,
spermatogonia, spermatocytes, Sertoli cells, and Leydig cells appeared more resistant to
AA-induced cell death than did spermatids.
Several additional studies have demonstrated reversible dominant lethal effects and
reversible effects on male fertility in animals orally exposed to AA for short time periods.
Working et al. (1987a, b) observed reversible male-mediated reproductive effects (dominant
lethal effects: increased implantation losses) in male F344 rats exposed to 30 mg/kg-day for 5
days. Sublet et al. (1989) observed dominant lethal effects (increased implantation losses) and
effects on male impregnation success in Long-Evans male rats exposed to oral doses as low as
15 mg/kg-day for 5 days. In this study, males were gavaged with 0, 5, 15, 30, 45, or 60 mg/kg
AA for 5 days prior to mating. Reduced fertility and increased preimplantation loss were found
in all dose groups except 5 mg/kg at week 1 posttreatment. Increased postimplantation loss was
seen at weeks 2 and 3 in the 15, 30, 45, and 60 mg/kg groups. In sperm samples collected from
the 45 mg/kg group, the percentage of motile sperm was modestly decreased to a statistically
significant degree (58% vs. 73% in controls) at week 3 but not at weeks 2 or 4. Sublet et al.
(1989) concluded that altered motility of sperm may have contributed to, "but can not completely
account for, the poorer reproductive performance of these males." Similarly, Tyl et al. (2000b)
observed significantly decreased fertility and increased postimplantation losses following mating
of untreated female rats with males that had been administered AA at gavage doses of 15, 30, 45,
or 60 mg/kg-day for 5 days prior to mating. No statistically significant effects were seen
regarding motility or concentration of epididymal sperm from AA-treated males, although sperm
beat cross frequency was significantly increased in the 60 mg/kg-day group. Clinical signs of
neurotoxicity, including unsteady movement and lethargy, were observed at the 45 and
122 DRAFT-DO NOT CITE OR QUOTE
-------
60 mg/kg-day dose levels. High-dose males exhibited significantly lower hindlimb grip strength
than controls, in the absence of microscopic evidence of sciatic nerve lesions.
In a summary paper, Bishop et al. (1997) reported tests of female "total reproductive
capacity" involving 29 chemicals tested over a 10-year period. Female mice were treated with a
single i.p. dose of AA (purity not stated) in Hanks' balanced salt solution (HBSS) at 0 or
125 mg/kg. The female mice were Fl hybrid SEC x C57BL6 and the males were Fl hybrid
C3H/R1 x C57BL10. The following day, females were paired with males for approximately
1 year. When litters were produced, pups were removed, counted, and killed. The number of
litters produced over either 347 or 366 days (the design changed during the course of these
studies, and the specific length for the AA study was not given) and the total number of offspring
produced was used to assess total reproductive capacity. There were no significant differences
between the AA- and vehicle-treated females in number of offspring/female (AA 142.6, control
146.2) or number of litters/female (AA 14.3, control 14.6). The paper lists 34 breeding pairs; it
is assumed (but not stated) that this number refers to the AA-treated animals. In a separate table
describing vehicle groups used for the 29 chemicals, the HBSS group with 146.2
offspring/female and 14.6 litters/female contained seven animals. (It was not stated that controls
were run concurrently. Neither standard error nor standard deviation were given.) Because this
is a summary of a large number of studies, the specifics of the AA study are neither available nor
presented, which represents a weakness, and it is difficult to ascertain the specifics of the AA
experiment or whether there were any characteristics that might flag the results as unusual or
give grounds for caution, another weakness in the AA portion of this study. The lack of
specifics and details moderate the conclusions that can be reached concerning AA's lack of
effect on female reproductive function.
4.3.2. Developmental Toxicity Studies
Developmental effects associated with oral exposure to AA are restricted to body weight
decreases in rats (Wise et al., 1995; Field et al., 1990; Zenick et al., 1986) and mice (Field et al.,
1990) and neurobehavioral changes in the offspring of female Sprague-Dawley rats exposed on
GDs 6-10 to 15 mg/kg-day, but not to 10 mg/kg-day (Wise et al., 1995) and in adolescent F344
rats exposed during gestation and lactation and extending through 12 weeks of age at an average
dose of 6 mg/kg-day, but not at 1.3 mg/kg-day (Garey and Paule, 2007). No exposure-related
fetal malformations or variations (gross, visceral, or skeletal) were found in Sprague-Dawley rats
exposed to doses up to 15 mg/kg-day on GDs 6-20 or in CD-I mice exposed to doses up to
45 mg/kg-day on GDs 6-17 (Field et al., 1990). These doses decreased the maternal weight
gain. No signs of hindlimb foot splay or other gross signs of peripheral or central neuropathy
were noted in suckling offspring of female Wistar rats that were given gavage doses of 25
mg/kg-day during the postnatal lactation period (Friedman et al., 1999a). The results of these
123 DRAFT-DO NOT CITE OR QUOTE
-------
studies are summarized in Section 4.7.1, and discussed below, except for the Zenick et al. (1986)
study, which has been discussed previously in Section 4.3.1. It is worth noting that many of the
adverse effects discussed in the mutagenicity and heritable germ cell sections can also be
considered adverse developmental effects (e.g., dominant lethality, heritable translocations,
specific locus mutations, abnormal conceptus).
Field et al. (1990) developmental toxicity study—gestational exposure
Field et al. (1990) administered AA (in distilled water) to groups of timed-mated
Sprague-Dawley rat dams (29-30/group) in gavage doses of 0, 2.5, 7.5, or 15 mg/kg-day on GDs
6-20 and to groups of timed-mated CD-I mice (30/group) at doses of 0, 3, 15, or 45 mg/kg-day
on GDs 6-17. Body weights were recorded on GD 0 and daily during treatment. Dosed animals
were observed daily for clinical signs of toxicity and sacrificed on the last treatment day.
Maternal body, liver, and intact uterus weights were recorded. Uteri were examined for number
of implant sites and resorptions. Live fetuses were counted, weighed, and examined for external
and visceral abnormalities, as well as skeletal variations and abnormalities.
Treatment-related effects are summarized in Table 4-23. Hindlimb splaying was
observed only in mice of the highest dose group (45 mg/kg-day). Statistically significant
adverse effects, relative to respective controls, included reduced maternal body weight gain
during treatment at high dose in both species, reduced weight gain corrected for gravid uterine
weight in rat dams of the 7.5 and 15 mg/kg-day groups (approximately 12 and 18% lower,
respectively), and reduced male and female fetal weights in the high-dose group of mice
(approximately 15% lower than controls). AA treatment did not adversely affect maternal liver
weight in rats or mice, percentages of pregnant rats or mice at sacrifice, number of implantations
in either species, or incidences of external, visceral, or skeletal malformations in rat or mouse
fetuses. The percentage of resorptions/litter did not differ significantly among treated and
control rats and mice, although a significantly increased percentage of litters with resorptions
was seen in mid-, but not high-dose mice. In rats, 15 mg/kg-day is the LOAEL and 7.5 mg/kg-
day is the NOAEL for maternal toxicity displayed as decreased weight gain. The highest dose
level, 15 mg/kg-day, is a NOAEL for fetal developmental effects (e.g., external, visceral, or
skeletal malformations or variations were not increased). In mice, 15 mg/kg-day is the NOAEL
and 45 mg/kg-day the LOAEL for maternal toxicity (decreased weight gain). The highest dose
level, 45 mg/kg-day, is a NOAEL for developmental effects in mouse fetuses.
124 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-23. Maternal and fetal effects in Sprague-Dawley rats and CD-I
mice following oral (gavage) administration of acrylamide to pregnant dams
Effects in rats
Number (%) dams pregnant at sacrifice
Maternal weight gain (g)a
Gestation period
Treatment period
Corrected weight gainb
Effects in mice
Number (%) dams pregnant at sacrifice
Maternal weight gain (g)a
Gestation period
Treatment period
Corrected weight gainb
Gravid uterine weight (g)
Number of litters
% resorptions/litter
% litters with resorptions
Mean male fetal body weight (g)/litter
Mean female fetal body weight (g)/litter
Dose (mg/kg-day)
0
23 (85)
151.1±4.1
107.7 ±4.0
78.6 ±2.3
2.5
26 (96)
152.0 ±4.2
111.0±3.5
75. 8 ±3.2
7.5
26 (90)
143.4 ±4.0
100.2 ±3.6
69.4±2.7C
15
24 (89)
139.2 ±3. 8
96.3±3.2C
64.3±3.7C
Dose (mg/kg-day)
0
28 (93)
23.6 ±0.7
21.2 ±0.7
4.7 ±0.4
18.8 ±0.6
28
3.5 ± 1.1
32.1
1.05 ±0.02
1.01 ±0.02
3
26 (87)
24.6 ±0.8
22.1 ±0.7
5.2 ±0.4
19.4 ±0.5
26
5.5± 1.5
46.2
1.03 ±0.02
0.97 ±0.02
15
29 (100)
21.5 ±1.1
19.5 ±1.0
5.0 ±0.4
16.5±0.8C
29
11.7±3.9
58.6C
1.02 ±0.01
0.99 ±0.01
45
25 (89)
19.9±0.7C
17.7±0.8C
3. 8 ±0.4
16.1±0.7C
25
3.4 ± 1.6
24.0
0.89±0.02C
0.86±0.02C
Includes all dams pregnant at sacrifice, mean ± SEM.
bWeight gain during gestation minus gravid uterine weight.
Significantly different from controls; p < 0.05.
Source: Field etal. (1990).
Wise et al. (1995) developmental neurotoxicity study—gestational exposure
Wise et al. (1995) investigated developmental neurotoxicity in pups of Sprague-Dawley
rat dams (12/group) that had been administered AA (in deionized water) at doses of 0, 5, 10, 15,
or 20 mg/kg-day from GD 6 to lactation day 10. Dams were observed daily for clinical signs.
Dam body weights were recorded periodically throughout gestation and lactation. All Fl pups
were counted, sexed, examined for external abnormalities, and weighed at birth. On PND 3,
each litter was reduced to five pups/sex. An additional four pups/sex/litter were retained for
behavioral assessment. Open-field behavior was tested on a single Fl rat/sex/litter on PNDs 13,
17, and 21 (the same animals were used for each session) and on PND 59 (Fl rats that had been
previously assessed for auditory startle habituation). Auditory startle habituation was tested on
PND 22 (naive Fl rats) and PND 59 (Fl rats previously subjected to open-field testing). Short-
term learning was assessed using a passive avoidance paradigm in previously untested Fl rats on
PNDs 24 and 59, and long-term retention was assessed in these rats 1 week later. The level of
significance was/? < 0.05 for results of statistical analyses.
Postsacrifice examinations were performed on one Fl pup/sex/litter following interim
sacrifice on PND 11 and on one Fl rat/sex/litter that had been used for passive avoidance testing
(sacrificed during postnatal week 11). Following sacrifice, body and brain weights were
recorded. Nervous tissues (brain, spinal cord, and unspecified peripheral nerve) were processed
125 DRAFT-DO NOT CITE OR QUOTE
-------
and stained with hematoxylin eosin. Histologic examinations were performed on these tissues
only from Fl rats of the control and 15 mg/kg-day treatment groups. All other Fl rats were
euthanized and discarded without further examination following completion of designated
testing.
Hindlimb splaying was observed in all FO dams of the two highest dose levels (15 and
20 mg/kg-day) during the first few days of lactation. No clinical signs of neurotoxicity were
seen in FO dams of lower dose groups. Statistically significant decreases in average maternal
weight gain between GDs 6 and 20 were observed in 15 and 20 mg/kg-day groups (14 and 26%
below controls, respectively). No adverse effects on maternal body weight gain during gestation
were seen at lower dose levels. All FO and Fl rats of the 20 mg/kg-day dose group were
euthanized between GD 24 and PND 4, due to high pup mortality (33% by PND 3) that was
likely the result of obvious maternal toxicity in this dose group. Between PNDs 4 and 21, pup
mortality (13%) was also seen in the 15 mg/kg-day dose group but not in other groups. Visceral
examination of dead pups did not reveal a cause of death. During the lactational dosing period
(PNDs 0-10), FO dams of the 10 and 15 mg/kg-day dose groups exhibited statistically significant
decreased average weight gain (45 and 90% lower than controls, respectively). No adverse
effect on maternal weight gain during lactation was seen in the 5 mg/kg-day group.
The study authors noted statistically significant, dose-related decreases in average pup
weights during the preweaning period. The effect was slight and transient in the 5 mg/kg-day
group (5-9% below controls and statistically significant only in female pups), moderate in the
10 mg/kg-day group (9-23% lower than controls), and still more severe in the 15 mg/kg-day
group. During the postweaning period, male and female Fl rats of the 15 mg/kg-day group
continued to exhibit significantly decreased average body weight (23 and 15% lower than
control at postnatal week 9). Body weight gain in Fl males (but not Fl females) was also
significantly depressed in the 15 mg/kg-day group. The average body weight of Fl males of the
10 mg/kg-day group was significantly less than controls (6% lower) at postnatal week 9, but
overall weight gain in this group was similar to that of controls during this period. No adverse
effects on postweaning Fl body weights were seen in the 5 mg/kg-day group. No deaths or
adverse clinical signs were seen in any group of Fl rats during the postweaning period.
No significant treatment-related effects were seen concerning open-field activity of Fl
rats tested on PNDs 13 or 17. When tested on PND 21, the only statistically significant effect
observed was that of increased overall average horizontal activity among female (but not male)
pups of the 15 mg/kg-day group. This effect was not seen in any groups that were tested as
adults. A decrease in the overall average peak amplitude of the auditory startle habituation test
was seen only in male and female Fl rats of the 15 mg/kg-day group tested on PND 22 and in
female Fl rats tested as adults. No apparent treatment-related effects were seen regarding
performance in passive avoidance testing.
126 DRAFT-DO NOT CITE OR QUOTE
-------
The results indicate that 5 mg/kg-day is the NOAEL and 10 mg/kg-day is the LOAEL for
maternal toxicity (decreased weight gain) in Sprague-Dawley rats (Wise et al., 1995). Higher
doses (15 and 20 mg/kg-day) produced hindlimb splaying and more severe effects on maternal
weight gain. The lowest dose, 5 mg/kg-day, is a developmental LOAEL for decreased body
weights in the offspring during the preweaning period. Neurodevelopmental effects in the
offspring (increased overall average horizontal activity and decreased auditory startle response)
were observed at 15 mg/kg-day but not at 5 or 10 mg/kg-day. Histologic examination of brain,
spinal cord, or peripheral nerve tissue samples collected on PND 11 and postnatal week 11
revealed no changes, relative to controls, in 15 mg/kg-day offspring.
Husain et al. (1987) developmental neurotoxicity study—lactational and postnatal exposure
Husain et al. (1987) assessed the potential for AA-induced neurotoxic effects on levels of
catecholamines (noradrenaline, dopamine, and 5-hydroxytryptamine) and activity of selected
enzymes in the brain of the developing rat. Two separate protocols were used in the study. In
one protocol, pups (number was not reported) were exposed during lactation via their nursing
mothers, which were administered AA orally at a dose level of 25 mg/kg-day (in 0.15 M NaCl)
throughout lactation. Brain levels of the catecholamines and enzymes of interest were measured
in selected pups that were serially sacrificed at 2, 4, 8, 15, 30, 60, and 90 days of age. The
second protocol involved the oral administration of AA (25 mg/kg-day) for 5 consecutive days to
rats of 12, 15, 21, or 60 days of age, followed by analysis of catecholamine levels in various
brain regions. Vehicle controls were included in both protocols. The level of significance was
p < 0.05 for results of statistical analyses.
No treatment-related effects on body or brain weights were seen in rats that had been
exposed via their mothers. Between the ages of 2 and 15 days, statistically significantly
decreased levels of noradrenaline, dopamine, and 5-hydroxytryptamine were observed in the
whole brains of offspring (5-hydroxytryptamine levels were also decreased in 30-day-old
offspring) but not at later time points. Compared with age-matched controls, the brain activity of
monoamine oxidase was significantly increased and that of acetylcholine esterase was
significantly decreased in offspring sacrificed at 2-30 days of age but not in 60- and 90-day-old
rats. Twelve-, 15-, and 21-day-old (but not 60-day-old) rats, treated according to the second
protocol, exhibited significantly decreased concentrations of noradrenaline in pons medulla and
basal ganglia, relative to age-matched controls. Noradrenaline was significantly decreased in the
mid-brain of all tested age groups. Other significant treatment-related alterations in brain
catecholamines included decreased levels of dopamine in cerebellum and midbrain at all ages
tested and in pons medulla of 12-, 15-, and 21-day-old rats and decreased levels of
5-hydroxytryptamine in pons medulla, hypothalamus, and cerebral cortex at all ages tested. The
study authors stated that decreased levels of catecholamines were associated with "progressive
127 DRAFT-DO NOT CITE OR QUOTE
-------
development of behavioral disorders leading to complete hindlimb paralysis," but the report does
not describe any specific observations of behavior in the rats. Thus, the report provides evidence
of neurochemical changes in the male offspring of rats exposed to 25 mg/kg-day during 21 days
of lactation but does not provide clear information that the male offspring had behavioral
disorders including hindlimb paralysis.
Friedman et al. (1999a) developmental neurotoxicity study—lactational exposure
Friedman et al. (1999a) administered AA to female Wistar rats (15/group) with litters at
gavage doses of 0 or 25 mg/kg-day in saline throughout the lactation period (PNDs 0-21). Dams
were weighed on PNDs 0, 4, 7, 14, and 21. Maternal food and water consumption were
measured for the intervals of PNDs 0-4, 4-7, 7-14, 14-21, and 0-21. Clinical observations
were made at least twice daily during the dosing period. On PNDs 7, 14, and 21, dams were
evaluated by an extensive functional observational battery that included observations of home
cage and open field behavior, clinical signs during handling, and sensory and neuromuscular
assessment (tail pinch response, hindlimb foot splay and grip strength, approach response, pupil
response, startle response, and pupil size). All live pups were individually counted, sexed,
weighed, and examined grossly at birth. Pups were examined at least twice daily for mortality
and morbidity.
At weaning on PND 21, maternal rats were weighed and sacrificed. Thoracic and
abdominal cavities and organs were examined, uterine implantation sites counted, and brain and
one sciatic nerve were fixed. Histopathologic examinations were performed on the sciatic nerve
preparation of each maternal rat, but details on tissue preparation and staining were not provided.
Female offspring were subjected to gross external examination and sacrificed on PND 21. Brain,
pituitary, and one sciatic nerve from one female pup of each litter were retained in fixative.
Male pups were weighed individually on PND 21 and weekly thereafter until PND 91. Ten male
pups/group were selected for grip strength measurements (forelimb and hindlimb) on PNDs 30,
60, and 90. Any selected male rat not available for grip strength assessment was replaced by
another male from the same litter, if possible. On PNDs 30, 60, and 91 (following grip strength
testing), one male rat/litter was sacrificed (when possible), and brain, pituitary, and one sciatic
nerve were retained in fixative. On PND 91, all remaining male pups were subjected to external
examination at terminal sacrifice.
For statistical analysis of results, the unit of comparison was the maternal female or the
litter. Statistical analysis of the data included Bartlett's test for homogeneity of variances,
general linear models procedures for ANOVA, the Kruskal-Wallis test, %2 test, and a test for
statistical outliers. Differences in results between treated and control groups were considered
significant at the level ofp < 0.05.
128 DRAFT-DO NOT CITE OR QUOTE
-------
Mean maternal body weight was similar between controls and treated groups just prior to
the beginning of dosing. Significantly lower body weight among AA-treated dams (relative to
controls) was noted as early as PND 4. Between PNDs 14 and 21, both controls and treated
dams exhibited weight loss, although the weight loss of treated dams was significantly greater
than that of controls. For the entire treatment period (PNDs 0-21), treated dams exhibited a
mean weight loss of 14 g, whereas a net mean weight gain of 47 g was seen in controls. Clinical
signs of toxicity were apparent in treated dams, beginning on PND 4; the range of clinical signs
broadened and increased in severity during the remainder of the treatment period. By PND 21,
two of the dams had been sacrificed moribund (PNDs 18 and 20), and there were numerous signs
(clinical, behavioral, and functional observational battery) of neurotoxicity in the surviving
dams. No histopathologic evidence of degeneration in sciatic nerve preparations from treated
dams was found.
Increased mortality and reduced body weights were observed in offspring of AA-treated
dams during the lactation period and were likely the result of maternal toxicity. Likewise,
clinical signs and gross examination of offspring during the lactation period were consistent with
inanition (i.e., little or no milk in the stomach). Body weight gain of postweaning males
paralleled that of controls, although the mean body weight in the AA-treated group remained
lower than that of controls throughout the postweaning observation period. Grip strength was
significantly lower in the AA-treatment group of male weanlings when tested on PND 30 but
was not significantly different from controls when tested on subsequent PNDs 60 and 90.
The study identifies 25 mg/kg-day for 21 days during lactation as a LOAEL producing
progressive signs of neurobehavioral disorders, including hindlimb foot splay in Sprague-
Dawley rat dams without histologic evidence of sciatic nerve damage. Nursing offspring of
exposed dams showed reduced weight gain, increased mortality, and little or no evidence of milk
in their stomachs. After weaning, surviving pups showed signs of recovery, including normal
weight gain and increasing grip strength over time. Characteristic signs of AA neurotoxicity,
such as hindlimb splaying, were not observed in the offspring.
Garey andPaule (2007) developmental neurotoxicity study—exposure during gestation,
lactation, and through 12 weeks of age
Garey and Paule (2007) evaluated performance in an operant test of cognitive motivation
in adolescent male and female F344 rats exposed to AA during gestation, lactation, and through
12 weeks of age. Pregnant F344 rats were exposed by gavage throughout gestation to AA doses
of 0, 0.1, 0.3, 1.0, or 5.0 mg/kg-day. On PNDs 1-22, offspring received the same gavage dose
received by their dam. On PNDs 22-85, offspring were provided AA in drinking water at
concentrations (0, 1, 3, 10, and 50 ppm) designed to maintain similar administered doses (based
on an assumed drinking water intake of 10% of body weight per day). Average daily doses
129 DRAFT-DO NOT CITE OR QUOTE
-------
between PNDs 40 and 85 are estimated at approximately 0, 0.2, 0.5, 1.6, or 7.8 mg/kg-day, from
reported average water intake values for offspring between PNDs 40 and 85 (15.5% of body
weight per day) and the reports that average water intakes for high-dose offspring between PND
40 and 50 was about 20% of body weight per day and average daily AA dose was 20 mg/kg-day
during this period. Offspring (one male and one female/litter; eight or nine/sex/exposure group)
were trained between 3 and 6 weeks of age to operate levers to obtain food pellets. Offspring
were subsequently evaluated in a progressive ratio task of motivation to obtain food in
14 10-minute sessions distributed between weeks 6 and 12 of age. In each session, food was
obtained with a progressively greater number of actions. For example, the first food pellet
("reinforcer") was obtained after one press of the lever; the second pellet was obtained after two
presses; and the third with three presses. Measured behavioral endpoints were number of food
pellets obtained per session (i.e., "number of food reinforcers earned"), post-reinforcement pause
(average duration from reinforcer delivery to the next lever press), and response rate (number of
presses/second over the entire 10-minute test session). No significant (p < 0.05) differences
were found between control and exposed groups in drinking water intake or body weight
throughout the testing period. Offspring in the high-dose group showed significant (p < 0.05)
decreases, compared with control values, in average reinforcers earned per session (5.5 ± SEM
0.2 vs. 7.2 ± SEM 0.2), and response rate (0.041 ± SEM 0.002 vs. 0.065 ± SEM 0.003). No
significant changes in these endpoints, compared with controls, were found in the other exposure
groups. No significant effect of exposure was found on post-reinforcement pause in any
exposure group. The high dose in this study, (about 6 mg/kg-day—an average of the gavage
dose and the estimated drinking water dose during testing) is a LOAEL for effects on measures
of cognitive motivation; the NOAEL is 1.3 mg/kg-day.
Other developmental toxicity studies
Genotoxic effects observed in the germ cells of mice following i.p. injection of
125 mg/kg AA included a weakly positive result for sperm head DNA dealkylation and a
positive result for sperm head protamine alkylation (Sega et al., 1989). Significant increases in
sperm head abnormalities were observed in epididymal samples taken from male ddY mice that
had received AA in the drinking water at a concentration of 1.2 mM for 4 weeks (Sakamoto and
Hashimoto, 1986).
Edwards (1976) treated Porton strain rats with AA (purity not specified) in the diet. In
the first experiment, eight females were given 200 ppm in powdered feed from the day a plug
was found until parturition. Offspring were apparently reared by their dams and were followed
until 6 weeks of age with weekly weights taken and observations made for abnormal gait. The
dams were described as showing "slight abnormalities of gait" at the times the litters were born.
There were no external abnormalities. The birth weights were similar to a control group (it is not
130 DRAFT-DO NOT CITE OR QUOTE
-------
clear if this control group was the same as the control group used in the second experiment,
described below), and litters were described as gaining weight normally until weaning, without
abnormalities of gait. No detailed information was presented.
In a second experiment by Edwards (1976), six pregnant females were given 400 ppm
AA in the diet from the day of mating until 20 days thereafter when they underwent cesarean
section. Six control dams were fed powdered diet without AA. Uteri were examined for
resorptions (presumably uteri: the text states that placentas were examined for resorptions). One
third of fetuses underwent Wilson sectioning, and the remaining fetuses were processed in
alizarin red for skeletal evaluation. Maternal feed intake was reduced in the AA group
(12.0 ± 0.8 g/rat/day, mean ± SEM) compared to the control group (23.0 ± 1.8 g/rat/day). The
weights of the rats were not given (assuming a 300 gram pregnant rat, 12 g/rat/day feed
containing 400 ppm AA represents a daily dose of 16 mg/kg-day). Fetal weights were reduced
by AA treatment (AA 2.4 ± 0.05 g, control 3.2 ± 0.05 g, mean ± SEM). (The/>-value reported by
the authors using the Student t-test was >0.2; however, the t-test performed by CERHR gave
p < 0.0001.) Four fetuses were found dead in one uterine horn in the AA-treated group, and
three resorptions were present in one litter in the control group. There were no fetuses with
abnormalities and "there was no increase in the occurrence (approx. 10%) of a naturally
occurring defect in the rib structure." No data were shown.
In a third experiment, Edwards (1976) administered 100 mg/kg AA in water i.v. to each
of four pregnant rats on GD 9 (plug date unspecified). The rationale for this timing was the
statement that GD 9 is when the nervous system is most susceptible to teratogenic effects. Pups
were apparently delivered and reared by their dams and on the third day of life, pups were
examined for external appearance and righting reflex. Offspring were followed for 6 weeks
during which the day of eye opening was noted and animals were evaluated for gait and were
weighed weekly. Two rats from each litter (sex unspecified) were perfused with
formaldehyde/acetic acid/methanol, and brains, spinal cord, and peripheral nerves were
evaluated by light microscopy (sectioning and staining unspecified). Two rats/litter (sex
unspecified) were killed with a barbiturate for dissection for gross abnormalities. Brain weight
was recorded. Four pregnant control rats were injected with saline and presumably handled in
the same manner. There were no differences among groups in birth weight, pup weight 24 hours
or 3 days after birth, righting reflex, or day of eye opening (data were not shown). There were
no abnormalities of nervous system tissues by gross examination or by light microscopy.
In summary of all three studies, due to the limited number of doses, very limited number
of pregnant rats/group, limited number of outcomes measured, and missing data necessary for
full evaluation of this report, the conclusions presented in the report are questionable.
Bio/dynamics Inc. (1979) administered AA in the feed to female Sprague-Dawley CD
rats at 0, 25, or 50 ppm for 2 weeks prior to mating, and on GDs 0-19. AA intake was estimated
131 DRAFT-DO NOT CITE OR QUOTE
-------
at 1.75-1.90 and 3.45-3.82 mg/kg-day in the 25 and 50 ppm dose groups, respectively. Litters
were standardized to three male and three female pups on PND 4 and pups were examined for
postnatal growth and mortality until weaning (PND 21). A slight but significant reduction in
body weight gain was observed in the 50 ppm dams during the premating period. No difference
among treatment groups was observed for mating and pregnancy indices, gestation length,
neonatal viability, live litter size at birth, pup survival throughout the lactation period, and pup
weights. Albert Einstein College of Medicine (1980) conducted a histopathologic evaluation of
brain and spinal cord and sciatic, tibial, and plantar nerves and reported that AA-associated
changes were confined to scattered nerve fiber degeneration in the sciatic and optic nerves. The
incidence and severity of these histologic effects were not provided.
In a study conducted at the National Institute for Environmental Health and Sciences,
Walden et al. (1981) evaluated the activity of five intestinal enzymes in the offspring of
AA-treated Sprague-Dawley rats. Dams were treated from GD 6 to 17 (insemination = GD 0)
with AA (purity not given) 20 mg/kg-day or water by gavage for a total cumulative dose of
200 mg/kg. There were 17 dams in each treatment group. On the day of birth (PND 0), pups in
each treatment group were pooled and divided among dams to produce four groups: control
dams with control pups (C-C); treated dams with treated pups (T-T); control dams with treated
pups (C-T); and treated dams with control pups (T-C). Four pups were removed from each litter
without regard to sex for intestinal enzyme analysis on PND 14, 21, and 60. The first 10-15 cm
of intestinal mucosa was scraped and homogenized (the report implies that the scrapings of the
four animals were pooled). Kinetic spectrophotometric assays were performed for alkaline
phosphatase, citrate synthase, and lactate dehydrogenase. Endpoint spectrophotometric assays
were performed for acid phosphatase and p-glucuronidase. Dams were killed on PND 24, after
weaning, and intestinal enzymes were measured by the same methods. The results of differences
(either increases or decreases) in enzyme activities for pups in the different groups were
indicative of prenatal effects (C-T compared with C-C), lactational effects (T-C compared with
C-C), or enhancement of prenatal effects (T-T compared with C-T) and are presented in
Table 4-24. Statistical analysis was performed by Mann-Whitney U-test (2p < 0.05). The results
indicate that prenatal exposure to AA in Sprague-Dawley dams at the doses stated above, and
lactational exposure to pups, significantly changed intestinal enzyme levels in pups during early
development. It is unknown whether these changes result in subsequent adverse structural or
functional effects. There were no differences in maternal body weight or in litter averages for
pup number, weight, or sex ratio. Dam intestinal enzyme levels did not differ from this exposure
level of AA.
132 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-24. Differences in marker enzymes in the small intestine of pups
cross-fostered to acrylamide-treated or control dams during postnatal
lactation
Intestinal enzyme
Alkaline phosphatase
Citrate synthase
Lactate dehydrogenase
Acid phosphatase
p-glucuronidase
Effect3
Prenatalb
Lactational0
Enhancement of prenatal effectd
Prenatal
Lactational
Enhancement of prenatal effect
Prenatal
Lactational
Enhancement of prenatal effect
Prenatal
Lactational
Enhancement of prenatal effect
Prenatal
Lactational
Enhancement of prenatal effect
Postnatal day
14
t
-
t
-
-
-
-
-
-
t
t
-
-
-
4
21
t
t
t
-
-
-
-
t
-
-
-
4
t
t
t
60
1
1
t
-
-
-
-
-
-
1
-
t
-
-
-
a| = Increase; J, = decrease; - = not significantly different. All reported effects are significant at the 2p < 0.05 level
using the Mann-Whitney U-test.
bC-T values compared with C-C values.
°T-C values compared with C-C values.
dT-T values compared with C-T values.
Source: Waldenetal. (1981).
A study by Rutledge et al. (1992) is unique in that female mice were dosed with AA
selectively during the perifertilization period at 125 mg/kg i.p. 1, 6, 9, or 25 hours after mating.
These times represented fertilization, the early pronuclear stage, pronuclear DNA synthesis, and
the two-cell stage, respectively. On GD 17, the uteri were inspected for resorptions, embryonic
death, and live fetuses. Live fetuses were inspected for external abnormalities. The number of
live fetuses was decreased and the number of resorptions was increased at all treatment times.
Among live fetuses, abnormalities were increased with treatment 6, 9, and 25 hours after mating.
In spite of the lack of important details in the paper and a discrepancy between text and table in
reporting the results, this study showed that an acute administration of AA at a high dose during
the perifertilization period can produce very early death or structural malformations.
Walum and Flint (1993) evaluated the effect of AA (purity not given) on rat midbrain
cells (obtained from embryos collected on day 13 postmating) in culture. This brain area is one
rich in both dopamine and gamma-aminobutyric acid (GABA) receptors developmentally. In
this assay, sometimes called micromass culture, neural epithelial cells in suspension aggregate
into foci of interconnected cells. A reduction in the number of such foci without a reduction in
cell number or viability is taken as evidence of disruption of developmental processes. In this
study, 10 |ig/mL AA was determined to reduce the number of foci by 25% without decreasing
133 DRAFT-DO NOT CITE OR QUOTE
-------
cell number, assessed by neutral red staining and protein content. Uptake of dopamine and
GABA were also decreased by AA exposure (the text indicates that GABA uptake was
"virtually" unaffected; the data table shows a statistically significant 8% reduction in GABA
uptake). The authors concluded that AA may reduce the "differentiation and development of
dopaminergic projections" in the developing rat brain. This study provides an in vitro
assessment of a potential mechanism of AA toxicity and a suggestion of how this mechanism
might be established. This approach is a good beginning for whole-animal researchers to follow-
up concerning these events within an in vivo model.
4.4. HERITABLE GERM CELL STUDIES
Five heritable translocation studies (Adler et al., 2004, 1994;Generoso et al., 1996; Adler,
1990; Shelby et al., 1987) and two specific mouse locus mutagenicity assays (Ehling and
Neuhauser-Klaus, 1992; Russell et al., 1991) are available. These studies all found positive
results following exposure of male mice to 40-100 mg/kg i.p. doses of AA, but do not provide
information for possible effects at lower exposure levels. No experiments have studied the
potential for AA to induce heritable germ cell effects in the female germ line. The heritable
germ cell effects in male mice are consistent with the extensive evidence supporting dominant
lethal effects in male murine test animals. In addition, there are two reports of increased
incidence of male-mediated stable chromosomal aberrations in two-cell mouse embryos
following exposure of male mice to 50 mg/kg AA and mating to unexposed females (Marchetti
et al., 2009, 1997). These studies found correlations between stable chromosomal aberrations
and the percentages of offspring with reciprocal translocations.
The seven heritable germ cell studies in mice are briefly discussed below, and the results,
as tabulated by Favor and Shelby (2005), are included in Tables 4-25, 4-26, and 4-27. These
studies are also listed in Appendix B, Table B-l that summarizes the mutagenicity assay results.
Recent reviews and discussions regarding the results of available heritable germ cell studies for
AA include Besaratinia and Pfeifer (2007), Carere (2006), Exon (2006), Shipp et al. (2006),
Favor and Shelby (2005), and NTP/CERHR (2004).
134 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-25. Frequency of translocation carriers in offspring derived from males exposed to acrylamide or glycidamide
Dose" (mg/kg)
Historical control
50 AA i.p.
100 AA i.p.
100 GA i.p.
5 x 40 AA i.p.
5 x 50 AA i.p.
5 x 50 AA i.p.
5 x 50 AA i.p.
5 x 50 AA dermal
Mating interval1"
progeny tested
Translocation carriers0
,e,f
7-16
7-16
3.5-7.5
7-10
7-10
7-11
36-42
1.5-8.5
Males
ll,292d
9,890
362f
367f
669
162
125
57
556
258
Females
48
449
217
Males
7 (0.06)
5 (0.05)f
2 (0.55)f
10 (2.72)f
135(20.17)
39 (24.07)
49 (39.20)
17 (29.82)
2 (0.36)
28(10.85)
Females
6(12.5)
0(0)
13 (5.99)
Reference
Generoso et al. (1996)
Adleretal. (2002)
Adleretal. (1994)
Adleretal. (1994)
Generoso et al. (1996)
Shelby etal. (1987)
Shelby etal. (1987)
Adler(1990)
Adler(1990)
Adleretal. (2004)
a5 x 40 and 5 x50 represent 40 or 50 mg AA/kg on 5 consecutive days.
bDays posttreatment.
°See text for methods to ascertain translocation carriers. Frequency (%) of translocation carriers given in parentheses.
laboratory historical control used for statistical comparisons of the translocation frequencies reported by Shelby et al. (1987) and Generoso et al. (1996).
laboratory historical control used for statistical comparisons of the translocation frequencies reported by Adler (1990) and Adler et al. (2004, 1994).
fBoth male and female Fl animals were tested but not reported separately.
Source: Favor and Shelby (2005).
Table 4-26. Results for specific locus mutations recovered in offspring of male mice exposed i.p to 50 mg/kg acrylamide
on 5 consecutive days
Mating interval (days posttreatment)
1-7
8-14
15-21
22-28
29-35
36-42
43-49
>49
Historical control
Number of offspring
113
1,506
5,077
5,191
5,312
5,353
6,419
17,112
801,406
Number of mutations"
0(0)
2(0.13)
1 (0.02)
0(0)
0(0)
1 (0.02)
1 (0.02)
0(0)
43 (0.01)
""Frequencies (%) of specific locus mutations given in parentheses.
13 5 DRAFT-DO NOT CITE OR QUOTE
-------
Sources: Data from Russell et al. (1991); table from Favor and Shelby (2005).
Table 4-27. Results for specific locus mutations recovered in offspring of male mice exposed to acrylamide as a single
100 or 125 mg/kg i.p. dose
Dose (mg/kg)
Historical control
100
125
Mating interval (days posttreatment)
—
1-4
5-8
9-12
13-16
17-20
21-42
>42
1-4
5-8
9-12
13-16
17-20
Number of offspring
248,413
1,362
2,226
2,421
2,453
2,574
2,925
23,489
771
1,924
1,948
2,419
2,598
Number of mutations"
22 (0.01)
0(0)
1 (0.04)
2 (0.08)
0(0)
0(0)
0(0)
6 (0.03)
0(0)
2(0.10)
1 (0.05)
0(0)
0(0)
""Frequencies (%) of specific locus mutations given in parentheses.
Note: Only the 100 mg/kg-treated males were used to establish a permanent monogamist mating starting on day 21 to assay for effects on spermatogonia (i.e., for effects
>43 days posttreatment).
Sources: Data from Ehling and Neuhauser-Klaus (1992); table from Favor and Shelby (2005).
136 DRAFT-DO NOT CITE OR QUOTE
-------
Heritable translocation studies
Shelby et al. (1987) administered AA i.p. at 40-50 mg/kg-day for 5 consecutive days to
male C3H/E1 mice. Matings on days 7-10 following the last injection yielded a high frequency
of translocation carriers in the Fl male population, demonstrating that AA is an effective inducer
of translocations in postmeiotic germ cells. The proportions of male progeny that were sterile or
semi-sterile after paternal treatment with 50 and 40 mg/kg-day for 5 days were 49/125 and
39/162, respectively, compared with 17/8,095 in the historical control. All 10 of the semi-sterile
males sampled from the 5 x 50 treatment for cytogenetic analysis of spermatocytes had
translocations.
Adler (1990) administered AA i.p. at 50 mg/kg-day for 5 consecutive days to male
C3H/E1 mice, which were then mated to untreated female 102/E1 mice on days 7-11 and again
on days 36-42 posttreatment. There were 23 translocation heterozygotes among 105 progeny
from the offspring of the 7-11 day mating interval. Among the offspring of the treated males,
there were 17 male translocation carriers among 57 male offspring and 6 female translocation
carriers among 48 female offspring (male vs. female,/* < 0.05). In the second mating interval
(36-42 days after treatment), 1,005 offspring were produced, of which 2 males were
translocation carriers. This rate did not differ from the historical control in the author's
laboratory when considered on a total-offspring basis but was significantly greater than the
historical control (p = 0.03) if considered on a male-offspring basis. All semi-sterile and sterile
mice from treated parental males were analyzed cytogenetically, with 22/25 semi-sterile mice
and 3/4 sterile mice confirmed as translocation carriers. This study provides further evidence for
AA-induced chromosomal damage in postmeiotic rodent germ cells.
Adler et al. (1994) administered AA i.p. as a single 50 or 100 mg/kg dose to male
C3H/E1 mice, which were then mated on days 7-16 posttreatment to untreated female 102/E1
mice. Translocation carriers among the Fl progeny were selected by a sequential procedure of
fertility testing and cytogenetic analysis, including G-band karyotyping, to determine the
chromosomes involved in the respective translocations. The frequency of confirmed
translocation carriers was 2/362 in the 50 mg/kg treatment group and 10/367 in the 100 mg/kg
treatment group. Both frequencies were significantly greater than the historical control, 5/9,890.
Clustering was not apparent, as indicated by the fact that all translocations were unique.
Adler et al. (2004) conducted heritable translocation tests with dermal exposure of male
mice to AA. Male C3FI/E1 mice were treated with five dermal exposures of 50 mg/kg AA and
mated 1.5-8.5 days after the end of exposure to untreated female 102/E1 mice. Pregnant females
were allowed to come to term and all offspring were raised to maturity. Translocation carriers
among the Fl progeny were selected by a sequential fertility testing and cytogenetic analysis
including G-band karyotyping and M-FISH. A total of 475 offspring were screened and
137 DRAFT-DO NOT CITE OR QUOTE
-------
41 translocation carriers were identified. The observed translocation frequency after dermal
exposure was 8.6% as compared to 21.9% after similar i.p. exposure (Adler, 1990). The
calculated ratio of end effects in this study of i.p. vs. dermal exposure is 0.39.
Favor and Shelby (2005) summarized the cytogenetic analysis from the Adler et al.
(2004, 1994) and Adler (1990) studies to emphasize the appearance of complicated chromosomal
rearrangements induced by AA. Among the 77 semi-sterile and sterile animals analyzed, 66
were carriers of reciprocal translocations between two chromosomes, 2 carried translocations
among three chromosomes, 6 were carriers of two independent reciprocal translocations each
between two chromosomes, 2 were carriers of a reciprocal translation between two chromosomes
plus an inversion on a third chromosome, and one animal carried a translocation among three
chromosomes plus a reciprocal translocation between two chromosomes.
Generoso et al. (1996) administered a single i.p. dose of GA at 100 mg/kg to male
(C3H/RL x 101/RL)F1 mice. Among the 669 male progeny of GA-treated sires, 135 (20.18%)
were confirmed as heterozygous translocation carriers, compared with 6% from the historical
controls. The GA treatment generated a much higher frequency of translocations in male
progeny than the comparable 100 mg/kg i.p. dose from AA reported in Adler et al. (1994)
(20.17 vs. 2.72%). Although the mating interval was different (3.5-6.5 days posttreatment for
GA and 7-10 days posttreatment for AA) and thus the spermatogonial stages were different and
the studies were conducted in two different laboratories, the results demonstrate that GA is a
potent inducer of chromosomal damage in postmeiotic rodent germ cells.
Specific locus studies
Russell et al. (1991) evaluated specific locus mutations, as well as fertility (measured as
litter size/fertile female) and dominant lethals resulting from AA exposure to male mice from an
i.p. 50 mg/kg-day dose for 5 consecutive days. Males were mated at specific intervals after
mating to T-stock females homozygous for a (non-agouti), b (brown), cch (chinchilla), p (pink-
eyed dilution), d (dilute), se (short ear), and s (piebald). AA was effective in the first 2 weeks
posttreatment, corresponding to germ cells exposed in the spermatozoa or spermatid stages. The
results confirmed previous dominant lethal studies and showed that germ cell stages in which the
treatment induced dominant lethals jointly yielded the highest frequency of specific locus
mutations. Specific locus mutations occurred in 5/28,971 offspring with exposures 1-7 weeks
after treatment, which was significantly higher than the historical control rate of
43/801,406 (p = 0.026 in a one-tailed Fisher Exact test). The two mutants arising from matings
1 and 2 weeks after treatment represented a significantly higher mutation rate than the three
mutants arising from matings in weeks 3-7; the rate in this latter period was not significantly
higher than the control rate. No mutations were recovered in 17,112 offspring derived from
treated stem cell spermatogonia (fertilizations occurring >49 days posttreatment). The major
13 8 DRAFT-DO NOT CITE OR QUOTE
-------
conclusions are that AA is mutagenically active in the late spermatid-spermatozoa stages, the
recovered mutations are associated with chromosomal aberration-type events (deletions and/or
translocations), and AA is not mutagenically active in stem cell spermatogonia. Russell et al.
(1991) reported that two specific locus mutations recovered in offspring derived from
fertilizations (in which the male gametes were exposed to AA at the spermatozoa and spermatid
stages) were homozygous lethal, of which one was associated with a cytogenetically visible
deletion, and concluded that the specific locus mutations were due to large, multilocus deletions.
Ehling and Neuhauser-Klaus (1992) exposed male mice to a single i.p. dose of AA at
100 or 125 mg/kg. Immediately after treatment, males were housed with untreated, test-stock
females homozygous for a (non-agouti), b (brown), cch (chinchilla), p (pink-eyed dilution),
d (dilute), se (short ear), and s (piebald). For the 100 mg/kg-treated males, a permanent
monogamist mating was established, starting on day 21. The offspring of the permanent mating
were classified according to their day of conception into those derived from treated
spermatocytes and differentiating spermatogonia (conception 21-42 days posttreatment), and
those from treated spermatogonia (>43 days posttreatment). Ehling and Neuhauser-Klaus (1992)
grouped their specific locus results for conceptions occurring in the intervals days 5-8 and 9-12
posttreatment, respectively, and reported an increased frequency of mutations due to exposure of
parental males to these levels of AA. They reported that, of the six specific-locus mutations
recovered following AA exposure of spermatids or spermatozoa, four had reduced viability, one
was sterile, and one was homozygous lethal. As in the Russell et al. (1991) study, the authors
concluded that the specific-locus mutations recovered in offspring derived from parental
exposure to AA were associated with multi-locus deletions. Unlike Russell et al. (1991), who
reported no increase in the frequency of specific-locus mutations in offspring derived from germ
cells exposed as stem-cell spermatogonia, Ehling and Neuhauser-Klaus (1992) observed a
significant increase in the frequency of specific-locus mutations following exposure of
spermatogonia to AA. Favor and Shelby (2005) reevaluated the mating intervals to more
directly compare the results and noted that in the results of Russell et al. (1991) for
spermatogonial exposure (days >42 posttreatment), the frequency of specific-locus mutations,
1/23,531, was not significantly higher than the frequency in the historical control. By contrast,
Ehling and Neuhauser-Klaus (1992) demonstrated a significantly higher specific-locus mutation
frequency in treated spermatogonia (6/23,489) than in their historical control. The difference in
the specific-locus mutation frequency for spermatogonia exposed to AA between Russell et al.
(1991) (higher total accumulated dose, 50 mg AA/kg on 5 consecutive days) and Ehling and
Neuhauser-Klaus (1992) (lower dose, 100 mg AA/kg) approached significance (p = 0.070,
Fisher's Exact test, two-tailed). Further, the intervals between treatment and conception for all
specific-locus mutations recovered in the spermatogonia exposure group were noted by Ehling
and Neuhauser-Klaus (1992). One mutation resulted from a conception 43 days posttreatment
139 DRAFT-DO NOT CITE OR QUOTE
-------
and represented an exposure at the differentiating spermatogonial stage. Russell et al. (1991)
also recovered one specific-locus mutation following exposure at this stage. The remaining five
mutations recovered for treatment of spermatogonia by Ehling and Neuhauser-Klaus (1992) all
had conceptions much later (70, 181, 201, 234, and 436 days posttreatment) and represented
exposures of stem-cell spermatogonia.
Synthesis and evaluation of heritable germ cell effects
Available heritable germ cell studies clearly demonstrate AA-induced heritable germ cell
effects. These findings are of particular concern because the human relevance has not been
determined. In the absence of experimental data from which to assess the potential for AA to act
as a human germ cell mutagen, the animal data must be considered potentially relevant to
humans.
Animal studies did not include adequate assessment of dose-response relationships for
the heritable germ cell effects, and doses employed in the various studies were relatively high.
Single or repeated i.p. doses of AA or GA ranged from 40 to 125 mg/kg and one study employed
five daily dermal applications of AA at 50 mg/kg. Heritable translocations appeared at high
frequency at the lowest doses tested, which indicates that lower doses may have also elicited
heritable translocations. Well-designed animal studies are needed to assess dose-response
relationships for AA- and GA-induced heritable germ cell effects, particularly in the low dose
region that is expected to be more relevant to human exposure.
The results of each of the above-described heritable germ cell studies suggest that AA
and/or GA act as clastogenic agents (Favor and Shelby 2005; Dearfield et al., 1988). Possible
mechanisms involved include (1) covalent modifications of protamines associated with DNA by
AA or GA, and (2) direct alkylation of DNA by GA or a combination of both modes of action
(Besaratinia and Pfeifer, 2007, 2004; Carere, 2006; Doerge et al., 2005a; Schmid et al., 1999;
Dearfield et al., 1995; Segerback et al., 1995; Moore et al., 1987).
Limited information is available regarding specific mechanisms of AA-induced heritable
germ cell effects in laboratory animals. Demonstrations that GA binds more strongly than AA to
DNA and some indication that the genetic damage in germ cells of mice is dependent on
metabolism of AA to GA by CYP2E1 (Ghanayem et al., 2005b) led Carere (2006) to suggest that
GA-DNA adducts may be responsible for gene mutations observed in the laboratory animal germ
cell studies.
Sega et al. (1989) proposed AA alkylation of protamine in late-stage spermatids as a
mechanism for AA-induced dominant lethal effects based on a parallel time course for protamine
alkylation and dominant lethal effects in spermatids of mice treated with AA. The involvement
of protamine binding as a mechanism of AA-induced heritable translocations is suggested
because both AA-induced dominant lethal mutations and heritable translocations appear to be
140 DRAFT-DO NOT CITE OR QUOTE
-------
late-stage germ cell effects and because AA has been shown to exert effects on the synaptoneural
complex and on the spindle. Furthermore, AA exhibits a relatively stronger binding to proteins
than does GA. Based on these observations, Carere (2006) suggested that AA-protamine
binding may explain some chromosomal effects in germ cells.
It is critical to determine mechanisms whereby AA induces heritable germ cell effects
and the critical germ cell stages at which the heritable germ cell effects occur because, as Favor
and Shelby (2005) note, if the mutagenic activity of AA is confined to postspermatogonial
stages, the risk of effects would be relative to the dose accumulated during the sensitive
postspermatogonial stages and this would be only a fraction of the lifetime accumulated
exposure. If, however, stem cell spermatogonia are sensitive to mutation induction by AA, the
risk would be relative to lifetime accumulated dose up to the time of fertilization.
4.5. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES
4.5.1. Neurotoxicity Studies
The oral toxicity animal studies described in detail in Sections 4.2 and 4.3 include those
most relevant to describing dose-response relationships for chronic exposure. Numerous
additional reports have been published in which AA-induced neurotoxicity has been assessed in
animal species following single or repeated oral exposure to AA. For example, both Fullerton
and Barnes (1966) and Tilson and Cabe (1979) observed clinical signs of neurotoxicity in rats
following single oral dosing with AA in the range of 100 to 200 mg/kg; repeated administration
at lower dose levels also resulted in neurotoxic signs. Aldous et al. (1983) reported overt signs
of neurotoxicity as early as day 4 in rats administered AA by gavage at a dose level of 50 mg/kg-
day.
Dixit et al. (1981) noted neurotoxicity in rats following 14 days of oral treatment at a
dose level of 25 mg/kg-day. Severe loss of hindlimb function was reported as early as day 21 in
rats administered AA in the diet for up to 90 days at a concentration that resulted in an estimated
dose of 30 mg/kg-day (McCollister et al., 1964). Fullerton and Barnes (1966) noted slight leg
weakness in rats after 40 weeks of dietary exposure at a concentration that resulted in a dose
ranging from approximately 6 to 9 mg/kg-day (according to the authors); the effect did not
appear to become more severe during the remaining 8 weeks of exposure.
Alterations in gait (home-cage and open-field assessment of neuromuscular function and
equilibrium) were reported in adult male and female Long-Evans rats administered i.p. injections
of AA at doses as low as 1 mg/kg-day for as little as 30 to 60 days (Moser et al., 1992). AA was
administered 5 days/week for 13 weeks and included dose levels of 1, 4, and 12 mg/kg-day.
Neurobehavioral observations were performed prior to dosing, at treatment days 29-31 and 58-
62, and immediately following treatment termination. Significantly increased foot splay was
observed at 4 mg/kg-day (females) and 12 mg/kg-day (males and females) at 60-day
141 DRAFT-DO NOT CITE OR QUOTE
-------
examination. All other signs of neurotoxicity (impaired mobility and righting reflex, decreased
grip strength, and axonal degeneration in peripheral nerves and spinal cord) were seen only at
the high dose (12 mg/kg-day).
Other investigators have reported AA-induced neurotoxicity in mice (Gilbert and
Maurissen, 1982; Hashimoto et al., 1981), cats (Post and McLeod, 1977; McCollister et al.,
1964), dogs (Hersch et al., 1989; Satchell and McLeod, 1981), and monkeys (Eskin et al., 1985;
Maurissen et al., 1983; McCollister et al., 1964).
4.5.2. Other Cancer Studies
The potential of AA to initiate skin tumors has been examined in female SENCAR mice
(40/group, 6 to 8 weeks of age) exposed via oral (gavage), i.p. injection, and dermal application
(Bull et al., 1984a). AA was dissolved in distilled water for oral and injection routes and in
ethanol for dermal applications. AA was administered at dose levels of 0, 12.5, 25, or 50 mg/kg-
day, 6 times during a 2-week period for each route (total AA doses of 0, 75, 150, or 300 mg/kg).
Two weeks later, dermal doses of a promoter, 1.0 jig 12-O-tetradecanoylphorbol-13-acetate
(TPA) (in 0.2 mL acetone) were applied to the shaved back 3 times/week for 20 weeks. Two
types of control groups (20-40 mice/group) were included for each route of administration:
(1) vehicle initiation with TPA promotion; and (2) 50 mg/kg-day AA plus vehicle promotion.
All animals were killed at 52 weeks, and all gross lesions in the skin were histologically
examined. The incidences of histologically confirmed squamous cell carcinomas or squamous
cell papillomas for the 0, 12.5, 25, or 50 mg/kg-day AA groups with TPA, followed by the
incidence for the 50 mg/kg-day group without TPA are shown in Table 4-28.
Table 4-28. Acrylamide initiation of squamous cell carcinomas or
papillomas in female SENCAR mice
Oral
Intraperitonea
1
Dermal
Skin carcinomas"
Skin papillomas1'
Dose (mg/kg-day)
With TPA
0
0/34
0/35
0/36
12.5
2/35
2/38
1/38
25
7/3 3b
4/36
2/35
50
6/3 8b
4/35
3/34
No
TPA
50
0/17
0/17
0/20
With TPA
0
0/34
0/35
5/36
12.5
3/35
2/38
3/38
25
8/3 3b
3/36
3/35
50
ll/38b
6/3 5b
2/34
No
TPA
50
0/17
0/17
0/20
Denominator is the number of surviving mice at 52 weeks with acceptable nonautolyzed tissues.
bSignificantly different (p < 0.05) from the vehicle initiation/TPA promotion group by Fisher's Exact test.
Source: Bulletal. (1984a).
Incidences were also reported for the number of skin tumor-bearing mice/total mice in
each group (Bull et al., 1984a). In this analysis, tumors were described as skin masses with
diameter >1 mm that were detected during a minimum of 3 consecutive weeks in the study.
142 DRAFT-DO NOT CITE OR QUOTE
-------
Incidences for the 0, 12.5, 25, or 50 mg/kg-day/+TPA promotion groups, followed by the
50 mg/kg-day/vehicle promotion group, for the three routes of administration are displayed in
Table 4-29.
Table 4-29. Acrylamide initiation of skin tumor masses >lmm in female
SENCAR mice
Oral
Intraperitoneal
Dermal
Skin tumor masses with diameter >1 mm
Dose (mg/kg-day)
With TPA
0
2/40
0/40
7/40
12.5
12/403
10/403
4/40
25
23/403
13/403
11/40
50
30/403
21/403
18/403
No TPA
50
0/20
0/20
0/20
""Significantly different (p < 0.05) from the vehicle initiation/+TPA promotion group by Fisher's Exact test.
Source: Bulletal. (1984a).
Overall, the data indicate that AA at oral dose levels of 25 or 50 mg/kg-day initiated
TPA-promoted skin tumors in SENCAR mice. However, the incidences of histologically
confirmed skin tumors were not statistically significantly elevated in mice receiving initiating
doses of AA by i.p. injection or dermal administration, with the exception of papillomas in mice
exposed to 50 mg/kg-day by i.p. injection followed by TPA promotion.
In another skin tumor initiation-promotion study, female Swiss-ICR mice (40/group)
were administered AA in oral doses of 0, 12.5, 25, or 50 mg/kg-day, 3 times/week for 2 weeks
(Bull et al., 1984b). Two weeks later, 2.5 jig TPA in acetone was applied to the shaved backs,
3 times/week for 20 weeks. Another group of 40 mice received 6 doses of 50 mg/kg-day AA
during 2 weeks, followed by dermal application in acetone without TPA for 20 weeks. Mice
were examined for skin papillomas on a weekly basis, until sacrifice at 52 weeks after start of the
initiation period. The skin and lungs were preserved for histologic examination of all gross
lesions. The combined incidence of mice with histologically confirmed skin papillomas or
carcinomas for the 0, 12.5, 25, or 50 mg/kg-day AA groups with TPA, followed by the incidence
for the 50 mg/kg-day group without TPA were as follows (* indicates significantly different
\p < 0.05] from the vehicle/+TPA promotion group by Fisher's Exact test; denominator is the
number of mice surviving to 52 weeks with acceptable nonautolyzed tissue): 0/35, 2/34, 3/32,
10/32*, and 1/36. Respective incidences for skin carcinomas alone were: 0/35, 1/34, 3/32,
4/32*, and 1/36. The data indicate that orally administered AA (50 mg/kg-day, 6 times during a
2-week period) initiated histologically confirmed mouse skin tumors promoted by TPA.
Support for the skin tumor initiation activity of AA is provided by an analysis in which
tumors were described as skin masses with diameter >1 mm that were detected during a
minimum of 3 consecutive weeks in the study (Bull et al., 1984b). In this analysis, incidences of
skin-tumor bearing animals were 0/40, 4/40, 4/40, and 13/40* for the 0, 12.5, 25, and 50 mg/kg-
143 DRAFT-DO NOT CITE OR QUOTE
-------
day /+TPA groups, respectively, and 10/40* for the 50 mg/kg-day/vehicle promotion group.
Incidences in the 50 mg/kg-day AA-exposed groups were statistically significantly elevated
(*p < 0.05 by Fisher's Exact test) compared with the vehicle/+TPA control group.
Lung tumors were also found in the Swiss-ICR mice that survived to 52 weeks (Bull et
al., 1984b). The combined incidences of mice with histologically confirmed alveolar bronchiolar
adenomas or carcinomas for the 0, 12.5, 25, or 50 mg/kg-day/+TPA promotion groups, followed
by the incidence for the 50 mg/kg-day/vehicle promotion were as follows: 4/36, 8/34, 6/36,
11/34*, and 14/36*. The respective incidences for carcinomas were: 1/36, 2/34, 1/36, 1/34, and
10/36*. The incidences for combined adenomas and carcinomas were statistically significantly
(Fisher's Exact test, * p < 0.05) elevated in both groups treated with 50 mg/kg-day 6 times
during 2 weeks, but only 1/11 lung tumors in the 50-mg/kg-day/+TPA group was a carcinoma, in
contrast to 10 carcinomas/14 lung tumors in the 50-mg/kg-day/-TPA group.
Bull et al. (1984a) also performed mouse lung adenoma bioassays on groups of 8-week-
old male and female A/J mice, a strain that is very susceptible to lung tumor formation. AA was
administered to mice (16/sex/group) via i.p. injection at doses of 1, 3, 10, 30, or 60 mg/kg-day,
3 times/week for 8 weeks. Untreated and vehicle control (distilled water) groups were also
employed. The mice injected with 60 mg/kg-day showed severe peripheral neuropathy and
deaths within the first 3 weeks of treatment and were not examined for lung tumor development.
Surviving mice in other groups were sacrificed at 8 months, lungs were fixed, and surface
adenomas were counted after 24 hours. AA exposure caused increased incidences of mice with
lung tumors at dose levels >3 mg/kg. Incidences were 12/30 and 3/31 for untreated and vehicle
controls, compared with 14/33, 15/33*, 21/31*, and 28/30* for the 1, 3, 10, and 30 mg/kg-day
groups, respectively (* indicates significantly different from combined control incidence by
Fisher's Exact test). Some evidence was also presented for increasing average number of lung
tumors/mouse ("tumor yield") with increasing AA exposure: 0.4 ± 0.5, untreated control;
0.1 ± 0.3, vehicle control; 0.6 ± 0.8, 1 mg/kg; 0.8 ± 1.0, 3 mg/kg; 1.2 ± 1.4, 10 mg/kg; and
2.2 ± 1.5, 30 mg/kg. In a later report, Bull et al. (1984b) reported that the tumor yield in this
study "displayed a reasonably strong relationship with dose (p < 0.03)" but did not provide
specific information on the statistical analysis performed.
Robinson et al. (1986) compared skin and lung tumor yields (number of tumors/mouse)
in several strains of mice (SENCAR, BALB/c, A/J, and ICR) injected i.p. with single 50 mg/kg
doses of AA followed by topical application of TPA 3 times weekly for 20 weeks. Groups of
60 mice of each strain received initiating injections with AA or water (vehicle); 40 mice in each
group then received TPA at the following dose levels: 1.0 jig for SENCAR, 5.0 jig for BALB/c,
and 2.5 jig for A/J and ICR. The mice were sacrificed at 36 weeks. Microscopic examinations
were conducted on all gross lesions found in lungs and skin and only lung adenomas and skin
papillomas were included in the tumor count and calculation of tumor yield. One experiment
144 DRAFT-DO NOT CITE OR QUOTE
-------
included all four strains, and a second experiment only examined SENCAR mice. Lung tumor
yields were statistically significantly increased by the AA treatment (0.42 tumors/mouse),
compared with vehicle controls (0.04 tumors/mouse), in the SENCAR strain but not in the
BALB/c, A/J, or ICR strains. However, in the other experiment with SENCAR mice, lung tumor
yields were not statistically significantly elevated (0.38 vs. 0.22 tumors/mouse). Skin tumor
yields were statistically significantly elevated in SENCAR mice in the two experiments (0.25 vs.
0.08 tumors/mouse and 0.38 vs. 0.05 tumors/mouse) but were not significantly elevated in the
other three strains. Robinson et al. (1986) only reported mean skin and lung tumor yield data, so
the value of the reported data are only of limited use for cancer hazard identification purposes.
Jin et al. (2008) assessed the potential for AA to induce thyroid tumors in female CD1
mice administered AA in drinking water. The study comprised both short-term (2-month) and
longer-term (up to 8 months) experiments. Each experiment included six groups of mice (20-
30/group); groups 1-3 received drinking water without AA and groups 4-6 were exposed to AA
in the drinking water. In addition, groups 2 and 5 received thyroxine in the drinking water to
depress activity of the thyroid and groups 3 and 6 received methimazole that causes thyroid
activation. Concentrations of AA in the drinking water were adjusted to deliver AA at an
intended dose of 3 mg/kg-day, although the concentration was increased during the later portion
of the longer-term experiment. The exposures designed to alter activity of the thyroid produced
the intended results. In the longer-term experiment, mice receiving thyroxine treatment with or
without AA exposure died or were sacrificed moribund after 6 months. Peripheral neuropathy
was noted in mice receiving AA in the longer-term experiment, particularly later in the
experiment when AA concentrations were increased. There were no indications of AA-induced
effects on thyroid weights or thyroid tumorigenesis, including those mice with hyper- or hypo-
stimulated thyroids. Although the AA in the drinking water of the mice resulted in AA doses
comparable to those associated with thyroid tumors in rats treated for 2 years (Friedman et al.,
1995), the mice were only treated and examined for up to 8 months (Jin et al. 2008).
01st0rn et al. (2007) assessed the tumorigenicity of subcutaneous administration of AA
or GA during early perinatal periods in the intestine of C57BL/6J Min/+ mice and their wild
type. The Min/+ mice are heterozygous for a mutation in the tumor suppressor gene (Ape),
which leads to the development of multiple intestinal neoplasms, particularly in the small
intestine. The study consisted of two experiments. In the first experiment, Min/+ and wild type
mice were subcutaneously injected with AA or GA (0 , 10, or 50 mg/kg-bw) at 1 and 2 weeks
postpartum and sacrificed after 8 (Min/+) or 32 (wild type) weeks. Respective numbers of mice
included in the control though high-dose AA groups were: 19, 19, and 17 for Min/+ mice, and
31, 12, and 24 for WT mice. Respective group sizes for the low- and high-dose GA groups
were: 26 and 21 for Min/+ mice and 24 and 24 for WT mice. At the 50 mg/kg dose level, GA
(but not AA) induced significantly (p < 0.05) increased number of small intestinal tumors per
145 DRAFT-DO NOT CITE OR QUOTE
-------
mouse in the Min/+ mice (1.3-fold increase compared with control value) and significantly (p <
0.05) increased incidence of WT mice with intestinal neoplasms (25 vs. 3% in controls). In the
second experiment, Min/+ and wild type mice were exposed to AA or GA either in utero via
single subcutaneous injections of their dams at 1 week prior to birth, postpartum subcutaneous
injections at 1 and 2 weeks, or both. In the second experiment, the number of small intestinal
and colonic tumors per mouse in the Min/+ mice correlated positively with the number of GA
injections. The results indicate that early life subcutaneous exposure of Min/+ and WT
C57BL/J6 mice to 50 mg/kg GA, but not 50 mg/kg AA, elicited tumorigenic responses in the
intestine. The tumorigenic response to GA, but not to AA, in experiment one of this study could
potentially be explained by relatively low expression of CYP2E1 (and thus low capability to
convert AA to GA) during the early life exposure period.
4.6. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION
4.6.1. Studies on the Hypothalamus-Pituitary-Thyroid Axis
Both of the available chronic oral exposure studies for AA in F344 rats reported
statistically significant increased incidences of thyroid follicular cell adenomas, or combined
adenomas and carcinomas, at the highest dose levels of 2-3 mg/kg-day (Friedman et al., 1995;
Johnson et al., 1986). Chemicals that alter thyroid hormone homeostasis by interfering with
synthesis or secretion of triiodothyronine (T3) or thyroxin (T4) or by increasing T3 or T4
metabolism can lead to compensatory release of TSH from the pituitary, which, if sustained, may
induce thyroid follicular cell hyperplasia that may progress to neoplasia (U.S. EPA, 1998c).
These findings have led to several investigations of effects of AA on hypothalamus-pituitary-
thyroid axis endpoints. To date, there is no clear and consistent evidence to support the
hypothesis that AA induces sustained follicular cell proliferation by altering thyroid hormone
homeostasis.
Exposure of female F344 rats to 2 or 15 mg/kg-day for 2 or 7 days induced follicular cell
morphometric changes (decreased colloid area and increased cell height) without significantly
changing circulating levels of T4 or TSH (Khan et al., 1999). In female F344 rats exposed to 2
or 15 mg/kg-day AA for 2 or 7 days, no statistically significant changes, compared with controls,
were found in plasma levels of T/t, TSH, or prolactin, in pituitary levels of TSH or prolactin, or
in body, pituitary, or adrenal weights, whereas thyroid gland morphometry showed statistically
significant decreased colloid area (56-57% decrease compared with control) and increased
follicular cell height (18-22% increase compared with control) (Khan et al., 1999).
In an unpublished study, blood levels of T3, T4, or TSH were evaluated in male or female
F344 rats exposed to AA in drinking water for 14 or 28 days at dose levels ranging from about
1 to 25 mg/kg-day (Table 4-30) (Friedman et al., 1999b). A significant decrease in T3 and T4 in
146 DRAFT-DO NOT CITE OR QUOTE
-------
high dose males is reported at 28 days, but T4 in high-dose males increased at 14 days, and
overall, there is inadequate support for a consistent, significant change in blood levels of T3, T4,
or TSH.
Table 4-30. Circulating thyroid hormone levels in F344 rats following
exposure to AA in drinking water for 14 or 28 days
Dose
(mg/kg-day)
Male
Female
T3
(ng/dL)
Male
Female
T4
(ng/dL)
Male
Female
TSH
(ng/mL)
Male
Female
14 days
0
1.4
4.1
12
19
25
0
1.3
4.3
9.0
19
24
85.2 ±14.4
75.2 ±16.0
80.3 ±7.7
81.6 ±10.2
92 ± 20.2
91.9 ±13.2
78.8 ±8.4
77.5 ±6.6
91.0 ±13
81.6 ±8.7
101.9 ±10.3a
89 ±15
3.5 ±0.5
3.3 ±0.3
3.8 ±0.3
3.6 ±0.3
4.0 ±0.5
4.1±0.4a
2.8 ±0.6
2.8 ±0.3
3.4±0.5a
3.2 ±0.5
3.2 ±0.3
3.0 ±0.8
2.7 ± .1
3.7 ± .7
3.1± .3
2.9 ± .4
3.7 ± .0
2.8 ±0.8
2.1 ±0.6
2.2 ±0.4
1.8 ±0.3
1.8 ±0.4
2.1 ±0.9
2.8±0.2a
28 days
0
1.4
4.1
12
19
25
0
1.3
4.3
9.0
19
24
90.8 ±13. 3
90.6 ±13. 8
82.0 ±13.1
80.3 ±11. 5
71.2±10.3a
61.4±32.4a
78.9 ±13. 5
75.5 ±13.0
79.6 ±8.2
84.9 ±4.4
81.6 ±7.9
65.2 ±23.6
3.9 ±0.6
4.0 ±0.5
3.9 ±0.5
3.7 ±0.4
3.3 ±0.5
2.6±1.0a
2.5 ±0.7
2.4 ±0.6
2.5 ±0.4
2.7 ±0.3
2.7 ±0.3
2.4 ±0.6
2.0 ±0.7
2.3 ± 1.2
2.1 ±0.9
2.1 ±0.4
1.9 ±0.4
2.8 ±1.2
1.5 ±0.4
1.8 ±0.6
1.6 ±0.2
1.7 ±0.4
1.9 ±0.9
1.6 ±0.4
"Statistically significantly different (p < 0.01) from control by an unspecified statistical test with unspecified
number. Available report does not specify if values are means ± SEM or SD.
Source: Friedman et al. (1999b).
In another unpublished study, no changes in plasma TSH levels were found in male
Sprague-Dawley rats exposed to 2 or 15 mg/kg-day AA for up to 28 days by an unspecified route
of administration, and evidence for a sustained statistically significant increase in DNA synthesis
in the thyroid of exposed rats, compared with control rats, was not found (Klaunig, 2000, as cited
in Environ, 2002). DNA synthesis in the thyroid was assayed as "BrdU incorporation and
proliferating cell nuclear antigen (PCNA) expression", but further methodological details were
not specified in the available report of this study. The results (as cited in Environ, 2002) are
shown in Table 4-31. The quality of these data, however, is poor due to lack of information on
methodological details and the fact that the data were neither published nor peer reviewed.
147 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-31. Plasma TSH, BrdU incorporation in thyroid, and PCNA
expression in thyroid in male Sprague-Dawley rats exposed to acrylamide by
an unspecified route for up to 28 days
Dose (mg/kg-day)
0
2
15
0
2
15
0
2
15
Day
7
14
28
TSH (ng/mL)
2.92 (0.90)
3.28(1.12)
4.09(2.16)
5.02 (2.44)
4.41(1.89)
4.72(2.10)
5.29 (2.44)
3.96(1.64)
4.90 (2.55)
BrdU (units not reported)
0.47(0.11)
4.09 (1.04)a
1.92(0.55)
2.31(0.18)
2.79(1.69)
5.60(1.73)
2.31(0.18)
3.13(1.53)
5.60(1.73)
PCNA (units not reported)
0.20 (0.07)
2.64 (1.39)a
2.29 (0.91) a
0.11 (0.05)
0.06 (0.04)
2.24 (0.59)a
0.04 (0.02)
1.21 (0.89)
3.13 (1.77)
"Reported as statistically significant (p < 0.05), by ANOVA followed by Fisher's Least Significant Difference
(LSD); values in parentheses were not specified. Methodological details concerning thyroid BrdU incorporation
and PCNA expression were not provided in Environ (2002).
Source: Klaunig (2000) as cited in Environ (2002).
Bowyer et al. (2008) examined a number of endpoints indicative of disruption of the
hypothalamus-pituitary-thyroid axis in male F344 rats (70 days of age) exposed for 14 days to
AA in drinking water delivering nominal doses of 0, 2.5, 10, or 50 mg/kg-day. Based on twice
weekly measurement of body weight and water intake, average measured doses for the 2.5-, 10-,
and 50-mg/kg-day groups were 93-100, 99-100, and 85-88% of nominal values. The following
endpoints were evaluated: (1) expression of genes related to thyroid hormone production and
cell proliferation in the hypothalamus, thyroid, and pituitary (using cDNA array and RT-PCR
analysis after isolation of total RNA from tissues from 20 rats per group); (2) levels of
neurotransmitters (and metabolites) in the brain and pituitary (from 10 rats per group) that affect
hormone homeostasis (dopamine, 3-methoxytyramine [3MT], homovanillic acid [HVA],
3,4,-dihydroxyphenylacetic acid [DOPAC]), serotonin, 5-hydroxytryptamine [5HT], and
5-hydroxyindoeacetic acid [5HIAA]); (3) serum levels of pituitary and thyroid hormones (from
10 rats/groups): TSH, T4, T3, and thyrotropin-releasing hormone (TRH); and (4) histology of
pituitary and thyroid glands. The low- and mid-dose groups showed no obvious effect on
locomotory activity or body weight gain. The high-dose group showed clear signs of locomotor
impairment (lethargy and hind limb paralysis) and decreased body weight (92-93% of control
values). No clear exposure-related effects were found on levels of dopamine and its metabolites
(DOPAC, 3MT, and HVA) or 5HT and 5HIAA in the hypothalamus or pituitary. AA exposure
caused a significant (p < 0.05) decrease in serum T4 only at the high dose, but had no effect on
serum T3 or TSH. Exposed and control rats showed no difference in response to a challenge
dose of TRH increase in serum levels of TSH and T4, 30 minutes following a challenge i.p. dose
of 2.5 mg/kg TRH. No significant exposure-related effects were found on mRNA levels in
hypothalamus or pituitary for TRH, TSH, or thyroid hormone receptor a and p. Control and
148 DRAFT-DO NOT CITE OR QUOTE
-------
high-dose rats showed no significant changes in mRNA levels for other important pituitary
hormones including growth hormone, opiomelancorticotropin, vasopressin, and luteinizing
hormone. Hematoxylin- and eosin-stained sections of thyroid and pituitary tissue from control
and high-dose rats showed no evidence for exposure-related changes in cell morphology (i.e.,
hypertrophy, hyperplasia, karyomegaly, or degeneration). Indices for cell proliferation in
pituitary and thyroid tissues (Mki67 mRNA and ki-67 protein levels) were not increased in
exposed rats, compared with controls. Expression of genes in the thyroid, which are typically
increased in response to anti-thyroid effects (e.g., thyroglobulin, thyroid peroxidase), were not
significantly increased in high-dose rats, compared with controls. This study found no evidence
that 14-day exposures of F344 rats to oral AA doses of 2.5, 10, or 50 mg/kg-day disrupt the
hypothalamus-pituitary-thyroid axis. Bowyer et al. (2008) concluded that these negative
findings are important to understanding AA's mode of action (MOA) in producing thyroid
tumors in rats with chronic oral AA exposure (see Section 4.2.1.2), because chronically elevated
TSH levels with resultant thyroid follicular cell hyperplasia are strongly associated with
exposure to other compounds that induce rodent thyroid tumors by a nongenotoxic mechanism.
4.6.2. Genotoxicity Studies
Appendix B (Table B-l) summarizes results of numerous published mutagenicity tests
for AA including the dominant lethal mutation assays discussed in a previous section. Results
from in vivo dominant lethal mutation assays involving i.p. exposure of mice (Adler et al., 2000;
Shelby et al., 1987), oral exposure of mice (Chapin et al., 1995; Sakamoto and Hashimoto, 1986)
or rats (Tyl et al., 2000a, b; Sublet et al., 1989; Working et al., 1987a, b; Smith et al., 1986;
Zenick et al., 1986), and dermal exposure of mice (Gutierrez-Espeleta et al., 1992) have been
consistently positive. Since the oral exposure studies were described in detail in Section 4.3.1,
results from dominant lethal mutation assays were generally not included in Appendix B.4
Heritable germ cell studies in male mice were consistently positive for heritable translocations
(Adler et al., 2004, 1994; Generoso et al., 1996; Adler, 1990; Shelby et al., 1987) and specific
mouse locus (Ehling and Neuhauser-Klaus, 1992; Russell et al., 1991). No experiments studied
the potential for AA to induce heritable mutations in the female germ line. The heritable germ
cell studies are listed in Appendix B and are discussed in Section 4.4.
Manjanatha et al. (2006) evaluated the somatic cell mutagenic potential of AA and GA in
an in vivo genotoxicity study in male and female Big Blue (BB) mice. BB mice were
administered 0, 100, or 500 mg/L of AA or equimolar doses of GA in drinking water for 3-
4 weeks. The estimated daily exposures to AA for males and females were 19 and 25 mg/kg-
4 It is further acknowledged that male-mediated dominant lethal effects can be mediated by effects on
altered male mating performance, sperm motility and/or morphology, as well as effects on genetic integrity of the
sperm (Perreault, 2003).
149 DRAFT-DO NOT CITE OR QUOTE
-------
day, respectively, for the low dose of 100 mg/L (4-week exposure) and 98 and 107 mg/kg-day
for the high dose of 500 mg/L (3 weeks only due to clinical signs of neurotoxicity). The
estimated daily exposure to GA for males and females were 25 and 35 mg/kg-day for the low
dose of 120 mg/L (4 weeks) and 88 and 111 mg/kg-day for the high dose of 600 mg/L (4 weeks).
Micronucleated reticulocytes (MN-RETs) were assessed in peripheral blood within 24 hours of
the last treatment, and lymphocyte Hprt and liver ell mutagenesis assays were conducted 21 days
following the last treatment. The types of ell mutations induced by AA and GA in the liver were
determined by sequence analysis. The frequency of MN-RETs was increased 1.7-3.3-fold in
males treated with the high doses of AA and GA (p < 0.05; control frequency = 0.28%). Both
doses of AA and GA produced increased lymphocyte Hprt mutant frequencies (MFs), with the
high doses producing responses that were 16-25-fold higher than those of the respective control
(p < 0.01; control MFs = [1.5 ± 0.3] x 10~6 and [2.2 ± 0.5] x KT6 in females and males,
respectively). Also, the high doses of AA and GA produced significant 2-2.5-fold increases in
liver ell MFs (p < 0.05; control MFs = [26.5 ± 3.1] x 10~6 and [28.4 ± 4.5] x 10"6). Molecular
analysis of the mutants indicated that AA and GA produced similar mutation spectra and that
these spectra were significantly different from that of control mutants (p < 0.001). The
predominant types of mutations in the liver ell gene from AA- and GA-treated mice were
G:C—>T:A transversions and -1/+1 frameshifts in a homopolymeric run of guanosines. The
results indicate that both AA and GA are mutagenic in mice. The MFs and types of mutations
induced by AA and GA in the liver are consistent with AA exerting its mutagenicity in BB mice
via metabolism to GA.
Ghanayem et al. (2005b) demonstrated the absence of AA-induced genotoxicity in
CYP2El-null mice as evidence of a GA-mediated genotoxic effect in somatic cells. Female
wild-type and CYP2El-null mice were administered AA (0, 25, 50 mg/kg) by i.p. injection once
daily for 5 consecutive days. Twenty-four hours after the final treatment, blood and tissue
samples were collected. Erythrocyte micronucleus frequencies were determined by flow
cytometry, and DNA damage was assessed in leukocytes, liver, and lung using the alkaline
(pH >13) single cell gel electrophoresis (Comet) assay. Results included significant dose-related
increases in micronucleated erythrocytes and DNA damage in somatic cells induced in
AA-treated wild-type mice but not CYP2El-null mice. These results were consistent with the
observations in a similar study in male germ cells, where dose-related increases in dominant
lethal mutations were detected in uterine contents of female mice mated to AA-treated wild-type
males but not CYP2El-null males (Ghanayem et al., 2005a) (discussed in Section 4.2.1).
Numerous previous tests were performed to evaluate AA-induced chromosomal
alterations in mammalian systems in vivo; most tests employed i.p. injection of AA at
concentrations in the range of 25 to 200 mg/kg. Tests for chromosomal aberrations in bone
marrow cells yielded both positive (Adler et al., 1988; Cihak and Vontorkova, 1988) and
150 DRAFT-DO NOT CITE OR QUOTE
-------
negative (Krishna and Theiss, 1995; Shiraishi, 1978) results. In one study, male B6C3Fi mice
were administered deionized water (control) or AA at doses ranging from 0.125 to 24 mg/kg-day
via gavage for 28 days and evaluated for micronuclei (MN) response in bone marrow cells
[reticulocytes (RETs) and normochromatic erythrocytes (NCEs)] by flow cytometry (Zeiger et
al., 2009). AA significantly (p < 0.05) induced MN at >4 mg/kg-day in RETs, and >6 mg/kg-
day in NCEs. MN were not induced at lower doses of AA (0.125-2 mg/kg-day).
Similar assays of mouse spleen lymphocytes, splenocytes, and spermatogonia were all
negative for chromosomal aberrations (Kligerman et al., 1991; Adler, 1990; Backer et al., 1989;
Adler et al., 1988). Significant increases in chromosomal aberrations were observed in
spermatocytes of mice that had been administered an i.p. dose of 100 mg/kg (Adler, 1990), but
the frequency of aneuploid sperm detected by fluorescence in situ hybridization (FISH) was not
increased by single i.p. injections of 60 or 120 mg/kg AA in male mice (Schmid et al., 1999).
Consistent with AA induction of chromosomal aberrations in sperm, the frequency of zygotes
with chromosomal aberrations was significantly elevated in zygotes from females mated to
males exposed to 50 mg/kg AA by i.p. injection for 5 days before mating (Marchetti et al.,
1997). Tests were positive for early cleavage stages of mouse zygotes (Pacchierotti et al., 1994)
and embryos (Valdivia et al., 1989), positive for polyploidy or aneuploidy (Shiraishi, 1978), and
negative for spindle disturbances (Adler et al., 1993) in mouse bone marrow cells.
AA-induced increases in micronuclei were seen in bone marrow cells, reticulocytes,
spleen lymphocytes, and splenocytes of mice and spermatids of rats and mice (Zeiger et al.,
2009; Paulsson et al., 2002; Lahdetie et al., 1994; Russo et al., 1994; Xiao and Tates, 1994;
Collins et al., 1992; Kligerman et al., 1991; Cihak and Vontorkova, 1990, 1988; Backer et al.,
1989; Adler et al., 1988; Knaap et al., 1988) but not in rat bone marrow cells (Paulsson et al.,
2002; Krishna and Theiss, 1995). Synaptonemal complex irregularities (asynapsis in meiotic
prophase) were slightly increased in germ cells of male mice following i.p. injection of AA,
without a significant increase in aberrations (Backer et al., 1989). Tests for heritable
translocations and reciprocal translocations in male mice yielded positive results (Adler et al.,
1994; Shelby et al., 1987).
AA was found to induce chromosomal alterations (chromosomal aberrations, cell
division aberration, chromosome enumeration, polyploidy, spindle disturbances) in a number of
in vitro mammalian cell test systems at concentrations as low as 0.01-1 mg/mL (Martins et al.,
2007; Adler et al., 1993; Tsuda et al., 1993; Warr et al., 1990; Knaap et al., 1988; Moore et al.,
1987). In human hepatoma G2 cells, concentrations of >0.625 mM induced micronuclei (Jiang
et al., 2007); however, a test for micronuclei in spermatids collected from Sprague-Dawley rats
yielded negative results at concentrations up to 0.05 mg/mL (Lahdetie et al., 1994).
Evidence for AA-induced DNA damage and repair includes positive results in a spore rec
assay (Tsuda et al., 1993), DNA breakage in vitro (Jiang et al., 2007) and in vivo in mice
151 DRAFT-DO NOT CITE OR QUOTE
-------
following i.p. injection of AA at doses >25 mg/kg (Dobrzynska, 2007; Sega and Generoso,
1990), oxidative DNA damage in human hepatoma G2 cells (Jiang et al., 2007), in vitro
unscheduled DNA synthesis (UDS) in human mammary epithelial cells (Butterworth et al.,
1992), and in vivo UDS in male mouse germ cells (Sega et al., 1990). Testing for UDS in male
rats in vivo/in vitro yielded positive results in spermatocytes and negative results in hepatocytes
(Butterworth et al., 1992).
AA tested positive for sister chromatid exchange in mammalian cells both in vitro
(Martins et al., 2007; Tsuda et al., 1993; Knaap et al., 1988) and in vivo (Russo et al., 1994;
Kligerman et al., 1991; Backer et al., 1989). Both positive (Park et al., 2002; Tsuda et al., 1993;
Banerjee and Segal, 1986) and negative (Raster et al., 1998; Abernethy and Boreiko, 1987)
results were obtained in cell transformation assays.
Results of reverse mutation assays in bacterial test systems did not indicate a mutagenic
response at AA concentrations ranging from 10 to 10,000 jig/plate with or without metabolic
activation (Muller et al., 1993; Tsuda et al., 1993; Jung et al., 1992; Knaap et al., 1988; Zeiger et
al., 1987; Hashimoto and Tanii, 1985; Lijinsky and Andrews, 1980). A fluctuation test in
Klebsiellapneumoniae was also negative for mutagenicity (Knaap et al., 1988).
Genotoxicity was not observed in a test for sex-linked recessive lethality in
Drosophila melanogaster following abdominal injection of a 50 mM solution of AA (Knaap et
al., 1988), but positive results were obtained when D. melanogaster larvae were fed
concentrations >1 mM (Tripathy et al., 1991). Somatic mutation and recombination assays were
positive for genotoxicity in D. melanogaster exposed by larval feeding at concentrations >1 mM
(Batiste-Alentorn et al., 1991; Tripathy et al., 1991; Knaap et al., 1988).
Positive results were obtained for gene mutation in mouse lymphoma cells in vitro at
concentrations as low as 0.3 mg/mL (Mei et al., 2008b; Barfknecht et al., 1988; Knaap et al.,
1988; Moore et al., 1987). This response was seen both with and without metabolic activation.
Negative results were obtained for gene mutation at the HPRT locus in Chinese hamster V79H3
cells at the highest concentration tested (7 mM) without activation (Tsuda et al., 1993); however,
positive results were obtained for gene mutation at the HPRT locus in human promyelocytic
leukemia HL-60 and NB4 cells at 700 mg/L without activation (Ao et al., 2008).
Additional studies on the genotoxic potential of GA include positive results to
Salmonella typhimurium strains TA100 and TA1535 (Hashimoto and Tanii, 1985) and mouse
lymphoma cells (Barfknecht et al., 1988) but not K. pneumoniae (Voogd et al., 1981). GA
induced unscheduled DNA synthesis in mouse spermatids in vivo (Sega et al., 1990), in human
epithelial cells in vitro (Butterworth et al., 1992), in one of two tests for unscheduled DNA
synthesis in rat hepatocytes in vitro (Butterworth et al., 1992; Barfknecht et al., 1988), and in
(C3H/RL x C57BL)F1 male mice given single i.p. injections of 150 mg/kg GA (Generoso et al.,
1996). GA (125 mg/kg by i.p. injection) induced dominant lethal mutations in male JH mice
152 DRAFT-DO NOT CITE OR QUOTE
-------
mated with nonexposed female SB mice (Generoso et al., 1996). GA treatment (100 mg/kg, i.p.
injection) of male (C3H x 101/RL)F1 mice (mated with nonexposed (SEC x C57BL)F1 female
mice) induced heritable translocations in male offspring at a frequency about twofold greater
than spontaneous frequencies in historical controls (Generoso et al., 1996). Synthetic GA
induced a similar frequency for micronuclei in erythrocytes per unit of in vivo dose in the mouse
as obtained in a study in the same laboratory where animals were treated with AA, and GA was
endogenously generated as a metabolite (Paulsson et al., 2003a). This equality in potency of
GA, whether its in vivo dose is established by injection of synthetic GA or through metabolism
of AA, supports the view that GA is the predominant genotoxic factor in AA exposure.
Formation ofDNA adducts and oxidative stress
GA forms DNA adducts in mice and rats (see Figure 3-2) (Doerge et al., 2005a; Gamboa
da Costa et al., 2003; Segerback et al., 1995). DNA adduct formation was seen in liver, lung,
kidney, brain, and testis of male mice and rats following i.p. injection of 46-53 mg/kg AA
(Gamboa da Costa et al., 2003; Segerback et al., 1995; Sega et al., 1990).
Doerge et al. (2005a) measured DNA adducts following a single i.p. administration of
AA and GA to adult B6C3Fi mice and F344 rats at 50 mg/kg AA or an equimolar dose of GA
(61 mg/kg), and reported GA-derived DNA adducts of adenine and guanine formed in all
relevant tissues in both males and females where tumors had been reported, including liver,
brain, thyroid, leukocytes, mammary gland, and testis in rats and liver, lung, kidney, leukocytes,
and testis in mice. Dosing rats and mice with an equimolar amount of GA typically produced
higher levels of DNA adducts than observed with AA. Kinetics of DNA adduct formation and
accumulation were measured following oral administration of a single dose of AA (50 mg/kg) or
from repeat dosing (1 mg/kg-day), respectively. The formation of these DNA adducts is
consistent with previously reported mutagenicity of AA and GA in vitro, which involved
reaction of GA with adenine and guanine bases. These results provide strong support for a
mutagenic mechanism of AA carcinogenicity in rodents.
AA has been observed to form DNA adducts in vitro, but the formation rate is very slow
(Solomon et al., 1985).
Besaratinia and Pfeifer (2004) treated normal human bronchial epithelial cells and Big
Blue mouse embryonic fibroblasts (that carry a lambda phage ell transgene) in vitro with AA, its
primary epoxide metabolite GA, or water (control) and then subjected the cells to terminal
transferase-dependent polymerase chain reaction to map the formation of DNA adducts within
the human gene encoding the tumor suppressor p53 gene (TP53) and the ell transgene. The
frequency and spectrum of GA-induced mutations in ell were examined by using a lambda
phage-based mutation detection system and DNA sequence analysis, respectively. All statistical
tests were two-sided. AA and GA formed DNA adducts at similar specific locations within
153 DRAFT-DO NOT CITE OR QUOTE
-------
TP53 and ell, and DNA adduct formation was more pronounced after GA treatment than after
AA treatment at all doses tested. AA-DNA adduct formation was saturable, whereas the
formation of most GA-DNA adducts was dose-dependent. GA treatment dose-dependently
increased the frequency of ell mutations relative to control treatment (p < 0.001). GA was more
mutagenic than AA at any given dose, and the spectrum of GA-induced ell mutations was
statistically significantly different from the spectrum of spontaneously occurring mutations in the
control-treated cells (p = 0.038). Compared with spontaneous mutations in control cells, cells
treated with GA or AA had more A~>G transitions and G—>C transversions and GA-treated
cells had more G—>T transversions (p < 0.001). These results support the hypothesis that the
mutagenicity of AA in human and mouse cells is based on the capacity of its epoxide metabolite
GA to form DNA adducts.
Martins et al. (2007) conducted a study that examined, side-by-side, the cytogenic
damage induced by AA and GA in V79 Chinese hamster cells, and compared this damage with
the extent of GA-DNA adduct formation in AA- and GA-treated cells. At the highest
concentration tested (2 mM), AA weakly induced chromosomal aberrations (CAs) and
significantly (p < 0.01) induced sister chromatid exchanges (SCEs). The levels of N7-(2-carb-
amoyl-2-hydroxyethyl) guanine (N7-GA-Gua), a well-characterized GA-DNA adduct, were only
detectable in AA-treated cells at 2 mM and at no other concentration. In contrast, treatment with
equimolar doses of GA produced a twofold higher clastogenic response. In GA-treated cells,
both the induction of SCEs and N7-GA-Gua levels increased linearly in response to GA
concentrations. The strong correlation (r = 0.987,/? = 1.25 x 10"12) between N7-GA-Gua levels
and SCE induction in both AA- and GA-treated cells provided evidence that the metabolic
conversion of AA to GA and the ensuing formation of DNA adducts may play a critical role in
the induction of SCEs. In support, the depurination of DNA is thought to generate abasic sites,
which are vulnerable to DNA breakage. Therefore, the higher clastogenicity of GA compared to
AA is likely due to higher levels of GA-DNA adduct formation, and AA probably induces SCEs
after its conversion to GA. Although these results support the hypothesis that the genotoxicity of
AA is based on its ability to form GA-DNA adducts, another mechanism must also be at play
since the induction of CAs did not correlate with N7-GA-Gua levels.
Additional research by Mei et al. (2008b) supports the hypothesis that AA and GA exert
their mutagenic effects through different mechanisms. In an in vitro mouse lymphoma assay,
mouse lymphoma cells (L5178Y/Tk+/") treated with AA or GA were examined for the frequency
and types of mutations incurred and for GA-DNA adduct formation. A significant increase in
mutation frequency was observed at concentrations >12 mM for AA and >2 mM for GA. In
GA-treated cells, GA-DNA adducts of both adenine and guanine were formed in a linear dose-
response manner; however, no GA-DNA adducts were detected in cells treated with 8-20 mM
AA. DNA analyses of the mutants revealed that the types of mutations incurred by AA and GA
154 DRAFT-DO NOT CITE OR QUOTE
-------
were significantly different (p < 0.018). The majority of AA and GA mutants displayed loss of
heterozygosity (LOH) at the Tk locus; however, AA mutants more frequently contained
mutations that resulted in LOH of more than half of the chromosome. Given these differences,
the authors concluded that AA and GA, although both clastogenic, may induce mutations
through different mechanisms. DNA breakage after GA treatment likely involves GA-DNA
adduct formation, whereas AA may incur DNA damage via oxidative stress. In support, AA is
known to interact with GSH, a nucleophile that protects cells from oxidants. In addition, AA has
been shown to interact with nucleophiles in DNA, lipids, and proteins and to be involved in the
production of reactive oxygen species.
4.7. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS
4.7.1. Oral
Neurological impairment has been established as a human health hazard from AA
exposure, predominantly based on studies of effects from occupational inhalation and dermal
exposure (see Section 4.5.2) (Tilson, 1981; Spencer and Schaumberg, 1974). There are few
reports of health effects in humans associated with oral exposure to AA. However, corroborative
case reports of neurological impairment from oral exposure include one of persistent peripheral
neuropathy in a subject who intentionally ingested 18 g AA crystals (Donovan and Pearson,
1987). In another report, signs of central and peripheral neurological deficits were observed in
family members exposed to AA in well water at a concentration of 400 ppm; both oral and
dermal exposure to AA were likely (Igisu and Matsuoka, 2002; Igisu et al., 1975).
Epidemiologic studies designed to evaluate noncancer health effects in groups of orally exposed
subjects have not been conducted.
Numerous studies in animals provide evidence of neurotoxic effects in males and females
and reproductive effects in males as the most sensitive noncancer effects associated with oral
exposure to AA (summarized in Table 4-32). The studies in Table 4-32 provided the
information needed to characterize the dose-response relationships for noncancer effects.
Table 4-32. Noncancer effects in animals repeatedly exposed to acrylamide
by the oral route
Reference/species
Bureketal., 1980
F344 rat, F
Field etal., 1986
F344 rat, M&F
Exposure
conditions
(mg/kg-day)
0,2.5,
or 20 by gavage
0,3, 15, or 45
2 years in D W
NOAEL
LOAEL
(mg/kg-day)
7.5
5
5
15
45
15
1
ND
ND
2
ND
45
Effect
Degenerative nerve changes (EM)
Degenerative nerve changes (LM)
Hindlimb splay, maternal
Degenerative nerve changes (LM)
Hindlimb splay, maternal
155 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-32. Noncancer effects in animals repeatedly exposed to acrylamide
by the oral route
Reference/species
Wiseetal., 1995
F344 rat, F
Tyletal, 2000a
F344 rat, F
Chapinetal., 1995
CD-I mouse, M&F
Zenicketal., 1986
Long-Evans rat, M
Zenicketal., 1986
Long-Evans rat, F
Smith etal., 1986
Long-Evans rat, M
Sakamoto and
Hashimoto, 1986
ddY mouse, M
Sakamoto and
Hashimoto, 1986
ddY mouse, F
Field etal., 1990
Sprague-Dawley
rat, F
Field etal., 1990
CD-I mouse, F
Exposure
conditions
(mg/kg-day)
0,0.1,0.5, or 20
GD 6-10 by gavage
0,0.5, 2.0, or 5.0
Two generations in
DW
0,0.8, 3.1, or 7.5
Two generations in
DW
0,4.6, 7.9, or 11.9
10 weeks in DW;
mated w/
nonexposed F
0,5.1, 8.8, or 14.6
9 weeks in DW;
mated w/
nonexposed M
0, 1.5, 2.8, or 5.8
80 days in DW;
mated w/
nonexposed F
0,3.3,9.0, 13.3,
or 16.3
4 weeks in DW;
mated w/
nonexposed F
0, 18.7
4 weeks in DW;
mated w/
nonexposed M
0,2.5, 7.5, or 15
GD 6-20 by gavage
0,3, 15, or 45
GD 6-17 by gavage
NOAEL
LOAEL
(mg/kg-day)
10
10
ND
10
ND
25
25
3.1
7.5
3.1
7.5
3.1(F)
ND
4.6
5.1
5.1
14.6
1.5
5.8
5.8
9.0
13.3
13.3
18.7
ND
7.5
15
15
15
45
15
15
15
5
15
25
ND
ND
7.5
ND
7.5
ND
7.5(F)
7.9
7.9
8.8
8.8
ND
2.8
ND
ND
13.3
16.3
16.3
ND
18.7
15
ND
ND
45
ND
45
Effect
Decreased maternal weight gain
Hindlimb splay
Decreased body weight (8-9%)
Early mortality after 60 weeks
Other nonneoplastic lesions
Hindlimb foot splay, maternal
Degenerative nerve changes (LM)
Hindlimb foot splay in offspring
MM implantation losses (FO&F1)
Degenerative nerve changes (F1,LM)
Mild grip strength deficits (F1&F2)
Hindlimb foot splay
Decreased body weight (8%, Fl only)
MM implantation losses
Hindlimb foot splay
Decreased maternal body weight (6%)
Decreased pup body weight (30-35%)
Other reproductive performance endpoints
(fertility, implantation loss)
MM postimplantation losses
Peripheral nerve changes (LM)
Hindlimb foot splay
MM decreased fetuses/dam
Slight hindlimb weakness
Decreased sperm counts, abnormal sperm
morphology
Female reproductive performance
Slight hindlimb weakness
Decreased maternal weight gain
Fetal malformations or variations
Hindlimb splay, maternal
Decreased maternal weight gain
Fetal malformations or variations
Hindlimb splay, maternal
156 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-32. Noncancer effects in animals repeatedly exposed to acrylamide
by the oral route
Reference/species
Wiseetal., 1995
Sprague-Dawley
rat, F
Friedman etal.,
1999a
Wistar rat, F
Garey and Paule,
2007
F344 rat, M and F
adolescents
Exposure
conditions
(mg/kg-day)
0,5, 10, 15, or 20
GD 6-10 by gavage
0, 25 (maternal
doses) PND 0-21
by gavage
Gavage to dams
during gestation;
PNDs 1-21 to
offspring at same
gavage dose;
PNDs 22-85 in
drinking water.
Average estimated
doses for offspring:
0,0.1,0.4, 1.3,or6.
NOAEL
LOAEL
(mg/kg-day)
10
10
ND
10
ND
25
25
6
1.3
15
15
5
15
25
ND
ND
ND
6
Effect
Decreased maternal weight gain
Hindlimb splay, maternal
Decreased body weight in offspring
Increased overall horizontal activity,
decreased auditory startle response in
offspring
Hindlimb foot splay, maternal
Degenerative nerve changes (LM), maternal
Hindlimb foot splay in offspring
Offspring body weight
Decreased cognitive motivation in adolescent
offspring
DW = drinking water LM = light microscopy ND = not determined
EM = electron microscopy LOAEL = lowest-observed-adverse-effect level
F = female M = male NOAEL = no-observed-adverse-effect level
GD = gestation days MM = male-mediated PND = postnatal days
157 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-32 indicates that the lowest effect levels are for degenerative peripheral nerve
changes in rats exposed to 1 mg/kg-day AA in drinking water for 90 days (Burek et al., 1980) or
2 mg/kg-day (Johnson et al., 1986) or 2 or 3 mg/kg-day (Friedman et al., 1995) for 2 years.
Comprehensive histologic examinations of all major organs and tissues in these rat studies
revealed no other exposure-related nonneoplastic lesions at dose levels below 5 mg/kg-day
(Friedman et al., 1995; Johnson et al., 1986; Burek et al., 1980) (see Table 4-32). Although
studies selected for inclusion in Table 4-32 only examined rats and mice, Table 4-33 lists reports
of AA neurological impairment in other species (cats, dogs, monkeys, and additional mouse
studies) exposed via intraparenteral administration or orally at higher dose levels.
Table 4-33. Neurological effects following exposure to acrylamide in species
other than the rat and mouse
Reference/Species
McCollisteretal., 1964
Cats (n = 2)
Post and McLeod, 1977
Cats (2-3 kg)
Herschetal., 1989
Dogs (greyhounds,
22-30 kg)
Satchell and McLeod 1981
Dogs (greyhound)
Eskinetal., 1985
Monkeys (macaque)
Maurissenetal., 1983
Monkeys (pigtail)
McCollisteretal., 1964
Monkeys (5.1 kg)
Gilbert and Maurissen, 1982
Mice (Balb/c)
Hashimoto etal., 1981
Mice (ddY strain)
Exposure conditions
(dose, route, duration)
Single 100 mg/kg i.p. dose
15 mg/kg in food for up to
16 weeks
5.7 mg/kg-day via ingested
capsule for 6-7 weeks
7 mg/kg-day in feed for 8
weeks
10 mg/kg-day in juice,
5 days/week for 6-10 weeks
10 mg/kg-day in juice,
5 days/week until appearance of
mild toxicity (n = 4; average
for 54 days; average total dose
400 mg/kg)
total of 200 mg/kg of four
consecutive 50 mg/kg i.v. doses
25.8 mg/kg-day (250 ppm) AA
in drinking water for 12 days
(total estimated dose
3 10 mg/kg)
1/5 to 1/2 of the LD50
(107 mg/kg) administered by
gavage twice weekly for 8-
10 weeks
Effect
After 24 hours, one was unconscious and was
sacrificed, the other had severe neurotoxicity.
Progressively increasing neurotoxicity; by 12-
16 weeks, severe poisoning, reduction in
conduction velocity, damage to large and small
myelinated fibers in peripheral nervous system.
Progressive, but reversible dysfunction of the
pulmonary stretch receptors and their
myelinated vagal afferents.
Sensorimotor peripheral neuropathy and
megaesophagus suggesting an axonopathy of
the vagus nerve.
Axonal swellings with neurofilament
accumulation in the distal optic tract and lateral
geniculate nucleus.
Loss of balance, impaired coordination, tremor
(these symptoms reversed relatively soon after
dosing); reduced vibration sensitivity and
remained impaired for several months after
dosing.
Death.
Decreased retention time and increased
hindlimb splay.
241 mg/kg was the total dose for half maximal
inhibition of rotarod performance.
LoPachin et al. (2002b) reported measures of gait characteristics as a sensitive behavioral
measure for the onset and progression of AA neurotoxicity, but the study protocols cited in
Table 4-32 were not oriented towards neurobehavioral endpoints and did not evaluate gait
158 DRAFT-DO NOT CITE OR QUOTE
-------
abnormalities. Instead, hindlimb foot splay, a gross characteristic sign of AA-induced peripheral
neuropathy, was measured in several of the studies cited in Table 4-32. Changes in foot splay
have been observed in most chronic or less-than-lifetime oral studies, but at levels above the
lowest dose associated with histologic signs of peripheral nerve damage (1-3 mg/kg-day). Only
one study reported statistically significantly increased incidences of FO-generation F344 rats
with hindlimb foot splay following exposure to a dose level as low as 0.5 mg/kg-day (Tyl et al.,
2000a) (Table 4-32). In the same study, however, Tyl et al. (2000a) did not observe hindlimb
foot splay in the Fl-generation rats exposed to doses as high as 5 mg/kg-day, nor was this
endpoint reported in F344 rats exposed to drinking water doses as high as 2-3 mg/kg-day for 2
years (Friedman et al., 1995; Johnson et al., 1986) or 5 mg/kg-day for 90 days (Burek et al.,
1980). Although adverse behavioral effects are not currently sufficiently supported to be
designated the most sensitive endpoint, there is a clear research need for additional
neurobehavioral studies with protocols and endpoints suitable for quantifying low dose-response
relationships, and efforts are ongoing to address this data need.
AA induces adverse reproductive and developmental effects, but study data suggest these
effects occur at higher doses than those resulting in neurotoxicity. Pre- and postimplantation
losses and decreased numbers of live fetuses have been observed following repeated prebreeding
oral exposure of rats and mice to AA at doses in the range of 3 to 8 mg/kg-day (Chapin et al.,
1995; Sakamoto and Hashimoto, 1986; Smith et al., 1986; Zenick et al., 1986) (see Table 4-32).
Dominant lethality testing (Tyl et al., 2000a, b; Chapin et al., 1995; Smith et al., 1986) and
crossover trials (Chapin et al., 1995; Sakamoto and Hashimoto, 1986; Zenick et al., 1986)
indicate male-mediated reproductive effects (Table 4-32). More gross effects on male
reproductive organs have been demonstrated at higher dose levels, e.g., exposure of F344 rats to
20 mg/kg-day AA in drinking water for 90 days produced severe testicular atrophy (Burek et al.,
1980). Male germ cell assays (e.g., sperm abnormalities, heritable translocations, specific locus
mutations) provide evidence of AA-induced male reproductive toxicity following drinking water
(Sakamoto and Hashimoto, 1986) or i.p. exposures (Adler et al., 2004, 2000, 1994; Generoso et
al., 1996; Ehling and Neuhauser-Klaus, 1992; Russell et al., 1991; Adler, 1990; Sega et al.,
1989; Shelby et al., 1987). No experiments have studied the potential for AA to induce heritable
mutations in the female germ line. Prebreeding exposure of female mice to doses of 18.7 mg/kg-
day (Sakamoto and Hashimoto, 1986) or female Long-Evans rats to doses up to 14.6 mg/kg-day
(Zenick et al., 1986) did not adversely affect reproductive performance variables such as fertility
or implantation when the animals were bred with nonexposed males (Table 4-32). In these
female-exposure studies, the only reproductive endpoint affected was body weight decreases in
offspring of female Long-Evans rats exposed to 8.8 and 14.6 mg/kg-day (Zenick et al., 1986).
Comparing the study LOAEL values listed in Table 4-32 suggests that the onset of
adverse effects for male reproductive toxicity results from lower levels of AA exposure (2.8-
159 DRAFT-DO NOT CITE OR QUOTE
-------
13.3 mg/kg-day) than those needed to produce clinical signs of neurotoxicity (15-20 mg/kg-day)
but higher than those that result in peripheral nerve damage following less-than-lifetime or
chronic exposures (1-2 mg/kg-day).
Developmental effects associated with oral exposure to AA are restricted to body weight
decreases and decreased auditory startle response in offspring of female Sprague-Dawley rats
exposed to 5 and 15 mg/kg-day, respectively, on GDs 6-10 (Wise et al., 1995) and decreased
performance in an operant test of cognitive motivation in adolescent F344 rats exposed during
gestation and lactation and extending through 12 weeks of age at an average dose of 6 mg/kg-
day, but not at 1.3 mg/kg-day (Garey and Paule, 2007). No exposure-related fetal malformations
or variations (gross, visceral, or skeletal) were found in Sprague-Dawley rats exposed to doses
up to 15 mg/kg-day on GDs 6-20 or in CD-I mice exposed to doses up to 45 mg/kg-day on GDs
6-17 (Field et al., 1990) (Table 4-32). These doses produced decreased maternal weight gains.
No signs of hindlimb foot splay or other gross signs of peripheral or central neuropathy were
noted in suckling offspring of female Wistar rats that were given gavage doses of 25 mg/kg-day
during the postnatal lactation period (Friedman et al., 1999a).
Subchronic or chronic exposure to AA doses in the 2-8.8 mg/kg-day range resulted in
small body weight deficits (4-9% decreased compared with controls) in F344 rats (Tyl et al.,
2000a; Friedman et al., 1995; Johnson et al., 1986), CD-I mice (Chapin et al., 1995), and Long-
Evans rats (Zenick et al., 1986). More pronounced decreases in body weight were seen at higher
doses, but these also produced overt neurotoxicity (e.g., Burek et al., 1980).
4.7.2. Inhalation
Numerous reports have associated human exposure to AA with neurological impairment
(Igisu and Matsuoka, 2002; Gjerl0ff et al., 2001; Hagmar et al., 2001; Mulloy, 1996; Calleman et
al., 1994; Bachmann et al., 1992; Myers and Macun, 1991; Dumitru, 1989; He et al., 1989;
Donovan and Pearson, 1987; Kesson et al., 1977; Mapp et al., 1977; Davenport et al., 1976;
Igisu et al., 1975; Takahashi et al., 1971; Fullerton, 1969; Auld and Bedwell, 1967; Garland and
Patterson, 1967). Most reports involved occupational exposure with potential for both inhalation
and dermal exposure. Although exposure concentrations of AA were measured in some
instances, studies describing reliable relationships between exposure concentrations and
neurological responses in humans are not available. However, cross-sectional health
surveillance studies of AA-exposed workers describe correlative relationships between
hemoglobin adduct levels of AA (an internal measure of cumulative dose) and changes in a
neurotoxicity index based on self-reported symptoms and clinical measures of neurological
impairment (Calleman et al., 1994) or increased incidences of self-reported symptoms alone
(Hagmar et al., 2001). These studies, however, do not provide reliable information on dose-
response relationships for chronic inhalation exposure to AA because (1) they involved mixed
160 DRAFT-DO NOT CITE OR QUOTE
-------
inhalation and dermal exposure (in both groups of workers dermal exposure was thought to have
been substantial); (2) the duration of exposure was less than chronic; (3) both groups of workers
were exposed to confounding chemicals (acrylonitrile in the first and NMA in the second study);
and (4) the internal measure of dose (N-terminal valine adducts of hemoglobin) is not specific
for AA alone (e.g., NMA can form the same adduct).
Data on AA-induced toxicity in animals exposed by inhalation are limited to a single
report of progressive signs of neuropathy and death in rats and dogs following acute-duration
repeated exposure to aerosols of AA dust at a concentration of 15.6 mg/m3 (Hazleton
Laboratories, 1953).
4.7.3. Mode-of-Action Information
4.7.3.1. Neurotoxic Effects
Since experimental AA neuropathy was first reported (Hazleton Laboratories, 1953), AA
has been extensively studied in efforts to understand its toxicological properties and MOA for
the functional deficits observed in animal studies, including alterations in gait, hindfoot splay,
impaired mobility and righting reflex, and decreased grip strength (Moser et al., 1992; Dixit et
al., 1981; Tilson and Cabe, 1979; Fullerton and Barnes, 1966; McCollister et al., 1964). Similar
muscle weakness and functional impairments have been observed in humans exposed to AA
(Hagmar et al., 2001; Calleman et al., 1994; He et al., 1989).
Early animal research associated AA functional neurotoxicity with central and peripheral
distal axonopathy and more specifically with histopathologic findings of neurofilamentous
accumulations in distal paranodal regions of large peripheral nerve fibers that appeared to cause
local axon swelling and subsequent degeneration of myelin (Spencer and Schaumberg, 1977,
1974). Axon degeneration was observed to progress proximally toward the cell body region, a
process known as "dying back." Based on these findings, neurofilaments were thought to be a
target for AA toxicity. Other potential pathways for AA-induced axonopathy included
interference with nerve cell body metabolism and delivery of nutrients to the axon (Spencer et
al., 1979; Cavanagh, 1964), interruption of axonal protein transport (Pleasure et al., 1969),
disruption of axon cytoskeleton (Lapadula et al., 1989), diminished axolemma Na+,K+-ATPase
activity (LoPachin and Lehning, 1994), and reduction of fast anterograde axonal transport
capacity (Harris et al., 1994; Padilla et al., 1993; Harry, 1992; Sickles, 1991).
Recent investigations serve as basis for two hypothetical MO As for AA neurotoxicity,
disruption of nitric oxide signaling at the nerve terminal and fast axonal transport disruption. A
third MOA hypothesis, which has received less research support, proposes that AA effects on
nerves may involve enhanced lipid peroxidation and decreased antioxidant status (decreased
GSH levels and anti-reactive oxygen species enzyme activity).
161 DRAFT-DO NOT CITE OR QUOTE
-------
Disruption of nitric oxide (NO) signaling at the nerve terminal hypothesis
The hypothesis that AA-induced neurotoxicity occurs at the nerve terminals as a primary
site of action possibly due to disruption of neuronal NO signaling has been supported by the
work of LoPachin and colleagues (LoPachin et al., 2008; LoPachin and Barber, 2006). AA is a
conjugated a,p-unsaturated carbonyl derivative in the type-2 alkene chemical class. Because
electrons in pi orbitals of a conjugated system are mobile, the a,p-unsaturated carbonyl structure
of AA is characterized as a soft electrophile according to the hard-soft, acid-base principle
(reviewed in Pearson, 1967). Characteristically, soft electrophiles will preferentially form
Michael-type adducts with soft nucleophiles, which in biological systems are primarily
sulfhydryl groups on cysteine residues (LoPachin and DeCaprio, 2005; Hinson and Roberts,
1992). Free sulfhydryl groups can exist in the reduced thiol-state or in the anionic thiolate-state,
and recent research indicates that the highly nucleophilic thiolate is the preferential adduct target
for AA (LoPachin et al., 2007a; see also Friedman et al., 1995). Based on the pKa of cysteine
(pH 8.5), at physiological pH (7.4), the thiolate state exists only in unique protein motifs called
catalytic triads, where proton shuttling through an acid-base pairing of proximal amino acids
(e.g., aspartic acid and lysine) regulates the protonation and deprotonation of the cysteine
sulfhydryl group. Indeed, both mass spectrometric and kinetic data have demonstrated the
selective adduction of cysteine residues on many neuronal proteins (Barber et al., 2007; Barber
and LoPachin, 2004). Furthermore, it is now recognized that the redox state or nucleophilicity of
cysteine sulfhydryl groups within catalytic triads can determine the functionality of these
proteins (reviewed in LoPachin and Barber, 2006; Stamler et al., 2001). In contrast to AA, the
epoxide metabolite GA, is a hard electrophile that preferentially forms adducts with hard
nucleophiles such as nitrogen, carbon, and oxygen. Nucleotide residues of DNA contain
abundant hard nucleophilic targets, which is consistent with the formation of GA adducts on
adenine and guanine bases in AA-intoxicated animals (Doerge et al., 2008, 2005; Ghanayem et
al., 2005a, b).
Based on the observation that the processes affected (e.g., neurotransmitter release and
storage) and corresponding kinetics (Km, Vmax) were similar in synaptosomes exposed in vitro
to AA and those isolated from AA-intoxicated rats (LoPachin et al., 2006, 2004; Barber and
LoPachin, 2004), LoPachin and colleagues have reasoned that the parent compound, AA, is
responsible for neurotoxicity. Moreover, cysteine thiolate groups have clear regulatory functions
in many critical neuronal processes (LoPachin and Barber, 2006), whereas protein valine, lysine,
and histidine residues, which are the likely hard nucleophilic targets for a hard electrophile such
as GA, have unclear functional and therefore toxicological relevance. Quantitative
morphometric and silver stain analyses of PNS and CNS of AA-intoxicated animals support the
hypothesis that axon degeneration is an epiphenomenon related to dose-rate; i.e., degeneration
occurs at lower but not higher dose-rates. In contrast, some studies indicate that nerve terminal
162 DRAFT-DO NOT CITE OR QUOTE
-------
degeneration occurs regardless of dose-rate and in correspondence with the onset and
development of neurological deficits (Lehning et al., 2003a, b, 2002; 1998; Crofton et al., 1996;
reviewed in LoPachin et al., 2003, 2002a; LoPachin and Lehning 1994), suggesting the nerve
terminals as a primary site of action. Subsequent neurochemical studies showed that both in
vitro and in vivo AA exposure produced early disruptions of neurotransmitter release, reuptake,
and vesicular storage (LoPachin et al., 2007b, 2006, 2004; Barber and LoPachin, 2004). Further,
proteomic analyses indicated that the inhibition of presynaptic function was due to the formation
of cysteine adducts on proteins that regulate neurotransmitter handling; e.g., Cys 264 of 7V-
ethylmaleimide sensitive factor, Cys 254 of v-ATPase (see LoPachin et al., 2009, 2007a, b;
Barber et al., 2007; Barber and LoPachin, 2004; Feng and Forgac, 1992; reviewed in LoPachin
and Barber, 2006). The anionic sulfhydryl state, which is only found in the catalytic triads of
regulatory proteins, is an acceptor for NO and, therefore, has led to the hypothesis that AA-
induced neurotoxicity results from disruption of neuronal NO signaling (LoPachin et al., 2008;
LoPachin and Barber, 2006).
Fast axonal transport disruption hypothesis
Sickles et al. (2002a) provide support for AA neurotoxicity resulting from inhibition of
the movement of materials in fast axonal transport from both AA and GA. According to this
"kinesin/axonal transport" hypothesis, toxicant inhibition of kinesin leads to reductions in the
axonal delivery of macromolecules eventually producing a deficiency of the essential proteins
required to maintain axon structure, function, or both. Distal axons and nerve terminals are
particularly vulnerable to transport defects based upon an exceptionally large axonal volume (as
much as 1,000 times the volume of the neuron cell body) and the dependence of these distal
regions on long distance transport (100-fold longer length than diameter of the cell body). This
regional sensitivity is consistent with the previously identified distal spatial distribution of
toxicant-induced damage (Cavanagh, 1964).
Microtubule motility assays using purified kinesin from bovine brain identified a dose-
dependent inhibition of kinesin as well as a less sensitive effect on microtubules (Sickles et al.,
1996). Preincubation of either kinesin or taxol-stabilized microtubules with AA produced a
reduction in the affinity between kinesin and microtubules, recognized as a reduced number of
microtubules bound or locomoting on an absorbed bed of kinesin. Microtubules that were
locomoting did so in a less directed or staggering type of progression. The inhibitions were
proposed to be due to covalent adduction, presumably through sulfhydryl alkylation by AA,
although adduction of other amino acid residues such as valine was possible. The non-
neurotoxic analog, propionamide, had no effect. Other investigators have identified kinesin
inhibition by sulfhydryl reagents such as N-ethylmaleimide and ethacrynic acid (Walker et al.,
1997). As with AA, inhibition by these sulfhydryl reagents produced the characteristic
163 DRAFT-DO NOT CITE OR QUOTE
-------
staggering movement of microtubules. The reaction was slow and temperature-dependent,
suggesting a sterically hindered cysteine residue as an important adduct target. Additional
studies have demonstrated a comparable effect of GA on kinesin (Sickles, unpublished data).
The predicted outcome of such an effect would be reduced quantity of flow, precisely the
outcome from several experiments where rate of transport versus quantity could be discriminated
(Stone et al., 1999; Sickles, 1989a, b).
Fast axonal transport has been studied in a variety of model systems using diverse
techniques. A comprehensive survey of AA effects on fast anterograde and retrograde axonal
transport (Sickles et al. 2002a) revealed that all studies measuring fast transport within 24 hours
of AA exposure demonstrated significant reductions, whereas longer postexposure delay was not
associated with changes in transport. Furthermore, a reduction in transport quantity (but not
rate) has been reported within 20 minutes of exposure. The duration of this effect was 16 hours,
with full recovery at 24 hours (Sickles, 1991). Quantitation of transport after multiple dosings
(i.e., 4, 7, or 10 doses) had a similar effect on transport in the proximal sciatic nerve (Sickles,
1991). The changes in transport were not due to an effect on protein synthesis and exposure of
only the axons confirmed that the target was axonal (Sickles, 1992, 1989a, b). Collectively,
these results suggest action on a target that is replaced via the fast transport system, which is
consistent with kinesin as that target. The actions of AA on fast axonal transport were
independent of effects on axonal neurofilaments, as similar reductions were observed in wild-
type and transgenic mice lacking axonal neurofilaments (Stone et al., 2000, 1999). The same
results were observed using radiolabeling of proteins in mouse optic nerves and differential
interference microscopy of isolated sciatic nerve axons. Other recent studies have identified a
parallel inhibition of retrograde axonal transport by AA (Sabri and Spencer, 1990), although it is
unclear whether this effect is due to inhibition of cytoplasmic dynein, the retrograde axonal
transport motor, or whether this is a result of indirect effects of kinesin motor inhibition (Brady
etal., 1990).
The predicted outcome from axonal transport compromise is a reduction in vital
macromolecules in the distal axons and an accumulation of transported material within the axon.
Morphological studies have consistently identified accumulations of tubulovesicular profiles and
neurofilaments in axons of AA-intoxicated animals (Spencer and Schaumburg, 1991), which are
morphological elements transported via kinesin along microtubules. Other studies have
identified reduced synaptic vesicles in neuromuscular junctions (DeGrandchamp and Lowndes,
1990; DeGrandchamp et al., 1990). A reduction in GAP-43 in the terminal neurites of cultured
primary spinal cord neurons following AA exposure has been observed (Clarke and Sickles,
1996). Future studies are required to quantitate reductions in specific axonal compartments
using a variety of neurotoxic and non-neurotoxic dosing regimens in vivo to confirm the loss of
physiologically or structurally important macromolecules.
164 DRAFT-DO NOT CITE OR QUOTE
-------
Additional supportive data for the axonal transport hypothesis come from several studies
of kinesin knockouts as well as similarity to human diseases. While most knockouts are lethal,
low level mutations of kinesin motors in Drosophila have identified an identical spatial pattern of
dysfunction and morphological similarity in axonal pathology (Kurd and Saxton, 1996; Gho et
al., 1992) as with AA intoxication. The group of neurological disorders classified as hereditary
spastic paraplegias has a spatial pattern of ataxia, spasticity, and muscle weakness as observed
with AA intoxication. Some of these types have been associated with mutations in kinesin
motors (Reid et al., 2002), while others are the result of either axonal or glial protein mutations.
However, the common theme is alteration in axonal transport (Gould and Brady, 2004; Reid,
2003).
Reactive oxygen species hypothesis
Zhu et al. (2008) provide some data supporting a third MO A for AA induced
neurotoxicity involving enhancement of lipid peroxidation and decreased antioxidative capacity,
as well as depletion of neural glutathione levels and antioxidant enzyme activities, resulting in
the key sequence of events of increased levels of reactive oxygen species, damage to cellular
macromolecules, and subsequent degeneration of neural tissues. In this study, adult male Wistar
rats were given AA (40 mg/kg i.p., 3 times/week) for 2, 4, 6, and 10 weeks. Time-dependent
changes in levels of malondialdehyde (an indicator of lipid peroxidation) and reduced
glutathione and enzyme activity levels of glutathione peroxidase, glutathione reductase,
superoxide dismutase, and anti-reactive oxygen species were examined along with several
electrophysiological indices (nerve conduction velocity and compound action potential duration,
amplitude, and latency). Time-dependent decreased glutathione levels and anti-reactive oxygen
species activities and increased malondialdehyde levels in sciatic nerve preparations were highly
correlated (p < 0.05, |r > 0.80) with changes in electrophysiological indices of AA-induced
neurotoxicity.
Summary and data needs
The respective adduct chemistries of AA and GA are well understood and could have
fundamental implications for neurotoxicity regardless of the proposed mechanism; i.e., kinesin
inhibition (Sickles et al., 2002a) or blockade of NO signaling (LoPachin et al., 2009, 2008;
LoPachin and Barber, 2006). Accordingly, an obvious data gap in the current mechanistic
understanding of AA neurotoxicity is the relative roles of the parent compound and GA. Thus,
although early research suggested that GA produced neurotoxicity both in whole animal (Abou-
Donia et al., 1993) and in vitro (Harris et al., 1994) model systems, other studies using similar
models failed to find neurotoxic effects associated with the GA metabolite (Costa et al., 1995,
1992; Brat and Brimijoin, 1993; Moser et al., 1992). Clearly, resolving the relative roles of AA
165 DRAFT-DO NOT CITE OR QUOTE
-------
vs. GA is an important issue that will require more research. Although the adduct chemistry of
these toxicants has been reasonably defined, the precise molecular mechanisms and sites of
neurotoxicity have not yet been clearly resolved.
4.7.3.2. Reproductive Effects
The MOA for AA-induced reproductive toxicity is poorly understood. Positive results of
germ cell mutagenicity assays and reproductive toxicity tests indicate that some aspects of
reproductive toxicity may be mediated by mutagenic effects on male germ cells (see Sections
4.3.1, 4.3.3, and 4.4.3) (Costa et al., 1992). Mechanistic proposals have also been made for a
common MOA for neurotoxic and male fertility effects (e.g., effects on mounting, sperm
motility, and intromission) involving modifications of kinesin and sulfhydryl groups of other
proteins by AA and/or GA and a separate mechanism for male dominant lethal mutations
involving clastogenic effects from AA and/or GA interactions with protamine or spindle fiber
proteins in spermatids and/or direct alkylation of DNA by GA (Perrault, 2003; Tyl and
Friedman, 2003; Adler et al., 2000; Tyl et al., 2000b; Sega et al., 1989).
Sega et al. (1989) proposed AA alkylation of protamine in late-stage spermatids as a
mechanism for AA-induced dominant lethal effects based on a parallel time course for
protamine alkylation and dominant lethal effects in spermatids of mice treated with AA. This
observation was repeated by Adler et al (2000), who further proposed that the GA metabolite is
the ultimate clastogen in mouse spermatids based on the results of enzyme inhibition studies.
Zenick et al. (1994) summarized the MOA as follows:
Protamines are highly basic (arginine and lysine rich) proteins that also contain
numerous cysteine residues. During epididymal transit and spermatozoal
maturation, the cysteine sulfhydryls are oxidized to form both inter- and
intramolecular disulfide bonds. These confer even greater stability on sperm
nuclei such that they become resistant to disruption by any means, including
anionic detergent treatment, unless a disulfide-reducing agent is applied. This
remarkably stable structure packages sperm DNA such that it remains
transcriptionally inert and protected from damage during transit through both the
epididymis and the female tract. Only after the sperm have entered the oocyte are
the disulfide bonds in its chromatin reduced, thus initiating the rapid
decondensation of the sperm nucleus with replacement of protamines by somatic
histones, and subsequent reactivation of the male genome. Chemicals that disrupt
sperm chromatin packaging by altering the synthesis of disposition of testis-
specific transitional proteins (which first replace somatic histones prior to
themselves being replaced with protamine) or protamines, or by binding to free
sulfhydryls and thus preventing protamine cross-linking, may contribute to
genetic damage, perhaps by an indirect mechanism or by making the chromatin
more vulnerable other DNA-binding chemicals.
The hypothesis that AA-induced germ cell and somatic mutations in male mice require
CYP2E1-mediated epoxidation of AA to GA received strong support from studies by Ghanayem
166 DRAFT-DO NOT CITE OR QUOTE
-------
et al. (2005a, b) where dose-responses for germ-cell and somatic mutagenicity were compared
between male CYP2El-null and wild-type mice treated with AA. In both studies, effects were
not observed in the CYP2El-null mice, while treated wild-type male mice responded with dose-
related increases in resorption moles (i.e., chromosomally aberrant embryos), decreases in the
numbers of pregnant females and the proportion of living fetuses, and somatic cell mutations.
These results support further evaluation of CYP2E1 polymorphisms in human populations as a
possible determinant of variability in, and susceptibility to, AA genotoxicity in the human
population.
Support for the occurrence of DNA alkylation in the MOA leading to dominant lethals
includes the detection of DNA adducts of GA in various tissues from mice and rats following
single i.p. injections of 50 mg/kg AA (Segerback et al., 1995). The mechanistic proposals
presented by Tyl and Friedman (2003) appear to be consistent with other proposals that the
primary direct biological reactivity of AA involves binding to proteins (in vitro direct binding of
AA to DNA is very slow), AA is converted to GA in rats and humans, and GA can react both
with proteins and with DNA (Dearfield et al., 1995).
4.8. EVALUATION OF CARCINOGENICITY
4.8.1. Summary of Overall Weight of Evidence
In accordance with the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a),
AA is characterized as "likely to be carcinogenic to humans." This characterization is based on
the following findings: (1) chronic oral exposure of F344 rats to AA in drinking water induced
statistically significant increased incidences of thyroid follicular cell tumors (adenomas and
carcinomas combined in both sexes), scrotal sac mesotheliomas (males), and mammary gland
fibroadenomas (females) in two bioassays; (2) oral, i.p., or dermal exposure to AA initiated skin
tumors that were promoted by TPA in SENCAR and Swiss-ICR mice; and (3) i.p. injections of
AA induced lung adenomas in strain A/J mice. In addition, CNS tumors were observed in both
of the chronic F344 rat bioassays. The elevation of the incidence for CNS tumors was
significant in the one bioassay and of uncertain statistical significance in the other. There are no
animal data on the carcinogenicity following chronic inhalation exposure to AA. EPA's
Guidelines for Carcinogen Risk Assessment (2005) indicate that for tumors occurring at a site
other than the initial point of contact, the weight of evidence for carcinogenic potential may
apply to all routes of exposure that have not been adequately tested at sufficient doses. In the
case of AA, there is evidence of rapid, nearly complete absorption from the oral route and rapid
distribution throughout the body (Kadry et al., 1999; Miller et al., 1982) and evidence that the
elimination kinetics of radioactivity from oral or i.v. administration of radiolabeled AA in rats is
similar (Miller et al., 1982). In addition, there is similar flux of AA through metabolic pathways
following either single dose oral or single 6 hr inhalation exposures in rats (Sumner et al., 2003)
167 DRAFT-DO NOT CITE OR QUOTE
-------
and while there are some route-to-route differences in the relative amounts of AA to GA, the
differences are within two fold of each other. For these reasons, acrylamide is considered likely
to be carcinogenic to humans by all routes of exposure.
The potential for AA carcinogenicity from dietary exposure has been assessed in a
number of case-control studies (Wilson et al., 2009a; Pelucchi et al., 2007, 2006; Michels et al.,
2006; Mucci et al., 2005, 2004, 2003) and several prospective studies (Larsson et al., 2009a, b, c,
d; Wilson et al., 2009b; Hogervorst et al., 2008a, b, 2007; Mucci et al., 2006). Two cohort
mortality studies (Collins et al., 1989; Sobel et al., 1986) with follow-up analyses (Marsh et al.,
2007, 1999; Swaen et al., 2007) have assessed associations with inhalation and dermal exposure
to AA in the workplace. In addition, two case-control studies examined relationships between
AA-Hb adducts and risks for breast cancer (Olesen et al., 2008) and prostate cancer (Wilson et
al., 2009a). These studies are judged as providing limited or no evidence of carcinogenicity in
humans.
In most of the case-control studies and prospective studies, no statistically significant
associations were found between frequent consumption of foods with high or moderate levels of
AA and cancer incidence for large bowel, bladder, kidney, renal cell, breast, colorectal, oral,
pharyngeal, esophageal, laryngeal, ovarian, or prostate cancer. Some of the sites observed in the
animal studies (thyroid, testicular, CNS) have not been evaluated, and there are limitations in
some of the study methods and cohort sizes. One case-control study reported a slightly increased
risk of breast cancer later in life associated with the consumption of French fries during
preschool (Michels et al., 2006). Olesen et al. (2008) reported a significant positive association
between AA-Hb adduct levels in red blood cells and ER+ breast in a case-control study of
Danish women (374 cases and 374 controls) only after adjusting for smoking; no significant
association was found between AA-Hb or GA-Hb adduct levels and total breast cancer either
with or without adjustment for smoking. Increased risks of postmenopausal endometrial,
ovarian, and renal cell cancer with increasing dietary AA intake were reported in prospective
studies of a Dutch population (Hogervorst et al., 2008a, 2007). Each of these studies were
limited by uncertainties in exposure assessments.
No statistically significant increased risks for cancer-related deaths were found in the
cohort mortality studies of AA workers with the exception that, in an exploratory dose-response
analysis of one follow-up assessment, an increased risk for pancreatic cancer was reported in a
subgroup with the highest cumulative AA exposure (Marsh et al., 1999). However, no increased
risk for pancreatic cancer was observed in the most recent follow-up analysis of this cohort
(Marsh et al., 2007).
The majority of the data support a mutagenic MOA for AA carcinogenicity. AA has
been reported to induce gene mutations and chromosomal aberrations in somatic and germ cells
of rodents in vivo and cultured cells in vitro, to transform cells of mouse cell lines, and to form
168 DRAFT-DO NOT CITE OR QUOTE
-------
adducts with protamines in germ cells. The mutagenic potential of GA is well- characterized in
studies of the induction of gene mutations in bacteria, unscheduled DNA synthesis in a variety of
test systems, and formation of DNA adducts. DNA adducts of GA have been observed in studies
of in vivo AA exposure of rodents and in vitro AA exposure of human cells. Demonstration of
AA tumor-initiation activity by several routes of exposure in mice provides additional support
for a mutagenic MOA. Furthermore, the multiple-site characteristic of AA carcinogenicity in
rats is consistent with other carcinogenic agents that are thought to act through mutagenic MO As
involving DNA alkylation. An alternative MOA of disruption of hormone levels or activity has
been proposed for some of the tumors observed in animal studies, but the data supporting such a
MOA are limited or lacking.
4.8.2. Synthesis of Human, Animal, and Other Supporting Evidence
Cohort mortality studies of AA workers at several locations in the United States and the
Netherlands (Marsh et al., 2007, 1999; Collins et al., 1989) and a location in Michigan (Swaen et
al., 2007; Sobel et al., 1986) have not found statistically significant increased risks for cancer-
related deaths compared with national cancer mortality rates in whole-cohort analyses.
Numerous case-control studies (Wilson et al., 2009a; Pelucchi et al., 2007, 2006; Michels
et al., 2006; Mucci et al., 2005, 2004, 2003) and prospective studies (Larsson et al., 2009a, b, c,
2008; Wilson et al., 2009b; Hogervorst et al., 2008b; Mucci et al., 2006) have found no
statistically significant associations between increased levels of AA in the diet and increased risk
for a variety of cancer types, including large bowel, bladder, kidney, renal cell, breast, colorectal,
oral, pharyngeal, esophageal, laryngeal, ovarian, or prostate cancers. These studies
predominantly evaluated Swedish, Danish, Dutch, or Italian populations; available assessment of
a U.S. population is restricted to the prospective study of Wilson et al. (2009b). Some of the
tumor sites observed in animal studies (thyroid, testis, CNS) have also not been evaluated, and
there are limitations in some of the study methods and cohort sizes. One case-control study
reported a slightly increased risk of breast cancer later in life associated with the consumption of
French fries during preschool (Michels et al., 2006), but there is considerable uncertainty in the
accuracy of the results from a recall questionnaire administered to mothers for diets in their
preschool children from an estimated 40-60 years earlier and no information on the AA content
of the foods in the diet. Increased risks of postmenopausal endometrial and ovarian cancer
(Hogervorst et al., 2007) and renal cell cancer (Hogervorst et al., 2008a) with increasing dietary
AA intake were reported in prospective studies of a Dutch population, but estimations of dietary
AA levels in foods on the market at baseline in 1986 were based on food samples analyzed since
2001 and questionnaires did not include details regarding specifics of food preparation. Olesen
et al. (2008) reported a significant positive association between AA-Hb adduct levels in red
blood cells and ER+ breast after adjusting for smoking, but this study is limited by the relatively
169 DRAFT-DO NOT CITE OR QUOTE
-------
small number of subjects (374 cases and 374 controls) and uncertainty regarding extrapolation of
AA exposure as assessed by a few months of AA-Hb adduct measurements to a lifetime of
exposure.
In an exploratory exposure-response analysis in which U.S. workers in one of the cohorts
were grouped into exposure categories, an increased risk for pancreatic cancer was calculated for
the group with the highest cumulative AA exposure category (>0.30 mg/m3-years: SMR 2.26,
95% CI 1.03-4.29, based on nine pancreatic cancer deaths) (Marsh et al., 1999). The risk for
pancreatic cancer in the four cumulative exposure categories did not increase monotonically
from the lowest to highest category. A monotonic increase in SMR with another measure of
exposure, duration of employment, was observed, but the SMRs for pancreatic cancer were not
statistically significantly elevated in any of the four duration categories. Furthermore, no
increased risk for pancreatic cancer was observed in the most recent follow-up analysis of this
cohort (Marsh et al., 2007).
Limitations in the epidemiology studies include small cohort size and limited follow-up
period (Swaen et al., 2007; Sobel et al., 1986); large proportion of short-term workers in the
cohort, low exposures, incomplete smoking habit information, and incomplete follow-up period
(Marsh et al., 2007, 1999; Collins et al., 1989); and relatively low dietary exposures, a relatively
short time frame for exposure information (5 years of recalled dietary habits), poor
characterization of AA levels in the food items, variability in levels among different brands, and
few food items in the diet known to have high levels of AA. Although a variety of cancer sites
in humans were evaluated in the case-control and prospective epidemiology studies that reported
no increased risk from dietary exposures (large bowel, kidney, renal cell, bladder, breast, ovary,
prostate, oral/pharyngeal), some of the sites observed in the animal studies have not yet been
evaluated (thyroid, testicular, CNS). The single case-control study (Michels et al., 2006) and
two prospective studies (Hogervorst et al., 2008a, 2007) that showed positive associations from
estimated dietary consumption of AA had questionable data on diet composition and AA content
in the diet.
Cancer studies in test animals include two 2-year drinking water administration studies in
F344 rats (Friedman et al., 1995; Johnson et al., 1986), skin tumor initiation assays involving
oral, i.p., or dermal initiating applications of AA and dermal promotion by TPA in SENCAR and
Swiss-ICR mice (Bull et al., 1984a, b), and a lung adenoma i.p. administration assay in strain A/J
mice (Bull et al., 1984a). The results from the two chronic oral exposure studies in rats are
presented in Table 4-34.
170 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-34. Incidence of tumors with statistically significant increases in
2-year bioassays with F344 rats exposed to acrylamide in drinking water
Reference/tumor type
Johnson et al., 1986; males
Follicular cell adenoma
Tunica vaginalis mesothelioma
Adrenal pheochromocytoma
Johnson et al., 1986; females
Follicular cell adenoma/carcinoma
Mammary adenocarcinoma
Mammary benign
Mammary benign + malignant3
CNS tumors of glial origin
Oral cavity malignant+benign
Uterus adenocarcinoma
Clitoral adenoma, benign
Pituitary gland adenoma
Friedman et al., 1995; males'3
Follicular cell adenoma/carcinoma
Tunica vaginalis mesothelioma0
Friedman et al., 1995; females'3
Follicular cell adenoma/carcinoma
Mammary benign + malignant
Dose (mg/kg-day)
0
1/60
3/60
3/60
1/58
2/60
10/60
12/60
1/60
0/60
1/60
0/2
25/59
3/100
4/102
1/50
7/46
0
—
-
-
-
-
-
_
-
_
_
-
2/102d
4/102
1/50
4/50
0.01
0/58
0/60
7/59
0/59
1/60
11/60
12/60
2/59
3/60
2/60
1/3
30/60
—
-
-
-
0.1
2/59
7/60
7/60
1/59
1/60
9/60
10/60
1/60
2/60
1/60
3/4
32/60
12/203
9/204
-
-
0.5
1/59
ll/60e
5/60
1/58
2/58
19/58
21/58
1/60
3/60
0/59
2/4
27/60
5/101
8/102
-
-
1.0
—
-
-
-
-
-
_
-
_
_
-
—
-
10/100
21/94e
2.0
7/59e
10/60e
10/60 e
5/60f
6/61
23/6 le
29/6 le
9/6 le
8/60e
5/60f
5/5f
32/60f
17/75e
13/75e
-
-
3.0
—
-
-
-
-
-
_
-
_
_
-
—
-
23/100e
30/95e
"Incidences of benign and adenocarcinoma were added herein, based on an assumption that rats assessed with
adenocarcinoma were not also assessed with benign mammary gland tumors.
bTwo control groups were included in the study design to assess variability in background tumor responses.
Incidences reported herein are those originally reported by Friedman et al. (1995) and not those reported in the
reevaluation study by latropoulos et al. (1998).
dThe data reported in Table 4 in Friedman et al. (1995) lists one follicular cell adenoma in the second control group;
however, the raw data obtained in the Tegeris Laboratories (1989) report (and used in the time-to-tumor analysis)
listed no follicular cell adenomas in this group. The corrected number for adenomas (0) and the total number (2) of
combined adenomas and carcinomas in the second control group are used in the tables of this assessment.
Statistically significantly (p < 0.05) different from control, Fisher's Exact test.
Statistically significantly (p < 0.05) different from control, after Mantel-Haenszel mortality adjustment.
Sources: Friedman et al. (1995); Johnson et al. (1986).
Tumor types that were consistently observed to increase in both chronic rat drinking
water bioassays included statistically significant increases in thyroid follicular cell adenomas or
carcinomas in male and female rats, tunica vaginalis testis (i.e., scrotal sac) mesotheliomas in
male rats, and mammary gland tumors (adenomas, fibroadenomas or fibromas) in female rats at
dose levels of 0.5 to 3 mg/kg-day but not at dose levels of 0.1 or 0.01 mg/kg-day (Friedman et
al., 1995; Johnson et al., 1986). Data from both studies are sufficient to describe relationships
between administered dose levels and cancer responses. The Friedman et al. (1995) bioassay
included 204 male rats in the 0.1 mg/kg-day group to increase statistical power sufficient to
171 DRAFT-DO NOT CITE OR QUOTE
-------
detect a 5% incidence of scrotal sac mesotheliomas over an expected background incidence of
this tumor in F344 rats of about 1%.
Findings of statistically significant increased incidences of adrenal pheochromocytomas
in male rats, oral cavity tumors in female rats, CNS tumors of glial origin, and clitoral or uterine
tumors in female rats in the earlier bioassay (Johnson et al., 1986) were not replicated in the
second bioassay (Friedman et al., 1995). With respect to the CNS tumors, Friedman et al. (1995)
reported no significant increase in glial tumors of brain and spinal cord, however, not all of the
animal brains or spinal cords in the treatment groups were examined (Rice, 2005), and seven
cases of a morphologically distinctive category of primary brain tumor described as "malignant
reticulosis" were reported but excluded from the authors' analysis (see Tables 4-13 and 4-14).
The Friedman et al. (1995) study therefore provides some support for AA induced CNS tumors,
even though the incomplete brain and spinal cord tumor data set from this study precludes a
quantitative analysis of CNS tumor incidence in the characterization of the dose-response
analysis. The data for the female uterine adenocarcinomas and pituitary gland adenomas
observed in Johnson et al. (1986) were not as strong as for the other tumor types because the
statistical significance of the elevated incidences in the high-dose group was only demonstrated
after Mantel-Haenszel mortality adjustment, there was no clear evidence for a trend for
increasing risk with increasing exposure level, and in the case of the pituitary gland adenomas,
there were very high control group levels (42% incidence) as well as incidence levels in all of
the dose groups, suggestive of a causal agent(s) other than acrylamide. The increased incidence
of clitoral adenomas is also less pursuasive because it is based on differences in a very small
number of animals (n = <5).
Results from the mouse skin tumor initiation assays add considerable weight to the
evidence for AA carcinogenicity in animals. Oral administration of AA, 6 times over a 2-week
period, followed by dermal application of the tumor promoter, TPA, for 20 weeks, induced
statistically significant increased incidences of histologically confirmed skin tumors (squamous
cell papillomas and carcinomas) at 52 weeks in two mouse strains, SENCAR and Swiss-ICR
(Bull et al., 1984a, b). Similar initiation treatments of the SENCAR strain involving i.p.
injections or dermal applications of AA (followed by TPA promotion) induced statistically
significant increased incidences of palpable skin masses during the course of the 52-week
observation period but were not as effective as oral administration (Bull et al., 1984a). These
findings provide evidence that AA can initiate tumor development in mice, a process that is
thought to involve a mutagenic MOA. These findings are consistent with the positive findings
for AA and GA genotoxicity in numerous tests.
Other evidence of the carcinogenicity of AA in mice is provided by the observations that
statistically significant increased incidences of lung tumors were found in A/J mice 8 months
after i.p. injection of AA 3 times/week for 8 weeks (Bull et al., 1984a) and in Swiss-ICR mice
172 DRAFT-DO NOT CITE OR QUOTE
-------
52 weeks after starting a 2-week oral administration AA initiation protocol followed by dermal
TPA application for 20 weeks (Bull et al., 1984b).
As discussed in Section 4.4.3 and tabulated in Appendix B, AA mutagenicity has been
extensively studied. Although AA did not induce mutations in bacterial assays (with or without
mammalian metabolic activation systems), results from certain other mutagenicity tests have
been predominantly positive and provide supporting evidence for the human carcinogenic
potential of AA. The positive results include demonstrations of chromosomal aberrations in
in vitro exposed mammalian cells (Tsuda et al., 1993; Warr et al., 1990; Moore et al., 1987);
in vitro cell transformation of Syrian hamster embryo cells (Park et al., 2002); chromosomal
aberrations or micronuclei in bone marrow of mice given i.p. injections of 50-100 mg/kg (Cihak
and Vontorkova, 1990, 1988; Adler et al., 1988); formation of DNA adducts of GA following
i.p. injection of 50 mg/kg of AA in mice and rats (Segerback et al., 1995); and dominant lethal
mutations in mice given one to five i.p. injections of 40-125 mg/kg AA (Shelby et al., 1987), in
rats exposed to 2.8 mg/kg-day in drinking water for 80 days (Smith et al., 1986), and in mice
exposed to five consecutive dermal doses of 50-125 mg/kg AA (Gutierrez-Espeleta et al., 1992).
In addition, in vitro exposure to AA induced micronuclei and DNA damage in human hepatoma
G2 cells (Jiang et al., 2007) and DNA adducts of GA in human bronchial epithelial cells
(Besaratinia and Pfeifer, 2004).
In addition, the epoxide metabolite of AA, GA, has been shown to be mutagenic to S.
typhimurium strains TA100 and TA1535 (Hashimoto and Tanii, 1985) and mouse lymphoma
cells (Barfknecht et al., 1988) but not to K. pneumoniae (Voogd et al., 1981). GA induced
unscheduled DNA synthesis in mouse spermatids in vivo (Sega et al., 1990), human epithelial
cells in vitro (Butterworth et al., 1992), in one of two tests for unscheduled DNA synthesis in rat
hepatocytes in vitro (Butterworth et al., 1992; Barfknecht et al., 1988), and in (C3H/RL x
C57BL)F1 male mice given single i.p. injections of 150 mg/kg GA (Generoso et al., 1996). GA
(125 mg/kg by i.p. injection) induced dominant lethal mutations in male JH mice mated with
nonexposed female SB mice, without producing discernible effects on mating performance
(Generoso et al., 1996). GA treatment (100 mg/kg by i.p. injection) of male (C3H x 101/RL)F1
mice mated with nonexposed females induced heritable translocations in male offspring
(Generoso et al., 1996).
4.8.3. Mode of Action for Carcinogenicity
The MOA discussion considers all of the tumor types observed in the animal assays and
the events that might lead to increased incidence in those tumors. The tumor types of interest
include the following: (1) the consistently observed increase in thyroid follicular cell adenomas
or carcinomas in male and female rats, tunica vaginalis testis (i.e., scrotal sac) mesotheliomas in
male rats, and mammary gland tumors (adenomas, fibroadenomas or fibromas) in female rats
173 DRAFT-DO NOT CITE OR QUOTE
-------
following chronic oral exposure (Friedman et al., 1995; Johnson et al., 1986); (2) the CNS
tumors reported in the Johnson et al. (1986) study, supported by the brain tumor data in
Friedman et al. (1995), although an incomplete analysis of all of the animals in the latter study
precluded the inclusion of brain tumors in the quantitative dose-response analysis; (3) the
initiated skin tumors following oral, i.p., or dermal exposure to AA in SENCAR and Swiss-ICR
mice (Bull et al., 1984a, b); and (4) the lung adenomas following i.p. doses of AA in A/J mice
(Bulletal., 1984a).
At present, the mechanistic sequence of events by which AA induces these tumor types is
not completely defined. The majority of the data, however, support a mutagenic MO A for AA
carcinogenicity. An alternative MOA has been proposed for some of the tumors observed in the
animal bioassays (i.e., disruption of hormone levels or activity), but data supporting this MOA
are limited or lacking.
4.8.3.1. Hypothesized Mode of Action—Mutagenicity
A number of study results support a mutagenic MOA for AA-induced carcinogenicity
(including Besaratinia and Pfeifer, 2007; Besaratinia and Pfeifer, 2005; Schmid et al., 1999;
Dearfield et al., 1995; Segerback et al., 1995; Moore et al., 1987). AA has been reported to
induce genotoxicity (gene mutations and some types of chromosomal aberrations [i.e.,
translocations]) in somatic and germ cells of rodents in vivo and cultured cells in vitro, to
transform cells of mouse cell lines, and to form DNA adducts in somatic cells. The mutagenic
potential of GA is well-characterized in studies of the induction of gene mutations in mammalian
cells, and in the formation of DNA adducts. The available data indicate that the major genotoxic
effects of AA are clastogenic, which may involve covalent modifications of proteins by AA and
GA, and that the mutagenic events that lead to tumors from exposure to AA are produced by GA
via direct alkylation of DNA.
Specifically, evidence in support of a mutagenic MOA for carcinogenicity includes the
following:
• AA is metabolized by CYP2E1 to the DNA-reactive epoxide, GA;
• AA and GA are genotoxic in the Big Blue mouse following oral exposures, significantly
increasing lymphocyte Hprt and liver ell mutation frequencies (MFs). Molecular
analysis of the mutants indicated that AA and GA produced similar mutation spectra that
were significantly different from controls consistent with AA exerting its genotoxicity in
BB mice via metabolism to GA. The predominant types of mutations in the liver ell gene
from AA- and GA-treated mice were G:C ->T:A transversions and -1/+1 frameshifts in a
homopolymeric run of Gs.
174 DRAFT-DO NOT CITE OR QUOTE
-------
• DNA adducts of GA have been detected in mice and rats exposed to AA and GA in all
relevant tissues in both males and females where tumors have been reported, including
liver, brain, thyroid, leukocytes, mammary gland, and testis in rats; and liver, lung,
kidney, leukocytes, and testis in mice.
• GA is mutagenic in short-term bacterial assays.
• GA is mutagenic in male and female mouse somatic cells following oral exposure and in
male mouse germ cells (heritable translocations) following i.p. exposure.
• AA induces heritable translocations in male mouse germ cells following i.p. or dermal
administration, and specific locus mutations in male germ cells following i.p.
administration.
• Positive mouse lymphoma assay results (with the caveat that it is not definitively known
whether these somatic cell mutations resulted from AA-induced chromosomal alterations
[chromatid and chromosome breaks and rearrangements] or GA-DNA adducts).
• Dominant lethal mutations have been demonstrated in rodents following subchronic oral
exposure at AA dose levels in the 2.8-13.3 mg/kg-day range, which is near the range of
chronic dose levels associated with carcinogenic effects in rats (0.5-3 mg/kg-day).
Description and identification of key events
The proposed sequence of events for a mutagenic MOA for AA is as follows:
(1) AA is metabolized to the relatively long-lived epoxide, GA, in rats and humans, and
GA reacts both with proteins and with DNA;
(2) GA binding to DNA results in mutations that persist in viable somatic cells; and
(3) GA's mutagenic activity leads to carcinogenicity and the formation of tumors
observed in the animal bioassays.
It is not known whether alterations in protein function due to the formation of both parent
compound- and reactive metabolite-protein adducts have an effect on cell replication or
proliferation or both. The primary mutagenic activity of AA, however, is proposed to result
from the direct binding of the GA metabolite to DNA. In vitro studies indicate that direct
binding of AA to DNA is slow.
Strength, consistency, and specificity of the association between exposure to AA and mutagenic
activity that could lead to the formation of tumors
There is ample evidence in the literature for the ability of AA and GA (administered via
different routes of exposure) to induce a variety of genotoxic effects in mammalian cells
(Besaratinia and Pfeifer, 2007; Rice, 2005; Doerge et al., 2005a; Ghanayem et al., 2005a;
Gamboa et al., 2003; Generoso et al., 1996; Dearfield et al., 1995; Segerback et al., 1995; Adler
et al., 1994; Ehling and Neuhauser-Klaus, 1992; Russell et al., 1991; Knaap et al., 1988; Moore
etal., 1987).
175 DRAFT-DO NOT CITE OR QUOTE
-------
Some genotoxic endpoints and cell assays may be considered to be less relevant to
carcinogenic potential than others. For example, genotoxicity results in germ cells are less
relevant than toxicity in somatic cells where tumors are formed. Further, some effects on germ
cells that appear to be transmitted via genetic alterations may be due to alternative causes.
Dominant lethals in males, for example, may be due not only to genotoxic events in the sperm
but alternatively to nongenetic interactions with proteins critical to the formation and function of
the sperm. Other genotoxic phenomena, such as chromosome breaks, are not heritable. Also,
alterations in chromosome numbers (aneuploidy) are usually due to protein effects and do not
involve a mutagenic MOA. Epidemiology studies that evaluated the association between
increased cytogenetic damage and enhanced cancer risk report no significant association
between the sister chromatid exchange or micronuclei frequencies and subsequent cancer
incidence or mortality (Hagmar et al., 1998; Bonassi et al., 2004). Other measures, such as
unscheduled DNA synthesis may be attributable to either DNA damage or general cytotoxicity
and, therefore, may not be directly attributable to mutagenicity.
The strongest direct evidence to supporting a mutagenic MOA for AA's carcinogenic
effects consists of positive findings of stable mutations in viable somatic cells. Such evidence,
and support that GA is the predominant mutagenic agent following exposure to AA, includes the
following:
1) significant increases in somatic cell mutations following in vivo oral exposures of the
Big Blue mouse to either AA and GA, and similar mutagenicity spectra between AA and GA
(Manjanatha et al., 2006);
2) formation of GA-DNA adducts at similar specific locations within the ell gene in Big
Blue mouse embryonic fibroblasts (that carry a lambda phage ell transgene) and the tumor
suppressor p53 gene (TP53) in normal human bronchial epithelial cells following exposure to
AA or GA in vitro (Besaratinia and Pfeifer, 2004);
3) detection of DNA adducts of GA in various mouse and rat tissues following single i.p.
administration of AA or GA (Doerge et al., 2005a; Segerback et al., 1995);
4) demonstration that AA-induced germ and somatic cell mutations in male mice require
CYP2E1-mediated epoxidation of AA (Ghanayem et al., 2005a, b);
5) positive results for GA in S. typhimurium strains TA100 and TA1535 (Hashimoto and
Tanii, 1985);
6) detection of heritable translocations in mice following single i.p. injections of GA
doses of 100-150 mg/kg (Generoso et al., 1996); and
7) positive results for gene mutation in mouse lymphoma cells in vitro at concentrations
as low as 0.3 mg/mL (Barfknecht et al., 1988; Knaap et al., 1988; Moore et al., 1987).
176 DRAFT-DO NOT CITE OR QUOTE
-------
The results of Manjanatha et al. (2006) studies on significantly increased in vivo
mutation frequencies in the Big Blue (BB) mouse following oral exposure to AA and GA are
consistent with AA's ability to induce heritable mutations in mammalian cells. Average daily
AA exposure from drinking water at the low dose of 100 mg/L (4-week exposure) was 19
mg/kg-day for male and 25 mg/kg-day for female BB mice; the high dose of 500 mg/L (3 weeks
only due to clinical signs of neurotoxicity) yielded average daily exposures of 98 mg/kg-day for
males and 107 mg/kg-day for females. GA exposures were 25 and 35 mg/kg-day for males and
females, respectively, administered the low dose of 120 mg/L (4 weeks), and 88 and 111 mg/kg-
day administered the high dose of 600 mg/L (4 weeks). Both doses of AA and GA produced
significantly increased lymphocyte Hprt mutant frequencies, with the high doses producing
responses that were 16-25-fold higher than those of the respective control. The high doses of
AA and GA also produced significant 2-2.5-fold increases in liver ell MFs. Molecular analysis
of the mutants indicated that AA and GA produced similar mutation spectra that were
significantly different from controls consistent with AA exerting its genotoxicity in the BB mice
via metabolism to GA. The predominant types of mutations in the liver ell gene from AA and
GA-treated mice were G:C ->T:A transversions and -1/+1 frameshifts in a homopolymeric run of
Gs.
AA and GA react with nucleophilic sites in macromolecules (including hemoglobin and
DNA) in Michael-type additions (Segerback et al., 1995; Bergmark et al., 1993, 1991; Solomon
et al., 1985). Solomon et al. (1985) conducted in vitro studies for the reaction of AA with calf
thymus DNA and with various deoxynucleosides including 2'-deoxyadenosine (dAdo),
2'-deoxycytidine (dCyd), 2'-deoxyguanosine (dGua), and 2'-deoxythymidine (dThd), and
demonstrated the formation of 2-formamidoethyl and 2-carboxyethyl adducts via Michael
addition. AA reacted extremely weakly with both the nucleosides and calf thymus DNA, even
under in vitro conditions, producing only small quantities of adducts only after incubations of
40 days even at high AA concentrations.
Segerback et al. (1995) reported much higher rates of DNA-adduct formation from
AA-generated GA than from AA itself. In analyzing either calf thymus DNA incubated with S-9
fraction in vitro or liver DNA from mice treated in vivo with radiolabeled AA, approximately
90% of the radioactivity released during hydrolysis co-chromatographed with a standard
synthesized from the reaction of GA and deoxyguanosine, N-7-(2-carbamoyl-2-hydroxyethyl)
guanine. The amount of this adduct formed in vivo was measured in a number of organs from
both rats and mice administered 46-53 mg AA/kg i.p., and was found to be in the range of 5-
62 pmol/mg DNA. The amount of guanine adduct that would have been formed solely from AA
at this dose was estimated to be much less, in the low fmol range, which would be negligible
compared with the observed levels.
177 DRAFT-DO NOT CITE OR QUOTE
-------
Besaratinia and Pfeifer (2004) treated normal human bronchial epithelial cells and Big
Blue mouse embryonic fibroblasts (that carry a lambda phage ell transgene) in vitro with AA, its
primary epoxide metabolite GA, or water (control) and then subjected the cells to terminal
transferase-dependent polymerase chain reaction to map the formation of DNA adducts within
the human gene encoding the tumor suppressor p53 gene (TP53) and the mouse embryonic
fibroblast ell transgene. AA and GA formed DNA adducts at similar specific locations within
TP53 and ell, and DNA adduct formation was more pronounced after GA treatment than after
AA treatment at all doses tested. AA-DNA adduct formation was saturable, whereas the
formation of most GA-DNA adducts was dose-dependent for all doses tested. GA formed more
adducts than AA at any given dose, and the spectrum of GA-induced ell mutations was
statistically significantly different from the spectrum of spontaneously occurring mutations in the
control-treated cells (p = 0.038). Compared with spontaneous mutations in control cells, cells
treated with GA or AA had more A—>G transitions and G—>C transversions and GA-treated
cells had more G—>T transversions (p < 0.001). These results support the hypothesis that the
mutagenicity of AA in human and mouse cells is based on the capacity of its epoxide metabolite
GA to form DNA adducts.
Doerge et al. (2005a) confirmed that GA-derived DNA adducts of adenine and guanine
were formed in all tissues examined from either AA or GA dosing, including target tissues
identified in rodent carcinogenicity bioassays and nontarget tissues including liver and
leukocytes in rats and liver, lung, kidney, leukocytes and testis in mice, indicating wide-spread
occurrence. They measured DNA adducts following a single i.p. administration of either AA or
GA to adult B6C3Fi mice and F344 rats at 50 mg/kg AA or an equimolar dose of GA (61
mg/kg). Kinetics of DNA adduct formation and accumulation were also measured following oral
administration of a single dose of AA (50 mg/kg) or from repeat dosing (1 mg/kg-day for up to
50 days). The formation of the DNA adducts was consistent with previously reported
mutagenicity of AA and GA in vitro involving reactions of GA with adenine and guanine bases.
Repeated dosing of rats and mice with AA administered in the drinking water resulted in
production of steady state serum levels of GA, and in accumulation of N7-GA-guanine adducts
in liver. Steady state levels of N7-GA-Gua were attained in approximately 14 days with a
formation half-life of about 4 days in male and female mice, and in female rats. Male rats
reached a maximum level at 14 days, but subsequently had an apparent slow decline in adduct
level. The findings indicate that DNA damage from exposure to AA can accumulate to a level
that is dependent on the frequency of consumption, the amount consumed, and depurination rate.
Ghanayem et al. (2005a) compared germ-cell mutagenicity in male CYP2El-null and
wild-type mice treated with AA, and provided the first unequivocal demonstration that
AA-induced germ cell mutations in male mice required CYP2E1-mediated epoxidation of AA to
GA. CYP2El-null and wild-type male mice were treated by i.p. injection with 0, 12.5, 25, or
178 DRAFT-DO NOT CITE OR QUOTE
-------
50 mg AA/kg bw in 5 mL saline/kg-day for 5 consecutive days. At defined times after exposure,
males were mated to untreated B6C3Fi females. Females were killed in late gestation, and
uterine contents were examined. Dose-related increases in resorptions (chromosomally aberrant
embryos), and decreases both in the numbers of pregnant females and the proportion of living
fetuses were seen in females mated to AA-treated wild-type mice. No changes in any fertility
parameters were seen in females mated to AA-treated CYP2El-null mice. Of importance to the
argument that GA is the putative mutagen in AA's mutagenic MO A, a further study by
Ghanayem et al. (2005b) demonstrated the absence of AA-induced genotoxicity in somatic cells
in CYP2El-null mice compared with wild-type mice treated with AA.
Generoso et al. (1996) had previously evaluated AA's ability to induce dominant lethal
mutations and heritable translocations in male mice spermatids, and demonstrated that GA
produced responses that were consistent with the proposal that in vivo conversion to GA is
responsible for the observed mutagenicity (e.g., heritable translocations) of AA in male mice.
Positive results for gene mutation were also observed in mouse lymphoma cells in vitro with
concentrations of AA as low as 0.3 mg/mL (Barfknecht et al., 1988; Knaap et al., 1988; Moore et
al., 1987). Moore et al. (1987) evaluated activity of AA without exogenous activation in
L5178Y/TK+/- -3.7.2C mouse lymphoma cells at the thymidine kinase locus, and noted AA
induced almost exclusively small-colony mutants, indicating clastogenic activity, including
chromatid and chromosome breaks and rearrangements. Thus, the positive results in these
assays, although relevant for heritable mutations cannot be definitively attributable to GA related
DNA mutations or AA related chromosomal alterations.
AA and 15 of its analogues have been tested for mutagenicity in five TA strains of
S. typhimurium (Hashimoto and Tanii, 1985). AA and most of its analogues were not mutagenic,
neither in the standard Ames assay either with or without Aroclor 1254-induced S9 liver
fraction, nor in the plate incubation or liquid preincubation procedures. However, three of the
epoxides including GA (the other two were N,N-diglycidyl AA and glycidyl methacrylamide)
were mutagenic in one or two strains both with and without the S9 fraction.
Overall, the available in vivo mutagenicity data indicate that AA, via conversion to its
active epoxide metabolite, GA, can form DNA adducts, point mutations, and frameshift
mutations that persist in viable mammalian (including human) somatic cells.
Mutations occur in target tissues where tumors have been observed
Doerge et al. (2005a) provide the strongest evidence that AA-induced mutagenicity (via
GA) can be associated with the target tissues where tumors are observed in the animal bioassays.
They report that GA-derived DNA adducts of adenine and guanine were formed in all target
tissues identified in rodent carcinogenicity bioassays as well as a number of nontarget tissues
179 DRAFT-DO NOT CITE OR QUOTE
-------
including liver, brain, thyroid, leukocytes, mammary gland, and testis in rats; and liver, lung,
kidney, leukocytes and testis in mice.
There is little information to causally associate the events between GA-DNA adduct
formation, the occurrence of a stable mutation, and the development of a tumor. It is also not
known why some tissues are more prone to tumor formation than others with similar levels of
GA-DNA adducts. Other tissue-specific events may be occurring. Klaunig and Kamendulis
(2005) reported the effects of AA reactivity with DNA and altered cell growth in the target
tissues identified in the chronic oral bioassays. DNA synthesis was examined in F344 rats
treated with AA at 0, 2, or 15 mg/kg-day for 7, 14, or 28 days. AA increased DNA synthesis in
the target tissues (thyroid, testicular mesothelium, adrenal medulla) at all doses and time points
examined. In contrast, in a nontarget tissue (liver), no increase in DNA synthesis was seen.
Examination of DNA damage using single cell gel electrophoresis (the Comet assay) showed an
increase in DNA damage in the target tissues but not in nontarget tissue (liver). In addition, a
cellular transformation model, the Syrian hamster embryo (SHE) cell morphological
transformation model, was used to examine potential mechanisms for the observed
carcinogenicity of AA. SHE cell studies showed that GSH modulation by AA was important in
the cell transformation process. Treatment with a sulfhydryl donor compound (N-acetyl
cysteine) reduced AA transformation, while depletion of GSH (buthionine sulfoximine) resulted
in an enhancement of transformation. AA was thus shown to increase both DNA synthesis and
DNA damage in mammalian tissues and cells, suggesting that DNA reactivity and cell
proliferation, in concert, may contribute to the observed AA-induced carcinogenicity in the rat
target tissues.
Dose-response concordance and temporal relationship
Empirical support for dose-response and temporal concordance between AA-induced
genotoxic events and tumor development comes from studies of DNA adduct formation in liver
of rodents following repeated oral exposure to doses at or below the dose range of the chronic rat
bioassays (0.5-3 mg/kg-day). N7-GA-guanine adducts increased to apparent steady-state levels
in livers of male and female F344 rats by 14 days of repeat dosing of approximately 1 mg
AA/kg-day in drinking water (Doerge et al., 2005a). A similar temporal pattern for increased
liver levels of N7-GA-guanine adducts was reported for male and female B6C3Fi mice exposed
to 1 mg/kg-day doses in drinking water for up to 40 days (Doerge et al., 2005a). In another
study, N7-GA-guanine adduct levels in livers increased with increasing dose level in mice
exposed for 28 days to gavage doses of AA ranging from 0.125 to 24.0 mg/kg-day; some
evidence for saturation at the higher dose levels was evident (Zeiger et al., 2009). In this study,
significantly (p < 0.05) increased micronuclei were observed in peripheral blood reticulocytes at
AA doses >6 mg/kg-day and in normochromatic erythrocytes at >4 mg/kg-day (Zeiger et al.,
180 DRAFT-DO NOT CITE OR QUOTE
-------
2009). These studies, which examined levels of DNA adducts at doses in the range of the
chronic bioassays, did not examine cancer target tissues (e.g., CNS, thyroid), but, as discussed
earlier, single i.p. doses of 50 mg/kg AA increased N7-GA-guanine adducts in the brain, thyroid,
testes, and mammary gland tissue of F344 rats (Doerge et al., 2005a).
There are also some data from mouse skin tumor initiation bioassays and several in vivo
genotoxicity assays (including dominant lethal mutation assays) that provide evidence of
mutagenicity from AA exposure in the range of 3 to 50 mg/kg-day.
AA's ability to initiate mouse skin tumors has been demonstrated at oral dose levels as
low as 12.5 mg/kg-day (Bull et al., 1984a, b). Oral administration of AA (3 times/week for
2 weeks, followed by dermal application of the cancer promoter, TPA) caused statistically
significant increased incidences of skin-tumor-bearing SENCAR mice at 12.5, 25, or 50 mg/kg-
day dose levels and statistically significant increased incidences of histologically confirmed skin
adenomas or carcinomas at 25 or 50 mg/kg-day (Bull et al., 1984a). In this study, oral
administration was more effective at initiating skin tumors than i.p. injection or dermal
application at equivalent dose levels. In Swiss-ICR mice, a similar initiation-promotion protocol
caused statistically significantly increased incidences of the same endpoints at oral doses of
50 mg/kg-day but not at 12.5 or 25 mg/kg-day (Bull et al., 1984b). The power to detect
statistically significant changes in these studies, however, is limited by the number of animals in
each exposure group (n = 40). For example, in the Swiss-ICR study, statistical significance
could not be demonstrated for the difference between the control incidence (0/40) and the
incidences of skin-tumor bearing animals in the 12.5 mg/kg-day (4/40) and 25 mg/kg-day groups
(4/40). Thus, the available data give some indication that AA tumor initiation activity increases
with increasing dose level, but these data are inadequate to determine whether oral dose levels of
0.5-3 mg/kg-day would also initiate mouse skin tumors.
Dominant lethal mutations following repeated exposure to AA in drinking water (e.g.,
implantation losses or decreased fetuses/dam) have been observed in male F344 rats exposed for
at least 12 weeks to 5 mg/kg-day, but not to 2 mg/kg-day (Tyl et al., 2000a); male Swiss CD-I
mice exposed for at least 15 weeks to 7.5 mg/kg-day, but not to 3.1 mg/kg-day (Chapin et al.,
1995); male Long-Evans rats exposed for 72 days to 2.8 mg/kg-day, but not to 1.5 mg/kg-day
(Smith et al., 1986); and male ddY mice exposed for 4 weeks to 13.3 mg/kg-day, but not to
9.0 mg/kg-day (Sakamoto and Hashimoto, 1986). There is currently insufficient information,
however, to determine if the events leading to the dominant lethals are relevant or not to a
mutagenic MOA.
Studies designed to examine in vivo clastogenic effects in mammals from subchronic or
chronic exposures at lower doses are limited to the reports of no chromosomal aberrations in
spermatogonia or spermatocytes in male Long-Evans rats exposed for 72 days to drinking water
doses between 1.5 and 5.8 mg/kg-day (Smith et al., 1986) and the dominant lethal effects
181 DRAFT-DO NOT CITE OR QUOTE
-------
described above with subchronic exposure to doses in the range of 2.8 to 7.5 mg/kg-day in
several studies (Tyl et al., 2000a; Chapin et al., 1995; Smith et al., 1986). These results,
however, indicate only that genotoxic effects on male germ cells can occur following subchronic
duration oral exposure to dose levels in the vicinity of the chronic dose levels that induced
carcinogenic effects in rats, and again it is uncertain whether or not the events are these results
are relevant to a mutagenic MOA for AA.
Allen et al (2005) attempted dose-response modeling of AA in vivo genotoxicity data to
extrapolate the response for chromosomal aberrations or sister chromatid exchange from the
relatively high administered doses in these assays (50-150 mg/kg) to the 2 mg/kg-day dose used
in the chronic oral bioassays that significantly increased thyroid tumors in F344 rats. The intent
of this approach was to move the analysis of genotoxicity assay results from qualitative
conclusions of "negative or positive" results (as listed in the table in Appendix B) to more useful
quantitative characterizations of the dose response that support or refute dose-response
concordance between mutagenic events and increased tumorigenicity. In their analysis of the
AA data (based on a variety of dose-response modeling approaches and a benchmark response
level of 10% for occurrence of chromosomal damage), the authors report that a 2 mg/kg-day
dose would result in levels indistinguishable from background (i.e., zero exposure), suggesting
little concordance between these studies and the observed tumorigenicity in rats. The analysis,
however, has a number of serious (if not fatal) flaws and assumptions, including some addressed
by the authors (e.g., comparing short-term, high-dose effects with long-term, low-dose effects,
comparing results in mice with results in rats, assuming low-dose response relationships based
on extrapolations from very high doses, and limited sample sizes), as well as others not well
addressed, including the assumption that chromosomal damage is the primary mutagenic event
(rather than DNA adducts or other DNA damage), not evaluating mutagenic events in target
tissues (i.e., not considering the toxicokinetics of AA) or at different life stages (not considering
the toxicodynamics of AA), and that very small increments above background are not important
(i.e., disregarding the one hit, one tumor hypothesis), or, alternately, that it is acceptable to apply
a benchmark response of 10% to mutagenic events assumed to lead to tumor formation when the
generally accepted "minimal" risk level for carcinogenicity is 0.0001% (i.e., one in a million, not
one in ten). Nonetheless, attempts to quantitate mutagenic dose response is clearly in the right
direction, and warrants further support and research.
In summary, the Doerge et al. (2005a, b) data demonstrate formation of GA-DNA
adducts in tissues throughout the body as a result of the rapid and wide distribution of AA and
GA from any route of exposure (i.e., a high volume of distribution). Additional indicators of
potential mutagenicity discussed above that occur within hours or days of treatment support
these events as precursor events to the formation of tumors, although the administered doses
were much higher than those given to the test animals in the chronic bioassays.
182 DRAFT-DO NOT CITE OR QUOTE
-------
Biological plausibility and coherence
The multiple-site characteristic of AA carcinogenicity in rats (i.e., CNS, thyroid, scrotal
sac, mammary gland) is consistent with other carcinogenic agents that are thought to act through
mutagenic MO As involving alkylation of DNA. For example, inhalation exposure to ethylene
oxide, a DNA reactive epoxide, induced lung, Harderian gland, uterine, mammary, and lymph
tumors in mice and leukemia, brain tumors, peritoneal mesotheliomas in the region of the testes,
and subcutaneous fibromas in rats (IARC, 2007a). Similarly, inhalation or oral exposure of rats
to acrylonitrile, which is metabolized to a DNA reactive epoxide, induced cancer in a range of
tissues, including CNS, Harderian gland, gastrointestinal tract, and mammary glands in rats
(IARC, 1999). Inhalation exposure of rats to 1,3-butadiene, which is metabolized to DNA-
reactive epoxides, induced tumors of the pancreas and testes in males and tumors of the thyroid,
uterus, Zymbal gland, and mammary glands in females (IARC, 2007b). In addition, oral
exposure to N-methylolacrylamide, which is metabolized to AA and GA (Fennell et al., 2003),
induced tumors in the liver, lung, and Harderian gland in mice (NTP, 1989). Thus, it is
biologically plausible that the formation of GA-DNA adducts is a key event in the
carcinogenicity of AA.
The fact that GA-DNA adducts have been detected in nontarget organs underscores the
importance of not assuming that adducts by themselves are sufficient to produce tumors. Only
certain DNA adducts lead to perturbed gene structure and function. Although biologically
plausible and coherent with other cancer agents acting through DNA-reactive epoxides, key
events that have not been firmly established for a mutagenic MOA for AA include AA-induced
DNA adducts in target tissues at tumor-inducing exposure levels and AA-induced DNA adducts
in cancer-critical genes in target tissues (e.g., proto-oncogenes/tumor-suppressor genes). Thus,
further research is needed to better assess dose-response and temporal concordance between
AA-induced DNA adduct formation, mutations in cancer-critical genes, and tumor responses.
Results from investigations of AA effects on hypothalamus-pituitary-thyroid endpoints in
rats provide no clear and consistent evidence to support an alternative cancer MOA involving
hormonal dysregulation by AA or its metabolites (see Section 4.8.3.2 for further discussion).
These results, especially the negative results from an examination of a comprehensive suite of
hypothalamus-pituitary-thyroid endpoints in rats exposed to 50 mg/kg-day for 14 days (Bowyer
et al., 2008; see Section 4.8.3.2), are consistent with a mutagenic MOA for AA.
Human relevance
A mutagenic MOA involving the key events of AA metabolic activation to GA, and GA
modification of DNA leading to mutation of cancer-critical genes is reasonably expected to be
relevant to humans. Observations of GA-Hb adducts (Vesper et al., 2008, 2007; Bjellaas et al.,
183 DRAFT-DO NOT CITE OR QUOTE
-------
2007a, b; Chevolleau et al., 2007) and GA urinary metabolites (Heudorf et al., 2009; Urban et
al., 2006) demonstrate that internal exposure to GA, the mutagenic AA metabolite, occurs in the
general population experiencing low levels of AA exposure. In addition, human cells exposed to
AA or GA have been shown to develop mutations in a cancer-critical gene. Besaratinia and
Pfeifer (2004) exposed normal human bronchial epithelial cells for 4 hours to AA or GA and
detected GA-DNA adducts within the tumor suppressor p53 gene, using terminal transferase-
dependent polymerase reaction to map the formation of DNA adducts within the gene.
Conclusion
There is evidence from a variety of studies of GA's mutagenicity in mammalian
(including human) somatic cells that supports a mutagenic MOA for AA that would be
operational in both test animals and humans. The mutagenicity of AA is indicated through its
ability to induce gene mutations and chromosomal aberrations in somatic and germ cells of
rodents in vivo and cultured cells in vitro and cell transformation in mouse cell lines, and its
ability to form adducts with protamines in germ cells. The mutagenicity of GA is characterized
by its induction of gene mutations in bacteria, unscheduled DNA synthesis in a variety of test
systems, and ability to form DNA adducts. The available data indicate that the major mutagenic
effects of AA are clastogenic, which may involve covalent modifications of proteins by AA and
GA, and direct alkylation of DNA by GA (Doerge et al., 2005a; Besaratinia and Pfeifer, 2004;
Schmid et al., 1999; Dearfield et al., 1995; Segerback et al., 1995; Moore et al., 1987). Support
for the genetic damage in somatic and germ cells of mice treated with AA being dependent upon
metabolism of the parent compound to GA by CYP2E1 comes from studies in CYP2El-null
male mice (Ghanayem et al., 2005a, b), and the similar mutation spectra that AA and GA
produced in the Big Blue male and female mice (Manjanatha et al., 2006).
There is some support for the temporal sequence in that mutagenic events (e.g., GA-DNA
adducts) have been observed in target tissues, and these occur soon after exposure to AA,
although most of these studies are at doses of AA higher than those of the bioassays. Additional
data are needed to further demonstrate the temporal sequence of events between the formation of
DNA adducts, the development of mutations, and the formation of tumors; and to establish dose-
response concordance to firmly establish that a GA-DNA adduct is an obligate precursor event in
tumor formation. Additional data are also needed to resolve why only hormonally responsive
tissues were observed to have increased tumors in the Friedman et al. (1995) chronic rat
bioassay, whereas GA-DNA adducts have been observed in a much wider array of tissues.
4.8.3.2. Alternative Mode of Action—Disruption of Hormone Levels or Signaling
An alternative MOA via disruption of hormone levels or hormone signaling has also been
suggested for the AA-induced tumors in hormonally sensitive tissues (mammary gland and
184 DRAFT-DO NOT CITE OR QUOTE
-------
thyroid) or in a tissue adjacent to hormonally sensitive tissue (tunica vaginalis, the scrotal sac
mesothelium) (Haber et al., 2009; Dourson et al., 2008; Klaunig, 2008; Shipp et al., 2006;
Environ, 2002; KS Crump Group, Inc., 1999a, b). Although this is a possible MO A, at present
there are only limited or absent supporting data.
The hypothesized sequence of events for the induction of tunica vaginalis and mammary
gland tumors is as follows: dopamine agonist activities promote age-related hormonal changes
that, in turn, stimulate sustained cell proliferation in the tunica vaginalis and mammary gland,
leading to progression to mesothelioma and fibroadenoma, respectively. For the thyroid tumors
the events are alteration of a signal transduction pathway, leading to persistent stimulation of cell
proliferation in thyroid follicular cells and eventual progression to follicular cell adenomas
(Dourson et al., 2008; Klaunig, 2008; Shipp et al., 2006; Environ, 2002; KS Crump Group, Inc.,
1999a, b).
In support of the hypothesis for dopamine agonist activity (at the D2 dopamine receptor),
AA has been shown to decrease circulating levels of prolactin in male F344 rats (Friedman et al.,
1999b; Khan et al., 1999; Ali et al., 1983; Uphouse et al., 1982). The relevance of the
carcinogenicity of chemicals that induce Leydig cell tumors in rats via dopamine agonist activity
is an issue of scientific debate, because human Leydig cells (as well as Leydig cells in other
animal species, except male rats) do not decrease their luteinizing hormone (LH) receptors in
response to decreased prolactin. Because of the evidence for dopamine agonist activity of AA in
male rats and evidence to suggest that the malignancy of the tunica vaginalis mesotheliomas in
F344 rats was linked to the extent of Leydig cell neoplasia, it has been proposed that the
mesotheliomas may not be relevant to humans. Additional supporting evidence would include
demonstration of a lack of mesotheliomas in other animal species chronically exposed to AA;
however, these data are not currently available.
In contrast to male rats, there is little empirical evidence to support this alternative MOA
in female rats. Marked changes in circulating levels of prolactin have not been observed in
female F344 rats exposed to AA for up to 28 days. There is also no direct evidence that AA
displays Dl dopamine agonist activity in female rats, which could enhance ovarian progesterone
secretion and subsequently stimulate cell proliferation in the stromal/fibroblast cells of the rat
mammary gland.
With respect to thyroid tumors, there is no clear and consistent evidence to support the
disruption of thyroid hormone homeostasis in acrylamide-exposed rats. Although there is one
published report of changes in thyroid follicular cell colloid area and cell height in female F344
rats exposed to 2 or 15 mg/kg-day for up to 7 days without any changes in circulating levels of
TSH or T4 (Khan et al., 1999), another study found no morphological changes or evidence of
increased cell proliferation in male F344 rats exposed to 50 mg/kg-day for 14 days (Bowyer et
al., 2008). Bowyer et al. (2008) found elevations of serum T4 levels at 50 mg/kg-day, but not at
185 DRAFT-DO NOT CITE OR QUOTE
-------
2.5 or 10 mg/kg-day, and no changes in serum levels of T3 or TSH at any of these dose levels.
Other unpublished studies indicated that AA doses as high as 25 mg/kg-day for up to 28 days did
not induce consistent, biologically significant changes in thyroid hormones or TSH levels
(Klaunig, 2000, as cited in Environ, 2002; Friedman et al., 1999b). Thus, current data do not
support a MOA by which AA alters thyroid hormone homeostasis. Direct evidence that AA may
cause follicular cell proliferation by an alternative MOA involving stimulation of a cAMP
cascade (without changes in TSH levels) is not currently available. TSH-induced mitogenic
activities are mediated largely by cAMP, which in turn may activate protein kinase (PKA)-
dependent and independent processes.
Tunica Vaginalis Mesotheliomas
Description and identification of key events
The events in the proposed hormonal pathway MOA for AA-induced formation of tunica
vaginalis mesotheliomas is as follows: (1) AA increases dopamine levels or functions as a
dopamine receptor agonist; (2) a dopamine agonist-induced decrease in prolactin levels then
down-regulates LH receptors on rat Ley dig cell membranes, leading to decreases in testosterone
production; (3) there is a subsequent compensatory increase in serum LH to maintain
testosterone at normal levels (Cook et al., 1999; Clegg et al., 1997; Prentice and Meikle, 1995);
and (4) the increase in LH stimulates sustained cell proliferation in the tunica vaginalis with
eventual progression to mesotheliomas.
Experimental support for the hormonal pathway MOA in male rats
Strength, consistency, and specificity of association
Serum prolactin levels have been observed to decrease in AA-exposed male rats, but not
females (Friedman et al., 1999b; Khan et al., 1999; Ali et al., 1983; Uphouse et al., 1982). These
studies were instigated because it is well known that dopamine plays a predominant role in
hypothalamic suppression of pituitary secretion of prolactin (Yamada et al., 1995; Neuman,
1991), and AA has been demonstrated to increase striatal dopamine receptors in rats (Agrawal,
1981a, b; Bondy et al., 1981; Uphouse and Russell, 1981). The results suggest that AA, in
inhibiting prolactin secretion by the pituitary, may act as a dopamine agonist, at least in male
rats.
In an unpublished study, male and female F344 rats (approximately 8 weeks of age at
beginning of exposure) were exposed to AA in drinking water providing doses of 0, 4.1, 12, 19,
or 25 mg/kg-day for up to 28 days (Friedman et al., 1999b). Serum prolactin levels in males
were decreased after 14 days of treatment: percentage decreases (compared with controls) were
17, 36, 81, and 87% for the 4.1 through 25 mg/kg-day groups, respectively. The values at the
two highest exposure levels were statistically significantly different from control values.
Percentage decreases in the mean values for the 4.1 through 25 mg/kg-day males at 28 days were
186 DRAFT-DO NOT CITE OR QUOTE
-------
0, 5, 44, and 33%, but none of the mean values were statistically significantly different from
control values at 28 days.
Circulating levels of prolactin in female F344 rats showed no consistent dose-related
changes, compared with controls, after 14 or 28 days of AA exposure (Friedman et al., 1999b,
unpublished) or, in another published study with 28-day-old females, after gavage administration
of 2 or 15 mg/kg-day AA for 2 or 7 days (Khan et al., 1999).
In earlier studies, serum prolactin levels were shown to be decreased in male F344 inbred
rats 24 hours after oral administration of 100 mg/kg AA (Uphouse et al., 1982). The decrease in
prolactin levels was statistically significant in rats that were not handled for 3 minutes/day for
7 days before AA administration but was not significant in rats that received this handling
pretreatment protocol. Serum prolactin levels were also decreased in male F344 rats (8-
10 weeks of age at the start of the study) following 20 daily i.p. injections of 10 or 20 mg/kg AA
(Alietal., 1983).
The available animal studies do not support a consistent AA effect on dopamine levels or
receptors in various brain regions.
AA has been shown to produce changes in the dopaminergic system in some short-term
oral exposures to AA (5, 10, or 20 mg/kg-day, 10 times during 14 days, or single doses of 50,
100, or 200 mg/kg) with increases in dopamine receptors (assayed as increased binding of
[3H]-spiroperidol) in the striatal brain region of young (6-week-old) Sprague-Dawley or F344
male rats (Agrawal, 1981a, b; Bondy et al., 1981; Uphouse and Russell, 1981). In contrast,
24 hours postdosing, rats orally exposed to 10 mg/kg AA for 10 consecutive days had a
decreased response to apomorphine (a dopamine receptor agonist) compared with nonexposed
controls (Bondy et al., 1981). Bondy et al. (1981) noted that similar, apparently paradoxical,
results were also reported for another neurotoxicant, haloperidol. It was proposed that AA might
induce damage to the dopaminergic pathways such that apomorphine would not elicit a response
even in the presence of an excess number of dopamine receptors.
Oral exposure of pregnant F344 rats to 20 mg/kg-day on GDs 7-16 was also reported to
induce decreased dopamine receptors in offspring assayed 2 weeks after birth but not at 3 weeks
(Agrawal and Squibb, 1981). Repeated oral exposure to AA (10 times during 14 days) also
caused an increase in other neurotransmitter receptors: acetylcholine striatal receptors (at 5, 10,
or 20 mg/kg-day), GABA cerebellar receptors (at 20 mg/kg-day), glycine medullar receptors (at
20 mg/kg-day), and serotonin frontal cortical receptors (at 20 mg/kg-day) (Bondy et al., 1981).
The biological and mechanistic significance of these findings of effects of AA on levels of
neurotransmitter receptors remains uncertain.
Exposure to AA also has been reported to cause changes in levels of dopamine in some
regions of the rat brain, but changes have been inconsistently observed across studies (Ali, 1983;
Ali et al., 1983; Rafales et al., 1983; Agrawal et al., 1981a). Mean striatal dopamine
187 DRAFT-DO NOT CITE OR QUOTE
-------
concentrations were higher than control values by about 22-31% in 6-week-old male Sprague-
Dawley rats, 24 hours after administration of single i.p. injections of 50, 100, or 150 mg/kg, but
the difference was not statistically significant (Agrawal et al., 1981a). Male 10-week-old F344
rats given single i.p. injections of 50 or 100 mg/kg AA showed no significant change in levels of
dopamine in the frontal cortex or striatum; in contrast, following 10 consecutive injections of
10 mg/kg-day, levels of dopamine and a metabolite, dihydroxyphenylacetic acid, were
significantly decreased in the frontal cortex but not changed in the striatum or hypothalamus (Ali
etal., 1983).
In another study, 8- to 10-week-old male F344 rats were administered 20 consecutive i.p.
injections of 10 or 20 mg/kg AA, resulting in significantly increased dopamine levels in the
caudate nucleus compared with controls; however, levels of dopamine in the frontal cortex or the
hypothalamus were not significantly affected (Ali, 1983). In male Long-Evans rats exposed to
100 mM AA in drinking water for 6 weeks, there were no changes in concentrations of dopamine
and its metabolites, dihydroxyphenylacetic acid and homovanillic acid in the nucleus
accumbens, septal area, corpus striatum, or thalamus compared with controls (Rafales et al.,
1983).
AA-exposed rats showed increased psychomotor stimulation from amphetamine,
compared with controls, that was associated with short-term elevations of 5-hydroxyindoleacetic
acid in several brain regions and a lesser elevation of dopamine in the nucleus accumbens but not
in the septal area, corpus striatum, or thalamus (Rafales et al., 1983).
Dose-response concordance
Only a few studies are available to support a dose-response relationship of AA on
circulating prolactin levels via an effect on the dopaminergic system in male rats and influence
on circulating levels of hormones. Serum testosterone levels in male F344 rats were statistically
significantly decreased following 28 days of exposure to AA in drinking water at dose levels of
19 and 25 mg/kg-day but not at lower dose levels (Friedman et al., 1999b). For groups exposed
to 0, 1.4, 4.1, 12, 19, or 25 mg/kg-day, respective mean testosterone values (±SD, in units of
ng/mL) were 1.1 ± 0.7, 2.1 ± 1.1, 2.2 ± 1.4, 0.5 ± 0.3, 0.3 ± 0.4, or 0.1 ± 0.1. Decreased serum
levels of testosterone have also been observed in male F344 rats exposed to 20 daily i.p.
injections of 10 or 20 mg/kg AA (Ali et al., 1983).
Temporal relationship
If AA-induced decreases in circulating levels of prolactin actually lead to physical or
hormonal changes in Leydig cell tumors, such changes may subsequently stimulate the
development of spontaneously initiated or AA-initiated mesothelial cells in the scrotal sac (i.e.,
tunica vaginalis) into mesotheliomas. These types of actions have been proposed by Tanigawa
188 DRAFT-DO NOT CITE OR QUOTE
-------
et al. (1987) to explain the higher spontaneous incidences of genital serosal mesotheliomas in
male F344 rats compared with other rat strains, such as Sprague-Dawley, that do not show high
spontaneous incidences of Ley dig cell tumors. Older male F344 rats, surviving between about
80 and 120 weeks, are well documented to display spontaneous Leydig cell tumors at high (80-
100%) incidences, and spontaneous mesotheliomas, predominantly in the genital serosa, at low
(3-4%) incidences (Tanigawa et al., 1987; Solleveld et al., 1984; Goodman et al., 1979). The
male F344 rats in the AA bioassays were not an exception to this occurrence. The appearance of
Leydig cell tumors in aging F344 rats shows a temporal relationship with age-related changes in
the synthesis or secretion of gonadal and adrenohypophyseal hormones (Amador et al., 1985;
Turek and Desjardins, 1979). In addition, persistently elevated levels of prolactin (produced by
transplantation of anterior pituitaries from adult females or by treatment with diethylstilbestrol)
have been shown to inhibit the development of spontaneous Leydig cell tumors in aging male
F344 rats (Bartke et al., 1985).
Biological plausibility and coherence
The mechanism by which AA may increase dopamine receptors or other neurotransmitter
receptors is unknown. One hypothesis that has been proposed involves AA down-regulation of
the microtubular system and disintegration of neurofilaments followed by blockage of
intracellular transport of receptors and their subsequent accumulation (Ho et al., 2002). This
hypothesis was based on observations that exposure of cultured brain neurons from chicken
embryos to 10 mM AA induced increased levels of GABAA receptors, decreased levels of
tubulin proteins, and decreased numbers of microtubules and neurofilaments in the neuron cell
body. Similar experiments examining AA effects on dopamine receptors and associated changes
in tubulin protein levels and numbers of neurofilaments in cultured brain neurons are not
available.
Human relevance
A reevaluation of the most recent of the two AA drinking water cancer bioassays for
tumors in reproductive tissues (latropoulos et al., 1998) in male rats originally assessed as
having tunica vaginalis mesotheliomas (Friedman et al., 1995) provides some support for the
proposal that AA-induced mesotheliomas in F344 rats may not be relevant to humans (Shipp et
al., 2006). In the reevaluation, all rats diagnosed with malignant mesothelioma were assessed as
having 75% or 100% of the testes occupied by Leydig cell neoplasia, whereas rats with
mesothelial hyperplasia or benign mesothelioma were assessed as having 50% or less of the
testes occupied by Leydig cell neoplasia (latropoulos et al., 1998).5 These observations suggest
5 In another study of the tunica vaginalis testis mesotheliomas reported in Friedman et al. (1995), it was
concluded, based on light and electron microscopy, that tumors in the acrylamide-exposed rats did not differ
189 DRAFT-DO NOT CITE OR QUOTE
-------
that the extent of Ley dig cell neoplasia and the development of malignant mesotheliomas in
these rats may have been linked.
Most of the possible mechanisms proposed for the chemical induction of Ley dig cell
hyperplasia and adenomas involve elevation of serum LH and/or a change in Ley dig cell
responsiveness to LH as the key event (Cook et al., 1999; Clegg et al., 1997). Several other
mechanisms involving elevations of LH or other disruptions of the hypothalamic-pituitary-testis
axis could possibly result in an adverse human response (Cook et al., 1999; Clegg et al., 1997).
Conclusion
In summary, there is some evidence to suggest that AA can promote or enhance age-
related decreases in serum prolactin and testosterone in older male F344 rats (Friedman et al.,
1999b; Khan et al., 1999; Ali et al., 1983; Uphouse et al., 1982) and that this enhancement may
lead to the development of tunica vaginalis mesotheliomas due to larger adjacent Ley dig cell
tumors (latropoulos et al., 1998). Because the response to decreased circulating levels of
prolactin in this sequence of events may be specific to male F344 rats (and not occur in humans
or other animal species), AA-induced tunica vaginalis mesotheliomas in older F344 rats may not
be relevant to humans. Additional support for this proposal, such as the lack of mesotheliomas
in other rat strains or other animal species exposed chronically to AA, however, is not available.
In conclusion, a hormone-mediated MOA for the observed mesotheliomas is possible but data
are lacking to link key events with tumor formation.
Mammary Gland Fibroadenomas
Description and identification of key events
The events in the proposed hormonal pathway MOA for AA induction of mammary
gland fibroadenomas in female F344 rats are as follows: an age-related decrease in dopamine,
leading to increased secretion of prolactin by the pituitary, followed by increased and sustained
release of progesterone from the ovary, leading to a sustained cell proliferative response in
stromal/fibroblast cells of the mammary gland and eventual progression to fibroadenomas (Shipp
et al., 2006).
Experimental support for the hormonal pathway MOA in female rats
Strength, consistency, specificity of association
The hypothesis proposes that AA acts as a dopamine agonist on Dl dopamine receptors
in the ovary to further enhance secretion of progesterone in aging rats. Direct in vitro or in vivo
morphologically from tumors in the control rats (Damjanov and Friedman, 1998). This study, however, did not
specifically compare morphological features of Leydig cell tumors between acrylamide-exposed and control rats.
190 DRAFT-DO NOT CITE OR QUOTE
-------
evidence showing that AA interacts with Dl dopamine receptors and subsequently enhances
progesterone secretion in female rats is not currently available.
Dose-response concordance
Circulating levels of prolactin in female F344 rats showed no consistent, dose-related
changes, compared with controls, after 14 or 28 days of AA exposure (Friedman et al., 1999b,
unpublished) or, in another published study with 28-day-old females, after gavage administration
of 2 or 15 mg/kg-day AA for 2 or 7 days (Khan et al., 1999).
Temporal relationship
No in vitro or in vivo evidence were available to support a temporal relationship between
AA interaction with Dl dopamine receptors, subsequent enhanced progesterone secretion in
female rats, and development of mammary tumors.
Biological plausibility and coherence
Although the proposed hormonal pathway MOA for AA-induced mammary
fibroadenomas in female F344 rats is possible, there are no empirical data directly linking AA to
an enhancement of any particular process in the proposed cascade of events (e.g., AA acting as
an agonist for Dl dopamine receptors, leading to enhanced progesterone secretion from rat, but
not human, ovary cells.
Human relevance
It has been proposed (Shipp et al., 2006) that the increased incidences of mammary gland
fibroadenomas in the AA bioassays are not relevant to humans because fibroadenomas in women
are associated with either an increase in estrogen or a decrease in progesterone or both (Smith,
1991) and not an increase in progesterone as in aging female rats; because increased prolactin
does not lead to increased progesterone secretion in humans or other primates (Neumann, 1991);
and because the dopamine agonist, SKF-38393, acting at Dl dopamine receptors in rat ovary
cells, stimulates progesterone secretion (Mori et al., 1994) but does not appear to stimulate
progesterone secretion in human ovary cells (Mayerhofer et al., 1999).
Conclusion
Although empirical support is inadequate or lacking for this proposed MOA, it is a
possible MOA, assuming that AA-induced fibroadenomas in female F344 rats are produced by
AA enhancement of the normal age-related mode of development of spontaneous fibroadenomas.
However, the possible human relevance of AA-induced mammary gland fibroadenomas cannot
be ruled out with confidence at this time, because there is no empirical evidence directly linking
AA to an enhancement of any particular process in the proposed cascade of events (e.g., AA
191 DRAFT-DO NOT CITE OR QUOTE
-------
acting as an agonist for Dl dopamine receptors, leading to enhanced progesterone secretion from
rat, but not human, ovary cells).
Thyroid Tumors
Description and identification of key events
The events in the proposed hormonal pathway MOA for AA-induced formation of
thyroid tumors in male and female F344 rats are alteration of a signal transduction pathway,
leading to persistent stimulation of cell proliferation in thyroid follicular cells and eventual
progression to follicular cell adenomas (Dourson et al., 2008; Environ, 2002; KS Crump Group,
Inc., 1999a, b).
Experimental support for the hormonal pathway MOA in male and female rats
Strength, consistency, specificity of association
Both of the available chronic exposure studies of F344 rats reported statistically
significant increased incidences of thyroid follicular cell adenomas, or combined adenomas and
carcinomas, at the highest dose levels of 2-3 mg/kg-day (Friedman et al., 1995; Johnson et al.,
1986). Chemicals that alter thyroid hormone homeostasis by interfering with synthesis or
secretion of T3 or T4 or by increasing T3 or T4 metabolism can lead to compensatory release of
TSH from the pituitary, which, if sustained, may induce thyroid follicular cell hyperplasia
possibly progressing to neoplasia (U.S. EPA, 1998c).
As discussed in more detail in Section 4.6.1, there is no clear and consistent evidence to
support the hypothesis that AA induces sustained follicular cell proliferation by altering thyroid
hormone homeostasis. Khan et al. (1999) reported statistically significant changes in follicular
cell colloid area and cell height in female F344 rats exposed to 2 or 15 mg/kg-day AA for 2 or
7 days without any significant changes in plasma levels of TSH or T4, but Bowyer et al. (2008)
found no morphological changes or evidence of increased cell proliferation (mRNA levels for
the Mki67gene and protein levels of Ki67) in the thyroid and pituitary of male F344 rats exposed
to 50 mg/kg-day AA for 14 days. Bowyer et al. (2008) found statistically significant elevations
of serum T4 levels in male rats at 50 mg/kg-day, but not at 2.5 or 10 mg/kg-day, and no changes
in serum T3 or TSH at any dose level. In unpublished studies, blood levels of T3, T4, or TSH
were not consistently changed in male and female F344 exposed to up to 25 mg/kg-day AA in
drinking water for up to 28 days (Friedman et al., 1999b), and blood levels of TSH and indices of
cell proliferation in the thyroid (BrdU incorporation into DNA) were not changed in male
Sprague-Dawley rats exposed to 2 or 15 mg/kg-day for up to 28 days (Klaunig, 2000, as cited in
Environ, 2002). Klaunig and Kamendulis (2005) reported that exposure of F344 rats to AA (0,
2, or 15 mg/kg-day) for 7, 14, or 28 days increased DNA synthesis in the target tissues (thyroid,
testicular mesothelium, adrenal medulla) at all doses and time points examined but not in
nontarget tissue (liver). They also reported increase in DNA damage in the target tissues but not
192 DRAFT-DO NOT CITE OR QUOTE
-------
in nontarget tissue (liver), which supports a mutagenic MOA. In summary, there is a lack of
consistent evidence for AA alteration of thyroid hormone homeostasis. The negative findings in
the comprehensive study by Bowyer et al. (2008) of hypothalamus-pituitary-thyroid axis
endpoints in male F344 rats exposed to AA for 14 days are particularly noteworthy as not being
consistent with the hypothesis that thyroid hormone dysregulation is a key event in AA induction
of thyroid tumors.
Dose-response concordance
No data are available to support dose-response concordance for the proposed effect on
circulating thyroid hormone levels.
Temporal relationship
No data are available to support the temporal relationship between AA exposure,
hormonal disruption, and formation of thyroid tumors to support this proposed MOA.
Biological plausibility and coherence
This hormonal pathway MOA is biologically plausible, and the occurrence of altered
thyroid hormone homeostasis leading to thyroid follicular cell hyperplasia with potential
progression to neoplasia is well established (U.S. EPA, 1998c).
Human relevance
If AA disruption of thyroid hormone homeostasis is supported by future studies, this
proposed MOA for thyroid tumorigenicity could call into question the human relevance of the
tumors.
Conclusion
Although this proposed MOA is possible for thyroid tumorigenicity in male and female
rats (and possibly humans), there is little empirical support for AA alteration of thyroid hormone
homeostasis.
4.8.3.3. Conclusion About the Mode of Action
The available data indicate that the most plausible MOA for the carcinogenicity of AA is
a mutagenic MOA based upon the numerous and consistent study results on the mutagenicity of
AA (or its GA metabolite) in both germ and somatic mammalian cells that support the events,
dose-concordance, and temporal relationship of a mutagenic MOA. There is relatively little
support for a hormonal pathway MOA for the tumor types observed in the animal studies,
although this is a possible MOA and warrants further evaluation. It is also possible that there is
193 DRAFT-DO NOT CITE OR QUOTE
-------
a mixed MOA, i.e., an increased mutagenic burden in hormonally-sensitive tissues with or
without disruption of the hormonal pathways.
4.9. SUSCEPTIBLE POPULATIONS
4.9.1. Possible Childhood Susceptibility
Neurotoxicity
No human data are available regarding age-related differences in susceptibility to
AA-induced neurotoxicity. Animal studies provide conflicting results. Some reports indicate
that young animals may be less susceptible than older ones (Kaplan et al., 1973; Fullerton and
Barnes, 1966), whereas other reports present evidence that young animals may be more sensitive
(Ko et al., 1999; Suzuki and Pfaff, 1973).
Fullerton and Barnes (1966) administered 100 mg/kg AA orally to groups of 5-, 8-, 26-,
and 52-week-old albino rats at weekly intervals and noted severe signs of peripheral neuropathy
in the oldest group after three treatments. The 26-week-old rats were severely affected after four
treatments, while rats whose treatment started at 5 weeks of age only showed "mild" clinical
signs of peripheral neuropathy after 4 weeks of treatment.
Kaplan et al. (1973) injected 50 mg/kg-day AA i.p. to rats ranging in age from 5 to
14 weeks. Impaired rotarod performance appeared earlier in the older rats, but the younger rats
recovered more slowly following the cessation of treatment.
Suzuki and Pfaff (1973) administered 50 mg/kg of AA to 1-day-old and adult rats
3 times/week for up to 18 injections. Signs of hindlimb weakness appeared several days earlier
in the young pups, and degenerative histopathologic changes in peripheral nerves were more
prominent in the pups than the adults.
Recently, Ko et al. (1999) demonstrated that mouse weanlings may be more susceptible
to the adverse neurological effects of AA than young adult mice. Groups of male ICR mice were
exposed to AA in the drinking water at concentrations of 0 or 400 ppm and observed for clinical
signs, rotarod performance, peripheral nerve growth and function, and histopathologic evidence
of peripheral neuropathy. Calculated AA doses were 91.8 ± 20.6 mg/kg-day for the 3-week-old
mice and 90.8 ± 10.9 mg/kg-day for the 8-week-old mice. The younger (3-week-old) mice
exhibited earlier onset (7.1 ± 1.1 days vs. 15.6 ± 4.0 days in 8-week-old mice) and more rapid
progression of AA-induced neuropathy.
Carcinogenicity
With respect to carcinogenicity, EPA has concluded by a weight-of-evidence evaluation
that AA is carcinogenic by a mutagenic MOA. According to the Supplemental Guidance for
Assessing Susceptibility from Early Life Exposure to Carcinogens (U.S. EPA, 2005b), those
exposed to carcinogens with a mutagenic MOA are assumed to have increased early life
194 DRAFT-DO NOT CITE OR QUOTE
-------
susceptibility. Data for AA, however, are not sufficient to develop separate risk estimates for
childhood exposure, thus the oral slope factor and inhalation unit risk (see Section 5.3.5) do not
reflect presumed early life susceptibility for this chemical, and age-dependent adjustment factors
(ADAFs) should be applied to this slope factor when assessing cancer risks for less than 16-year-
old subpopulations or for lifetime exposures that begin in less than 2-year-olds. Example
evaluations of cancer risks based on age at exposure are given in Section 6 of the Supplemental
Guidance.
Aside from the assumption that early life stages are more susceptible to mutagens, there
are limited data on early-life susceptibility to AA-induced carcinogenicity. Gamboa da Costa et
al. (2003) measured DNA adduct formation in selected tissues of adult and whole body DNA of
3-day-old neonatal mice treated with AA and GA. In adult mice, DNA adduct formation was
observed in liver, lung, and kidney with levels of N7-GA-Gua around 2,000 adducts/108
nucleotides and N3-GA-Ade around 20 adducts/108 nucleotides. Adduct levels were modestly
higher in adult mice dosed with GA as opposed to AA; however, treatment of neonatal mice with
GA produced five- to sevenfold higher whole body DNA adduct levels than with AA. The
authors suggest that this is due to lower oxidative enzyme activity in newborn mice. DNA
adduct formation from AA treatment in adult mice showed a supralinear dose-response
relationship, consistent with saturation of oxidative metabolism at higher doses.
Increased incidences of tumors in hormonally responsive tissues (thyroid gland,
mammary gland, and tunica vaginalis mesothelium) have been noted in rats chronically exposed
to AA in the diet (Friedman et al., 1995; Johnson et al., 1986). Since AA induced disruption of
hormonal pathways or homeostasis is a possible MO A, additional studies are needed to evaluate
this MOA and whether there is an increased susceptibility to AA induced hormonal disruption
during early developmental stages.
As discussed in Section 3.3, CYP2E1 catalyzes the initial oxidation of AA to the epoxide
derivative, GA, and there are age-related increases in CYP2E1 expression in humans as reported
by Johnsrud et al. (2003). CYP2E1 was detected as early as the second trimester (0.35 pmol/mg
microsomal protein), increasing approximately fivefold from neonatal levels (median =
8.8 pmol/mg microsomal protein) to post-90-day levels (41.4 pmol/mg microsomal protein).
Levels in older infants (>90 days old), children, and young adults up to 18 years old were
relatively similar. A fourfold or greater intersubject variation was observed among samples from
each age group, with the greatest variation, 80-fold, seen among neonatal samples. These results
suggest that infants less than 90 days old would have decreased clearance of CYP2E1 substrates
compared with older infants, children, and adults. However, the delivery rate of the substrate
relative to the value of the Michaelis-Menten constant (Km) for CYP2E1 is an important
determinant of the total amount metabolized (or parent compound cleared) (Lipscomb, 2004;
Lipscomb et al., 2003), such that the higher the substrate concentration is relative to Km, the
195 DRAFT-DO NOT CITE OR QUOTE
-------
more profound the influence of enzyme level and differences in the enzyme's maximum velocity
(Vmax) on total clearance for a saturable enzyme like CYP2E1. There is no reason to suspect that
the Km value of CYP2E1 in <90-day-old infants would be any different than the Km for
CYP2E1 in older infants, so that a difference in susceptibility in neonates would mostly depend
on levels of CYP2E1 and delivery rates of AA. There is therefore a research need to develop
quantitative estimates of differences in clearance due to different levels of CYP2E1 for less than
90-day-old infants at high or low levels of AA exposure.
4.9.2. Possible Gender Differences
No data are available regarding gender-related differences in sensitivity to AA in
humans.
AA-induced adverse reproductive effects (male-mediated implantation losses and
reduced number of fetuses, testicular atrophy) have been demonstrated in male rodents at dose
levels that do not affect female reproductive performance (see Sections 4.3.1 and 4.5.1; see also
Table 4-32). Part of the gender difference may be due to the AA or GA alkylation of sperm
protamines late during spermiogenesis and resultant genetic damage (Perrault, 2003; Adler et al.,
2000; Generoso et al., 1996; Sega et al., 1989; Sublet et al., 1989). Other modes may involve
neurotoxic actions impairing copulatory behavior (Zenick et al., 1986) and sperm motility (Tyl et
al., 2000b; Sublet et al., 1989), both of which are key determinants of male reproductive
performance (see Section 4.3 for a more detailed discussion).
AA-induced neurological effects have been observed in both male and female rats at
similar dose levels. Light microscopic examination of peripheral nervous tissue revealed
evidence of distal axonal neuropathy in both sexes at doses of 2-3 mg/kg-day for up to 2 years
(Friedman et al., 1995; Johnson et al., 1986, 1985; Burek et al., 1980). Male and female rats also
exhibited similar clinical signs of neurotoxicity following repeated exposure to doses of 20 or
50 mg/kg-day (Burek et al., 1980; Fullerton and Barnes, 1966).
Chronic exposure of F344 rats to AA in drinking water induced increased incidences of
thyroid follicular cell tumors (adenomas and carcinomas combined) in males and females, scrotal
sac mesotheliomas in males, and mammary gland fibroadenomas in females (Friedman et al.,
1995; Johnson et al., 1986). These results show that both male and female rats are susceptible to
AA-induced carcinogenic effects.
4.9.3. Other
No data are available regarding the effects of AA on other potentially susceptible
populations.
Variability in internal dose following exposure to acrylamide in the diet or the
environment is also an area of active research. Genetic polymorphisms in the AA metabolizing
196 DRAFT-DO NOT CITE OR QUOTE
-------
P-450 enzyme CYP2E1 have been identified in humans (Hanioka et al., 2003) and studied for
the impact of a susceptible population to alcohol toxicity (Verlaan et al., 2004) and to
acrylonitrile, a chemical with similar metabolism to AA (Thier et al., 2002). The polymorphisms
result in differences in the Vmax of the enzyme (Hanioka et al., 2003) that could result in greater
or lesser production of the GA metabolite and make some people more or less sensitive to
adverse effects. The epidemiology evidence is not strong. There is some suggestive (i.e., not
statistically significant) evidence that polymorphisms in CYP2E1 might confer a differential risk
to alcohol-induced chronic pancreatitis (Verlaan et al., 2004) and that a slower CYP2E1-
mediated metabolism of acrylonitrile might result in higher acrylonitrile-hemoglobin adducts
(and lower N-(cyanoethyl)valine adducts from the metabolite) (Thier et al., 2002). As discussed
for childhood susceptibility, however, the delivery rate of the substrate relative to the values of
Km and Vmax for CYP2E1 is an important determinant of the total amount metabolized (or parent
compound cleared) (Lipscomb, 2004; Lipscomb et al., 2003).
Estimates of daily AA intake in a nonsmoking general population in Germany based on
hemoglobin adduct levels in blood and mercapturic acid excretion in urine indicated that
children take up approximately 1.3-1.5 times more AA per kilogram of body weight than adults.
The ratio GAMA/AAMA was also significantly higher in the group of young children (6-10
years) with a median level of 0.5 (Hartmann et al., 2008). In this same study there were no
observed gender-related differences in internal exposure and metabolism.
Vesper et al (2008) reported highly variable estimates of acrylamide exposure in
subgroups of the European Prospective Investigation into Cancer and Nutrition (EPIC) study
population (510 subjects from 9 European countries, randomly selected and stratified by age,
gender, and smoking status) based upon levels of hemoglobin adducts of acrylamide (HbAA)
and its primary metabolite glycidamide (HbGA). A large variability in acrylamide exposure and
metabolism between individuals and country groups was observed with HbAA and HbGA values
ranging between 15-623 and 8-377 pmol/g of Hb, respectively. Both adducts differed
significantly by country, sex, and smoking status.
Clearly, any assessment of health effects for a potentially susceptible population must
consider the impact of factors such as age, country of origin, BMI, alcohol consumption, sex,
and smoking status on the internal dose, as well as the susceptibility of any subpopulation to the
biological activity of AA or GA at a target site. It is important to note, however, that since both
the parent AA and the metabolite GA have adverse effects, different catalytic activities of
CYP2E1 or other factors that lead to high variability in the internal dose may result in different
spectra of adverse effects, providing an even greater challenge to simple classification schemes
for susceptible subpopulations.
197 DRAFT-DO NOT CITE OR QUOTE
-------
5. DOSE-RESPONSE ASSESSMENTS
5.1. ORAL REFERENCE DOSE
5.1.1. Choice of Principal Study and Critical Effect—with Rationale and Justification
As discussed in Section 4.7.1, there are only a few reports of noncancer health effects in
humans associated with oral exposure to AA, but occupational experiences involving inhalation
and dermal exposures firmly establish neurological impairment as a potential human health
hazard from acute and chronic exposure to AA. In contrast, the oral toxicity database for
laboratory animals is robust and contains (as shown in Figure 5-1 and Table 5-1): two 2-year
carcinogenicity/toxicology drinking water studies in F344 rats; two two-generation reproductive
toxicity studies, one in F344 rats and one in CD-I mice; several single-generation reproductive
toxicity studies involving prolonged prebreeding drinking water exposure of Long-Evans rats
and ddY mice; and several developmental toxicity studies involving gestational exposure of
Sprague-Dawley and Wistar rats and CD-I mice.
198 DRAFT-DO NOT CITE OR QUOTE
-------
50 -
45 -
A r\
40 •
35 -
30 -
s-
•9
£> 25 -
0)
E
20 -
15 -
10 -
5 •
o -
j
1
J
1 '
t
i 1
I
i
1
1
i
1 i
i
j
i
i
i
A
i
1
tii
^- ^- T- CN CO O) O CN CO CO
ro ro ro Q, .i-. J
!". !". !". (/) 2 :
O O W g • - ;
» -jj- ra ra ra
3 in ......
2 g O O O
Drop-lines span highest to lowest doses tested
1
ITTT!
i
1
1
t
i
1
1
i
1
t
I'
i
I
, A
' i
•
1
\
|
1
^~ CN co co in ^~ 51- ^~ in CD oo o o r —
ra ra ra ra
O CO CO CO
—r 1 1 1 1 1 1
3
i £
2 -
J ra ra <
3 . . . . :
> ^ -
3 W ,
^ D L
I O
TJ
_
^1
i
1
1
i
I
f^ T
.
-> -.
^
1
1
1 • NOAEL
A LOAEL
•
^
II
• A
'lift
- CN co in m ^-" in co in
---—'-—'-—' 0 -~ ' 0 -~ ' -— -•
£ IS IS w •£ w "S "S
0 {/) (u (u -3 TO 3 TO TO
:" o > > i 61" E '-" L-
1 ti
£00
= 00
ro Q Q
•— ro ro
o •§ •§
o -° -°
000
Q ^ ^
Subchronic
Chronic
Repro
Develop-other
Develop-neuro
bw = body weight; Deer = decreased; Develop = developmental study; DNC = degenerative nerve changes; HLFS = hindlimb foot splay; LOAEL = lowest-observed-adverse-
effect level; MM IL = male mediated implantation losses; Malform = fetal malformations and variations; Neuro = neurological; NOAEL = no-observed-adverse-effect level;
Repro = reproductive study. References: 1-Burek et al., 1980; 2-Johnson et al., 1986; 3-Friedman et al., 1995; 4-Tyl et al., 2000a; 5-Chapin et al., 1995; 6-Zenick et al, 1986-M;
7-Zenicketal., 1986-F; 8-Smithet al., 1986; 9-Sakamoto and Hashimoto, 1986-M; 10-Sakamoto and Hashimoto, 1986-F; 11-Field etal., 1990-rat; 12-Field et al., 1990-Mouse;
13-Wiseetal., 1995; 14-Friedman et al., 1999a; 15-Garey and Paule, 2007.
Figure 5-1. Acrylamide oral exposure: selected NOAELs and LOAELs.
199 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-1. Acrylamide oral exposure: selected NOAELs and LOAELs
Effect/study
Lowest
dose tested
NOAEL
LOAEL
Highest
dose
tested
Reference
Subchronic
DNC; M&F rat EM, 90 days
DNC;M&Frat LM, 90 days
HLFS; M&F rat, 90days
HLFS; F mouse, GD 6-20
HLFS; F rat, GD 6-17
HLFS; M mouse, 4 weeks before mating
HLFS; F mouse, 4 weeks before mating
0.05
0.05
0.05
3
5
3.3
—
0.2
1
5
15
10
13.3
—
1
5
20
45
15
16.3
18.7
20
20
20
45
20
16.3
18.7
1
1
1
12
13
9
10
Chronic
DNC; M&F rat, LM, 2 years
DNC; M rat, LM, 2 years,
DNC; F rat, LM, 2 years
DNC; M rat, LM, 2-generation
HLFS; M&F rat, 2 years
HLFS; F rat, 2 years
HLFS; F rat, 2 years
HLFS; M&F mouse, 2 generations
0.01
0.1
1
0.5
0.01
0.1
1
0.8
0.5
0.5
1
—
2
2
o
J
7.5
2
2
3
5
~
—
~
~
2
2
3
5
2
2
3
7.5
2
3
3
4
2
3
3
5
Reproductive effects
Testes atrophy; M rat
Sperm changes; M mouse
Male-mediated Implantation Losses
(MM IL) ; M rat
MM IL; M mouse
MM IL; M rat
MM IL; M rat
MM IL; M mouse
Fertility; F mouse, 4 weeks before mating
Fertility; F rat, 9 weeks before mating
0.05
3.3
0.5
0.8
7.9
1.5
3.3
18.7
5.1
5
13.3
2
3.1
~
1.5
9
18.7
14.6
20
16.3
5
7.5
7.9
2.8
13.3
~
~
20
16.3
5
7.5
7.9
5.8
16.3
18.7
14.6
1
9
4
5
6
8
9
10
7
Developmental effects - other
Decreased pup bw; rat
Malformations; rat
Malformations; mouse
Decreased pup bw; rat
Decreased pup bw; rat
Develo
DNC; mouse
HLFD; rat
Decreased grip strength; mouse
Neurobehavior; rat, GD 6-10
Neurobehavior; rat, gestation, lactation, to
PND85
5.1
2.5
3
5
0.1
5.1
15
45
~
6
8.8
—
—
5
~
14.6
15
45
20
6
7
11
12
13
15
pmental effects - neurological
0.8
25
0.8
5
0.1
7.5
25
3.1
10
1.3
_
—
7.5
15
6
7.5
25
7.5
20
6
5
14
5
13
15
200 DRAFT-DO NOT CITE OR QUOTE
-------
As shown in Figure 5-1 and Table 5-1, the most sensitive effects noted in animals are
degenerative peripheral nerve changes ("DNC" in Figure 5-1) and male-mediated implantation
losses (i.e., male-mediated dominant lethal mutations — "MM IL" in Figure 5-1). The lowest
observed exposure levels associated with peripheral nerve changes are: 1 mg/kg-day for
ultrastructural changes associated with degenerative nerve changes (0.2 mg/kg-day NOAEL) in
male F344 rats exposed for 90 days (Burek et al., 1980) and 2 mg/kg-day for degenerative nerve
changes detected by light microscopy (0.5 mg/kg-day NOAELs) in male F344 rats exposed for 2
years in two separate bioassays (Friedman et al., 1995; Johnson et al., 1986). The lowest
exposure levels associated with male-mediated implantation losses are somewhat higher than
those associated with degenerative nerve changes: 2.8 mg/kg-day (1.5 mg/kg-day NOAEL) in
Long-Evans rats exposed for 80 days (Smith et al., 1986); 5 mg/kg-day (2.0 mg/kg-day NOAEL)
in FO and Fl F344 rats exposed over 10 weeks plus breeding (Tyl et al., 2000a); 7.5 mg/kg-day
(3.1 mg/kg-day NOAEL) in FO and Fl CD-I mice exposed over 14 weeks of breeding (Chapin et
al., 1995); and 13.3 mg/kg-day (9.0 mg/kg-day NOAEL) in ddY mice exposed for 4 weeks
(Sakamoto and Hashimoto, 1986). Testicular atrophy in rats and sperm abnormalities in mice
have been observed only at oral doses > 15 mg/kg-day, and female fertility and reproductive
performance in rats were unaffected at doses in the 15-20 mg/kg-day range (Figure 5-1).
Comprehensive histologic examinations of all major organs and tissues in the chronic and
subchronic rat bioassays found no exposure-related nonneoplastic lesions at other sites at dose
levels below 5 mg/kg-day (Table 5-1). Hindlimb splaying, a gross characteristic sign of
peripheral neuropathy, has been observed in most studies at oral exposure levels (about 9-
25 mg/kg-day) well above the lowest doses (1-2 mg/kg-day) associated with microscopically
detected degenerative peripheral nerve changes (Table 5-1; HLFS in Figure 5-1). As discussed
in Section 4.7.1, an exception is one report that exposure to 0.5 mg/kg-day AA induced hindlimb
splaying in FO male F344 rats (Tyl et al., 2000a), but this report is not consistent with other
findings and was not included in Figure 5-1. In the same study, Tyl et al. (2000a) did not
observe hindlimb foot splay in the Fl-generation rats exposed to doses as high as 5 mg/kg-day,
nor was this endpoint reported in the other F344 rats exposed to drinking water doses as high as
2-3 mg/kg-day for 2 years (Friedman et al., 1995; Johnson et al., 1986) or 5 mg/kg-day for 90
days (Burek et al., 1980) (Table 5-1; Figure 5-1). No increases in fetal malformations or
variations were observed in rats or mice following maternal exposure to oral doses as high as 15
or 45 mg/kg-day, respectively (Field et al., 1990; Table 5-1; Figure 5-1). Neurobehavioral
assessments of rat offspring found evidence for subtle effects at >15 mg/kg-day (decreased
auditory startle response, Wise et al., 1995) and 6 mg/kg-day (decreased cognitive motivation,
Gary and Paule, 2007); NOAELs for these effects were 10 and 1.3 mg/kg-day, respectively
(Table 5-1; Figure 5-1). In conclusion, microscopically detected degenerative peripheral nerve
201 DRAFT-DO NOT CITE OR QUOTE
-------
changes appear to be the most sensitive effect from oral exposure and are selected as the critical
effect for deriving the RfD.
AA induces transmissible genetic damage in male germ cells of mice in the form of
reciprocal translocations and gene mutations. Such effects can lead to genetic disorders and
infertility in subsequent generations. As discussed in Section 4.4, dose-response relationships
for heritable germ cell effects in animals (exposed dermally or by i.p. injection to AA) are not
well described, particularly at dose levels below 50 mg/kg-day, and possible associations
between human exposure to acrylamide and altered sperm characteristics have not been
adequately studied. Thus, although the available data indicate that degenerative nerve changes
are the critical effect in animals from chronic oral exposure to AA, additional research may find
that heritable germ cell effects may occur at chronic oral exposure levels comparable to those
inducing degenerative nerve cell changes.
Two chronic (2-year) drinking water studies (Friedman et al., 1995; Johnson et al., 1986)
were selected as co-principal studies for deriving the RfD, and the final quantitative RfD value is
based on the dose-response data from only the Johnson study. These studies are the most
appropriate datasets to derive the chronic RfD, rather than the subchronic study by Burek et al.
(1980), primarily due to more appropriate durations of exposure (lifetime vs. 90 days) and
greater numbers of animals/exposure group (a range of 20 to 88/sex/group in the chronic studies
vs. 10/sex/group in the subchronic study). All three studies included multiple dose groups,
thereby providing information on characteristics of the dose-response relationship.
The subchronic, 90-day study (Burek et al., 1980) used a more sensitive electron
microscopic technique to detect degenerative nerve changes vs. the light microscopy used in the
2-year bioassays. The chronic drinking water study by Johnson et al. (1986) examined nerves
sampled at 18 and 24 months by electron microscopy but reported that the background of
ultrastructural changes in aging rats was too high to discern differences between control and
exposed groups. The Burek et al. (1980) study evaluated sciatic nerves from only three
rats/exposure group (about 150 fields/rat)6, and the changes noted were reported only as the total
numbers of fields (per group) with ultrastructural changes as axolemma invaginations or
Schwann cells without axons and/or with degenerating myelin (see Table 4-8). This reporting of
the electron microscopy data does not support a statistical comparison of the incidence of
changes between the exposed and control groups because it is unknown within any exposure
group how the numbers of changes were distributed among the three rats (i.e., whether the
apparent increase in incidence of fields with changes was due to one, two, or all three rats in the
6 The incidences of fields with any alterations were: 68/450, 39/450, 44/350, 108/453, 149/443, and
239/435 for the 0, 0.05, 0.2, 1, 5, and 20 mg/kg-day groups. Approximately 150 fields were examined for each rat;
however, further statistical analysis was not possible because the numbers of fields with changes observed were not
reported for each of the three rats singly from each group.
202 DRAFT-DO NOT CITE OR QUOTE
-------
1, 5, and 20 mg/kg-day groups). The 1 mg/kg-day LOAEL and 0.2 mg/kg-day NOAEL from
this subchronic study were, therefore, based on a semi quantitative assessment of the electron
microscopy data, i.e., the incidences of electron microscopic fields with any ultrastructural
changes were higher in the 1, 5, and 20 mg/kg-day groups than in the 0, 0.05, and 0.2 mg/kg-day
groups, and light microscopy of sciatic nerves revealed no signs of degeneration in the 0, 0.05,
0.2, or 1 mg/kg-day groups, equivocal to very slight degeneration in 15/20 5 mg/kg-day rats, and
moderate to severe degeneration in 20/20 20 mg/kg-day rats. The 1 mg/kg-day LOAEL,
however, was for only very slight changes that were reversible by day 25 posttreatment, and the
NOAEL from this study was limited to the selection of dose levels (i.e., there was no 0.5 mg/kg-
day group as in the 2-year studies). The raw data for the ultrastructural changes in the
subchronic study were not available for benchmark analysis, but the results clearly support the
findings from the chronic studies.
The two chronic studies provided sufficient data for benchmark analysis, and
reproducible NOAELs of 0.5 mg/kg-day and LOAELs of 2 mg/kg-day for persistent
microscopically-detected AA-induced degenerative nerve changes from lifetime exposures.
These studies are lacking functional testing of neurotoxicity and use a relatively insensitive
measure (peripheral axonopathy detected by light microscopy) as the primary index of
neurotoxicity, and leave yet unanswered the possibility that lower doses might result in adverse
terminal degeneration or other functional deficit prior to the onset of axonal degeneration.
5.1.2. Methods of Analysis—Including Models (BMD, equivalent AUCs, in vivo rate
constants, etc.)
The methods and models used to derive an RfD include benchmark dose models;
methods to estimate the serum area under the concentration-time curve (AUC) based upon
hemoglobin adduct levels, serum time-course data; and second order rate constants for adduct
formation; and a model developed by EPA that estimates rat in vivo adduct formation rate
constants based on rat adduct time-course data and various measures of rat serum in vivo AUCs
for a given dose rate.
All available models in the EPA Benchmark Dose Software (BMDS version 1.3.1) were
fit to the incidence data for microscopically-detected degenerative nerve changes in male and
female F344 rats from the two 2-year drinking water studies (Friedman et al., 1995; Johnson et
al., 1986). The modeled data are shown in Table 5-2. The benchmark response (BMR)
predicted to affect 5% of the population, BMR5, was selected for the point of departure (POD).
A BMR of 5% extra risk was selected for the following reasons: (1) this effect level is
considered to be a minimal biologically significant change given the critical effect of
degenerative nerve changes; (2) the BMDL5 remained near the range of observation; and (3) the
203 DRAFT-DO NOT CITE OR QUOTE
-------
5% extra risk level is supportable given the relatively large number of animals used in the
principal studies.
Table 5-2. Incidence data for degenerative changes detected by light
microscopy in nerves of male and female F344 rats exposed to acrylamide in
drinking water for 2 years
Reference
Johnson et al, 1986
(incidence of rats with changes in tibial
nerves: see Table 4-9)
Males (moderate to severe)3
Females (slight to moderate)3
Friedman et al., 1995
(incidence of rats with minimal to mild
changes in sciatic nerves: see Table 4-12)
Males
Females
Dose (mg/kg-day)
0
9/60
3/60
30/83
7/37
0
-
29/88
12/43
0.01
6/60
7/60
-
0.1
12/60
5/60
21/65
0.5
13/60
7/60
13/38
1.0
-
2/20
2.0
16/60b
16/6 lc
26/49c
3.0
-
38/86c
"Reported severity classes were very slight, slight, moderate, and severe. Males showed a high background of very
slight and slight lesions; females showed a high background of very slight lesions.
bStatistically significant trend test (Mantel-Haenszel extension of the Cochran-Armitage test, p < 0.05) for pooled
moderate and severe degeneration. Note: no statistical significance for the high dose group. Incidence for severe
degeneration with dose level in parentheses (in mg/kg-day) was 1 (control), 1 (0.01), 0 (0.1), 0 (0.5), and 4 (2.0).
Statistically significantly different from control incidences (p < 0.05).
dTwo control groups were included in the study design to assess variability in background tumor responses;
degeneration was reported to be characterized by vacuolated nerve fibers of "minimal-to-mild severity."
As shown in Appendix C, all models provided adequate fits to the data for changes in
tibial nerves of male and female rats in the Johnson et al. (1986) study, as assessed by a
X2 goodness-of-fit test. The log-logistic model was the best fitting model for the male rat data as
assessed by Akaike's Information Criterion (AIC). The probit model was the best fitting model
for the female rat data as assessed by Akaike's Information Criterion (AIC). The log-logistic
model was thus selected to estimate a benchmark dose (BMD) from the Johnson et al. (1986)
data. The probit model was selected to estimate the BMD for the female rat data. Table 5-3
(same as Table C-4 in Appendix C) lists the predicted doses associated with 10, 5, and 1% extra
risk for nerve degeneration in female and male rats in the Johnson et al. (1986) study. The
BMD5 is the predicted dose associated with a 5% extra risk for degenerative lesions in either
tibial or sciatic nerves, the BMDLs is the lower 95% confidence limit for the 5% extra risk. For
male rats, the BMD5 is 0.58 mg/kg-day, and the BMDL5 is 0.27 mg/kg-day. For female rats, the
BMD5 is 0.67 mg/kg-day, and the BMDL5 is 0.49 mg/kg-day.
204 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-3. Predictions (mg/kg-day) from best-fitting models for doses
associated with a 10, 5, and 1% extra risk for nerve degeneration in male and
female rats exposed to acrylamide in drinking water (Johnson et al 1986)
Model
Male
Log-logistic
Female
Probit
BMD10
(ED10)
1.22
1.19
BMDL10
(LED10)
0.57
0.88
BMD5
(ED5)
0.58
0.67
BMDLS
(LED5)
0.27
0.49
BMDi
(EDO
0.11
0.15
BMDLi
(LEDi)
0.05
0.11
Data source: Johnson et al. (1986). See Appendix C for model description and results.
Several models in the software provided adequate fits to the data for minimal to mild
changes in sciatic nerves of male and female rats in the Friedman et al. (1995) study, as assessed
by a x2 goodness-of-fit test (Appendix C). The quantal-quadratic and gamma models provided
the best fit of the male and female rat data, respectively, as assessed by AIC. Table 5-4 (same as
Table C-7 in Appendix C) lists the predicted doses associated with 10, 5, and 1% extra risk for
nerve degeneration in female and male rats in the Friedman et al. (1995) study. The BMDs for
minimal to mild changes in sciatic nerves for male rats is 0.77 mg/kg-day and the BMDL5 is
0.57 mg/kg-day. For female rats, the BMD5 is 2.25 mg/kg-day and the BMDL5 is 0.46 mg/kg-
day.
Table 5-4. Predictions (mg/kg-day) from best-fitting models for doses
associated with 10, 5, and 1% extra risk for sciatic nerve changes in male and
female rats exposed to acrylamide in drinking water (Friedman et al 1995)
Model
Male
Quantal quadratic
Female
Gamma3
BMD10
(ED10)
1.11
2.48
BMDL10
(LED10)
0.82
0.93
BMD5
(ED5)
0.77
2.25
BMDLS
(LED5)
0.57
0.46
BMDi
(EDO
0.34
1.86
BMDLi
(LEDO
0.25
0.09
aRestrictpower>l.
Data source: Friedman et al. (1995). See Appendix C for model description and results.
5.1.3. RfD Derivation—Including Application of Uncertainty Factors
The results of BMD analysis for the male and female rats appeared to be similar in the
Johnson et al. (1986) and Friedman et al. (1995) studies with BMDLs for female rats of 0.49 and
0.46 mg/kg-day, respectively, and for male rats of 0.27 and 0.57 mg/kg-day, respectively. The
lowest of the BMDLs is from the Johnson et al. (1986) study for male rats (0.27 mg/kg-day for
5% extra risk for mild-to-moderate lesions), and was chosen as the POD for the most sensitive
effect to derive the RfD.
As discussed in Section 3-5, the internal dose (area under a time-concentration curve,
AUC) of acrylamide and glycidamide in a rat can be derived for an external exposure to the
205 DRAFT-DO NOT CITE OR QUOTE
-------
BMDL5 of 0.27 mg/kg-day based on the relationships among hemoglobin adducts, serum levels,
and administered dose that have been evaluated in a number of studies (Doerge et al., 2005 a,b,c,
Tareke et al., 2006; Fennell et al., 2005; Bergmark et al., 1993). These data can then used to
estimate the daily human intake of acrylamide needed to produce a comparable internal serum
AUC level in humans (i.e., the humam equivalent concentration [HEC]), and with further
adjustment of the HEC with uncertainty factors, an RfD.
The equations needed to derive the rat AUC and the human equivalent concentration are
as follows:
1. Estimate of the rat serum AUCpoo (uM-hr) for AA or GA based on the POD dose (i.e, the
BMDL in mg AA/kg bw/d).
F344Rat AUCmn (uM -... ,
rUL) \ / / 7 A A l 1 l
kg bw mg AA I kg bw
where the AUCs are for either A A or GA , and the conversion factor of uM-hrp344 rat / nig AA/kg
bw is the measured (or estimated) AUC values in rats normalized to an administered dose of AA
in mg / kg bw.
2. Estimate of the human equivalent concentration or the daily intake (in mg AA/kg bw/d)
needed to generate a compable AUC in humans to that of the rat AUC from the POD.
mrr A A llM — hj",
HEC in -£ = F344 Rat A UCPOD (uM - hr) + ^^-
kg bw mg AA I kg bw
where the AUCs are for either A A or GA , and the conversion factor of uM-hrhuman / nig AA/kg
bw is the measured (or estimated) AUC values in humans normalized to an administered dose of
AA in mg / kg bw.
The tables discussed in Section 3-5 (Tables 3-5 through 3-7) that summarize the second order
rate constants and various measured or estimated uM-hr AUCs per mg AA/kg bw for F344 rats
and humans are reproduced below for convenience as Tables 5-5 through 5-7.
206 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-5. Second order rate constants for reaction of acrylamide or glycidamide with the
N-terminal valine residue of hemoglobin.
Second Order Rate Constant for Formation of
Hemoglobin Adducts
(l/g globin/h) x 106
Average
or pooled Gender
Male and Not Pooled
Male Female Female Specified Rat and
Source Rat Rat Rat Rat Mouse Human
AA -Val In Vivo Adduct Formation Rate3
Based on all rat and mice Tareke et al. (2006)
adduct data and measured serum AUCs in 7.5
Doerge et al. (2005 b, c) single dose studies.
Based on gender specific rat Tareke et al.
(2006) adduct data and measured serum AUCs
in Doerge et al. (2005c) single dose studies 8.9 5.9 7.4
Based on all rat Tareke et al. (2006) adduct
data and measured serum AUCs in Doerge et
al. (2005c) single dose studies. 7.5
AA -Val In Vitro Rate Adduct Formation Rate
As reported by Fennell et al. (2005) 3.82 4.27
As reported by Bergmark et al. (1993) 4.4
As reported by Tareke et al. (2006) 2.9 7.4
As reported by Tb'rnqvist et al. (2008) 4.6
GA -Val In Vivo Adduct Formation Rate3
Based on all rat and mice Tareke et al. (2006)
adduct data and measured serum AUCs in 32.5
Doerge et al. (2005 b, c) single dose studies.
Based on gender specific rat Tareke et al.
(2006) adduct data and measured serum AUCs
in Doerge et al. (2005c) single dose studies 35.3 20.0 27.6
Based on all rat Tareke et al. (2006) adduct
data and measured serum AUCs in Doerge et
al. (2005c) single dose studies. 34.0
GA -Val In Vitro Rate Adduct Formation Rate
As reported by Fennell et al. (2005) 4.96 6.72
As reported by Bergmark et al. (1993)b 12.0 11.0
As reported by Tareke et al. (2006) 9.5 59.0
As reported by Tb'rnqvist et al. (2008) 13.6
a See Appendix E for a complete description of the derivation of the in vivo adduct formation rates.
b Note: Bergmark derived the rat GA-Val residue such that kval = (GA-Val *kcys)/GA cys; the human GA-Val adduct was measured
directly.
207 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-6. Measured and estimated AA AUCs normalized to dose in humans
and F344 rats.
AA ADC in uM-hr per mg AA/kg bw
Average of
Male and Unspecifie
Male Female Female d Gender
Rat Rat Rat Rat Human
AA in Humans
Measured
Kopp and Dekant 2009 - human serum data AA
(single dose of 20 ug/kg, n=3F,3M) 2.83
Estimated using human adduct data and test
animal in vivo rate constants
Fennell et al 2005 - human adduct data and in
vivo rate constants derived from Tareke et al.
(2006) adduct data for all rat and mice in Doerge
et al. (2005 b, c) single dose AUCs. 140.1
Estimated using human adduct data and
human in vitro rate constants
Fennell et al 2005 - human adduct data and
Fennell in vitro rate constants 246.0
Fennell et al 2005 - human adduct data and
Bergmark et al. 1993 in vitro rate constants 238.8
Estimated using human adduct data and rat in
vitro rate constants
Fennell et al 2005 - human adduct data and
Tb'rnqvist et al. 2008 in vitro rate constants 228.5
AAin F344Rats
Measured
Doerge et al. 2005 c - time course data from a
single dietary exposure 18.0 15.0 16.5
Doerge et al. 2005 c - time course data from a
single gavage exposure 24.0 45.0 34.5
Estimated using rat adduct data and rat in vivo
rate constants
Tareke et al (2006) adduct data for the Doerge et
al. (2005a) 42 day drinking water study, and
gender specific in vivo derived rate constants from
Tareke et al. (2006) and Doerge et al. (2005c) 22 48 35
Estimated using rat adduct data and rat in
vitro rate constants
Tb'rnqvist et al. 2008 - adduct data from drinking
water studies and in vitro rate constants 34.0 48.0 41.0
Fennell et al. 2005 - adduct data single dose
gavage studies and in vitro rate constant. 80.2
208 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-7. Measured and estimated GA AUCs normalized to Dose in
Humans and F344 rats.
GA ADC in uM-hr per mg AA/kg bw
Average of Unspecified
Male Female Male and Gender
Rat Rat Female Rat Rat Human
GA in Humans
Estimated using human adduct data and test
animal in vivo rate constants
Fennell et al 2005 - human adduct data and in
vivo rate constants derived from Tareke et al.
(2006) adduct data for all rat and mice in Doerge
et al. (2005 b, c) single dose AUCs. 12.5
Estimated using human adduct data and in
vitro rate constants
Fennell et al 2005 - human adduct data and
Fennell in vitro rate constants 60.4
Fennell et al 2005 - human adduct data and
Bergmark et al. 1993 in vitro rate constants 37.0
Estimated using human adduct data and rat in
vitro rate constants
Fennell et al 2005 - human adduct data and
Tb'rnqvist et al. 2008 in vitro rate constants 29.9
GAin F344Rats
Measured
Doerge et al. 2005 c - time course data from a
single dietary exposure 19.0 15.0 17.0
Doerge et al. 2005 c - time course data from a
single gavage exposure 13.0 44.0 28.5
Estimated using rat adduct data and rat in
vivo rate constants
Tareke et al (2006) adduct data for the Doerge et
al. (2005a) 42 day drinking water study, and
gender specific in vivo derived rate constants
from Tareke et al. (2006) and Doerge et al.
(2005c) 15.0 48.0 31.5
Estimated using rat adduct data and rat in
vitro rate constants
Tb'rnqvist et al. 2008 - adduct data from drinking
water studies and in vitro rate constants 18.0 34.0 26.0
Fennell et al. 2005 - adduct data single dose
gavage studies and in vitro rate constant. 52.1
209 DRAFT-DO NOT CITE OR QUOTE
-------
Choice of the rat AUC /mg AA/kg bw conversion factor to derive the ratAUCpoo.
The best AUC conversion factor (i.e., AUC per administered dose) value to use in the
derivation of a rat AUC based on the POD (BMDL) from the Johnson et al. (1986) study would
be a directly measured in vivo gender specific serum AUC following a known dose of
acrylamide in a similar dosing regimen (i.e., a drinking water exposure). For the F344 rat, the
only directly measured AUC values from serum time course data for male and female rats are
from single doses of AA administered by iv, gavage, or in the diet (Doerge et al., 2005c). Some
serum data are available from the Doerge et al. (2005a) drinking water study in F344 rats,
however, the published report only provides average serum levels apparently based on point
samples from a limited number of days during the 42 day dosing period, and taken at unknown
times relative to the intake of AA in the drinking water. Because of the rapid clearance of AA
and GA from the blood, a single point sample with unknown time relative to intake is not
sufficiently accurate to derive an AUC. A daily AUC from the Doerge et al. (2005a) drinking
water study, however, can be derived from the measured hemoglobin adduct levels reported by
Tareke et al. (2006) divided by a second order adduct formation rate. Table 5-8 below list the
available adduct formation rates for rat AA-Val and GA-Val based on in vitro or in vivo data.
The most relevant rates are those derived from in vivo data.
The derivation of in vivo adduct formation rates requires three critical types of data from
a single study: 1) the administered dose, 2) time course serum levels, and 3) time course adduct
levels (including sufficient post dosing sample times to determine elimination rates) if longer
than one day of exposure. The only studies meeting this requirement are those from Doerge et
al. (2005c) and Tareke et al. (2006). EPA has derived in vivo gender specific adduct formation
rates based upon these single dose in vivo studies in male and female F344 rats. Table 5-8 list
the raw serum AUCs that were available in numerical tables from Doerge et al. (2005c) and the
levels of hemoglobin that were taken from bar chart compilations in Tareke et al. (2006).
210 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-8. Serum AUC from Doerge et al. (2005c) and hemoglobin adduct levels
from Tareke et al (2008) for a 0.1 mg/kg single dose of AA in male and female F344
rats.
Type of
adduct from
AA dosing
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
sex-
route
M-control
M-Diet
M-
gavage
M-IV
F-control
F-Diet
F-gavage
F-IV
M-control
M-IV
M-
gavage
M-Diet
F-control
F-IV
F-Diet
F-gavage
Hb adduct level
(fmole/mg
globin)
9
19.5
20
46.5
12
23
29
49.5
32.5
36
64
98.5
45
48.5
102
131
AUC
(uM-hr)
0
1.8
2.4
4.1
0
1.5
4.5
4.6
0
0.58
1.3
1.9
0
0.6
1.5
4.4
A linear regression of the hemoglobin adduct levels against the AUCs resulted in the
following correlations coefficients (r), equations, and slopes.
Regression of hemoglobin adduct levels to AUC to derive in vivo second order rate constants for
adduct formation (i.e., the slope of the regression line)
y- Ratio of
Intercep Slope GA-
Gender Adduct Slope t r2 Val / AA-Val r
Male
Female
Male
Female
AA-Val
AA-Val
GA-Val
GA-Val
8.92
5.90
Average = 7.4
35.33
19.98
Average = 27.7
5.24
12.73
24.36
49.16
0.94
0.85
0.96
0.93
0.97
0.92
3.96 0.98
3.38 0.96
Example equation:
AA-Val (fmoles/mg globin) for males = 8.9 AA AUC (uM-h) + 5.2
GA-Val (fmoles/mg globin) for females = 20.0 AA AUC (uM-h) + 49.2
The slopes from these linear regressions represent the in vivo derived second order rate
constants for AA-Val listed in Tabel 5-5 as 8.9 for male F344 rats, 5.9 for female F344 rats, and
211 DRAFT-DO NOT CITE OR QUOTE
-------
7.4 as the average of both genders (in units of [1/g globin/h] multiplied by 106 in Table 5-5 for
ease of presentation). The GA-Val formation rate constants are 35.3 for males, 20.0 for females,
and 27.6 as the average of both genders.
Tareke et al. (2006) do not report gender specific slopes for F344 rats, but they do report
a slope of 7.5 for AA-Val (assumed to be pooled data for both genders), and 34 for GA-Val
(both genders). The higher GA-Val slope for both genders from Tareke et al. (2006) of 34
compared with the average from EPA's derivation (27.6) may have resulted if Tareke et al.
included the serum and adduct data from administered doses of GA in their analysis, data that
were not included in EPA's regression analysis, i.e., EPA only used the adduct data from the AA
dosing. EPA chose to derive the gender specific rates to more accurately estimate the internal rat
AUC from the POD because Doerge et al. (2005c) observed lower serum AA-AUC and GA-
AUC in male rats compared to females following the same gavage dose of AA, and Tornqvist et
al. (2008) also report a gender difference in estimated AUCs from a drinking water exposure (see
Tables 5-6 and 5-7).
EPA developed a simple model to fit the hemoglobin adduct data that Tareke et al.
(2006) collected from the Doerge et al. (2000a) drinking water study to derive a conversion
factor for the internal AA-AUC or GA-AUC per dose of AA taken in via drinking water. The
model was parameterized with the the gender specific rat in vivo adduct formation rate constants
derived by EPA, and the adduct elimination rate constants reported by Tareke et al. (2006).
Appendix E provides the details of the model, the model code, the parameters and supporting
data, and examples of the fits to the hemoglobin adduct data. The model simulations resulted in
the following AUC conversion factors (also listed in Tables 5-6 and 5-7). These values were
used to estimate the rat internal AUC that would result from the BMDL (i.e., the POD) for
neurotoxicity as the basis for the RfD:
AA AUC in |iM-hr per mg AA/kg bw = 22 (for males) and 48 (for females)
GA AUC in |iM-hr per mg AA/kg bw =15 (for males) and 48 (for females)
The POD for neurotoxicity is a BMDLs of 0.27 mg AA/kg-day in male F344 rats from
the Johnson et al. (1986) study. As discussed in the section on the mode of action of AA induced
neurotoxicity, there is uncertainty as to whether AA or GA or both are responsible for the
observed effects, although the current evidence tends to favor acrylamide as the putative
neurotoxin. Based on a choice of the parent AA as the putative neurotoxin, and using the male
(i.e., gender specific) AUC conversion factor of 22 |iM-hr per mg AA/kg bw, the estimated
F344 male rat AUCPOD from exposure to a BMDL5 of 0.27 mg AA/kg-day is 5.94 uM-hr :
212 DRAFT-DO NOT CITE OR QUOTE
-------
.ra
F344Male Rat AUCmn (uM - hr) = x f = 5.94MM - hr
rULJ \ / 77 A A I 1 1
kg bw mg AAI kg bw
Deriving the HEC
The human equivalent concentration is the administered dose in humans that would result
in the same internal serum AA-AUC as produced in the rat from the rat BMDL dose, i.e., an
internal AUC of 5.94 uM-hr. No direct measurement of human serum levels of AA are available
to derive an AA-AUChuman /AA mg/kg bw conversion factor for humans except from one study
by Kopp and Dekant (2009) where human time course serum concentrations of AA were
measured for 2 hours following a single oral dose of 20 ug AA/kg bw. As can be seen in Table 5-
6, however, the resulting AUC /mg AA/kg bw of 2.8 jiM AA-hrhuman per mg AA/kg bw is
grossly under the other estimates of from 140 to 246 jiM AA-hrhuman per mg AA/kg bw, based
on hemoglobin adduct levels and various second order rate constants. Hemoglobin adduct levels
were not measured by Kopp and Dekant to help resolve why there is such discordance with other
estimates, and until resolved, the Kopp and Dekant AUC data are not considered sufficient for
use in deriving the human AUC conversion factor.
The other options for a human conversion factor listed in Table 5-6, however, are
sufficient to estimate the AA-AUC per administered dose for use in this toxicological review.
These are based on the Fennell et al. (2005) data for human AA-Val levels per administered dose
divided by various in vivo or in vitro based second order rate constants. There is under a two
fold range in values from the lowest to the highest formation rate (140 to 246), which would
result in a comparable spread of HECs. EPA considers the most relevant and accurate adduct
formation rate constant to be the one derived from a linear regression of the in vivo adduct level
versus AUC data for all male and female mice and rats combined from single dose studies of
Doerge et el. (2005 b,c) and Tareke et al. (2006). The resulting in vivo AA-Val formation rate is
7.5 x 10"6 1/g globin/h, and is included in Table 5-5, as well as discussed in greater detail in
Appendix E. The choice of a non-gender specific in vivo formation rate for humans is supported
by the epidemiology results of Hartmann et al. (2009) who did not observe a gender-related
difference in internal exposure and metabolism of AA in a study of a nonsmoking general
population especially designed for an even distribution of age and gender.
The human AA-AUC conversion factor is calculated by dividing the measured human
AA-Val adduct level of 74.7 nmol of AA-Val/g globin/mmol AA/kg bw (Fennell et al., 2005) by
the in vivo AA-Val adduct formation rate of 7.5 x 10"6 1 /g globin/h resulting in a conversion
factor of 140.1 jiM AA-hrhuman per mg AA/kg bw (see Tables 5-5 and 5-6).
213 DRAFT-DO NOT CITE OR QUOTE
-------
74.7nmolAA-Val 7.5jd(T6/ _9.96mMAA-hr
gglobin ' g globin - h ~ mMAAIkgbw
mM AAI kg bw
The above human serum AA AUC of 9.96 mMoles AA-hr /mMoles of AA /kg bw is
converted to 140.1 jiM AA-hr per mg AA/kg bw using the following unit conversions (i.e., the
units in Fennel et al. (2005) are divded by the molecular weight of acrylamide (71.08), and
multiplied by 1000 to convert mMoles to jiMoles):
9.96mMAA-hr ImMAA 0.1401mMAA -hr 1/wwx 140.1wM AA -hrhuman
-x = xlOOO = -
mMAAIkgbw ll.OSmgAA mgAAIkgbw mgAAIkgbw
The HEC based on this conversion factor and the rat AUCpoo from exposure to a BMDL5
of 0.27 mg AA/kg-day is 0.042 mg AA/kg bw:
mgAA v-LAA-D * ATTn f AS 1 \
HEC m -^ — = F344 Rat A UCPOD (uM -hr)+-
_ — -- _ _ ^(jjj \— - / j j 11 i
kg bw mg AA I kg bw
HECinmgAA =5.94uM-hr+ — r±^L = Q,Q42mgAA/kgbw
kg bw mg AA I kg bw
Derivation of the RfD
The HEC adjusted POD of 0.042 mg AA/ kg bw is divided by a total uncertainty factor
(UF) of 30 to derive the RfD: 3 for extrapolation for interspecies toxicodynamic differences
(UFA_TD: animal to human) and 10 for consideration of intraspecies variation (UFH: human
variability).
Total UF =30
= 3 (UFA.TD) x 1 (UFA.TK) x 10 (UFH) x 1 (UFS) x 1 (UFL) x 1 (UFD)
An UF of 3 (101/2= 3.16, rounded to 3) was selected to account for uncertainties in
extrapolating from rats to humans for toxicodynamic differences (UFA-TD)- It is reasonable to
assume that the neuropathic effects observed in rats are relevant to humans since peripheral
neuropathy in humans has been widely associated with occupational (inhalation and dermal)
exposure to AA, and cases of peripheral neuropathy associated with oral exposure have been
reported. Available information is inadequate to quantify potential differences between rats and
214 DRAFT-DO NOT CITE OR QUOTE
-------
humans in the toxicodynamics of orally administered AA. The lack of a mechanistic basis or
any quantitative information on toxicodynamic differences between rats and humans provides
support for the UFA-TD of 3. The equivalent AUC method was used to account for intraspecies
toxicokinetic differences, and thus the UFA-TK = 1 instead of the default value of 3.16 (101/2).
An UF of 10 was used to account for interindividual variability in toxicokinetics and
toxicodynamics to protect potentially sensitive populations and lifestages (UFn). Although male
rats appear to be slightly more sensitive than female rats to AA-induced neurotoxicity and were
the basis of the POD for the RfD, the extent of variation in sensitivity to AA within the human
population is unknown. In the absence of this information, the default value of 10 was selected.
An UF for extrapolating from a subchronic exposure duration to a chronic exposure
duration (UFS) was not needed, because the point of departure was derived from a study with
chronic exposure (i.e., the UFs = 1).
An UF to account for the extrapolation from a LOAEL to a NOAEL (UPi,) was not
applied because the current approach is to address this extrapolation as one of the considerations
in selecting a BMR for BMD modeling (i.e., UFL =1). In this case, EPA concluded a 5% increase
in response, is appropriate for use in deriving the RfD under the assumption that it represents a
minimal biologically significant change.
An UF to account for database deficiency is not necessary (i.e., UFD = 1). The oral
toxicity database for laboratory animals repeatedly exposed to AA is robust and contains two
2-year carcinogenicity/toxicology drinking water studies in F344 rats and numerous shorter-term
oral toxicity studies in animals; two two-generation reproductive toxicity studies, one in F344
rats and one in CD-I mice; several single-generation reproductive toxicity studies involving
prolonged prebreeding drinking water exposure of Long-Evans rats and ddY mice; and several
developmental toxicity studies involving gestational exposure of Sprague-Dawley and Wistar
rats and CD-I mice. The database identifies nerve degeneration as the critical effect from
chronic oral exposure. There are unresolved issues that warrant further research including the
MOA of AA-induced neurotoxicity, the potential for behavioral or functional adverse effects not
detected in the assays to date, and the uncertainty that heritable germ cell effects may occur at
doses comparable to those inducing degenerative nerve lesions with chronic oral exposure.
These issues, however, do not warrant applying an UF for database deficiencies.
Functional neurotoxic deficits have been observed in both animal and human studies, and
at least two MOA precursor events have been proposed (i.e., central nerve terminal damage or
reduction in fast axonal transport). Either of these precursor events might result in other serious
behavioral or functional neurological deficits that were not detected in the bioassays. More
research is needed to further evaluate more subtle irreversible adverse behavioral or functional
effects in humans and laboratory animals. As discussed in Section 4.4, the magnitude of
response at low doses, and the shape of the low dose-response curve for potentially serious
215 DRAFT-DO NOT CITE OR QUOTE
-------
heritable germ cell effects, is also a research need. Some of these data needs are currently being
addressed.
The RfD for AA was calculated as follows:
RfD = HEC •*• UF
= 0.042 mg/kg-day ^-30
= 0.001 mg/kg-day (rounded to one significant digit)
5.1.4. Previous RfD Assessment
This RfD replaces the previous RfD for AA of 0.0002 mg/kg-day entered into the IRIS
database on September 26, 1988. The previous RfD was based on nerve damage (NOAEL of
0.2 mg/kg-day; LOAEL of 1 mg/kg-day) observed in a rat subchronic drinking water study
(Burek et al., 1980). The RfD was derived by dividing the NOAEL by an UF of 1,000: 10 for
uncertainty in extrapolating from animals to humans, 10 for intrahuman variability, and 10 for
uncertainty in extrapolating from a subchronic to a chronic exposure. The new RfD is based on
a more recent chronic exposure studies (Friedman et al., 1995; Johnson et al., 1986), as well as
current methodology for characterizing the dose-response curve, for determining the POD (i.e.,
the BMDL), and for deriving the HEC.
5.2. INHALATION REFERENCE CONCENTRATION (RfC)
5.2.1. Choice of Principal Study and Critical Effect—with Rationale and Justification
As discussed in Section 4.5.2, neurological impairment is a well-established human
health hazard associated with acute and repeated occupational exposure involving inhalation of
airborne AA and dermal contact with AA-containing materials. Studies describing reliable
relationships, however, between exposure concentrations and neurological responses in humans
or animals are not available. Two cross-sectional health surveillance studies of AA-exposed
workers describe correlative relationships between hemoglobin adduct levels of AA (an internal
measure of dose) and changes in a neurotoxicity index based on self-reported symptoms and
clinical measures of neurological impairment (Calleman et al., 1994) or increased incidences in
self-reported symptoms of neurological impairment and eye and respiratory irritation (Hagmar et
al., 2001). These studies, however, provide limited information on dose-response relationships
for chronic inhalation exposure to AA, because they involved mixed inhalation and dermal
exposure (in both groups of workers, dermal exposure was thought to have been substantial), the
duration of exposure was less than chronic, workers in both studies were exposed to confounding
chemicals (acrylonitrile in the first study and NMA in the second), and the internal measure of
dose (N-terminal valine adducts of hemoglobin) is not specific for AA alone (i.e., NMA can
form the same adduct). Although these data are limited, EPA did derive an RfC from the
Calleman et al (1994) data for comparison purposes (see Section Appendix F). The preferred
216 DRAFT-DO NOT CITE OR QUOTE
-------
derivation for the RfC, however, is based on a route-to-route extrapolation directly from the oral
exposure POD.
The justification for deriving an RfC directly from the oral exposure POD used as the
basis for the RfD includes: (1) a well characterized dose-response and identification of the most
sensitive noncancer endpoint from an adequate database of oral exposure studies; 2) considerable
evidence from occupational experience that dermal and inhalation exposures to AA induce
peripheral neuropathies, including development of the types of degenerative lesions observed in
nerves of rats exposed via drinking water; (3) evidence of rapid, nearly complete absorption
from the oral route and rapid distribution throughout the body (Kadry et al., 1999; Miller et al.,
1982); 4) evidence that the elimination kinetics of radioactivity from oral or i.v. administration
of radiolabeled AA in rats is similar (Miller et al., 1982); 5) similar flux of AA through
metabolic pathways following either single dose oral or single 6 hr inhalation exposures in rats
(Sumner et al., 2003); 6) some route-to-route differences in the relative amounts of AA to GA,
however, the differences are within two fold of each other; and 7) lack of support for portal of
entry effects.
In the only animal inhalation kinetic study (i.e, no human inhalation kinetic information
is available) Sumner et al. (2003) report a statistically significantly larger percentages of urinary
metabolites associated with GA formation following an inhalation exposure compared with an
i.p. and gavage exposure. GA-Val levels are also higher and AA-Val levels lower (as indicators
of serum AUCs), following the single 6 hr inhalation exposures versus the single gavage dose in
rats, however, statistical significance was not reported for the adduct level differences, and the
numbers are within two fold of each other. Doerge et al. (2005b, 2005c) report an increased
percentage of GA formation observed in mice and F344 rats from a gavage or dietary exposure
compared to an i.v. exposure that, in conjunction with the Sumner et al. (2003) results, indicate
that there is first pass metabolism in the lungs following an inhalation exposure similar to the
first pass metabolism in the liver from an oral exposure, but apparently the lungs may have a
larger percent of oxidative metabolism of AA to GA.
Although in this only available inhalation kinetic study by Sumner et al. (2003) there do
appear to be some route-to-route differences in the relative amounts of AA to GA, the
differences are within two fold of each other, and the metabolic paths and total disposition are
similar, supporting the derivation of the RfC based upon the oral POD used as the basis for the
RfD.
The level of AA in the air that would result in a comparable intake to the oral exposure POD is
based on a 70 kg person who breathes 20 m3 of air/day. The benchmark response (BMR)
predicted to affect 5% of the population, BMR5, was selected for the point of departure (POD).
A BMR of 5% extra risk was selected for the following reasons: (1) this effect level is
considered to be a minimal biologically significant change given the critical effect of
217 DRAFT-DO NOT CITE OR QUOTE
-------
degenerative nerve changes; (2) the BMDL5 remained near the range of observation; and (3) the
5% extra risk level is supportable given the relatively large number of animals used in the
principal studies.
5.2.2. Methods of Analysis — Including Models (BMD, equivalent AUCs, in vivo rate
constants, etc.)
See Section 5.1 for derivation of the chronic oral RfD for A A, Section 3.5 for a
discussion of the use of hemoglobin adducts and AUCs to derive an HEC, and Appendix E for
details on the model used to estimate of in vivo second order rate constants for the formation of
hemoglobin adducts.
5.2.3. RfC Derivation — Including Application of Uncertainty Factors
The BMDLs for degenerative nerve lesions in male rats exposed to AA in drinking water
for 2 years is taken as the POD for deriving the RfC. The internal dose metric remains the AUC
of AA in male rat blood, and the human equivalent daily inhalation intake required to produce
that same AUC value in human blood would be 0.042 mg/kg-day. The air concentration that
would provide a 70 kg person who breathes 20 m3 of air that amount of daily exposure is 0.15
mg/ m3.
Air Concentration HEC_POD = 0.042 mg I kg - day x 70 kg -=- - - =0.15 mg I m3
20m
This HEC-adjusted POD for a continuous inhalation exposure of 0.15 mg/m3 is divided
by a total UF of 30: 3 for extrapolation for interspecies toxicodynamic differences
animal to human) and 10 for consideration of intraspecies variation (UFn: human variability).
Total UF = 30
= 3 (UFA.TD) x 1 (UFA.TK) x 10 (UFH ) x 1 (UFS) x 1 (UFL) x 1 (UFD)
An UF of 3 (101/2 = 3.16, rounded to 3) was selected to account for uncertainties in
extrapolating from rats to humans for toxicodynamic differences (UFA-TD). It is reasonable to
assume that the neuropathic effects observed in rats are relevant to humans since peripheral
neuropathy in humans has been widely associated with occupational (inhalation and dermal)
exposure to AA, and cases of peripheral neuropathy associated with oral exposure have been
reported. Available information is inadequate to quantify potential differences between rats and
humans in toxicodynamics of orally administered AA. The lack of a mechanistic basis or any
quantitative information on toxicodynamic differences between rats and humans provides
support for the UFA-TD of 3 . The equivalent AUC method was used to account for intraspecies
toxicokinetic differences, and thus the UFA.TK = 1 instead of the default value of 3.16 (101/2).
2 1 8 DRAFT-DO NOT CITE OR QUOTE
-------
An UF of 10 was used to account for interindividual variability in toxicokinetics and
toxicodynamics to protect potentially sensitive populations and lifestages (UFH). Although male
rats appear to be slightly more sensitive than female rats to AA neurotoxicity and were the basis
of the POD for the RfD, the extent of variation in sensitivity to AA within the human population
is unknown. In the absence of this information, the default value of 10 was selected.
An UF for extrapolating from a subchronic exposure duration to a chronic exposure
duration (UFS) was not needed because the point of departure was derived from a chronic
exposure study (i.e., the UFS = 1).
A UF to account for the extrapolation from a LOAEL to a NOAEL (UFL) was not applied
because the current approach is to address this extrapolation as one of the considerations in
selecting a BMR for BMD modeling (i.e., UFL =1). In this case, EPA concluded a 5% increase in
response, is appropriate for use in deriving the RfD under the assumption that it represents a
minimal biologically significant change.
An UF to account for database deficiency is not necessary for this derivation (i.e., UFo =
1) because an AUC equivalence method was used to conduct the route-to-route extrapolation
based on an oral POD, and the oral POD was based on an adequate database. The oral toxicity
database for laboratory animals repeatedly exposed to AA is robust and contains two 2-year
carcinogenicity/toxicology drinking water studies in F344 rats and numerous shorter-term oral
toxicity studies in animals; two two-generation reproductive toxicity studies, one in F344 rats
and one in CD-I mice; several single-generation reproductive toxicity studies involving
prolonged prebreeding drinking water exposure of Long-Evans rats and ddY mice; and several
developmental toxicity studies involving gestational exposure of Sprague-Dawley and Wistar
rats and CD-I mice. The database identifies nerve degeneration as the critical effect from
chronic oral exposure. There are unresolved issues that warrant further research, including the
MOA of AA neurotoxicity, the potential for behavioral or functional adverse effects not detected
in the assays to date, and the uncertainty that heritable germ cell effects may occur at lower than
previously reported doses. These issues, however, do not warrant applying a UF for database
deficiencies.
The RfC for AA is calculated as follows:
RfC = Air Concentration HEC-POD ^
= 0.15mg/m3-30
= 0.005 mg/m3 (rounded to one significant digit)
5.2.4. Previous RfC Assessment
The previous IRIS assessment did not derive an RfC for acrylamide.
219 DRAFT-DO NOT CITE OR QUOTE
-------
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE AND INHALATION
REFERENCE CONCENTRATION
The following discussion identifies uncertainties in the derivation of the RfD and RfC for
AA. Uncertainties in key aspects of the AA assessment include: (1) the completeness of the
database for identifying potentially adverse effects, (2) the choice of the critical effect and its
relevance for humans, (3) the biological rationale supporting the choice of the dose-response
model and determination of the point of departure (POD), (4) the use of the oral POD to derive
the RfC (i.e., the route-to-route extrapolation, and (5) the uncertainties in the derivation of the
HEC based on human hemoglobin adduct data and second order rate constants.
U.S. EPA has developed default uncertainty factors to account for uncertainties in an RfD
or RfC due to missing or inadequate data (U.S. EPA, 2002, 1994b). The default uncertainty
factors address the following areas of uncertainty: (1) variation in susceptibility among the
members of the human population (i.e., inter-individual or intraspecies variability); (2) in
extrapolating animal data to humans (i.e., interspecies uncertainty); (3) in extrapolating from
data obtained in a study with less-than-lifetime exposure (i.e., extrapolating from subchronic to
chronic exposure); (4) in extrapolating from a LOAEL rather than from a NOAEL; and (5)
associated with extrapolation when the database is incomplete. Uncertainty factors are used in
the derivation of the RfD and RfC to adjust the POD downward and thus reduce the potential
risk of adverse effects to public health.
The specific uncertainty factors used in deriving the AA RfD and RfC were previously
discussed in Sections 5.1.3 and 5.2.3, respectively. A methodology to extrapolate an internal
AUC was available to account for interspecies toxicokinetic differences. Default uncertainty
factors were therefore used to account for toxicodynamic differences when extrapolating the
dose-response relationship from test animals to humans, and to account for intrahuman
variability in toxicokinetics and toxicodynamics to protect susceptible subpopulations.
In the case of AA, the uncertainties in the underlying data and methods used are similar
for the RfD and the RfC since the RfC is based on the same data as the RfD. The following
discussion, therefore, addresses the main areas of uncertainty relevant to both the RfD and the
RfC in Section 5.3.1. Section 5.3.2 provides a more detailed look at the uncertainty factors used
in the derivation of the RfD and RfC. Key points in the discussion are summarized in Table 5-9.
220 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-9. Summary of uncertainty in the acrylamide noncancer risk assessment
Consideration/
Approach
Impact on noncancer risk
estimate
Decision
Justification
Completeness of the
database
Alternative endpoints not
identified in the current
database could t the
estimated risk in humans
from exposure to AA.
The available AA database is
sufficiently robust and adequate to
identify commonly known endpoints
for adverse effects, and to not warrant
aUFD>l.
The animal database is robust and complete by IRIS
assessment standards. Although the human data are limited,
they clearly demonstrate neurotoxicity as the predominant
observable noncancer adverse effect. Although animal
studies for inhalation exposures are limited, kinetic studies in
animals and humans indicate no critical route specific
endpoints. Heritable germ cell effects have been reported in
animal studies at much higher levels of exposure (i.p. or
dermal, 40-125 mg/kg), and further research is warranted to
evaluate the potential for these effects at lower doses.
Selection of the most
sensitive endpoint
relevance to humans
If a more sensitive endpoint
than histological changes
were demonstrated (e.g.,
functional or behavioral
effects, heritable germ cell
effects), there could be an f
in the proposed risk to
humans.
The available data support
neurotoxicity (as determined by
histological changes) as the most
sensitive endpoint.
Limited human data support neurotoxicity as the most
sensitive noncancer endpoint, and this endpoint is well
supported by numerous animal studies. Heritable germ cell
effects have been reported in animal studies at much higher
levels of exposure (i.p. or dermal, 40-125 mg/kg), and
further research is warranted to evaluate the potential for
these effects at lower doses. Other reproductive effects have
been observed in animals, but at levels 3-5 fold higher than
neurotoxic effects, and no reports were identified of
reproductive effects in humans.
Dose-response
modeling
Alternative approaches to
determining a POD could
either f or J, the estimated
risks to humans.
A BMD analysis with mulitple model
choices resulted in adequate fits to the
AA dose-response data and provided
valid estimates of the POD.
A number of BMD models provided reasonable fits to the
AA dose-response data from both bioassays. The model with
the best AIC and the lowest POD were chosen as the basis for
the RfD. There was reasonably good concordance in the
estimated PODs from the best fitting models to the available
chronic bioassay data supporting a relatively hgh degree of
confidence in the BMD approach.
221 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-9. Summary of uncertainty in the acrylamide noncancer risk assessment
Consideration/
Approach
Impact on noncancer risk
estimate
Decision
Justification
Use of the
equivalent AUC
method to estimate
an oral exposure
HEC to derive the
RfD
An alternate approach (e.g..
using default uncertainty
factors) could either f or J,
the estimated risks to
humans.
The AUC method used to estimate the
human equivalent dose used in the
derivation of the RfD.
The development of an HEC based on estimates of the
internal AUC/ mg AA/kg bw provides a better estimate of
interspecies toxiokinetic differences than uncertainty factors,
and is more scientifically supportable. The choice of a non-
gender specific in vivo formation rate for humans is
supported by the epidemiology results of Hartmann et al.
(2009) who did not observe a gender-related difference in
internal exposure and metabolism of AA in a study of a
nonsmoking general population especially designed for an
even distribution of age and gender. Additonal human serum
data and in vivo adduct formation rate data, however, are
needed to reduce uncertainty in the estimate of human AUC
per intake of AA using the equivalent AUC method, or to
develop a PBPK model that would provide additional
capability to evaluate different dose metrics or dosage
regimens.
Estimate of the HEC
(route-to-route
extrapolation) to
derive the RfC
An alternate method (e.g.,
multiple assumptions about
absorption and distribution of
an inhaled dose) could either
t or J, the estimated risks to
humans.
Use an HEC for the inhalation
exposure comparable to the daily
intake level derived using the AUC
method for an oral exposure to derive
an RfC.
Justification for deriving an RfC from the oral RfD is based
on animal kinetic data suggesting some differences in relative
levels of GA and AA between the inhalation and oral route,
but sufficient similarities in metabolic pathways and internal
disposition to support the extrapolation based on the oral
POD. Additional animal or human inhalation kinetic data are
needed to verify the limited available data, and to reduce
uncertainty in the route-to-route extrapolation, as well as to
develop a PBPK model that would provide additional
capability to evalute different dose metrics or dosage
regimens. The alternate RfC based on the Calleman et al.
(1994) data is comparable to the RfC based on the route-to-
route extrapolation, and provides some additional scientific
support for this value.
222 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-9. Summary of uncertainty in the acrylamide noncancer risk assessment
Consideration/
Approach
Impact on noncancer risk
estimate
Decision
Justification
Default uncertainty
factor used to
account for
interspecies
differences in
toxicodynamics
(UFA.TDof3.16;
rounded to 3)
The magnitude of possible
over- or underestimation in
the default uncertainty factor
for interspecies differences in
toxicodynamics could t or J,
the estimated risks to
humans.
The default toxicodynamic
uncertainty factor was used in
conjunction with the AUC method for
deriving an HEC in the derivation of
the RfD and RfC.
The default uncertainty factor for toxicodynamic differences
is used in the absence of adequate chemical or species
specific data to support a more informed extrapolation.
Default uncertainty
factor used to
account for
intrahuman
variability:
UFH = 10
The magnitude of possible
over- or underestimation in
the default factor for
intrahuman differences could
t or J, the estimated risks to
humans.
The default uncertainty factor for
human variability was used.
The default factor for intrahuman variability was used in the
absence of an adequately developed and tested PBPK/PD
model (or other chemical and human data) that would support
a more informed estimate of intrahuman variability.
223 DRAFT-DO NOT CITE OR QUOTE
-------
5.3.1. Areas of Uncertainty
Completeness of the database
The human data for potential noncancer adverse effects from exposure to AA are limited
to occupational case reports for neurological effects following inhalation and/or dermal exposure
(with no data on levels of exposure), two cross-sectional health surveillance studies of
AA-exposed workers that correlate AA-hemoglobin adduct levels and measures of neurological
impairment in AA workers (Hagmar et al., 2001; Calleman et al., 1994), and one kinetic study in
24 human volunteers who were exposed to either a single low-level oral exposure with no
observed toxicity, or to a dermal exposure with adverse effects reported for only one individual
who responded with a mild reversible contact dermatitis (delayed hypersensitivity reaction)
(Fennell et al., 2005). No human studies were identified on the potential for adverse
reproductive or developmental effects from exposure to AA via inhalation or dermal exposure,
and no human repeated oral exposure studies were identified that evaluated any adverse
noncancer effect.
The animal database for repeated oral exposures, however, is robust, and includes two
2-year carcinogenicity/toxicology drinking water studies in F344 rats, numerous shorter-term
toxicity studies in various species, two two-generation reproductive toxicity studies (one in F344
rats and one in CD-I mice), several single-generation reproductive toxicity studies involving
prolonged prebreeding drinking water exposures (in Long-Evans rats and ddY mice), and several
developmental toxicity studies with gestational exposures to dams of Sprague-Dawley rats,
Wistar rats, and CD-I mice. Animal studies for inhalation exposures are limited to three
subchronic studies in cats, dogs, and rats from the mid-1950s (Hazleton Laboratories, 1954,
1953) that report neurotoxicity dependent on the dose and species tested. No chronic animal
inhalation studies for exposure to AA were identified.
With respect to the route of exposure versus the observed adverse effect, animal studies
indicate that AA is rapidly absorbed and distributed when it enters the body from either an oral
or inhalation exposure (Sumners et al., 2003). Moreover, the neurological effects reported in
human occupational studies and case reports following inhalation or dermal exposure are similar
to the effects observed in a broad range of oral exposure animal studies, and neurological effects
appear to be the most sensitive effect (see Section 4). Thus there is good support for the
hypothesis that the neurological effects observed in humans from an inhalation exposure would
likely be observed from an oral exposure that produced a comparable internal level of parent AA
(or metabolite) at an internal target site. As a result, the absence of animal inhalation studies
does not compromise the completeness of the database as it would if the spectrum of effects
were very much different for different routes of exposure.
In summary, there is a substantial animal database to assess the noncancer effects of AA.
The oral toxicity database for laboratory animals repeatedly exposed to AA is robust and
adequate to support the derivation of the RfD, and the validity of conducting a route-to-route
extrapolation from the oral data to derive an RfC is well supported by the available kinetic data.
224 DRAFT-DO NOT CITE OR QUOTE
-------
Although a database deficiency uncertainty factor does not appear to be warranted in the
derivation of the RfD and RfC (UFoB = 1), there are unresolved issues that warrant further
research including the MOA of AA-induced neurotoxicity, the potential for behavioral or
functional adverse effects not detected in the assays to date, and the uncertainty that heritable
germ cell effects may occur at doses lower than those identified in currently available animal
studies. As discussed in Section 4.4, single or repeated i.p. doses of AA or GA ranged from 40
to 125 mg/kg and one study employed five daily dermal applications of AA at 50 mg/kg.
Heritable translocations appeared at high frequency at the lowest doses tested, which indicates
that lower doses may have also elicited heritable translocations. Well-designed animal studies
are needed to assess oral exposure dose-response relationships for AA- and GA-induced
heritable germ cell effects, particularly in the low dose region. Any future studies of possible
associations between AA exposure and sperm characteristics in humans should adjust for
smoking history and alcohol consumption, especially due to the growing evidence of
associations between cigarette smoking and altered sperm endpoints.
Selection of the most sensitive endpoint
The available human and animal data clearly support the choice of neurotoxicity as the
most sensitive endpoint. The human occupational studies and case studies report neurotoxicity,
and both oral exposure animal chronic bioassays report nerve degeneration as the most sensitive
adverse effect.
Reproductive toxicity (e.g., reduced number of live pups per litter) has been observed in
rodent studies, but the no effect level was approximately 3-5 fold higher (i.e., a less sensitive
response) than observed for neurotoxicity. Germ cell effects (e.g., heritable translocations or
mutations, dominant lethals) have also been reported in animal studies, and are a potentially
more serious adverse event than neurotoxicity, because heritable germ cell effects can occur not
only in the exposed individual, but also in their offspring and subsequent generations. Heritable
germ cells effects, however, have only been observed at relatively high levels of AA exposure in
animal studies (40-125 mg/kg, i.p. or dermal) and there are no oral or inhalation exposure
studies examining heritable germ cell effects.
Another area of uncertainty is the possibility that functional or behavioral neurotoxic
endpoints might occur at lower dose levels than the morphological changes that were used as the
measure of neurotoxicity in the animal chronic assays. Functional neurotoxic deficits have been
observed in shorter term animal studies, and in humans occupationally exposed to AA. Two
precursor events have been proposed for the MOA leading to functional neurotoxicity - central
nerve terminal damage and reduction in fast axonal transport. Either of these precursor events
might result in serious behavioral or functional neurological deficits at doses lower than those
needed to produce histologically observable morphological changes. The U.S Food and Drug
Administration is conducting studies to address this issue. If adverse functional changes were, in
fact, determined to occur at dose lower than those for histologically observable nerve tissue
225 DRAFT-DO NOT CITE OR QUOTE
-------
damage, the values of the RfD and RfC could potentially be lower. In addition, the use of a
relatively insensitive measure of neurotoxicity in the chronic principal studies (peripheral
axonopathy detected by light microsopy, as opposed to ultrastructural changes detected by
electron microscopy), raises concern for the possibility that, in looking at axonal degeneration,
preceding terminal degeneration may have been missed, particularly at lower doses.
Dose-response modeling and determination of the point of departure
BMD modeling was used to estimate the POD for the AA RfD. BMD modeling has
advantages over a POD based on a NOAEL or LOAEL because all of the data are used to
characterize the dose-response relationship, and because NOAELs/LOAELs are a reflection of
the particular exposure concentration or dose at which a study was conducted.
All available models in the EPA Benchmark Dose Software (BMDS version 1.3.1) were
fit to the incidence data for microscopically-detected degenerative nerve changes in male and
female F344 rats from the two 2-year drinking water studies (Friedman et al., 1995; Johnson et
al., 1986). The BMR predicted to affect 5% of the population, BMR5 was selected for the POD
rather than the more commonly chosen BMR of 10% for the following reasons (1) the 95%
lower bound of the benchmark dose (BMD), BMDL5, remained near the range of observation;
(2) the 5% extra risk level is supportable given the relatively large number of animals used in the
critical studies; and (3) the use of BMDL5 is consistent with the technical guidance for BMD
analysis (U.S. EPA, 1995).
BMD models provide empirical fits to the dose-response data, and no data or valid
arguments were available to support a biological rationale for selecting one model over the other.
The best model to use for estimating the POD was therefore selected based on Akaike's
Information Criterion (AIC). The AIC is a measure of the goodness of fit of an estimated
statistical model within the context of the complexity of the model, i.e., between models with
comparable fits, the best model is the one with the lowest number of parameters (the simpler
model). Once the model with the lowest AIC score for each data set is identified, the resulting
PODs are compared, and the lowest POD is used to derive the RfD. For AA, the log-logistic
model provided the best fit for the male rat data and resulted in the lowest POD, and was thus
used to derive the RfD in the current assessment. As seen in Table 5-10, all of the final POD
estimates are within twofold of each other, supporting a relatively high degree of confidence that
the estimated BMDL5 in this analysis is a valid estimate of the no effect level for mild
histological changes from a lifetime of exposure as a measure of AA induced neurotoxicity.
With respect to the impact that additional data or a new biological rationale would have on the
rank ordering of the BMD models, there is no way to predict whether the revised estimate of risk
to humans would go up or down.
226 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-10. Estimated POD (mg/kg-day) from best-fitting models for doses
associated with a 5% extra risk for nerve degeneration in male and female
rats exposed to acrylamide in drinking water.
BMD
Model (EDS)
BMDL
(LED5)
Johnson etal. (1986)
Male
Log-logistic 0.58
Female
Probit 0.67
0.27
0.49
Friedman etal. (1995)
Male
Quantal quadratic 0.77
Female
Gamma3 2.25
0.57
0.46
"Restrict power >1
Uncertainty in the animal to human extrapolation A UC method used to estimate of the oral
human equivalent exposure (HEC)
The AUC methodology used to estimate the oral human equivalent concentration (i.e.,
extrapolate the animal dose-response relationship to humans) to derive the RfD is dependent
upon the accuracy of the measured or estimated conversion factors for estimating the rat and
human AUCs /mg AA/kg bw. Currently there is a lack of sufficient data to accurately estimate
human in vivo rate constants for the formation of hemoglobin adducts. The HEC results using a
variety of alternate rate constants to estimate the human AA-AUC (including in vivo constants
based on rat data or constants based on in vitro human data) are reasonably concordant with a
range of values only two fold different from the lowest to the highest estimate, but a wider five
fold range (and thus greater uncertainty) exist for the rate constants and conversion factors
needed to estimate the HECs based on the human GA-AUC. Additional data are clearly needed
for these critical rate constants and conversion factors not only for the derivation of reference
standards, but for the considerable on-going effort to estimate daily intake levels in the general
public based on hemoglobin adduct levels as a biomarker of exposure.
Uncertainty in the route-to-route extrapolation to derive the RfC
A route-to-route extrapolation (oral-to-inhalation) of the dose-response relationship was
performed to derive the RfC based upon the daily intake based on the oral POD. Justification for
deriving an RfC based on the oral POD comes from animal kinetic studies that observed some
differences in relative levels of GA and AA between the inhalation and oral route, but sufficient
227
DRAFT-DO NOT CITE OR QUOTE
-------
similarities in metabolic pathways and internal disposition to the route-to-route extrapolation.
More specifically, there is: 1) considerable evidence from occupational experience that dermal
and inhalation exposures to AA induce peripheral neuropathies, including development of the
types of degenerative lesions observed in nerves of rats exposed via drinking water; 2) evidence
of rapid, nearly complete absorption from the oral route and rapid distribution throughout the
body (Kadry et al., 1999; Miller et al., 1982); 3) evidence that the elimination kinetics of
radioactivity from oral or i.v. administration of radiolabeled AA in rats is similar (Miller et al.,
1982); 4) similar flux of AA through metabolic pathways following either single dose oral or
single 6 hr inhalation exposures in rats (Sumner et al., 2003); 5) some route differences in
relative GA and AA serum levels, but with numbers that are within two fold of each other, and
6) lack of support for portal of entry effects. Additional animal or human inhalation kinetic data
are needed to reduce the uncertainty in quantitating the internal disposition of AA or GA
following different routes of exposure.
The alternate RfC based on the Calleman et al. (1994) data is comparable to the RfC
based on the route-to-route extrapolation, and provides some additional scientific support for this
value.
Since there are no credible default methods to estimate a safe daily inhaled intake level in
the absence of inhalation study data or relevant animal and human kinetic data, the level of
uncertainty in the RfC based on sufficient similarities in the disposition of AA and GA
regardless of exposure route must be compared to the complete uncertainty of having no RfC.
Additional animal or human inhalation kinetic data are needed to reduce the uncertainty in
quantitating potential route differences in disposition.
Use of default factors for the interspecies differences in toxicodynamics in conjunction with the
equivalent A UC method to derive the HEC
The equivalent AUC method to estimate the HEC replaced the default factor for
interspecies toxicokinetic differences of 3 (UFA-TK = 3.16 as a default; UFA-TK = 1 with the
model). A default factor of 3 was used to account for toxicodynamic difference between animals
and humans (UFA-TD of 3; 3.16 rounded down to 37). Thus the overall default factor for
interspecies differences using the model was 3 (UFA = 3 = UFA-TK of 1 x UFA-TD of 3). This
compares to a default factor of 10 without the the use of the AUC method (UFA = 10 = UFA-TK of
3.16 x UFA-TD of 3.16). In the case of AA, using the default approach to derive the RfD8 would
result in a value 0.003 mg/kg-day as the RfD. One interpretation of this similarity is that the
7 The factor of 10 is actually split into the two toxicokinetic and toxicodynamic components by taking the
square root of 10 = 3.16. For convenience when a model is used leaving only the toxicodynamic factor, it is rounded
down to 3.
8 The RfD using the default approach is 0.003 mg/kg/day. RfDdeMtapproach = POD of 0.27 mg/kg/day ^
UFA of 10 - UFH of 10 = 0.0027 mg/kg/day; rounded up to 0.003 mg/kg/day.
228 DRAFT-DO NOT CITE OR QUOTE
-------
interspecies differences for parent AA toxicokinetics might scale roughly to the ratio of body
weight to the % power which for extrapolating between an average rat (250 - 350 grams) and
human (70kg) is approximately a four fold reduction in dose on a mg/kg basis.
How much the default factor over- or underestimates interspecies differences cannot be
determined.
Intrahuman variability
Heterogeneity among humans is another source of uncertainty. In the absence of
AA-specific data on human variation, a default UFH of 10 was used to account for uncertainty
associated with human variation in the derivation of the RfD and RfC. How much the default
factor over- or underestimates human variability cannot be determined.
Subchronic to chronic exposure extrapolation
Chronic oral toxicity studies for AA were available and acceptable for use in the
assessment, precluding the need to use a default factor for extrapolating from a subchronic study
(i.e., UFS = 1).
5.4. CANCER ASSESSMENT
5.4.1. Choice of Study/Data—with Rationale and Justification
As summarized in Section 4.8.1, AA is likely to be carcinogenic to humans based on
findings of increased incidences of thyroid follicular cell tumors (combined adenomas and
carcinomas in either sex), scrotal sac mesotheliomas (males), mammary gland tumors (females)
in two chronic drinking water exposure bioassays with F344 rats (Friedman et al., 1995; Johnson
et al., 1986); increased incidences of skin tumors in SENCAR and Swiss-ICR mice given oral,
i.p., or dermal initiating doses of AA followed by tumor-promoting doses of TPA (Bull et al.,
1984a,b); and increased incidences of lung tumors in strain A/J mice following i.p. injection of
AA (Bull et al., 1984a). In addition, one of the F344 rat chronic drinking water bioassays also
found increased incidences of adrenal pheochromocytomas in males and CNS tumors of glial
origin and oral cavity tumors in females (Johnson et al., 1986).
Human studies provide very limited evidence of AA carcinogenicity (as discussed in
Sections 4.1, 4.8.1, and 4.8.2). No statistically significant increased risks for cancer-related
deaths were consistently found in the cohort mortality studies of AA workers (Marsh et al.,
2007; Swaen et al., 2007). In most case-control studies and prospective studies, no statistically
significant associations were found between frequent consumption of foods with high or
moderate levels of AA and cancer incidence for large bowel, bladder, kidney, renal cell, breast,
colorectal, oral, pharyngeal, esophageal, laryngeal, ovarian, or prostate cancer. One case-control
study reported a slightly increased risk of breast cancer later in life associated with the
consumption of French fries during preschool (Michels et al., 2006), but there is considerable
229 DRAFT-DO NOT CITE OR QUOTE
-------
uncertainty in the accuracy of the exposure assessment methods. Increased risks of
postmenopausal endometrial and ovarian cancer (Hogervorst et al., 2007) and renal cell cancer
(Hogervorst et al., 2008a) with increasing dietary AA intake were reported in prospective studies
of a Dutch population, but estimations of dietary AA levels in foods on the market at baseline in
1986 were based on food samples analyzed since 2001 and questionnaires did not include details
regarding specifics of food preparation. Olesen et al. (2008) reported a significant positive
association between AA-Hb adduct levels in red blood cells and ER+ breast cancer after
adjusting for smoking, but this study is limited by the relatively small number of subjects
(374 cases and 374 controls) and uncertainty regarding extrapolation to lifetime exposure from
AA exposure as assessed by a few months of AA-Hb adduct measurements.
The mechanisms by which AA induces cancer in animals are not fully understood,
however, the weight of the scientific evidence strongly supports a mutagenic MO A, as discussed
in Section 4.8.3.1. An alternative MOA has been proposed for the development of AA-induced
thyroid follicular cell tumors, scrotal sac mesotheliomas, and mammary gland tumors in rats,
however, the available evidence in support of these hypotheses is judged to be inadequate to rule
out human relevance9. Therefore, the cancer dose-response relationships for tumors with
statistically significantly elevated incidences in both of the available rat bioassays (thyroid
tumors in both sexes, mammary gland tumors in females and tunica vaginalis mesotheliomas in
males) are the best available basis for deriving an oral cancer slope factor and inhalation unit risk
for AA.
The two chronic bioassays with F344 rats provide appropriate data to describe dose-
response relationships for tumors induced by chronic oral exposure to AA. Strengths in both
assays include sufficient numbers of animals in control and multiple exposure groups for
statistical analysis of dose-response relationships, histological examinations of most tissues, and
sufficient reporting of experimental details and results. Johnson et al. (1986) reported increased
tumor incidences at sites in females (CNS, oral cavity, uterus, and pituitary) and males
(adrenals), which were reported to not be elevated in the Friedman et al. (1995) bioassay.
However, the Johnson et al. (1986) study had abnormally high CNS and oral cavity tumors in
control males and possible confounding effects from a viral infection. The Friedman et al.
(1995) study was designed to include different dose spacings to support better characterization of
9As discussed in detail in Section 4.8.3.2, the evidence that acrylamide-induced mesotheliomas in male
F344 rats may not be relevant to humans includes observations that acrylamide caused decreased circulating levels
of prolactin in male F344 rats (presumably through dopamine agonist activity at the D2 dopamine receptor); that
chemicals that induce Leydig cell tumors in rats are generally not considered relevant to humans because, unlike rat
Leydig cells, human Leydig cells do not decrease luteinizing hormone receptors in response to decreased prolactin;
and the extent of Leydig cell neoplasia has been linked to the development of malignant mesotheliomas in control
and acrylamide-exposed male F344 rats. However, additional support for this proposal, such as the lack of
mesotheliomas in other animal species exposed to acrylamide, is not currently available. In the absence of
additional support for this proposal, the male rat mesotheliomas are assumed to be relevant to humans.
230 DRAFT-DO NOT CITE OR QUOTE
-------
dose-response relationships in the low-dose region and substantially larger control (n = 204) and
0.1 mg/kg-day male rat (n = 204) groups to increase the statistical power in the study to detect
significantly increased tumor incidence. Although glial tumors of brain and spinal cord were
reported by Friedman et al. (1995) not to be increased, not all of the brains and spinal cords in
the test animals were examined, and seven cases of a morphologically distinctive category of
primary brain tumor described as "malignant reticulosis" were reported but excluded from the
Friedman et al. (1995) analysis of the data. In addition incidences of oral cavity tumors, clitoral
gland adenomas and uterine adenomas were reported not to be increased, but the number of these
tumors was not reported.
Dose-response data from both bioassays were therefore analyzed for potential points of
departure in the derivation of an oral slope factor (see Sections 5.4.2, 5.4.3, and 5.4.4).
5.4.2. Dose-Response Data
As discussed in the previous section, incidence data for tumors in both studies were
evaluated to determine the best basis for the oral slope factor. These data included thyroid
tumors in male and female rats, tunica vaginalis mesotheliomas in male rats, and mammary
gland tumors in females from the Friedman et al. (1995) bioassay; and thyroid tumors in male
and female rats, tunica vaginalis mesotheliomas, and adrenal pheochromocytomas in male rats,
and mammary gland tumors, CNS tumors of glial origin, and oral cavity tumors in female rats
from the Johnson et al. (1985) bioassay. Incidences of tumors with statistically significant
increases in the 2-year bioassays with F344 rats exposed to AA in drinking water are shown in
Table 5-11.
231 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-11. Incidence of tumors with statistically significant increases in
2-year bioassays with F344 rats exposed to acrylamide in drinking water
Reference/tumor type
Johnson et al., 1986; males
Follicular cell adenoma
Tunica vaginalis mesothelioma
Adrenal pheochromocytoma
Johnson et al., 1986; females
Follicular cell adenoma/carcinoma
Mammary adenocarcinoma
Mammary benign
Mammary benign + malignant3
CNS tumors of glial origin
Oral cavity malignant+benign
Uterus adenocarcinoma
Clitoral adenoma, benign
Pituitary gland adenoma
Friedman et al., 1995; males'3
Follicular cell adenoma/carcinoma
Tunica vaginalis mesothelioma0
Friedman et al., 1995; females'3
Follicular cell adenoma/carcinoma
Mammary benign + malignant
Dose (mg/kg-day)
0
1/60
3/60
3/60
1/58
2/60
10/60
12/60
1/60
0/60
1/60
0/2
25/59
3/100
4/102
1/50
7/46
0
-
—
-
-
_
-
-
—
-
—
_
-
2/102d
4/102
1/50
4/50
0.01
0/58
0/60
7/59
0/59
1/60
11/60
12/60
2/59
3/60
2/60
1/3
30/60
—
-
—
-
0.1
2/59
7/60
7/60
1/59
1/60
9/60
10/60
1/60
2/60
1/60
3/4
32/60
12/203
9/204
—
-
0.5
1/59
ll/60e
5/60
1/58
2/58
19/58
21/58
1/60
3/60
0/59
2/4
27/60
5/101
8/102
—
-
1.0
-
—
-
-
_
-
-
—
-
—
_
-
—
-
10/100
21/94e
2.0
7/59e
10/60e
10/60e
5/60f
6/61
23/6 le
29/6 le
9/6 le
8/60e
5/60f
5/5f
32/60f
17/75e
13/75e
—
-
3.0
-
—
-
-
_
-
-
—
-
—
_
-
—
-
23/100e
30/95e
"Incidences of benign and adenocarcinoma were added herein, based on an assumption that rats assessed with
adenocarcinoma were not also assessed with benign mammary gland tumors.
bTwo control groups were included in the study design to assess variability in background tumor responses.
Incidences reported herein are those originally reported by Friedman et al. (1995) and not those reported in the
reevaluation study by latropoulos et al. (1998).
dThe data reported in Table 4 in Friedman et al. (1995) lists one follicular cell adenoma in the second control group;
however, the raw data obtained in the Tegeris Laboratories (1989) report (and used in the time-to-tumor analysis)
listed no follicular cell adenomas in this group. The corrected number for adenomas (0) and the total number (2) of
combined adenomas and carcinomas in the second control group are used in the tables of this assessment.
Statistically significantly (p < 0.05) different from control, Fisher's Exact test.
Statistically significantly (p < 0.05) different from control, after Mantel-Haenszel mortality adjustment.
Sources: Friedman et al. (1995); Johnson et al. (1986).
5.4.3. Dose Adjustments and Extrapolation Method(s)
The current EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a) indicate
that the method used to characterize and quantify cancer risk from a chemical is determined by
what is known about the MOA of the carcinogen and the shape of the cancer dose-response
curve. The dose response is assumed to be linear in the low dose range, when evidence supports
a mutagenic MOA because of DNA reactivity, or if another MOA that is anticipated to be linear
is applicable. The linear approach is used as a default option if the MOA of carcinogenicity is
not understood. (U.S. EPA, 2005a). In the case of AA, there are data available that support a
mutagenic mode of carcinogenic action. Thus, a linear-low-dose extrapolation approach was
used to estimate human carcinogenic risk associated with AA exposure.
232
DRAFT-DO NOT CITE OR QUOTE
-------
Modeling of tumor incidence data from the Friedman et al. (1995) bioassay
Data for both the individual incidence and the incidence of tumor bearing animals in the
Friedman et al. (1995) drinking water bioassays were modeled to derive potential points of
departure for an oral slope factor and inhalation unit risk. For males, the tumor types were
tunica vaginalis mesotheliomas or thyroid follicular cell (adenoma/carcinoma). For females, the
tumor types were mammary gland tumors (malignant and benign combined) or thyroid follicular
cell (adenoma/carcinoma).
Details of the modeling are described in Appendix D. Briefly, the female data were fit
with the multistage model to estimate the BMD, which is the same as the effective dose (ED),
and the 95% lower confidence limit on the BMD, the BMDL (or 95% lower bound of the ED
[LED]). Because male rats in the highest dose group in the Friedman et al. (1995) study showed
early mortalities (75 vs. 53% and 44% in control groups 1 and 2; statistical analysis not
reported), the multistage-Weibull model—which adjusted for early mortality—was fit to the data
for tunica vaginalis mesotheliomas and thyroid follicular cell adenoma and carcinoma, using the
licensed software MULTI-WEIB (KS Crump and Company, Ruston LA). The model includes
two explanatory variables, dose and time to death with tumor, for predicting probability of tumor
occurrence; the mathematical function for dose is a polynomial exponential (i.e., multistage)
function and time to death is described as a Weibull function. Pathology reports for individual
rats in the study (Tegeris Laboratories, 1989) were examined to extract time-to-death and tumor
occurrence data for each animal. The incidence of mortality rate in female rats between the high
dose (49%) and the two control groups (40 and 28%) was similar. Consequently, it was judged
that the multistage-Weibull model would not provide an appreciably different estimate of risk for
either tumor site, and a time-to-tumor modeling approach was not applied.
The POD results for modeling the female mammary tumor and thyroid tumor incidence
data separately are presented in Table 5-12. In addition, the results for considering female rats
with either tumor are also presented in Table 5-12. The rat slope factors corresponding to
mammary tumors and to follicular cell thyroid tumors in female F344 rats were very similar,
0.13 vs. 0.11 (mg/kg-day)"1. The BMR was selected so as to use a low benchmark response
level as a point of departure for a cancer response while maintaining the BMD close to the
empirical data. For the female rat data, the BMR of 10% was chosen for both tumor types when
analyzed separately. Given that there was more than one tumor site, basing the unit risk on one
tumor site may underestimate the carcinogenic potential of AA. The EPA cancer guidelines
(U.S. EPA, 2005a) suggest two approaches for calculating the risks when there are multiple
tumor sites in a data set to assess the total risk. The simpler approach suggested in the cancer
guidelines would be to estimate cancer risk from the incidence of tumor-bearing animals. EPA
traditionally used this approach until the NRC (1994) Science and Judgment document indicated
233 DRAFT-DO NOT CITE OR QUOTE
-------
that evaluating tumor-bearing animals would tend to underestimate overall risk when tumor
types occur in a statistically independent manner. The NRC recommended an approach that
added distributions of the individual tumor incidence to obtain a distribution of the summed risk
for all etiologically different tumor types. Consistent with the 2005 cancer guidelines, both
approaches were considered for this assessment (see Table D-3 for the summed risk of mammary
or thyroid tumors in female F344 rats). The method used to derive the summed risk is as follows:
1. The central tendency or maximum likelihood estimates of unit potency (i.e., risk per unit
of exposure) are estimated by R/BMDR, and the upper confidence limit on the unit risk
estimated by R/BMDLR.
2. The central tendency or maximum likelihood estimates of unit potency (i.e., risk per unit
of exposure estimated as R/BMDR), are summed across the multiple sites.
3. An estimate of the 95% upper bound on the summed unit risk is calculated by assuming a
normal distribution for the individual risk estimates, and deriving the variance of the risk
estimate for each tumor site from its 95% upper confidence limit (UCL), according to the
formula:
95% UCL = MLE + (1.645 x SD)
where 1.645 is the t-statistic corresponding to a one-sided 95% confidence interval and
>120 degrees of freedom, and the standard deviation (SD) is the square root of the
variance of the MLE. The variances are summed across tumor sites to obtain the
variance of the sum of the MLE. The 95% UCL on the sum of the individual MLEs is
then calculated from the variance of the sum of the MLE.
The approach that used a point of departure for the combined incidence data was based
on 20% extra risk, because 20% was the lowest extra risk consistent with the lower end of the
observed data range. The BMD2o is 1.2 mg/kg-day, and the BMDL2o is 0.88 mg/kg-day. For
linear low-dose extrapolation, the rat slope factor associated with this combined risk is
0.2/0.88 (mg/kg-day)"1, or 0.23 (mg/kg-day)"1, approximately two fold higher than either of the
risks estimated from the individual sites (see Appendix D for more details). Both approaches
yielded a similar result when rounded to one significant digit, 0.2 (mg/kg-day)"1.
234 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-12. Points of departure and oral slope factors derived from
Friedman et al. (1995) tumor incidence data for female rats exposed to
acrylamide in drinking water.
Incidence modeled
Mammary tumors
Follicular cell thyroid tumors
Mammary or thyroid tumorsb
BMDRa
(mg/kg-day)
1.2
1.3
1.2
BMDLRa
(mg/kg-day)
0.78
0.94
0.88
Oral Slope Factor
based on rat BMDL
[risk level/BMDL]
(mg/kg-day)"1
1.3 x 10"1
1.1 x 1Q-1
2.3 x 10"1
aR = 10% extra risk for mammary tumors, thyroid tumors; 20% for the incidence of either tumor type.
bTumor-bearing animal method: Individual rats that had more than one of the tumor types were counted only once
(see Table D-l for incidences). For summed risk, EPA used an approach that was consistent with the NRC (1994)
recommendation, resulting in a rat slope factor 0.21 (see Appendix D).
Data Source: Friedman et al. (1995). See Appendix D for derivation of BMDs and BMDLs.
Because of mortality issues in the male rat data, time-to-tumor modeling was used (see
Appendix D). The time-to-tumor results for the male tunica vaginalis mesothelioma (TVM) and
thyroid tumor incidence data evaluated separately or combined are presented in Table 5-13. For
the male rat data, the BMDs and BMDLs were linear with risk in the range of 1-10% risk (see
model output in Appendix D). Consequently, the BMR of 10% was chosen for estimating rat
slope factors. As with the female rats, two methods were considered for estimating total cancer
(see Table D-5 for the summed risk of tunica vaginalis mesotheliomas or thyroid tumors in male
F344 rats). Both approaches (tumor-bearing and summed risk) yielded a similar result for risks
from multiple tumor sites when rounded to one significant digit, 0.3 (mg/kg-day)"1.
Table 5-13. Points of departure derived from Friedman et al. (1995) tumors
incidence data for male F344 rats exposed to acrylamide in drinking water.
Incidence modeled
TVM
Follicular cell thyroid tumors
TVM or thyroid tumorsb
BMDRa
(mg/kg-day)
1.2
0.71
0.70
BMDLRa
(mg/kg-day)
0.75
0.45
0.30
Oral Slope Factor
based on rat BMDL
[risk level/BMDL]
(mg/kg-day)"1
1.3 x 10"1
2.2 x lO"1
3.3 x 1Q-1
aR = 10% extra risk.
bTumor-bearing animal method: Individual rats that had more than one of the tumor types were counted only once
(see Table D-l for incidences). For summed risk, EPA used an approach that was consistent with the NRC (1994)
recommendation, resulting in a rat slope factor 0.32 (see Appendix D).
Data Source: Friedman et al. (1995). See Appendix D for derivation of BMDs and BMDLs.
For linear low-dose extrapolation, the rat slope factor associated with the BMDLio of
0.3 mg/kg-day for combined TVM and thyroid tumor incidence is 0.1/(0.3 mg/kg-day), or
0.33 (mg/kg-day)"1, approximately 50% higher than the risk for just thyroid tumors,
235
DRAFT-DO NOT CITE OR QUOTE
-------
0.22 (mg/kg-day) l and 2.5-fold higher than for testicular tumors, 0.13 (mg/kg-day) l (see
Appendix D for more details).
Modeling of tumor incidence data from the Johnson et al. (1986) bioassay
Incidence data for tumors showing statistically significant elevations in the Johnson et al.
(1985) drinking water bioassays were modeled to derive potential points of departure for an oral
slope factor and inhalation unit risk. For males, the tumor types were tunica vaginalis
mesotheliomas, thyroid follicular cell (adenoma/carcinoma), and adrenal pheochromocytomas.
For females, the tumor types were mammary gland tumors (malignant and benign combined),
thyroid follicular cell (adenoma/carcinoma), CNS tumors of glial origin, and oral cavity tumors
(malignant and benign combined). The data for uterine adenocarcinomas and pituitary gland
adenomas were not analyzed because the statistical significance of the elevated incidences in the
high-dose group was only demonstrated after Mantel-Haenszel mortality adjustment and there is
no clear evidence for a trend for increasing risk with increasing exposure level for the incidence
data for these tumor sites shown in Table 5-11. The data for clitoral adenomas were not
analyzed because the number of tissues examined in each group was very small (n = <5, Table 5-
11).
Details of the modeling are described in Appendix D. The tumor data for each sex and
tumor site were fit with the multistage model to estimate the BMD and the BMDL. During the
last 4 months of the Johnson et al. (1986) study, there were increased mortalities in high-dose
males and females, compared with controls, but no adjustment (e.g., excluding from incidence
denominators animals that died before the time of first appearance of tumors) or special
modeling was done for early mortalities because individual animal data for the time of death
were not available.
The POD results (and oral slope factors) for separately modeling the female mammary,
thyroid, CNS, and oral cavity tumor incidence data, and for the summed risks for multiple tumor
types are presented in Table 5-14. The summed risks were calculated using a method consistent
with the NRC (1994) recommended approach, and discussed in detail in Appendix D. Risks for
tumors in different organs were summed to allow for the possibility that different tumor types
can have different dose-response relationships. Consequently, the modeling for each of the
tumor types was used as a basis for estimating a statistically appropriate upper bound on the
summed risk. This estimate of overall risk describes the risk of developing any combination of
the tumor types considered, not just the risk of developing all tumor types simultaneously.
236 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-14. Points of departure and oral slope factors derived from Johnson
et al. (1986) tumor incidence data for female F344 rats exposed to acrylamide
in drinking water.
Incidence data modeled
Mammary tumors
Follicular cell thyroid tumors
CNS tumors of glial origin
Oral cavity, malignant or benign
BMDRa
(mg/kg-day)
0.44
2.93
1.80
1.80
BMDLRa
(mg/kg-day)
0.30
1.47
1.03
0.99
Mammary or thyroid tumorsb
Mammary, thyroid or CNS tumorsb
Mammary, thyroid, CNS or oral cavity tumorsb
Oral Slope Factor
based on rat BMDL
[risk level/BMDL]
(mg/kg-day)"1
3.4 x 10"1
6.8 x 10~2
1.0 x 10"1
1.0 x 1Q-1
3.8 x 10"1
4.4 x 1Q-1
5.0 x 10"1
aR = 10% extra risk for mammary, thyroid, CNS and oral cavity tumors.
b Summed risk were derived with a method that is consistent with the NRC (1994) recommendation. See Appendix
D for a detailed discussion.
Data Source: Johnson et al. (1986). See Appendix D for derivation of BMDs, BMDLs, and slope factors.
The POD results (and oral slope factors) for modeling of the male tunica vaginalis,
thyroid, and adrenal tumor incidence data separately and for summed risks are presented in
Table 5-15.
Table 5-15. Points of departure and oral slope factors derived from Johnson
et al. (1986) tumor incidence data for for male F344 rats exposed to
acrylamide in drinking water
Incidence data modeled
TVM
Follicular cell thyroid tumors
Adrenal pheochromocytomas
BMDRa
(mg/kg-day)
0.27
2.04
2.55
BMDLRa
(mg/kg-day)
0.16
1.12
1.08
TVM or thyroid tumorsb
TVM, thyroid or adrenal tumorsb
Oral Slope Factor
based on rat BMDL
[risk level/BMDL]
(mg/kg-day)"1
6.1 x 10"1
8.9 x 10~2
9.3 x 10~2
6.7x 10"1
7.1 x 1Q-1
aR = 10% extra risk for mammary, thyroid, and adrenal tumors.
b Summed risk were derived with a method that is consistent with the NRC (1994) recommendation. See Appendix
D for a detailed discussion.
Data Source: Johnson et al. (1986). See Appendix D for derivation of BMDs and BMDLs.
Comparison of modeling results
The results of summing the risks for tumors observed in the two F344 rat bioassays are
very similar as shown in the comparison of rat cancer slope factors listed in Table 5-16.
237
DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-16. Comparison of oral slope factors based on summed risks for tumors at
several sites in two bioassays of F344 rats exposed to acrylamide in drinking water.
Bioassay/sex/tumor sites
Friedman / female / mammary or thyroid
Johnson / female / mammary or thyroid
Johnson / female / mammary, thyroid or CNS
Johnson / female / mammary, thyroid, CNS, or oral cavity
Friedman / male / TVM or thyroid
Johnson / male / TVM or thyroid
Johnson / male / TVM, thyroid, or adrenal
Oral Slope Factor based
on rat BMDL
[risk level/BMDL]
(mg/kg-day)'1
2.1 x 10'1
3.8 x ID'1
4.4 x 10'1
5.0 x 10'1
3.2 x ID'1
6.7 x 10'1
7.1 x ID'1
Note: Oral slope factors (= risk level/BMDL) are used to compare summed risks because the BMDLs in the
summed risk analysis did not all have the same benchmark response level (BMR), i.e., it is difficult to readily rank
order BMDLs with different BMRs.
For tumors showing significantly elevated incidences in both bioassays (mammary or
thyroid tumors in females and TVM and thyroid tumors in males), the oral slope factors based on
the rat BMDLs are within a two-fold range for each sex. The oral slope factor of 6.7 x 10"1
(mg/kg-day)"1 derived from the Johnson et al. (1986) male rat data for the summed incidence of
thyroid tumors or tunica vaginalis mesotheliomas was selected as the best estimate of cancer
risks as this estimate represented effects seen in the most sensitive species and sex among the
various summed risks for tumors that were reproducibly observed in both assays.
This rat oral slope factor of 6.7 x 10"1 (mg/kg-day)"1 is the upper bound on the summed
risk, and can be used to derive a point of departure analogous to the BMDLio as follows:
BMDL,R = Benchmark Response Level/ Rat Oral Slope Factor (i.e., the upper bound on
the summed risk)
BMDLio = 0.1/6.7 x 10'1 (mg/kg-day)'1 = 1.5 x 10'1 mg AA/kg bw/d
This BMDLio of 1.5 x 10"1 mg AA/kg bw/d represents the point of departure (POD)
used as the basis for the human equivalent concentration, and the subsequent derivation of the
human oral slope factor, as discussed in the next sections (Section 5.4.4 and 5.4.5).
23 8 DRAFT-DO NOT CITE OR QUOTE
-------
The corresponding BMDio of 2.38 x 10"1 mg AA/kg bw/d is calculated from the BMR
divided by the sum of the central tendency for risks or throid or TVM tumors in male rats (4.2
10"1 (mg/kg-day)"1; see Appendix D, Table D-10) as follows
= Benchmark Response Level/ Sum of the central tendency for risks
BMLio = 0.1/4.2 x 10'1 (mg/kg-day)'1 = 2.38 x 10'1 mg AA/kg bw/d
5.4.4. Human Equivalent Concentration (HEC) - based on equivalent areas under the
time-concentration curve (AUC) for serum AA or GA.
Section 3.5 and Section 5.3.1 previously demonstrated how an internal dose (area under
the time-concentration curve, AUC) of serum acrylamide or glycidamide in a rat can be derived
for an external exposure based on the relationships among hemoglobin adducts, serum levels,
and administered dose as reported in studies by Doerge et al. (2005 a,b,c), and Tareke et al.
(2006). The administered dose in humans that would produce a comparable internal serum AUC
level in humans (i.e., the human equivalent concentration [HEC]) is then derived based on the
relationship between human hemoglobin adduct data and administered dose reported by Fennell
et al. (2005), and the use of adduct formation rates to relate a human internal AUC to human
hemoglobin adduct levels.
The dose metric used to estimate the HEC for the cancer reference values is the GA-
AUC, since GA is considered to be the putative mutagenic carcinogen. To estimate the internal
GA AUC from a BMDio of 2.38 x 10'1 mg AA/ kg bw/d, and a BMDLio of 1.5 x 10'1 mg AA/ kg
bw/d, the male AUC conversion factor of 15 uM GA-hr /mg AA/kg bw (see Table 5-7) was
chosen as the best factor, based on the Doerge et al. (2005a) 42 day drinking water study adduct
data, and the gender specific in vivo derived rate constants from Tareke et al. (2006) and Doerge
et al. (2005c). The equations used to derive a GA-AUCeMD of 1.44 x 10"4 uM-hr, and a GA-
AUCBMDL-POD of 1.03 x 10"4 uM-hr are as follows uM-hr:
' -hr
F344 Female Rat A UCBMn (uM - hr) = — '"6 — x ™*2_
DML) \ / / -i A A I 1 1
kg bw mg AA I kg bw
= 3.57uM-hr
and
Z7?/MZ7 1 D+ATJ^ (I* 1 \ — '"6 — \5-QuM ~ hrF344rat
F344 Female Rat AUCBMDL_POD (uM - hr) = ;—;— x •
kgbw mgAAIkgbw
= 2.25uM-hr
239 DRAFT-DO NOT CITE OR QUOTE
-------
Deriving the HEC
Similar to the HEC derived for the AA-AUC dose metric discussed in Section 5.3.1, the
HEC derivation here for the GA-AUC dose metric requires a human GA-AUC conversion factor.
This factor is calculated by dividing the measured human GA-Val adduct level of 28.9 nmol of
GA-Val/g globin/mmol AA/kg bw from Fennell et al. (2005) by an in vivo GA-Val adduct
formation rate of 32.5 x 10"6 GA-Val 1 /g globin/h (see Table 5-5) resulting in a conversion
factor of 12.5 jiM GA-hrhuman per mg AA/kg bw (see Table 5-7) as follows:
28. 9 nmol GA-Val 32.5 JclO'6 GA- Val I _ 0.889 mMGA-hr
gglobin gglobin-h mMAAIkgbw
mM AA I kg bw
The GA AUC of 0.889 mM GA-hr /mM AA /kg bw is converted to 12.5 |iM GA-hr / mg
AA/kg bw with the following unit conversions; molecular weight of acrylamide = 71.08,
and multiplied by 1000 to convert mMoles to jiMoles:
Q.8896mM-hrGA ImM AA 0.0125 mMGA-hr 1rtrtrt \2.5uM GA-hr,
x - = - xlOOO =
mMAAIkgbw ll.QKmgAA mgAAIkgbw mgAAIkgbw
The range of values that could be used for the GA-AUC conversion factor is wider (five
fold range; from 12.5 to 60.4) than for acrylamide (two fold), because of the wider difference in
the in vitro and in vivo based second order rate constants for GA-Val adduct formation that
could be used to derive the factor. As in the derivation of the RfD, the more scientifically
supportable adduct formation rate constant is the in vivo rate of 32.5 x 10"6 GA-Val 1 /g globin/h
based on all of the male and female mice and rat data from single dose studies of Doerge et el.
(2005 b,c) and Tareke et al. (2006).
The HECs for the male rat GA-AUCeMD of 3.57 uM-hr, and the GA-AUCBMDL-poo of
2.25 uM-hr, using the 12.5 jiM GA-hr / mg AA/kg bw conversion factor, are 2.85 x 10"1 mg
AA/kg bw and 1.80 x 10"1 mg AA/kg bw, respectively, as follows:
HECRMn in -2 = F344 Rat GA A UCRMn (uM - ,
DJVIL) 7 7 DML) \ s A A I 1 7
kg bw mg AA I kg bw
„„„ . mgAA \2.5uM-hrhuman
HECm4nm—2- = 3.57uM -hr -r- = 2.85x10 mgAA kg bw
DM.L) 77 A A I 1 1 OO
kg bw mg AA I kg bw
240 DRAFT-DO NOT CITE OR QUOTE
-------
and
„„„ . mgAA j-,-,,„,,„„ , ,, , , uMGA-hrhuman
HECBMDL_POD m ° = F344 Rat A UCBMDL_POD (uM -hr)+-
i 7 __.. -_ „ „ BML>L-rUL> \ / j j i i 7
kg bw mg AAI kg bw
. mgAA \2.5uM-hrhuman ,_.,._! A A 11 i.
HECm/m, pnnm—^ = 2.25 uM-hr^ = 1.80x10 mgAAIkgbw
tiMLJL—rULJ 77 A A I 1 1 OO
kg bw mg AA I kg bw
5.4.5. Oral Slope Factor and Inhalation Unit Risk
5.4.5.1. Oral Slope Factor
A linear extrapolation approach is taken based on the assumption that AA likely induces
cancer through a mutagenic MOA at dose levels below the POD. Support for this approach
includes observations of: (1) strong evidence of mutagenicity in somatic cells and male germ
cells from in vivo assays; (2) male-mediated dominant lethal mutations following subchronic
oral exposure at dose levels (2.8 to 13.3 mg/kg-day) in the vicinity of chronic oral dose levels
that induced carcinogenic effects in rats (0.5 to 3 mg/kg-day); (3) initiation of skin tumors
(presumably via a genotoxic action) in mice by short-term exposure to oral doses as low as
12.5 mg/kg-day followed by TPA promotion; (4) metabolism of AA by CYP2E1 to the
DNA-reactive metabolite, GA; (5) following an i.p. dose of AA or GA, DNA adducts of GA
observed in all tissues where tumors have been observed in rats and mice.
The daily intake of AA used to derive an HEC POD as the basis for cancer risks in
humans orally exposed to AA, are the BMD of 2.38 x 10"1 mg AA/ kg bw/d, and the BMDL of
1.50 x 10"1 mg AA/ kg bw/d from the Johnson et al. (1986) date for summed risk of thyroid
tumors or tunica vaginalis mesotheliomas in male F344 rats. The equivalent AUC method was
used to derive an HECBMD of 2.85 x 10"1 mg AA/kg bw and HECBMDLPOD of 1.80 x 10"1 mg
AA/kg. The human oral slope factor is derived by linear extrapolation from the HECBMDL-POD of
1.80 x 10"1 mg AA/kg to the origin, corrected for background, and is calculated as the response
rate (10"1) divided by the HECBMDL resulting in a value of 0.56 [mg/kg-day]"1 (response rate of
0.1 /HECBMDL of 1.80 x 10"1 mg AA/kg bw = 0.56 [mg/kg-dayf1).
With rounding to one significant figure, the human oral slope factor based on the
HECBMDL for a BMR of 10' is 0.6 [mg/kg-day]'1.
The human slope factor for AA should not be used with exposures exceeding 0.24 mg
AA/kg bw/d (an approximate estimate of the BMDio for the summed risk), because above this
level the fitted dose-response model better characterizes what is known about the carcinogenicity
241 DRAFT-DO NOT CITE OR QUOTE
-------
of AA. Age-dependent adjustment factors (ADAFs) combined with age-specific exposure
estimates should be applied to this slope factor when assessing cancer risks to individuals <16
years old or for lifetime exposures that begin in less than 2-year-olds (U.S. EPA, 2005b) [see
Section 5.4.6]. The most current information on the application of ADAFs for cancer risk
assessment can be found at www.epa.gov/cancerguidelines/.
5.4.5.2. Inhalation Unit Risk
No human or animal inhalation cancer dose-response data were available for acrylamide
to directly derive an inhalation unit risk. The IUR was thus derived in a route-to-route
extrapolation of the dose-response relationship (oral-to-inhalation exposure) by converting the
oral daily intake POD developed for the oral slope factor to an equivalent air concentration.
Support for use of the oral daily intake to derive an inhalation unit risk value comes from:
1) a characterized dose-response and indentification of tumor types and incidence from two
chronic oral bioasssays; 2) evidence of rapid, nearly complete absorption from the oral route and
rapid distribution throughout the body (Kadry et al., 1999; Miller et al., 1982); 3) evidence that
the elimination kinetics of radioactivity from oral or i.v. administration of radiolabeled AA in
rats is similar (Miller et al., 1982); 4) similar flux of AA through metabolic pathways following
either single dose oral or single 6 hr inhalation exposures in rats (Sumner et al., 2003); 5) some
route-to-route differences in the relative amounts of AA to GA, however, the differences are
within two fold of each other; and 6) lack of support for portal of entry effects.
In the only animal inhalation kinetic study (i.e, no human inhalation kinetic information
is available) Sumner et al. (2003) report a statistically significantly larger percentages of urinary
metabolites associated with GA formation following an inhalation exposure compared with an
i.p. and gavage exposure. GA-Val levels are also higher and AA-Val levels lower (as indicators
of serum AUCs), following the single 6 hr inhalation exposures versus the single gavage dose in
rats, however, statistical significance was not reported for the adduct level differences, and the
numbers are within two fold of each other. Doerge et al. (2005b, 2005c) report an increased
percentage of GA formation observed in mice and F344 rats from a gavage or dietary exposure
compared to an i.v. exposure that, in conjunction with the Sumner et al. (2003) results, indicate
that there is first pass metabolism in the lungs following an inhalation exposure similar to the
first pass metabolism in the liver from an oral exposure, but apparently the lungs may have a
larger percent of oxidative metabolism of AA to GA. Although in this single study with
inhalation kinetic data, there do appear to be some route-to-route differences in the relative
amounts of AA to GA, the differences are within two fold of each other, and the metabolic paths
and total disposition are similar, supporting the derivation of the inhalation unit risk based upon
the oral POD (i.e., the BMDL).
242 DRAFT-DO NOT CITE OR QUOTE
-------
The inhalation unit risk for AA is based on adult exposures and is derived by dividing
the risk (as a fraction) by the BMDLio which is the 95% lower bound on the exposure associated
with a 10"1 extra cancer risk. In this case, the BMDL is the oral HECBMDL based on the oral
exposure rat bioassay BMDL. The inhalation unit risk represents an upper bound risk estimate
for continuous lifetime exposure without consideration of increased early life susceptibility due
to AA's mutagenic MO A.
The oral HECBMDL is the dose that results in a human internal AUC of GA in blood that is
comparable to that which occurs in rats following an oral exposure to the rat BMDL, in this case
of 1.50 x 10"1 mg AA/kg-day based on the male F344 rats for the summed risk of thyroid tumors
or tunica vaginalis mesotheliomas in the Johnson et al. (1986) study. An equivalent AUC
method ios used to estimate an oral HECBMDL of 1.80 x 10"1 mg AA/kg-day as the POD for the
derivation of the oral slope factor, with a central estimate (HECeMo) of 2.85 x 10"1 mg AA/kg-
day. The calculation used to derive an air concentration that will provide a daily intake
comparable to this oral HECBMDL (as the POD) and the HECeMD (as the central estimate) is
straightforward as shown below, and assumes a continuous 24-hour inhalation exposure for a 70
kg person who breathes 20 m3/day air. The equivalent air concentrations to the HECBMDL is 6.3 x
10"1 mg/m3, and to the HECsMD is 1 mg/m3.
Air Concentration equivalent to the HEC BMDL = \.8x\Q~1 mg/kg-day xlQkg + ^-— = 6.3x\Q~lmg/m3
20m
Air Concentration equivalent to the HECBMD = 2.85x10^ mg / kg -day xlQkg + ^—- = \.0mg /m3
20m
This HECBMDL equivalent air concentration is the lower 95% bound on exposure at a 10"1
response, and is used to derive an inhalation unit risk of 1.6 x 10~4 (jig/m3)"1 as follows:
Inhalation unit risk based on the HECBMDL for a BMR of 10"1 in (jig/m3)"1
= 0.1/6.3 x 10'1 mg/m3= 1.6 x 10~4 (jig/m3)"1
With rounding to one significant figure, the inhalation unit risk is 2 x 10~4 (ug/m3)"1.
As noted in the discussion on the oral slope factor, age-dependent adjustment factors
(ADAFs) combined with age-specific exposure estimates should be applied to this inhalation
unit risk when assessing cancer risks to individuals <16 years old or for lifetime exposures that
begin in less than 2-year-olds (U.S. EPA, 2005b) [see Section 5.3.6].
243 DRAFT-DO NOT CITE OR QUOTE
-------
5.4.6. Application of Age-Dependent Adjustment Factors
Because a mutagenic MOA for AA carcinogenicity is sufficiently supported in laboratory
animals and relevant to humans (Section 4.8.3), and in the absence of chemical-specific data to
evaluate differences in susceptibility, increased early-life susceptibility is assumed and the age-
dependent adjustment factors (ADAFs) should be applied, as appropriate, along with specific
exposure data in accordance with the Supplemental Guidance for Assessing Susceptibility from
Early-Life Exposure to Carcinogens (U.S. EPA, 2005b). The oral slope factor of 0.1 per mg/kg-
day and the inhalation unit risk of 3 x 10"5 per ug/m3, calculated from data for adult exposures,
do not reflect presumed early-life susceptibility for this chemical. Example evaluations of cancer
risks based on age at exposure are given in Section 6 of the Supplemental Guidance.
The Supplemental Guidance establishes ADAFs for three specific age groups. The
current ADAFs and their age groupings are 10 for <2 years, 3 for 2 to <16 years, and 1 for
16 years and above (U.S. EPA, 2005b). The 10-fold and 3-fold adjustments in slope factor are to
be combined with age specific exposure estimates when estimating cancer risks from early life
(<16 years age) exposure to AA. The most current information on the application of ADAFs for
cancer risk assessment can be found at www.epa.gov/ cancerguidelines/. In estimating risk, EPA
recommends using age-specific values for both exposure and cancer potency; for AA, age-
specific values for cancer potency are estimated using the appropriate ADAFs. A cancer risk is
derived for each age group, and these are summed across age groups to obtain the total risk for
the exposure period of interest.
Oral exposure
To illustrate the use of the ADAFs established in the 2005 Supplemental Guidance
(U.S.EPA, 2005b), some sample calculations are presented for three exposure duration scenarios,
including full lifetime, assuming a constant AA exposure of 1 ug/kg-day.
70-year exposure to 1 ug/kg-day AA from ages 0-70:
Age group ADAF Unit risk Exposure Duration
(per ug/kg- concentration adjustment
day) (ug/kg-day)
1 2 years/
70 years
1 14 years/
70 years
>16 years 1 6.0 x 10"4 1 54 years/
70 years
0-<2 years 10 6.0 x 10'4
2-<16 years 3 6.0 x 10"4
1
Partial risk
(per ug/kg-
day)
1.7 x 10
-4
3.6 x 10
-4
4.6 x 10
-4
244
DRAFT-DO NOT CITE OR QUOTE
-------
Total risk = 1 x 10
,-3
Note that the partial risk for each age group is the product of the values in columns 2 thru
5 [e.g., 10 x (6.0 x 10"4) x 1 x 2/70 = 1.7 x 10"4], and the total risk is the sum of the partial risks.
Thus, a 70-year risk estimate for a constant exposure of 1 ug/kg-day is equivalent to a lifetime
unit risk estimate of 1 x 10"3 per ug/kg-day, adjusted for early-life susceptibility, assuming a
70-year lifetime and constant exposure across age groups.
If calculating the cancer risk for a 30-year exposure to a constant AA exposure level of
1 ug/kg-day from ages 0-30, the duration adjustments would be 2/70, 14/70, and 14/70, and the
partial risks would be 1.7 x 10"4, 3.6 x 10"4, and 1.2 x 10"4, resulting in a total risk estimate of
7 x 1Q-4.
If calculating the cancer risk for a 30-year exposure to a constant AA exposure level of 1
ug/kg-day from ages 20-50, the duration adjustments would be 0/70, 0/70, and 30/70, and the
partial risks would be 0, 0, and 2.6 x 10"4, resulting in a total risk estimate of 2.6 x 10"4.
Inhalation Exposure
To illustrate the use of the ADAFs established in the 2005 Supplemental Guidance
(U.S.EPA, 2005b), some sample calculations are presented below for three exposure duration
scenarios assuming a constant AA exposure of 1 ug/m3.
70-year exposure to 1 ug/m AA from ages 0-70:
Age group ADAF Unit risk Exposure
0-<2 years
2-< 16 years
>16 years
Duration
10 2 x 10
-4
2x 10
-4
1 2 x 10
-4
(ug/m3)
1
1
1
2 years/
70 years
14 years/
70 years
54 years/
70 years
Partial risk
(per ug/m ) concentration adjustment (per ug/m )
5.7
1.2 x 10
-4
1.5 x 10
-4
Total risk = 3.3 x 10
-4
Note that the partial risk for each age group is the product of the values in columns 2-5
[e.g., 10 x (2 x 10'4) x 1 x 2/70 = 5.7 x 10"5], and the total risk is the sum of the partial risks.
This 70-year risk estimate for a constant exposure of 1 ug/m3 is equivalent to a lifetime unit risk
estimate of 3 x 10"4 per ug/m3, adjusted for early-life susceptibility, assuming a 70-year lifetime
and constant exposure across age groups.
245
DRAFT-DO NOT CITE OR QUOTE
-------
If calculating the cancer risk for a 30-year exposure to a constant AA exposure level of
3
1 ug/m from ages 0-30, the duration adjustments would be 2/70, 14/70, and 14/70, and the
p
2
partial risks would be 5.7 x 10"5, 1.2 x 10"4, and 4.0 x 10"5, resulting in a total risk estimate of
If calculating the cancer risk for a 30-year exposure to a constant AA exposure level of
1 ug/m3 from ages 20-50, the duration adjustments would be 0/70, 0/70, and 30/70, and the
partial risks would be 0, 0, and 8.6 x 10"5, resulting in a total risk estimate of 9 x 10"5.
Other subgroups that may be more or less susceptible to AAs carcinogenic effects include
people with DNA repair deficiencies (increased sensitivity to mutagenic events), or who have
lower levels or activity of CYP2E1 enzymes due to genetic polymorphisms or age related
developmental differences. Those with lower enzyme activity levels could have potentially
decreased susceptibility to carcinogenicity due to the lower production of the putative mutagen,
the GA active metabolite (see Section 4.8). At present, there are no methods to develop
quantitative adjustments in risk for these potential subpopulations.
5.4.7. Uncertainties in Cancer Risk Values
The following discussion identifies uncertainties associated with the estimated risk of
cancer in humans from exposure to AA, specifically the cancer oral slope factor (CSF) and the
inhalation unit risk (IUR). These uncertainties arise either from incomplete knowledge about
AA's toxic effects and mode of action in humans, or because of insufficient or absent data to
support key steps in the quantitation of risks.
Uncertainties in the AA cancer risk assessment include: (1) the completeness of the
database for identifying AA carcinogenic potential, (2) the choice of the tumor types and their
relevance for humans, (3) the choice of methods for modeling the dose-response relationship and
estimating the cancer risks, (4) the use of the AUC method in deriving the oral slope factor, (5)
derivation of the inhalation unit risk based on the oral POD (i.e, the route-to-route extrapolation),
and (6) the choice of the low-dose linear method of extrapolation from the POD to estimate the
CSF and IUR.
In the case of AA, the uncertainties in the underlying data and methods used to derive the
CSF and IUR are similar since the IUR is based on the same oral dose-response data used to
derive the CSF. The following discussion on uncertainty is therefore applicable to both the CSF
and IUR values. The discussion is accompanied by a summary of the main points in Table 5-15.
246 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-15. Summary of uncertainty in the acrylamide cancer risk assessment
Consideration/
approach
Impact on cancer risk
estimate
Decision
Justification
Completeness of the
database
New data could t or J, the
estimate of risks for AA
induced cancer in humans.
Based on the currently available data,
EPA classified AA as "likely to be
carcinogenic to humans" (U.S. EPA,
2005a)
The available human epidemiology studies as of 2009 provide
limited to inadequate support for definitive statements. Animal
bioassays, however, clearly demonstrate multi-site carcinogenicity,
and provide good support for AA being classified as likely to be
carcinogenic to humans.
Selection of bioassay
Analysis based on
alternative bioasssys or
human data could t or J,
the estimated risks of AA
related cancer in humans.
The Friedman et al. (1995) and
Johnson et al. (1986) studies were
chosen for use in the derivation of the
CSF and IUR.
In the absence of direct human data, the Friedman et al. (1995) and
the Johnson et al. (1986) chronic rat drinking water studies were
the only available cancer bioassays. Uncertainty in the risk values
based on these bioassay arises because there was only one species
tested, data are only available for the oral route of exposure (albeit
the most relevant to humans), and the two studies were not
conducted by completely independent laboratories (i.e., the
primary author of the Friedman et al. [1995] study was also an
author for the]). On-going National Toxicology Program (NTP)
studies will add considerable new chronic bioassay data on tumor
types in rats and mice for both AA and GA (U.S. FDA [2009]).
Selection of tumor
types, and relevance to
humans
A different selection of
tumor types from the
Johnson et al. (1986) study
could t or J, the estimated
risks of AA related cancer
in humans.
Tumor types used in the derivation of
the CSF and IUR included
reproducible and statistically
significant increases in thyroid and
testicular tumors in male rats, and
thyroid, mammary gland, and CNS
tumors in female rats.
The choice of tumor types used in the analysis was based on those
tumor that were consistently observed to increase in both of the
available chronic rat drinking water bioassays. As to relevance to
humans, currently available information indicate that GA directly
alkylates DNA, which is the most likely mutagenic event leading
to tumorigenicity. The basic biology of DNA adduct formation
and subsequent perturbation of gene structure and function is
believed to be similar between test animals and humans. Thus, a
mutagenic MOA for AA related carconogenicity is considered
likely, and is a biologically relevant MOA in humans.
Methods used for the
dose-response modeling
and estimate of cancer
risks.
Alternative approaches to
determining a POD could
either f or J, the
estimated risks of AA
related cancer in humans.
A BMD analysis was used to fit to
the AA dose-response data and
provided valid estimates of the POD.
The BMD approach used to develop the POD is in accordance with
EPA guidance (U.S. EPA, 2005a, 1995). Model and parameter
uncertainty at the BMD was assessed by comparing the BMD with
the BMDL, and indicated a relatively low level of uncertainty in
the model results. The BMD to BMDL ratios reflect a relatively
low level of uncertainty in the model results for these data sets.
EPA cancer guidelines (U.S. EPA, 2005a) were followed to
calculate risks for multiple tumor sites.
247
DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-15. Summary of uncertainty in the acrylamide cancer risk assessment
Consideration/
approach
Impact on cancer risk
estimate
Decision
Justification
Use of the equivalent
AUC method in the
derivation of the CSF
Alternative methods and
additional kinetic data
could t or I the estimate of
risks to humans.
A method to estimate AUCs
normalized to administered dose was
used to estimate the oral human
equivalent concentration in the
derivation of the CSF.
The equivalent AUC method was especially important to
estimating the oral human equivalent concentration in the CSF
derivation because the putative toxin was the AA metabolite, GA.
The default uncertainty factor for interspecies toxicokinetic
differences would not account for differences in the internal levels
of GA, while the AUC method did. The choice of a non-gender
specific in vivo formation rate for humans is supported by the
epidemiology results of Hartmann et al. (2009) who did not
observe a gender-related difference in internal exposure and
metabolism of AA in a study of a nonsmoking general population
especially designed for an even distribution of age and gender.
Additonal human serum data and in vivo adduct formation rate
data, however, are needed to reduce uncertainty in the estimate of
human GA AUC per intake of AA using the equivalent AUC
method, or to develop a PBPK model that would provide
additional capability to evaluate different dose metrics or dosage
regimens.
Rotue-to-route
extrapolation to derive
thelUR
Additional animal or
human inhalation kinetic
data could t or J, the
estimate of risks to
humans.
The oral POD used to derive the CSF
was also used to derive the IUR.
Justification for using the oral HEC and POD to derive the IUR is
based on animal kinetic data suggesting some differences in
relative levels of GA and AA between the inhalation and oral
route, but sufficient similarities in metabolic pathways and internal
disposition to support the extrapolation based on the oral POD.
Additional animal or human inhalation kinetic data are needed to
verify the limited available data, and to reduce uncertainty in the
route-to-route extrapolation, as well as to develop a PBPK model
that would provide additional capability to evalute different dose
metrics or dosage regimens.
Choice of low-dose
extrapolation approach
An low-dose extrapolation
that assumed a nonlinear
dose-respopnse
relationship at lower doses
would likely J, the
estimated risks.
A linear low-dose extrapolation from
the POD was used to estimate the risk
of cancer in humans.
In accordance with the Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005a), a mutagenic MOA prompts the
use of a linear low-dose extrapolation from the POD. The mode of
action discussion in Section 4.8.3 concludes that the majority of
the data support a mutagenic MOA for AA carcinogenicity. An
alternative MOA has been proposed for some of the tumors
observed in the animal bioassays (i.e., disruption of hormone
levels or activity), but data supporting this MOA are limited or
lacking.
248
DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-15. Summary of uncertainty in the acrylamide cancer risk assessment
Consideration/
approach
Impact on cancer risk
estimate
Decision
Justification
Method used to protect
sensitive subpopulations
Alternative methods could
t or I the estimated risk for
susceptible subpopulations.
ADAFs are to be applied to the slope
factors when assessing cancer risks for
less than 16-year-old subpopulations
or for lifetime exposures that begin in
less than 2-year-olds. ADAF's should
only be applied as appropriate and in
conjunction with site specific exposure
information.
Neither the extent of interindividual variability in AA metabolism
nor human variability in response to AA has been well
characterized. The uncertainties associated with this lack of data
and knowledge about human variability can, at present, only be
discussed in qualitative terms, however, EPA has developed age-
dependent adjustment factors (ADAFs) to quantitatively account
for some of the potential differences in age-dependent response to
carcinogens with a mutagenic MOA (U.S. EPA, 2005b).
249
DRAFT-DO NOT CITE OR QUOTE
-------
5.4.7.1. Areas of Uncertainty
Completeness of the database
Uncertainty in the risk assessment due to lack of completeness of the database is
primarily due to the lack of human data. The available human epidemiology studies as of 2007
provide limited to inadequate support for definitive statements. Animal bioassays, however,
clearly demonstrate multi-site carcinogenicity, and provide good support for classifying AA as
"likely to be carcinogenic to humans" (U.S. EPA, 2005a). The uncertainty in the database is
being actively addressed in on-going studies sponsored by US FDA and other national and
international public and private sector organizations. The impact of new data could be to either
increase or decrease the estimate of risks of AA induced cancer in humans.
Selection ofbioassay(s), tumor types, and relevance to humans (i.e., the MO A)
In the absence of direct human data, the most appropriate animal bioassays to use in the
derivation of cancer risk values are chronic (i.e., lifetime) studies in two species of rodents for
the most relevant route of exposure. Only two chronic bioassays were available for AA
exposure via the drinking water, both in the F344 rats (Friedman et al., 1995; Johnson et al.,
1986). Strengths in both assays include sufficient numbers of animals in control and multiple
exposure groups for statistical analysis of dose-response relationships, histological examinations
of most tissues, and sufficient reporting of experimental details and results. Uncertainty in the
choice of bioassay arises because there was only one species tested, data are only available for
the oral route of exposure (albeit the most relevant to humans), and the two studies were not
conducted by completely independent laboratories (i.e., the primary author of the Friedman et al.
[1995] study was also an author for the Johnson et al. [1986]). On-going National Toxicology
Program (NTP) studies at US FDA research laboratories will add considerable new chronic
bioassay data in rats and mice for both AA and GA (U.S. FDA, 2009). The impact of these new
data could be either to increase or decrease the estimate of risks of AA induced cancer in
humans.
Tumor types that were consistently observed to increase in both of the available chronic
rat drinking water bioassays included statistically significant increases in thyroid follicular cell
adenomas or carcinomas in male and female rats, tunica vaginalis testis (i.e., scrotal sac)
mesotheliomas in male rats, and mammary gland tumors (adenomas, fibroadenomas or fibromas)
in female rats. Johnson et al. (1986) reported increased tumor incidences at sites in females
(CNS, oral cavity, uterus, and pituitary) and males (adrenals), which were reported to be not
elevated in the Friedman et al. (1995) bioassay. However, the Johnson et al. (1986) study had
abnormally high CNS and oral cavity tumors in control males and possible confounding effects
from a viral infection. Although glial tumors of brain and spinal cord were reported by Friedman
et al. (1995) not to be increased, not all of the brains and spinal cords in the test animals were
250 DRAFT-DO NOT CITE OR QUOTE
-------
examined, and seven cases of a morphologically distinctive category of primary brain tumor
described as "malignant reticulosis" were reported but excluded from the Friedman et al. (1995)
analysis of the data. In addition incidences of oral cavity tumors, clitoral gland adenomas and
uterine adenomas were reported not to be increased, but the number of these tumors was not
reported. Rice (2005) raised the issue of under-reporting of CNS tumors in the Friedman et al
(1995) study, and this is a significant source of uncertainty. The impact of the new data to be
reported from the NTP studies may resolve this issue of types of animal tumors consistently
induced, however it is not known whether the incidence data will increase or decrease the
estimate of risks of AA induced cancer in humans.
The relevance of the tumor types observed in animals to humans based on a proposed
mode of action was considered in Section 4.8.3. The available limited human data do not
provide any support for AA induced thyroid, mammary, scrotal sac, or brain tumors in humans.
The precise mechanism(s) by which the multi-site carcinogenicity occurs in animal models is not
well-established, however, currently available information indicate that AA and GA covalently
bind and modify proteins, and that the mutagenic events that lead to tumors from exposure to AA
are most likely produced by GA via direct alkylation of DNA. The basic biology of DNA adduct
formation and subsequent perturbation of gene structure and function is believed to be similar
between test animals and humans. Thus, a mutagenic MOA is considered a biologically relevant
MOA in humans. Qualitatively, there is considerable evidence in test animal and mammalian
cells to support the relevance of a mutagenic MOA for AA in humans. Quantitative data are
only available in one in vitro assay that measured mutagenicity directly in human bronchial
epithelial cells (Besaratinia and Pfeifer, 2004). The uncertainty in the MOA and significance of
the animal tumor types to humans will require additional data to resolve. Additional data are
also needed to resolve why only hormonally responsive tissues were observed to have increased
tumors in the Friedman et al. (1995) chronic rat bioassay, whereas GA-DNA adducts have been
observed in a much wider array of tissues.
Methods for the dose-response modeling and estimate of cancer risks
For AA, there is a lack of knowledge about the underlying biology, but extensive
guidance (U.S. EPA, 2005a, 1995) and expert judgment to support a BMD analysis, the choice
of the most appropriate model, BMR, and approach for calculating risks when there are multiple
tumor types. The male rat incidence data (tunica vaginalis mesotheliomas and/or thyroid
tumors) were fit with the multistage-Weibull model that accounts for early mortality because the
highest male dose group in the Friedman et al. (1995) study had increased early mortalities
compared with controls. Mortality rates among high dose and control female rats were similar,
so the female incidence data (mammary gland and/or thyroid tumors) were fit with the
multistage model. During the last 4 months of the Johnson et al. (1986) study, there were
251 DRAFT-DO NOT CITE OR QUOTE
-------
increased mortalities in high-dose males and females, compared with controls, but no adjustment
(e.g., excluding from incidence denominators animals that died before the time of first
appearance of tumors) or special modeling was done for early mortalities because individual
animal data for the time of death were not available. For the benchmark response level (BMR) as
a point of departure for the cancer dose-response, the lowest BMR was selected consistent with a
resulting bench mark dose (BMD) that remained close to the empirical data (U.S. EPA, 1995).
Model and parameter uncertainty at the BMD can be assessed by comparing the BMD, a
central estimate of risk, with the BMDL, which corresponds to the lower statistical confidence
limit of a one-sided 95% confidence interval on the BMD. The multistage modeling of the
Johnson et al. (1986) male rat data yielded a summed incidence (thyroid or TVM tumors)
BMDio of 0.238 mg/kg-day, and BMDLio of 0.150 mg/kg-day, an approximately 1.6 fold
difference. The multistage modeling of the female rat data yielded a summed incidence
(mammary or thyroid or CNS tumors) BMDio of 0.44 mg/kg-day, and BMDLio of 0.32 mg/kg-
day, an approximately 1.4-fold difference. The BMD to BMDL ratios for the Johnson et al.
(1986) male female individual tumor types ranged from the 1.5 to 2.4. These numbers reflect a
relatively low level of uncertainty in the model results for these data sets. The use of the BMD
central estimate would decrease the estimated risk of cancer by decreasing the value of the slope
factor.
EPA cancer guidelines (U.S. EPA, 2005a) suggest two approaches for calculating the
risks when there are multiple tumor sites in a data set to assess the total risk: (1) estimate cancer
risk from the incidence of tumor-bearing animals; and (2) adding distributions of the individual
tumor incidence to obtain a distribution of the summed risk for all etiologically different tumor
types. Both approaches were considered in this assessment. For the Friedman et al. (1995) male
rat data, both approaches (tumor-bearing and summed risk) yielded a similar result for risks from
multiple tumor sites when rounded to one significant digit, 0.3 (mg/kg day)"1. Analysis of the
female rat data with both approaches also yielded a similar result for the cancer slope factor
when rounded to one significant digit, 0.2 (mg/kg-day)"1. The summed risks based on the
Johnson et al. (1986) male and female rat data yielded cancer slope factors ranging from 0.4 to
0.7 (mg/kg-day)"1. The relatively similar results for the Friedman et al. (1995) analysis using
different approaches to calculating risks, and the relatively narrow range of slope factors based
on summed risk for various combinations of male or female tumor incidence in the Johnson et al.
(1986) study increases the confidence in the results. The impact of additional knowledge about
the underlying biological processes or availability of other data sets on the estimated risks of
cancer in humans is unknown, and could either increase or decrease the estimated risks.
Adequacy oftheAUC method for use in deriving the cancer slope factor
The AUC methodology used to estimate the oral human equivalent concentration and
252 DRAFT-DO NOT CITE OR QUOTE
-------
derive the cancer slope factor is dependent upon the accuracy of the measured or estimated
conversion factors used to estimate the rat and human AUCs /mg AA/kg bw. Currently there is a
lack of sufficient data to accurately estimate human in vivo rate constants for the formation of
hemoglobin adducts. The estimated human AA-AUC based upon a variety of alternate rate
constants (including in vivo constants for rats and in vitro constants for human) are reasonably
concordant with a range of values only four fold different from the lowest to the highest
estimate. A wider range (and thus greater uncertainty), however, exist for the rate constants and
conversion factors needed to estimate the human GA-AUC. Addditional data are clearly needed
for these critical rate constants and conversion factors not only for the derivation of reference
standards, but for the considerable effort going into estimating daily intake levels based on
hemoglobin adduct concentrations in the general public.
Uncertainty in the route-to-route extrapolation to derive the IUR
A route-to-route extrapolation (oral-to-inhalation) of the dose-response relationship was
performed to derive the IUR based upon the daily intake from the oral POD. Justification for
deriving an IUR from the oral POD comes from animal kinetic studies that observed some
differences in relative levels of GA and AA between the inhalation and oral route, but sufficient
similarities in metabolic pathways and internal disposition to support the route-to-route
extrapolation. More specifically, there is: (1) a well characterized dose-response and
identification of the tumor type and incidence from two chronic oral bioassays ; (2) evidence of
rapid, nearly complete absorption from the oral route and rapid distribution throughout the body
(Kadry et al., 1999; Miller et al., 1982); 3) evidence that the elimination kinetics of radioactivity
from oral or i.v. administration of radiolabeled AA in rats is similar (Miller et al., 1982); 4)
similar flux of AA through metabolic pathways following either single dose oral or single 6 hr
inhalation exposures in rats, and similar GA-Val levels (an indicator of serum GA AUC)
following an oral or inhalation exposure, although AA-Val levels were somewhat lower
(statistical significance not reported) (Sumner et al., 2003); 5) some route differences in relative
GA and AA serum levels, however the numbers are within two fold of each other, and (6) lack of
support for portal of entry effects. Additional animal or human inhalation kinetic data are
needed to reduce the uncertainty in quantitating the internal disposition of AA or GA following
different routes of exposure.
Choice of low-dose extrapolation approach
The mode of action discussion in Section 4.8.3 concludes that at present, the mechanistic
sequence of events by which AA induces the tumor types observed in the animal studies is not
completely defined, however, the majority of the data, support a mutagenic MO A for AA
carcinogenicity. An alternative MOA has been proposed for some of the tumors observed in the
253 DRAFT-DO NOT CITE OR QUOTE
-------
animal bioassays (i.e., disruption of hormone levels or activity), but data supporting this MOA
are limited or lacking.
In accordance with the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), a
mutagenic MOA prompts the use of a linear low-dose extrapolation from the POD to estimate
the risk of cancer in humans. The value of the cancer slope factor is accompanied with the
caveat that it should not be used with human equivalent exposures greater than those
corresponding to the highest exposure in the male rat bioassay (2.0 mg/kg-day) because above
this level the dose-response relationships of the observed tumor types are not likely to continue
linearly and there are no data to indicate where the nonlinearity would begin to occur. If a new
data or a re-analysis of the extant data were to conclude that the MOA for AA carcinogen!city
was not a mutagenic MOA or that there were nonlinearities (i.e., specifically sublinearities) in
the low level dose-response than the estimated risk of cancer to humans would be decreased.
Conversely, if new cancer incidence data supported a steeper dose-response and a linear low
dose-response relationship, then the estimate of risk would increase.
Human population variability and sensitive subpopulations
Neither the extent of interindividual variability in AA metabolism nor human variability
in response to AA has been well characterized. Factors that could contribute to a range of
human response to AA include variations in CYP450, epoxide hydrolase, or glutathione
transferase levels (or activity) because of age-related, gender, or genetic differences or other
factors including exposure to other chemicals that induce or inhibit enzyme levels, nutritional
status, alcohol consumption, or the presence of underlying disease that could alter metabolism of
AA or antioxidant protection systems. Incomplete understanding of the potential differences in
metabolism and susceptibility across exposed human populations represents a considerable
source of uncertainty. The uncertainties associated with this lack of data and knowledge about
human variability can, at present, only be discussed in qualitative terms, however, EPA has
developed age-dependent adjustment factors (ADAFs) to quantitatively account for some of the
potential differences in age-dependent response to carcinogens with a mutagenic MOA. ADAFs
are to be applied to the slope factors when assessing cancer risks for less than 16-year-old
subpopulations or for lifetime exposures that include early-life exposure (U.S. EPA, 2005b, also
see Section 5.4.6).
5.4.8. Previous Cancer Assessment
A cancer assessment for AA was previously entered into the IRIS database on September
26, 1998. Using the EPA cancer classifications at that time, AA was classified as Group B2, a
probable human carcinogen, based on inadequate human data and sufficient evidence of
carcinogenicity in animals (significantly increased incidences of benign and/or malignant tumors
254 DRAFT-DO NOT CITE OR QUOTE
-------
at multiple sites in both sexes of rats and carcinogenic effects in a series of 1-year limited
bioassays in mice by several routes of exposure). The classification was supported by positive
genotoxicity data, adduct formation activity, and structure-activity relationships to vinyl
carbamate and acrylonitrile. An oral slope factor of 4.5 (mg/kg-day)"1 and a drinking water unit
risk of 1.3 x 10^ (jig/L)"1 were derived using a linearized multistage procedural analysis (extra
risk) of combined incidence data for tumors in the CNS, mammary and thyroid glands, uterus,
and oral cavity in female F344 rats exposed to AA in drinking water for 2 years (Johnson et al.,
1986), with the external AA exposure as the dose metric. The current derivation of the oral
slope factor of 0.6 (mg/kg-day)"1 is based on the summed risks for increased incidence of thyroid
tumors and tunica vaginalis mesotheliomas in male F344 rats exposed to AA in drinking water
for 2 years (Johnson et al., 1986). The dose metric used in the current estimation of the HEC is
GA-AUC rather than the external AA exposure. GA is considered to be the putative toxin for
the hypothesized mutagenic MOA leading to carcinogenicity, and thus a better internal dose
metric to correlate to response than the internal (or external) level of AA. Differences in the
metabolism of GA between rats and humans had a considerable impact on the resulting slope
factors.
The previous inhalation unit risk of 1.3 x 1CT3 (jig/m3)"1 was calculated from the oral data
and an external exposure level of AA, based on the assumption that the tissue distribution of AA
appeared to be quantitatively the same regardless of route of exposure (Dearfield et al., 1988).
This assumption was supported by the data on the distribution of AA following oral or i.v.
administration in rats (Miller et al., 1982). The current inhalation unit risk of 2.0 x 10"4 (jig/m3^1
is based on EPA's subsequent methodology for inhalation dosimetry (U.S. EPA, 1994), a human
equivalent internal level of GA in blood to the POD calculated from the rat oral study data, and
animal toxicokinetic data for AA and GA following different routes of exposure that overall
indicate sufficient similar internal disposition of GA or AA to support a route-to-route
extrapolation. The oral HEC daily intake derived from animal oral exposure data was converted
to the air concentration for a continuous inhalation exposure that would result in the same daily
intake as the basis for the inhalation unit risk.
5.5. QUANTITATING RISK FOR HERITABLE GERM CELL EFFECTS
U.S. EPA's Guidelines for Mutagenicity Risk Assessment (1986) describe procedures for
the qualitative and quantitative assessment of risk of heritable mutations in human germ cells.
Although no studies that directly reported the effects of AA on human germ cells were identified
to support a definitive statement about AA's heritable mutagenic effects, there are sufficient
animal toxicity data and other supporting data (e.g., toxicokinetics, mechanistic studies in germ
and somatic cells) to support the hypothesis that AA is a potential human germ-cell mutagen. In
accordance with the Guidelines, the data is sufficient to prompt both a qualitative and
255 DRAFT-DO NOT CITE OR QUOTE
-------
quantitative assessment of risk. The qualitative assessment of AA's heritable germ cell effects
has been previously discussed in Section 4.4. Presented in Section 5.5 are the results of different
approaches to quantitate AA's potential heritable germ cell effects in humans, along with the
uncertainties in the underlying assumptions. With the caveat concerning the overall uncertainty
in the quantitation, there is further discussion of the estimated incidence of heritable effects
given different exposure scenarios including exposure at the levels of the proposed IRIS
reference values. Finally, there is a discussion of the data needed to reduce uncertainties in the
qualitative and quantitative risk assessment of risk of AA's heritable effects.
5.5.1. Quantitative Approaches
In 1993, a European Commission (EC)/ U.S. EPA workshop was convened to identify the
methodology, data requirements, and mechanistic research that was being used to understand and
quantitate the human health risk for germ cell mutagens from exposure to genotoxins. The
workshop results were published in a special edition of Mutation Research (EC/US EPA
Workshop, 1995), and included four case studies, one of which addressed AA's effects
(Dearfield et al., 1995). AA has, perhaps, more quantitative data on genetic and heritable germ
cell effects than any other chemical under evaluation in the IRIS Program, yet important data
gaps remain that add considerable uncertainty to the human quantitative risk assessment.
Dearfield et al. (1995) summarized the data up to 1995, and evaluated several approaches to
quantitate the human dose-response for AA induced heritable germ cell effects, including a
parallelogram approach, a modified direct approach, and a doubling dose approach. A
discussion of each approach are provided below along with the results, key assumptions, and
uncertainties in those assumptions.
Parallelogram approach
The parallelogram approach was originally formulated by F.H. Sobels (1989, 1982, 1977)
to derive an estimate (corrected by DNA adduct dosimetry) of the risk of chemically-induced
heritable effects in human germ cells. The method consisted of first measuring a common
endpoint in human and test animal somatic cells (such as gene mutation in lymphocytes), and in
test animal germ cells; then extrapolating the test animal somatic to germ cell mutation rate ratio
to estimate the "analogous" mutation rate in human germ cells (which are not directly
measurable). A schematic of the original concept is presented in Figure 5-3. The key
assumption in this approach is that the ratio of the somatic to germ cell mutation rate in the test
animal is the same as the ratio in man for a specified dose range (Waters and Nolan, 1995).
256 DRAFT-DO NOT CITE OR QUOTE
-------
Animal somatic cells
measured mutations
and adducts •
Human somatic
cells
-> measured mutations
Animal germ cells
measured mutations
and adducts
Human germ cells
estimated mutations
Comparisons
Estimates
Figure 5-3. Original parallelogram approaches for estimating risk of
heritable germ cell effects.
Dearfield et al. (1995) evaluated two modification to the original parallelogram approach
for use in quantitating the risk for AA, as presented in Figure 5-4. The first modification
(Figure 5-3a) incorporates somatic in vivo data into the parallelogram approach, since by 1995, it
was possible to measure mutations in somatic cells in vivo, and to determine the relationship
between specific DNA adducts (or other alterations) and outcomes, and whether these
relationships are the same among somatic and germ cells treated in vitro and between in vitro
and in vivo exposures. The technology was also available to determine the relationship between
the applied dose and specific DNA adduct production. A representation of the modifications is
shown in Figure 5-4a. The EC/US EPA workshop participants who evaluated this case study
concluded, however, that the modified parallelogram approach in Figure 5-4a was not relevant
for AA, because AA appeared to act primarily via a clastogenic mechanism (e.g., aneuploidy or
via protein [e.g., protamine] adduction), and aside from specific-locus mutations suggestive of a
point mutation mechanism, there were very few other related data to implement
257
DRAFT-DO NOT CITE OR QUOTE
-------
a)
In Vitro
Somatic In Vivo
Germ (mouse) in vivo
Specific DNA adducts
Measured Mutations
Specific DNA adducts
Measured Mutations
Specific DNA adducts
Measured Mutations
b)
Animal somatic cell
measured potency
(2)
Animal germ cell
measured potency
(1)
(1)
Human somatic cell
measured potency
(2)
-*• Human germ cell
measured potency
Figure 5-4. Two modifications in the parallelogram approach for estimating
risk of heritable germ cell effects from exposure to AA.
the parallelogram approach in Figure 5-4a. Furthermore, there is no representation of human
germ cell effects in this modification, nor was information available at the time that related
specific DNA-adduct formation to a measured mutational outcome, which remains true as of
mid-2009.
A second parallelogram approach shown in Figure 5-4b addresses effects in human germ
cells, and assumes that the mathematical relationship "(2)" between the somatic cell and the
germ cell effect is the same in rodents and humans. It further assumes that the mouse-to-human
somatic cell outcome relationship "(1)" i§ the same as the mouse-to-human germ cell outcome
relationship, and that all three measures of potencies are equivalent. The measured potency, in
each case, is derived from a dose-effect relationship, and for example, could be based on specific
DNA adduct formation. As with the approach in Figure 5-4a, however, the types of data needed
to implement the approach in Figure 5-4b are not available for AA. Specifically, the only
information on AA's effects in human somatic cells, is hemoglobin adduction (Boettcher et al.,
2005; Fennell et al, 2005; Bergmark et al., 1993), and GA induced unscheduled DNA synthesis
in human epithelial cells in vitro (Butterworth et al., 1992). Other deficiencies in the AA
258
DRAFT-DO NOT CITE OR QUOTE
-------
database that preclude implementation of the Figure 5-4b parallelogram approach include: (1)
no comparative endpoints in germ cells to establish a similar biological endpoint dosimetry, and
(2) no standardized procedures to measure potency of effects in human germ cells following
chemical exposure. The parallelogram approach also does not provide a means to estimate
increased incidence of genetic disease(s).
As an alternative to the parallelogram approach, the EC/US EPA workshop participants
determined that enough information was available on AA's heritable effects in mice, and dose-
response relationships to chemical mutagens in general, to support quantitation of heritable germ
cell effects in humans using either a direct approach (or modified direct approach) or a doubling
dose approach.
Uncertainty in the quantitation of heritable germ cell effects
Both of the approaches discussed below are based on a number of assumptions about the
similarity or differences between mice and human responses and variability of critical processes
in the MOA leading to heritable disease. The assumptions as discussed by Ehling (1988)
include:
1) The amount of genetic damage induced by a given type of exposure under a given
set of conditions is the same in the germ cells of mice and humans.
2) The various biological and application factors affect the magnitude of the induced
mutation frequency in similar ways and to similar extents in mice and in humans.
The parallelogram approach (i.e., relationships "(1)" and "(2)" in Figure 5-4b) was then
used to identify data to support estimates of the extrapolation factors for key events in the MOA
leading to genetic diseases that could be used to extrapolate from a mouse dose-response to a
human dose-response. An International Commission for Protection Against Environmental
Mutagens and Carcinogens (ICPEMC) Workgroup in 1993 developed risk extrapolation factors
(REFs) to quantitate risk from exposure to AA, and to extrapolate risk from rodent (e.g., mice)
experimental models to humans (ICPEMC, 1993a):
259 DRAFT-DO NOT CITE OR QUOTE
-------
Parameter REF REF
(FOR AA) (DEFAULT)
Locus specificity 2 2-5
DNA repair variability 0.1 0.1
Metabolic variability 1 1
Dose rate variability 1 0.1
Exposure route 1 0.5
Germ cell stage specificity 1 1
Dose-response kinetics 1 1
Overall REF 0.2
* An REF of 1 indicates equivalency between the animal and human.
There is considerable uncertainty in the above assumptions and risk extrapolation factors.
The REFs for AA shown above are mostly in accordance with default values proposed by the
ICEPMC Taskgroup, as described by Favor et al. (1995). The default REF for locus specificity
(2-5) is based on information that, on average, human genes exhibit higher spontaneous
mutability. The default REF for DNA-repair variability (0.1) is based on information that the
overall human repair capability is 10-100 times more efficient than that of mice. The default
REF for metabolic variability (1) was proposed because, in lieu of sufficient data linking
metabolic and mutagenic heterogeneity in humans, or the lack of animal data, it is impossible to
make an informed approximation of the magnitude of the default factor. The default REF for
dose-rate variability (0.1) is based on information for several chemicals indicating that at least an
order of magnitude reduction in mutations occurred following split dose or chronic dosing. The
default REF for exposure route (0.5) is based on the frequency of i.p. data for genetic studies and
the presumption that the i.p. route is the most efficient route, but for other routes (or where
routes are the same in mice and humans), a higher default REF was recommended (Favor et al.,
1995). The default REF for germ-cell specificity (1) is based on information that
spermatogenesis and spermiogenesis in mice and humans are similar and the protamines of mice
and human spermatozoa have high homology. The default REF for dose-response kinetics (1) is
recommended when the dose-response curve appears to be linear or if no data exist (Favor et al.,
1995). Dearfield et al. (1995) did not clearly explain the basis for recommending deviations
from the default REFs for dose-rate variability and exposure route for AA.
It was assumed in 1995 that any effects seen in germ cells represented an integration of
effects from both the parent AA and its metabolite GA. Although GA has been reported to be as
effective as AA in inducing dominant lethal mutations for similar germ cell stage sensitivity
(post-meiotic), more recent research has demonstrated that GA is a much more potent inducer of
dominant lethal mutations in germ cells (Adler et al., 2000; Generoso et al., 1996) compared to
AA, and is also the primary inducer of DNA-adducts in somatic cells (Besaratinia and Pfeifer,
260 DRAFT-DO NOT CITE OR QUOTE
-------
2005). The dominant lethal effect has been shown to require the conversion of AA to GA, using
wild type and CYP2E1 knockout mice (Ghanayem et al., 2005a, b). The AA REFs specified
above of 1 for metabolic variability and dose-response kinetics (i.e., indicating equivalency),
therefore, may not accurately reflect interspecies toxicokinetic differences for GA production
and the resulting estimated interspecies extrapolation of the external dose to mutation rate
relationship. These uncertainties in the assumptions and data gaps warrant further research to
improve the usefulness of the following quantitative estimates of risk for AA induced heritable
effects.
Direct and modified direct approach
In the "direct approach" to estimating genetic disease rates based on mutation rates, a
dominant mutation and endpoint, such as dominant skeletal or cataract alteration is used. In
contrast, the "modified direct approach" uses a recessive mutation rate to predict dominant
disease rates. A modified direct approach was used for AA based on an estimate of the per locus
mutation rate in the mouse relative to the number of loci in humans capable of mutating to
dominantly expressed disease alleles. Although the value for the number of human loci capable
of mutating to dominantly expressed disease alleles is critical to the derivation of the estimated
risk to exposed humans, this number is not known and was assumed to be 1,000 for dominant
single gene diseases, and 10 for dominant chromosomal diseases (i.e., this assumption represents
another major source of uncertainty in these calculations). The modified direct approach
incorporates these estimates into the following equation to derive the number of new diseases in
offspring descendent from exposed parents (ICPEMC, 1993a,b):
Number of new diseases in the offspring descendent from exposed parents
= REF x Mmouse x Lhuman x D x N where:
Mmouse = induced per locus mutation rate per unit dose exposure estimated in the mouse;
Lhuman = number of loci in humans that mutate to dominant disease alleles; D = exposure dose; N
= number of offspring descendent from exposed parents; REF = risk extrapolation factor (see
above for A A).
Doubling dose approach.
The doubling dose approach does not require a specific estimate of the number of human
loci that mutate to dominant disease alleles as does the modified direct approach. Instead, the
doubling dose approach is based on an estimate of the overall spontaneous mutation frequency in
humans that leads to dominant disease alleles. The doubling dose (DD) is the dose which
induces a mutation rate equal to the spontaneous mutation rate. This dose can be evaluated in
animal studies and extrapolated to humans based on the assumptions discussed above. Dearfield
et al. (1995) state that data for spontaneous mutation rates in humans are more available than the
261 DRAFT-DO NOT CITE OR QUOTE
-------
number of disease associated loci in humans thus making the doubling dose approach preferable
to (i.e., less uncertain than) the modified direct approach. For an estimate of the spontaneous
mutation rate and the spontaneous chromosomal aberration rate in humans, Dearfield et al.
(1995) used numbers developed by UNSCEAR (1986) and Sankaranarayanan (1982) of
1.5 x 10"3 and 6.2 x 10"8, respectively. These mutation frequencies in humans were used in the
following equation (ICPEMC, 1993a,b) to derive the number of new diseases in the offspring
descendent from exposed parents:
Number of new diseases in the offspring =
REF X SponhumanS x D/DD x N
where: REF = risk extrapolation factor (see above discussion of REFs); Sponhumans=
overall spontaneous mutation rate to dominant disease alleles in humans; D = exposure dose; DD
= doubling dose estimated in the mouse (the DD is calculated as the mouse spontaneous rate per
unit dose); and N = number of offspring descendent from exposed parents.
Dearfield et al. (1995) derived a doubling dose in mice based on four data sets (Adler et
al., 1994; Ehling and Neuhauser-Klaus, 1992; Adler, 1990; Shelby et al., 1987) using the
following equation:
DD = Spontaneous mutation rate
Induced mutation rate / unit exposure
As an example using data from Ehling and Neuhauser-Klaus (1992):
DD = 22/248.413 = 53.1 mg/kg
[(6/23,489) - (22 /248,413)] / 100 mg/kg
The other estimates were 1.8 mg/kg, 3.3 mg/kg, and 0.39 mg/kg for the Shelby et al.
(1987), Adler et al. (1994), and Adler (1990) data, respectively. Aside from the wide range of
values derived from the different data sets, a major assumption in these calculations is that the
doubling doses increases linearly with dose. The gene mutation rates are based on a single data
point and no other dose-response data were available in 1995 to suggest a nonlinear response.
Dearfield et al. (1995) note that from an empirical examination of AA data at doses of 100
mg/kg and lower, most of the data from the dominant lethal studies have a linear component
(e.g., based on data from the dermal dominant lethal study), and that the Adler et al. (1994) data
from the control and the 50 and 100 mg/kg doses could be fitted with a linear equation. As an
alternate model, Adler et al.(1994) combined both of their data sets and fit the resulting dose-
response curve with a Weibull model to derive a human DD estimate of about 25 mg/kg based
262 DRAFT-DO NOT CITE OR QUOTE
-------
on a human background translocation frequency of 1.9 per 1,000 newborns (Lyon et al., 1983).
Nevertheless, the accuracy of extrapolation of these high exposure rates to the expected human
exposure scenarios presented in Table 5-16 is another major uncertainty in the calculations.
263 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-16. Heritable genetic risk estimates for humans exposed to acrylamide
Number of induced genetic diseases/106 offspring
Endpoint
Gene mutation
Chromosomal
alterations
Mouse dose,
mg/kg
(dose schedule)
100a (single)
200b (5 x 40)
50C (single)
250d (5 x 50)
Combinedc'd
Approach
Doubling dose
Modified direct
Doubling dose
Modified direct
Doubling dose
Modified direct
Doubling dose
Modified direct
Doubling dose
Doubling
dose,
mg/kg
53.1
1.8
3.3
0.39
25
Ingestion
1.3 x 1Q-5
mg/kg-
day
7.3 x 10~5
4.3 x IQ-3
3.0 x 10~2
3.1 x 10~2
1.7 x 10~3
2.7 x 10~3
1.4 x 1Q-2
2.3 x 10~2
2.2 x IQ-4
0.027
mg/kg-day
OSHA
PEL
0.15
9.0
6.3
64.4
3.4
6.0
29.1
47.2
0.45
Inhalation
0.00072
mg/kg-day
grout
worker
0.004
0.24
0.17
1.7
0.09
0.15
0.78
1.3
0.01
Dermal
0.011
mg/kg-day
grout
worker
0.062
3.7
2.6
26.3
1.4
2.3
11.8
19.2
0.18
0.016
mg/kg-day
grout
worker
0.09
5.3
3.7
38.2
2.0
o o
J.J
17.2
28.0
0.27
0.13
mg/kg-day
grout
worker
0.73
43.4
30.3
310
16.5
27.0
140
221
2.2
"Ehling and Neuhauser-Klaus (1992).
bShelbyetal. (1987).
cAdleretal. (1994).
dAdler(1990).
Source: Dearfieldetal. (1995).
264
DRAFT-DO NOT CITE OR QUOTE
-------
Quantitative assessment for various exposure routes and levels
The results of the Dearfield et al. (1995) quantitative analysis for risk of heritable germ
cell effects from different routes and levels of exposure are presented in Table 5-16. In these
derivations, N is set at 1 million (1 x io6), the total REF is set to 0.2, and a range of values are
presented using the two approaches (modified direct and doubling dose) for each of four mouse
data sets (Adler et al., 1994; Ehling and Neuhauser-Klaus, 1992; Adler, 1990; Shelby et al.,
1987). For example, the estimated risk for heritable mutations that could potentially lead to
induced genetic disease in offspring from fathers exposed to 1.3 x 1CT5 |ig AA/kg-day in
drinking water range from 7.3 x 1CT5 /IO6 offspring for gene mutations leading to disease (using
the doubling dose approach and the Ehling and Neuhauser-Klaus [1992] data) to 3 x 10~2/106 for
chromosomal alterations (using the modified direct approach and the Shelby et al.[1987] data).
The oral exposure level that Dearfield et al. (1995) used was derived from estimates of drinking
water consumption and AA levels in drinking water. By using the Fennell et al. (2005) updated
upper estimate of daily oral exposure to an average adult male based on background hemoglobin
adduct levels (i.e., 1.26 jig/kg-day instead of Dearfield et al.'s [1995] estimate of 1.3 x 10~2
|ig/kg-day), the upper range of the estimated risk for heritable mutations potentially leading to
induced genetic disease would be 3/106 offspring for chromosomal alterations using the modified
direct approach and the Shelby et al.(1987) data. Table 5-16 also presents risk for induced
genetic disease in offspring from fathers exposed via inhalation or dermal exposures in
occupational settings that are considerably higher.
Conclusions on the utility of the quantitation of heritable germ cell effects and identification of
data needs
The quantitation of heritable germ cell effects described in Dearfield et al. (1995) is
based primarily on male translocation data and one gene mutation study, and accounts only for
dominant genetic diseases induced by either gene mutations or chromosomal alterations. The
estimates do not take into account other potential genotoxic mechanisms such as effects in
spermatogonia stem cells, effects in female germ cells, or induction of recessive mutations that
would not appear in the first generation, but could lead to additional adverse effects in
subsequent generations. Thus, the Dearfield et al. (1995) risk estimates may be an underestimate
of the total effects on heritable germ cells.
The uncertainties in the assumptions used to quantitate risks for heritable germ cell
effects (discussed above), however, reduce the utility of the Dearfield et al. (1995) quantitative
results for risk assessment purposes. A National Toxicology Program (NTP) expert panel
(NTP/CERFIR, 2004), charged with evaluating the evidence for AA's adverse reproductive and
developmental effects, reviewed the Dearfield et al. (1995) quantitation of heritable germ cell
265 DRAFT-DO NOT CITE OR QUOTE
-------
effects, and concluded that little weight could be placed on the estimated risks due to the
uncertainties associated with the assumptions employed in the quantitation.
The lack of knowledge about the timing of an AA exposure relative to the most affected
germ cell stage also confound how the results would be used for risk assessment. For example,
short-term exposures that induce mutations in spermatogonia stem cells could result in potential
adverse outcomes (increased risk) for the remainder of a male's reproductive life, while
comparable exposures that induce damage only during the post-meiotic stages of the germ cell
cycle (as reported in most of the studies to-date), would increase risks levels only while the
affected sperm are viable, i.e., before they are reabsorbed and replaced by unaffected sperm. In
this scenario, exposures at earlier stages would result in little, if any, risks. Continuous
exposures would result in some weighted combination of risk depending on the sensitivity of
each germ cell stage to damage.
Given the uncertainties in the current quantitative characterization of heritable germ cell
effects, EPA does not consider the quantitative results from Dearfield et al. (1995) sufficient to
support derivation of a toxicity value. EPA does, however, agree with the NTP Expert Panel
conclusion that, "considering the incidence in treated and control animals of the response
detected for heritable translocations at the lowest dose level tested (40 mg/kg bw/day * 5 days),
it is likely that such effects would occur at lower dose levels" (NTP/CERHR, 2004). Thus,
further research and data are clearly needed to fill the critical data gaps and reduce uncertainties
in the characterization of risks for heritable germ cell effects including gaps in the interspecies
extrapolation factors, in the quantitative relationship between genetic alterations in germ cells
and heritable disease, and in the shape of the low-dose response relationship.
266 DRAFT-DO NOT CITE OR QUOTE
-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD
AND DOSE RESPONSE
6.1. HUMAN HAZARD POTENTIAL
AA (CASRN 79-06-1) has the chemical formula C3H5NO (structural formula CH2=CH-
CONH2) and a molecular weight of 71.08. AA is an odorless, white, crystalline solid at room
temperature with a melting point of 84.5°C. It is soluble in water (2.155 g/mL at 30°C) and is
used in photopolymerization systems, adhesives and grouts, and polymer cross-linking. The
primary commercial use of AA is in the production of polyacrylamides, which are used in the
coagulation process of water treatment; as thickening agents for agricultural sprays,
papermaking, textile printing paste, and consumer products; and as water retention aids. Release
of AA to the environment can occur during the manufacturing process and from polyacrylamide
materials containing residual AA. AA forms during the high-temperature heating of starchy
foods. AA is expected to be highly mobile in water and soils but is not expected to accumulate
in the environment due to fairly rapid physical and biological degradation.
Neurological impairment (including peripheral neuropathy involving nerve tissue
damage) has been repeatedly observed in case reports, and health surveillance studies, as well as
extensive laboratory animal studies clearly establishing this endpoint as a potential human health
hazard associated with acute and repeated occupational exposure via inhalation of airborne AA
or dermal contact with AA-containing materials. There are only a few case reports of similar
effects in humans orally exposed to AA, and the human data are inadequate to develop a
quantitative characterization of the dose-response, however there are many laboratory animal
studies that have quantitatively examined the general toxicity, neurotoxicity, reproductive
toxicity, and developmental toxicity of chronic and less-than-lifetime oral exposure to AA. The
animal studies indicate that microscopically-detected degenerative peripheral nerve changes are
the most sensitive health effect from repeated oral exposure to AA, with LOAELs in chronic rat
studies in the 1-2 mg/kg-day range. Early animal research associated AA functional
neurotoxicity with central and peripheral distal axonopathy and, more specifically, with
histopathologic findings of neurofilamentous accumulations in distal paranodal regions of large
peripheral nerve fibers that appeared to cause local axon swelling and subsequent degeneration
of myelin. Axon degeneration was observed to progress proximally toward the cell body region,
a process known as "dying back." Based on these findings, neurofilaments were thought to be a
target for AA toxicity. Other potential pathways for AA-induced axonopathy include
interference with nerve cell body metabolism and delivery of nutrients to the axon, interruption
of axonal protein transport, disruption of axon cytoskeleton, diminished axolemma
Na+/K+-ATPase activity, and reduction of fast anterograde axonal transport capacity.
267 DRAFT-DO NOT CITE OR QUOTE
-------
Impaired male reproductive performance (i.e., male-mediated implantation losses) has
been observed in laboratory animals orally exposed to AA, and the lower end of the range of
animal oral doses associated with germ cell effects, particularly male-mediated implantation
losses, is close to the lowest exposure levels associated with neurotoxicity in orally exposed
animals. To date, associations between human exposure to AA and reproductive effects
(including possible effects on sperm characteristics) have not been adequately studied.
Studies in mice exposed dermally or by i.p. injection show that AA induces transmissible
genetic damage in male germ cells of mice in the form of reciprocal translocations and gene
mutations. No experiments have studied the potential for AA to induce heritable mutations in
the female germ line. The heritable germ cell effect in male mice is consistent with the extensive
evidence supporting dominant lethal effects in male murine test animals. The main adverse
effects are summarized as follows: (1) AA is mutagenic in spermatozoa and spermatid stages of
the male germ line; (2) in these spermatogenic stages AA is mainly or exclusively a clastogen;
(3) per unit dose, i.p. exposure is more effective than dermal exposure; and (4) per unit dose, GA
is more effective than AA. Since stem cell spermatogonia persist and may accumulate mutations
throughout the reproductive life of males, assessment of induced mutations in this germ cell
stage is critical for the assessment of genetic risk associated with exposure to a mutagen.
Mechanistic proposals have been made for a common MOA for neurotoxic and male
fertility effects (e.g., effects on mounting, sperm motility, and intromission) involving
modifications of kinesin and sulfhydryl groups of other proteins by AA and/or GA and a separate
mechanism for male dominant lethal mutations involving clastogenic effects from AA and/or GA
interactions with protamine or spindle fiber proteins in spermatids and/or direct alkylation of
DNA by GA. As reviewed by LoPachin (2008b), there is potential for cumulative effects from
exposure to AA and other type-2 alkenes (which can produce similar noncancer effects via
common mechanisms of action), since human exposure can be related to environmental pollution
(e.g., acrolein, acrylonitrile), contamination of food (e.g., AA, methyl acrylate), and endogenous
generation (e.g., acrolein, 4-hydroxy-2-nonenal).
In accordance with the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a),
AA is characterized as "likely to be carcinogenic to humansin
This characterization is based on the following findings: (1) chronic oral exposure of
F344 rats to AA in drinking water induced statistically significant increased incidences of
thyroid follicular cell tumors (adenomas and carcinomas combined in both sexes), scrotal sac
mesotheliomas (males), and mammary gland fibroadenomas (females) in two bioassays; (2) oral,
i.p., or dermal exposure to AA initiated skin tumors that were promoted by TPA in SENCAR
and Swiss-ICR mice; and (3) i.p. injections of AA induced lung adenomas in strain A/J mice. In
addition, CNS tumors were found in both of the chronic F344 rat bioassays. The elevation of the
incidence for CNS tumors was significant in the first bioassay and of uncertain statistical
268 DRAFT-DO NOT CITE OR QUOTE
-------
significance in the second. There are no animal data on the carcinogenicity of chronic inhalation
exposure to AA.
The available human studies on potential AA carcinogenicity are for either dietary
exposures—numerous case-control studies and prospective studies—or occupational exposures
from inhalation and/or dermal exposure—two cohort mortality studies with follow-up analyses.
These studies are judged as providing limited or no evidence of carcinogenicity in humans.
Although the precise mechanism(s) by which the multi-site carcinogenicity occurs in
animal models cannot be well-established, currently available information indicates that
mutagenicity plays an important role in AA-induced carcinogenicity. The evidence consists of
AA induced genotoxicity in somatic and germ cells of rodents in vivo and cultured cells in vitro
including gene mutations and some types of chromosomal aberrations (i.e., translocations),
formation of GA-DNA in mammalian somatic cells, the positive mouse lymphoma assay
response, and mutagenicity of GA in short-term bacterial assays. The available data indicate that
the major genotoxic effects of AA may involve covalent modifications of proteins by AA and
GA, and that the mutagenic events that lead to tumors from exposure to AA are most likely
produced by GA via direct alkylation of DNA. Errors in base sequence during DNA replication,
especially for the DNA adduct component, may be involved in the MOA.
An alternative MOA involving altered hormonal responses has also been proposed for the
carcinogenicity of AA, but the available data are insufficient to make a determination as to the
likelihood of this MOA. It should be noted that that AA-induced carcinogenicity may have a
mixed MOA involving a mutagenic component and another component, such as an altered
hormonal response or some as yet unknown MOA.
On-going studies at the U.S. Food and Drug Administration
The US Food and Drug Administration's National Center for Toxicological Research
(NCTR) under the auspices of the National Toxicology Program (NTP), are conducting long-
term carcinogenicity bioassays of carcinogenicity for AA and GA in male and female F344 rats
and male and female B6C3Fi mice. The projected schedule for the NTP Technical Report
Subcommittee approval is by November 2009. NCTR is also conducting a developmental
neurotoxicity study of AA in F344 rats under the auspices of the NTP Program. EPA will
continue to monitor new science to inform future directions.
Suggestions for additional studies
To further resolve if there is dose-concordance and temporal sequence in the mutagenic
MOA, a study could be conducted with the same regimen as in a cancer bioassay with
measurement of gene mutations in the tumor target tissues, employing sampling times that would
establish the temporal induction of mutation. A study that would help resolve the difference
269 DRAFT-DO NOT CITE OR QUOTE
-------
between AA and GA mutagenicity leading to tumors would breed wild type lacl mice with
knockouts for CypIIEl, and evaluate mutations in the target tissue. Additional studies to identify
the types of mutations in oncogenes or tumor suppressor genes from the tumors induced in
rodents by AA (or GA) are needed. A treatment-specific tumor mutational spectrum that
matched the mutational signature of AA/GA would be powerful evidence of a mutagenic MO A,
especially if the mutational signature were developed in the tumor target tissue (e.g., using the
lacl transgene).
Additional studies are warranted to evaluate the potential for hormonal disruption, and
the interaction of hormonal disruption and increased levels of DNA adducts in the tumor bearing
tissues observed in the animal studies. Additional studies are also warranted to further evaluate
the low-dose response relationship for heritable germ cell effects in orally exposed animals and
to examine possible associations between AA exposure and sperm characteristics in humans
(adjusting for smoking history and alcohol consumption). Such studies should reduce
uncertainty in the interspecies extrapolation for the dynamic events in the MOA for heritable
effects, and to improve estimates of the quantitative relationship between genetic alterations in
germ cells and heritable disease.
Because of the direct correlation of hemoglobin adducts to administered dose as a
biomarker of exposure, there is considerable potential to develop more accurate measures of
exposure of acrylamide to the general population, and to develop accurate estimates of the
internal dose of both AA and GA for a given daily intake. These estimates, however, depend
upon accurate values for the formation rate of hemoglobin adducts in humans and test species,
and currently there are limited data to derive in vivo formation rates. Additional studies are
needed in humans and test species that measure all three of the cirtical variables to support
derivation of in vivo foramation rate constants - administered dose, serum and adduct levels
over time following a single dose, and time course data for adduct levels at a sufficiently long
enough time post dosing to allow derivation of adduct elimination rates.
6.2. DOSE RESPONSE
6.2.1. Noncancer/Oral
An increased incidence of degenerative lesions of peripheral nerves was selected as the
critical effect for derivation of the RfD for AA, because the doses associated with this effect in
subchronic and chronic drinking water studies with rats were lower than the lowest doses
associated with other AA-induced noncancer effects in animals, including male-mediated
implantation losses. The two 2-year drinking water bioassays with F344 rats were selected as
co-principal studies for deriving an RfD (Friedman et al., 1995; Johnson et al., 1986), and the
270 DRAFT-DO NOT CITE OR QUOTE
-------
final quantitative RfD value is based on the dose-response data from only the Johnson et al.
study. A BMD analysis of the incidence data for microscopically-detected degenerative nerve
lesions in rats indicated that male rats were slightly more sensitive than female rats in these
studies. The 95% lower confidence limits on estimates of doses associated with 5% extra risk
(BMDLs) for nerve lesions were 0.49 and 0.46 mg/kg-day for female rats and 0.27 and
0.57 mg/kg-day for male rats in the Johnson et al. (1986) and Friedman et al. (1995) studies,
respectively. The lowest of the BMDLs from the Johnson et al. (1986) study (0.27 mg/kg-day
for 5% extra risk for mild-to-moderate lesions) reflects the most sensitive response, and was
selected as the POD for deriving the RfD. A equiavelent AUC method was used to derive an
FffiC based on an internal dose metric of AA-AUC in the blood for a drinking water exposure.
An FffiC of 0.42 mg AA/ kg bw/d was used as the POD. The POD was then divided by a total
UF of 30 (3 for animal-to-human extrapolation to account for toxicodynamic differences; 10 for
intra-individual variability in human toxicokinetics and toxicodynamics) to derive the RfD of
0.001 mg/kg-day.
The overall confidence in this RfD assessment is medium to high based on medium to
high confidence in the studies and medium to high confidence in the database. The animal
database is robust. Although no data were available to characterize the neurotoxic dose-response
relationships from chronic oral exposure in humans, neurotoxicity from inhaled or dermal
occupational exposures to AA are well documented. Two co-principal studies provide adequate
characterization of the dose-response relationship for degenerative nerve lesions from a chronic-
duration oral exposure, and for neurotoxicity as the most sensitive endpoint. There might be
behavioral or functional effects that were not evaluated in these bioassays, and that would result
in lower LOAELs than those for the histological effects used to derive the RfD. There is also
uncertainty as to the dose-reponse relationship for heritable germ cell effects. These two
uncertainties lower the overall confidence in the RfD from high to medium-to-high. There are
ongoing studies sponsored by the NTP and FDA that may address these data needs.
6.2.2. Noncancer/Inhalation
An inhalation RfC for AA was derived based upon the same daily intake as derived for
the oral POD. The internal dose metric (AUC for AA in the blood) from an oral exposure in rat
was used to estimate an oral exposure in humans based upon conversion factors that were
developed from adduct data and serum levels in rats, and adduct data and adduct formation rate
constants in humans. The human equivalent concentration were used to derive both the RfD and
the RfC, assuming continuous inhalation exposure over 24 hours for the later. Results from
studies of occupationally exposed workers are sufficient to firmly establish neurological
impairment as a potential health hazard from inhalation and dermal exposure to AA but are
limited in characterizing the dose-response relationships for inhalation exposure.
271 DRAFT-DO NOT CITE OR QUOTE
-------
The justification for deriving an RfC directly from the oral exposure POD used as the
basis for the RfD includes: (1) a well characterized dose-response and identification of the most
sensitive noncancer endpoint from an adequate database of oral exposure studies; 2) considerable
evidence from occupational experience that dermal and inhalation exposures to AA induce
peripheral neuropathies, including development of the types of degenerative lesions observed in
nerves of rats exposed via drinking water; (3) evidence of rapid, nearly complete absorption
from the oral route and rapid distribution throughout the body (Kadry et al., 1999; Miller et al.,
1982); 4) evidence that the elimination kinetics of radioactivity from oral or i.v. administration
of radiolabeled AA in rats is similar (Miller et al., 1982); 5) similar flux of AA through
metabolic pathways following either single dose oral or single 6 hr inhalation exposures in rats
(Sumner et al., 2003); 6) some route-to-route differences in the relative amounts of AA to GA,
however, the differences are within two fold of each other; and 7) lack of support for portal of
entry effects.
The oral BMDLs was converted to a human equivalent daily intake, and then to an air
concentration that would result in comparable internal level assuming a 70 kg person who
breathes 20 m3/day air. The resulting air concentration of 0.15 mg/m3 was used as a POD. The
POD was then divided by a total UF of 30 (3 for animal-to-human extrapolation to account for
toxicodynamic differences; and 10 for intra-individual variability in human toxicokinetics and
toxicodynamics) to derive the RfC of 0.005 mg/m3.
The human data from the Calleman et al. (1994) study are limited both in terms of
number of subjects and by a number of confounding factors (exposure to acrylonitrile, exposure
via dermal and inhalation), however, they were used to derive an RfC for comparison purposes,
and to support an improved design of future studies. The results of this derivation are presented
in Appendix F.
The RfC is based on a route-to-route extrapolation of the oral exposure data using the
same AUC method to develop an inhalation HEC. The overall confidence in the RfC study is
similar to that for the RfD, with additional uncertainty as to the toxicokinetics for an inhalation
exposure, specifically concerning different internal disposition of AA and GA due to
qualitatively similar but possibly quantitatively different first pass effects in lung versus the
liver. The similar RfC derived from the Calleman et al. (1994) data provide some additonal
confidence in RfC with an overall confidence of medium. Additional kinetic data (e.g., serum
data) or improved estimates of the AA and GA AUC from different exposure routes in humans
or test animals based on hemoglobin adduct levels would help improve the confidence in the RfC
based on an HEC derived from oral data. There is low to medium confidence in the database
because inhalation studies are lacking. The overall confidence in the RfC is medium, i.e., less
than the confidence in the RfD.
272 DRAFT-DO NOT CITE OR QUOTE
-------
6.2.3. Cancer/Oral
Two bioassays with F344 rats provide appropriate data to describe dose-response
relationships for tumors induced by chronic oral exposure to AA. Dose-response data for tumors
observed in both bioassays were analyzed for potential points of departure for deriving an oral
slope factor (see Sections 5.4.2, 5.4.3, and 5.4.4). Linear extrapolation to the origin, corrected
for background, from BMDLs (as POD) was used to derive the oral slope factors. Support for a
linear extrapolation comes from evidence of a mutagenic MOA for AA, including observations
of: (1) strong evidence of mutagenicity from in vitro assays and somatic cells from in vivo
assays; (2) male-mediated dominant lethal effects following subchronic oral exposure at dose
levels (2.8-13.3 mg/kg-day) in the vicinity of chronic oral dose levels that induced carcinogenic
effects in rats (0.5-3 mg/kg-day); (3) initiation of skin tumors (presumably via a mutagenic
action) in mice by short-term exposure to oral doses as low as 12.5 mg/kg-day followed by TPA
promotion; (4) metabolism of AA by CYP2E1 to the DNA-reactive metabolite, GA; and
(5) DNA adducts of GA in various tissues in rats and mice exposed to AA. Although proposals
have been made that AA induction of scrotal sac mesotheliomas in male rats and mammary
gland tumors in female rats may be caused by hormonally based MO As that may not be relevant
to humans, the available evidence in support of these hypotheses is judged to be inadequate.
Oral slope factors were calculated based on summed risks for increased incidence of
tumor types that were reproducibly observed in both of the F344 rat bioassays (mammary or
thyroid tumors in females and TVM and thyroid tumors in males). The resulting slope factors
were all within four fold range across studies, and within a two fold range within studies.
The male rat oral slope factor of 0.6 (mg/kg-day)"1 derived from the Johnson et al. (1986)
male rat BMDL for the summed risk of thyroid tumors or testicular tumors (tunica vaginalis
mesotheliomas) was selected for calculating a POD. The choice was based on the following: 1)
there were reproducible thyroid and TVM tumors in both studies, and 2) this choice represented
the most sensitive species, sex and tumor type among the other mammary, thyroid, CNS, and
TVM tumor summations. The male rat BMDLio of 1.5 x 10"1 that was calculated from the
summed risks for thyroid or TVM tumors was used as the POD for deriving a human oral slope
factor.
The rat BMDLio was converted to a human equivalent BMDLio based on comparable
levels of GA-AUC in blood between the rat and human relative to their respective administered
doses. GA has been shown to be the primary reactive mutagenic agent, and the total amount in
blood is the most appropriate and supportable dose metric to use as a correlate to increased
incidence of tumors. The resulting HECBMDL (1.8 x 10"1 mg/kg-day) at the benchmark response
rate of 0.1 was used to derive a human oral slope factor of 0.6 (mg/kg-day)"1.
Because a mutagenic MOA for AA carcinogenicity is sufficiently supported in laboratory
animals and relevant to humans (Section 3.4.1), and in the absence of chemical-specific data to
273 DRAFT-DO NOT CITE OR QUOTE
-------
evaluate differences in susceptibility, increased early-life susceptibility is assumed and the age-
dependent adjustment factors (ADAFs) should be applied to the slope factor based on specific
exposure data, as appropriate, in accordance with the Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA, 2005b).
The overall confidence in the oral cancer slope factor is medium based on medium
confidence in the principal studies and medium confidence in the database. The principal 2-year
studies (Friedman et al., 1995; Johnson et al., 1986) provided corroborative results for most, but
not all, tumor types. There remain some uncertainties concerning the differences between the
two study tumor types and incidence data, in particular for the CNS tumors, and in the
histopathological interpretation of the male TVMs. The database is also incomplete with only
one animal species tested, and little human data to support AA's carcinogenic potential in
humans. At this time, the preponderance of evidence supports a mutagenic MOA for AA-
induced tumors observed in the F344 rat bioassays (thyroid, mammary gland, and TVMs).
Although an alternate nonmutagenic MOA has been proposed involving hormonal pathway
disruption for tumors specific to F344 rats, supporting data are limited or nonexistent.
Additional MOA data would be useful in this regard.
6.2.4. Cancer/Inhalation
No animal or human cancer data were available to directly derive an inhalation unit risk.
An AA PBTK model is needed that simulates both oral and inhalation first pass effects.
However, studies are available that support a direct extrapolation of the dose-response
relationship from the oral exposure POD using a few assumptions for an average human weight
and daily volume of inhaled air. The extrapolation assumes that an exposure (oral or inhalation)
will yield a comparable internal dose using the internal dose metric of area under the time-
concentration curve for GA in blood. As for the derivation of the oral slope factor, the AA
metabolite, GA, is considered to be the putative mutagen and most directly related to AA's
carcinogenicity.
Support for use of the oral daily intake to derive an inhalation unit risk value comes from:
1) a characterized dose-response and identification of tumor types and incidence from two
chronic oral bioasssays; 2) evidence of rapid, nearly complete absorption from the oral route and
rapid distribution throughout the body (Kadry et al., 1999; Miller et al., 1982); 3) evidence that
the elimination kinetics of radioactivity from oral or i.v. administration of radiolabeled AA in
rats is similar (Miller et al., 1982); 4) similar flux of AA through metabolic pathways following
either single dose oral or single 6 hr inhalation exposures in rats (Sumner et al., 2003); 5) some
route-to-route differences in the relative amounts of AA to GA, however, the differences are
within two fold of each other, and the metabolic paths and total disposition are similar (Sumner
et al., 2003); and 6) lack of support for portal of entry effects.
274 DRAFT-DO NOT CITE OR QUOTE
-------
An inhalation unit risk is derived from the BMDLio, the 95% lower bound on the
exposure associated with an 10"1 extra cancer risk, by dividing the risk (as a fraction) by the
BMDL. The inhalation unit risk thus represents an upper bound risk estimate for continuous
lifetime exposure without consideration of increased early life susceptibility due to AA's
mutagenic MOA.
The inhalation unit risks for AA is based on the air concentration that would produce a
comparable daily intake to that resulting from exposure to the oral HECBMDL as the point of
departure. The oral HECBMDL itself is an estimate of the oral dose of AA for a human that would
result in a level of GA in blood comparable to what was observed in the male F344 rats
following exposure to the rat oral BMDL. An air concentration of 6.3 x 10"1 mg/m3 was
calculated as being the concentration needed to achieve the same daily intake level as wold be
achieved with exposure to the oral H£CBMDL of 1.8 x 10"1 mg/kg-day. The conversion assumes a
continuous 24-hour inhalation exposure for a 70 kg person who breathes 20 mVday air. The air
concentration of 6.3 x 10"1 mg/m3 ios considered the POD to derive the inhalation unit risk of 2 x
10~4 (jig/m3)"1 based on a response level of 10"1.
As noted above, because a mutagenic MOA for AA carcinogenicity is sufficiently
supported in laboratory animals and relevant to humans (Section 3.4.1), age-dependent
adjustment factors (ADAFs) should be applied to the inhalation unit risk based on specific
exposure data, as appropriate, in accordance with the Supplemental Guidance for Assessing
Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA, 2005b).
The overall confidence in the inhalation unit risk is medium based on a medium to high
confidence in the the oral slope factor (see discussion in the previous section) with uncertainty as
to the impact that differences in first pass effects between an oral and an inhalation exposure
(i.e., lung versus liver first pass effects) might have on overall disposition.
275 DRAFT-DO NOT CITE OR QUOTE
-------
7. REFERENCES
Abdelmagid, HM; Tabatabai, MA. (1982) Decomposition of acrylamide in soils. J Environ Qual 11(4):701-704.
Abernethy, DJ; Boreiko, CJ. (1987) Acrylonitrile and acrylamide fail to transform C3H/10T1/2 cells. Environ
Mutagen 9(suppl 8):2.
Abou-Donia, MB; Ibrahim, SM; Corcoran, JJ; etal. (1993) Neurotoxicity of glycidamide, an acrylamide metabolite,
following i.p. injections in rats. J Toxicol Environ Health 39:447-464.
Adler, I-D. (1990) Clastogenic effects of acrylamide in different germ-cell stages of male mice. In: Biology of
mammalian germ cell mutagenesis, Banbury Report Vol. 34. Allen, J; Bridges, B; Lyon, M; eds. Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory Press, pp. 115-131.
Adler, I-D; Ingwersen, I; Kliesch, U; et al. (1988) Clastogenic effects of acrylamide in mouse bone marrow cells.
Mutat Res 206:379-385.
Adler, I-D; Zouh, R; Schmid E. (1993) Perturbation of cell division by acrylamide in vitro and in vivo. Mutat Res
301:249-254.
Adler, I-D; Reitmer, P; Schmoller, R; et al. (1994) Dose response for heritable translocations induced by acrylamide
in spermatids of mice. Mutat Res 309:285-291.
Adler, I-D; Baumgartner, A; Gonda, H; et al. (2000) 1-Aminotriazole inhibits acrylamide-induced dominant lethal
effects in spermatids of male mice. Mutagenesis 15:133-136.
Adler, I-D; Kliesch, U; Jentsch, I; et al. (2002) Induction of chromosomal aberrations by decarbazine in somatic and
germinal cells of mice. Mutagenesis 17:383-389.
Adler, I-D; Gonda, H; Hrabe de Angelis, M; et al. (2004) Heritable translocations induced by dermal exposure of
male mice to acrylamide. Cytogenet Genome Res. 104:271-276.
Agrawal, AK; Squibb, RE. (1981) Effects of acrylamide given during gestation on dopamine receptor binding in rat
pups. Toxicol Lett 7:233-238.
Agrawal, AK; Squibb, RE; Bondy, SC. (1981a) The effects of acrylamide treatment upon the dopamine receptor.
Toxicol Appl Pharmacol 58:89-99.
Agrawal, AK; Seth, PK; Squibb, RE; etal. (1981b) Neurotransmitter receptors in brain regions of acrylamide-
treated rats. I: Effects of a single exposure to acrylamide. Pharmacol Biochem Behav 14:527-531.
Albert Einstein College of Medicine. (1980) A fetal toxicity study of acrylamide in rats. Wayne, NJ.
Aldous, CN; Farr, CH; Sharma, RP. (1983) Evaluation of acrylamide treatment on levels of major brain biogenic
amines, their turnover rates, and metabolites. Fundam Appl Toxicol 3:182-186.
Ali, SF. (1983) Acrylamide-induced changes in the monoamines and their acid metabolites in different regions of
the rat brain. Toxicol Lett 17:101-105.
Ali, SF; Hong, J-S; Wilson, WE; et al. (1983) Effect of acrylamide on neurotransmitter metabolism and
neuropeptide levels in several brain regions and upon circulating hormones. Arch Toxicol 52:35-43.
Allen, B; Zeiger, E; Lawrence, G; et al. (2005) Dose-response modeling of in vivo genotoxicity data for use in risk
assessment: some approaches illustrated by an analysis of acrylamide. Regul Toxicol Pharmacol 41(l):6-27.
276
-------
Amador, A; Steger, RW; Bartke, A; et al. (1985) Testicular LH receptors during aging in F344 rats. J Androl 6:61-
64.
Ao, L; Liu, SX; Yang, MS; etal. (2008) Acrylamide-induced molecular mutation spectra at HPRT locus in human
promyelocytic leukaemia HL-60 and NB4 cell lines. Mutagenesis 23(4):309315.
Augustsson, K; Skog, K; Jagerstad, M; et al. (1999) Dietary heterocyclic amines and cancer of the colon, rectum,
bladder, and kidney: a population-based study. Lancet 353(9154):703-707.
Auld, RB; Bedwell, SF. (1967) Peripheral neuropathy with sympathetic overactivity from industrial contact with
acrylamide. Can Med Ass J 96:652-654.
Bachmann, M; Myers, JE; Bezuidenhout, BN. (1992) Acrylamide monomer and peripheral neuropathy in chemical
workers. Am JInd Med 21:217-222.
Backer, LC; Dearfield, KL; Erexson, GL; et al. (1989) The effects of acrylamide on mouse germ-line and somatic
cell chromosomes. Environ Mol Mutagen 13:218-226.
Bailey, E; Farmer, PB; Bird, I; et al. (1986) Monitoring exposure to acrylamide by the determination of S-(2-
carboxyethyl)cysteine in hydrolyzed hemoglobin by gas chromatography-mass spectrometry. Anal Biochem
157(2):241-248.
Banerjee, S; Segal, A. (1986) In vitro transformation of C3H/10T1/2 andNIH/3T3 cells by acrylonitrile and
acrylamide. Cancer Lett 32:293-304.
Barber, DS; Hunt, JR; Enrich, MF; et al. (2001) Metabolism, toxicokinetics and hemoglobin adduct formation in
rats following subacute and subchronic acrylamide dosing. NeuroToxicology 22:341-353.
Barber, DS; LoPachin, RM. (2004) Preteomic analysis of acrylamide-protein adduct formation in rat brain
synaptosomes. Toxicol Appl Pharmacol 201:120-136.
Barber, DS; Stevens, S; LoPachin, RM. (2007) Proteomic analysis of rat striatal synaptosomes during acrylamide
intoxication at a low dose rate. Toxicol Sci 100(1): 156-167.
Barfknecht, TR; Mecca, DJ; Naismith, RW. (1988) The genotoxic activity of acrylamide. Environ Mol Mutagen
ll(supplll):9.
Bartke, A; Sweeney, CA; Johnson, L; et al. (1985) Hyperprolactenemia inhibits development of Leydig cell tumors
in aging Fischer rats. Exper Aging Res 11:123-128.
Batiste-Alentorn, M; Xamena, N; Creus, A; et al. (1991) Genotoxicity studies with the unstable Zeste-White (UZ)
system ofDmsophila melanogaster: Results with ten carcinogenic compounds. Environ Mol Mutagen 18:120-125.
Bergmark, E. (1997) Hemoglobin adducts of acrylamide and acrylonitrile in laboratory workers, smokers and
nonsmokers. Chem Res Toxicol 10(l):78-84.
Bergmark, E; Calleman, CJ; Costa LG. (1991) Formation of hemoglobin adducts of acrylamide and its epoxide
metabolite glycidamide in the rat. Toxicol Appl Pharmacol 111:352-363.
Bergmark, E; Calleman, CJ; He, F; et al. (1993) Determination of hemoglobin adducts in humans occupationally
exposed to acrylamide. Toxicol Appl Pharmacol 120:45-54.
Besaratinia, A; Pfeifer, GP. (2004) Genotoxicity of acrylamide and glycidamide. J Natl Cancer Inst.
96(13):1023-1029.
Besaratinia, A; Pfeifer, GP. (2005) DNA adduction and mutagenic properties of acrylamide. Mutat Res
580(1-2):31-40.
277
-------
Besaratinia, A; Pfeifer, GP. (2007) A review of mechanisms of acrylamide carcinogenicity. Carcinogenesis
28(3):519-528.
Bio/Dynamics-Inc. (1979) A fetal toxicity study of acrylamide in rats. Submitted under TSCA. EPA Document
No. 878211679; NTIS No. OTS0206055.
Bishop, JB; Morris, RW; Seely, JC; et al. (1997) Alterations in the reproductive patterns of female mice exposed to
xenobiotics. Fundam Appl Toxicol 40:191-204.
Bjellaas, T; Janak, K; Lundanes, E; et al. (2005) Determination and quantification of urinary metabolites after
dietary exposure to acrylamide. Xenobiotica 35:1003-1018.
Bjellaas, T; Stolen, LH; Haugen, M; et al. (2007a) Urinary acrylamide metabolites as biomarkers for short-term
dietary exposure to acrylamide. Food Chem Toxicol 45(6): 1020-1026.
Bjellaas, T; Oleson, PT; Frandsen, H; et al. (2007b) Comparison of estimated dietary intake of acrylamide with
hemoglobin adducts of acrylamide and glycidamide. Toxicol Sci 98(1): 110-117.
Boettcher, MI; Schettgen, T; Kutting, B; et al. (2005) Mercapturic acids of acrylamide and glycidamide as
biomarkers of the internal exposure to acrylamide in the general population. Mutat Res 580(1-2): 167-176.
Boettcher, MI; Bolt, HM; Drexler, H; et al. (2006a) Excretion of mercapturic acids of acrylamide and glycidamide
in human urine after single oral administration of deuterium-labelled acrylamide. Arch Toxicol. 80(2):55-61.
Boettcher, MI; Bolt, HM; Angerer, J. (2006b) Acrylamide exposure via the diet: influence of fasting on urinary
mercapturic acid metabolite excretion in humans. Arch Toxicol. 80(12):817-819.
Bondy, SC; Tilson, HA; Agrawal, AK. (1981) Neurotransmitter receptors in brain regions of acrylamide-treated
rats. II. Effects of extended exposure to acrylamide. Pharmacol BiochemBehav 14:533-537.
Bowyer, JF; Latendresse, JR; Delongchamp, RR; et al. (2008) The effects of subchronic acrylamide exposure on
gene expression, neurochemistry, hormones, and histopathology in the hypothalamus-pituitary-thyroid axis of male
Fischer 344 rats. Toxicol Appl Pharmacol 230(2):208215. (See Erratum in: Toxicol Appl Pharmacol 232(3):498).
Brady, ST; Pfister, KK; Bloom, GS. (1990) A monoclonal antibody against kinesin inhibits both anterograde and
retrograde fast axonal transport in squid axoplasm. Proc Natl Acad Sci USA 87:1061-1065.
Brat, DJ; Brimijoin, S. (1993) Acrylamide and glycidamide impair neurite outgrowth in differentiating N1E. 115
neuroblastoma without disturbing rapid bidirectional transport of organelles observed by video microscopy. J
Neurochem 60(6):2145-2152.
Brown, L; Rhead, MM; Bancroft, KCC; et al. (1980) Model studies of the degradation of acrylamide monomer.
Water Res 14(7):775-778.
Budavari, S, ed. (2001) The Merck index - An encyclopedia of chemicals, drugs, and biologicals. 13th ed.
Whitehouse Station, NJ: Merck and Co., Inc., p. 24.
Bull, RJ; Robinson, M; Laurie, RD; et al. (1984a) Carcinogenic effects of acrylamide in SENCAR and A/J mice.
Cancer Res 44:107-111.
Bull, RJ; Robinson, M; Stober, JA. (1984b) Carcinogenic activity of acrylamide in the skin and lung of Swiss-ICR
mice. Cancer Lett 24:209-212.
Burek, JD; Albee, RR; Beyer, JE; et al. (1980) Subchronic toxicity of acrylamide administered to rats in the
drinking water followed by up to!44 days of recovery. J Environ Pathol Toxicol 4:157-182.
Butterworth, BE; Eldridge, SR; Sprankle, CS; et al. (1992) Tissue-specific genotoxic effects of acrylamide and
acrylonitrile. Environ Mol Mutagen 20:148-155.
278
-------
Calleman, CJ. (1996) The metabolism and pharmacokinetics of acrylamide: implications for mechanisms of toxicity
and human risk estimation. Drag Metab Rev 28(4)527-590.
Calleman, CJ; Wu, Y; Tian G; et al. (1994) Relationships between biomarkers of exposure and neurological effects
in a group of workers exposed to acrylamide. Toxicol Appl Pharmacol 126:361-371.
Carere, A. (2006) Genotoxicity and carcinogenicity of acrylamide: a critical review. Annali deU'lstituto Superiore
diSanita 42(2): 144-155.
Cavanagh, JB. (1964) The significance of the "dying-back" process in experimental and human neurological
disease. Int Rev Exp Pathol 3:219-267.
CFR (Code of Federal Regulations). (2005). Chapter 21 Code of Federal Regulations, Section 173. Acrylate-
acrylamide resins. 21CFR173.5. Available online at http://www.gpoaccess.gov/cfr/index.html.
Chapin, RE; Fail, PA; George, JD; et al. (1995) The reproductive and neuronal toxicities of acrylamide and three
analogues in Swiss mice, evaluated using the continuous breeding protocol. Fundam Appl Toxicol 27:9-24.
Chevolleau, S; Jacques, C; Canlet, C; et al. (2007) Analysis of hemoglobin adducts of acrylamide and glycidamide
by liquid chromatography-electrospray ionization tandem mass spectrometry, as exposure biomarkers in French
population. JChromatogrA 1167(2):125-134.
Cihak, R; Vontorkova, M. (1988) Cytogenetic effects of acrylamide in the bone marrow of mice. Mutat Res
209:91-94.
Cihak, R; Vontorkova, M. (1990) Activity of acrylamide in single-, double-, and triple-dose mouse bone marrow
micronucleus assays. Mutat Res 234:125-127.
CIR (Cosmetic Industry Review Expert Panel). (1991) Final report on the safety assessment of polyacrylamide. J
Am Coll Toxicol 10:193-203.
Clark, CH; Sickles, DW. (1996) Decreased GAP-43 accumulation in neurite tips of cultured hippocampal neurons
by acrylamide. NeuroToxicology 17:397-406.
Clegg, ED; Cook, JC; Chapin, RE; et al. (1997) Leydig cell hyperplasia and adenoma formation: mechanisms and
relevance to humans. Reprod Toxicol 1:107-121.
Collins, BW; Howard, DR; Allen, JW. (1992) Kinetochore-staining of spermatid micronuclei: studies of mice
treated with X-radiation or acrylamide. Mutat Res 281:287-294.
Collins, JJ; Swaen, GMH; Marsh, GM; et al. (1989) Mortality patterns among workers exposed to acrylamide. J
OccupMed31(7):614-617.
Cook, JC; Klinefelter, GR; Hardisty, JF; et al. (1999) Rodent Leydig cell tumorigenesis: a review of the physiology,
pathology, mechanisms, and relevance to humans. Crit Rev Toxicol 29:169-261.
Costa, LG; Deng, H; Gregotti, C; et al. (1992) Comparative studies on the neuro- and reproductive toxicity of
acrylamide and its epoxide metabolite glycidamide in the rat. NeuroToxicology 13:219-224.
Costa, LG; Deng, H; Calleman, CJ; et al. (1995) Evaluation of the neurotoxicity of glycidamide, an epoxide
metabolite of acrylamide: behavioral, neurochemical and morphological studies. Toxicology 98:151-161.
Crofton, KM; Padilla, S; Tilson, HA; et al. (1996) The impact of dose rate on the neurotoxicity of acrylamide: the
interaction of administered dose, target tissue concentrations, tissue damage, and functional effects. Toxicol Appl
Pharmacol 139:163-176.
Croll, BT; Arkell, GM; Hodge, RPJ. (1974) Residues of acrylamide in water. Water Res 8:989-993.
279
-------
Damjanov, I; Friedman, MA. (1998) Mesotheliomas of tunica vaginalis testis of F344 (F344) rats treated with
acrylamide: a light and electron microscopy study. In Vivo 12:495-502.
Davenport, JG; Farrell, MD; Sumi, SM. (1976) 'Giant axonal neuropathy' caused by industrial chemicals:
neurofilamentous axonal masses in man. Neurology 26:919-923.
Dearfield, KL; Abernathy, CO; Ottley, MS; et al. (1988) Acrylamide: its metabolism, developmental and
reproductive effects, genotoxicity, and carcinogenicity. MutatRes 195:45-77.
Dearfield, KL; Douglas, GR; Ehling, UH; et al. (1995) Acrylamide: a review of its genotoxicity and an assessment
of heritable genetic risk. Mutat Res 330:71-99.
DeGrandchamp, RL; Lowndes, HE. (1990) Early degeneration and sprouting at the rat neuromuscular junction
following acrylamide administration. Neuropathol Appl Neurobiol 16:239-254.
DeGrandchamp, RL; Reuhl, KR; Lowndes, HE. (1990) Synaptic terminal degeneration and remodeling at the rat
neuromuscular junction resulting from a single exposure to acrylamide. Toxicol Appl Pharmacol 105:422-433.
Deng, H; He, F; Zhang, S; et al. (1993) Quantitative measurements of vibration threshold in healthy adults and
acrylamide workers. Int Arch Occup Environ Health 65:53-56.
Diembeck, W; Dusing, H-J; Akhiana, M. (1998) Dermal absorption and penetration of acrylamide ([14C]
acrylamide as tracer) in different cosmetic formulations and polyacrylamide-solution after topical application to
excised pig skin. Beiersdorf 9.
Dixit, R; Husain, R; Mukhtar, H; et al. (1981) Effect of acrylamide on biogenic amine levels, monoamine oxidase,
and cathepsin D activity of rat brain. Environ Res 26:168-173.
Dixit, R; Seth, PK; Mukhtar, H. (1982) Metabolism of acrylamide into urinary mercapturic acid and cysteine
conjugates in rats. Drug Metab Dispos Biol Fate Chem 10:196-197.
Dobrzynska, MM. (2007) Assessment of DNA damage in multiple organs from mice exposed to x-rays or
acrylamide or a combination of both using the comet assay. In Vivo 21(4):657-662.
Dobrzynska, M; Lenarczyk, M; Gajewski, AK. (1990) Induction of dominant lethal mutations by combined X-ray-
acrylamide treatment in male mice. Mutat Res 232:209-215.
Doerge, DR; da Costa, GG; McDaniel, LP; et al. (2005a) DNA adducts derived from administration of acrylamide
and glycidamide to mice and rats. MutatRes 580(1-2):131-141.
Doerge, DR; Young, JF; McDaniel, LP; et al. (2005b) Toxicokinetics of acrylamide and glycidamide in B6C3F!
mice. Toxicol Appl Pharmacol 202(3):258-267.
Doerge, DR; Young, JF; McDaniel, LP; et al. (2005c) Toxicokinetics of acrylamide and glycidamide in Fischer 344
rats. Toxicol Appl Pharmacol 208:199-209.
Doerge, DR; Twaddle, NC; Boettcher, MI; et al. (2007) Urinary excretion of acrylamide and metabolites in Fischer
344 rats and B6C3F! mice administered a single dose of acrylamide. Toxicol Lett 169(l):34-42.
Doerge, DR; Young JF; Chen, JJ; et al. (2008) Using dietary exposure and physiologically based
pharmacokinetic/pharmacodynamic modeling in human risk extrapolations for acrylamide toxicity. J Agric Food
Chem 56(15):6031-6038.
Donovan, JW; Pearson, T. (1987) Ingestion of acrylamide with severe encephalopathy, neurotoxicity and
hepatotoxicity. Vet Hum Toxicol 29:462.
Doroshyenko, O; Fuhr, U; Kunz, D; et al. (2009). In vivo role of cytochrome P450 2E1 and glutathione-S-
transferase activity for acrylamide toxicokinetics in humans. Cancer Epidemiol Biomarkers Prev. 18(2):433-43.
280
-------
Dourson, M; Hertzberg, R; Allen, B; et al. (2008) Evidence-based dose-response assessment for thyroid
tumorigenesis from acrylamide. Regul Toxicol Pharmacol 52(3):264-289.
Dow Chemical Co. (1984) Acrylamide: toxicodynamics in rats. Submitted under TSCA. NTIS No. OTS0507270.
Dumitru, DL. (1989) Occupational intoxication with acrylamide: discussion of causes with acute and chronic
intoxication. Rev Ig Med Muncii Med Soc Bacteriol Virusol Parazitol Epidemiol Pneumofiziol Ser Ig Med Muncii
MedSoc 35:359-364.
Dybing E; Farmer PB; Andersen M; et al. (2005) Human exposure and internal dose assessments of acrylamide in
food. Food Chem Toxicol 43(3):365-410.
Dybing, E; Sanner, T. (2003). Risk assessment of acrylamide infoods. Toxicol Sci 75:7-15.
Edwards, PM. (1975) The distribution and metabolism of acrylamide and its neurotoxic analogues in rats. Biochem
Pharmacol 24:1277-1282.
Edwards, PM (1976) The insensitivity of the developing rat foetus to the toxic effects of acrylamide. Chem Biol
Interact 12:13-18.
Ehling, UH. (1988) Quantification of the genetic risk of environmental mutagens. Risk Anal 8:45-57.
Ehling, UH; Neuhauser-Klaus, A. (1992) Reevaluation of the induction of specific-locus mutations in
spermatogonia of the mouse by acrylamide. Mutat Res 283:185-191.
Environ. (2002) Toxicological review of acrylamide (CAS No. 79-06-01). Prepared by Environ International
Corporation (Ruston, Louisiana) for SNF S.A. (Saint-Etienne Cedex, France). November 2002.
Eskin, TA; Lapham, LW; Maurissen, JPJ; et al. (1985) Acrylamide effects on the macaque visual system. Invest
Ophthalmol Vis Sci 26:317-329.
Exon, JH. (2006) A review of the toxicology of acrylamide. J Toxicol Environ Health B 9(5):397-412.
FAO/WHO. (2005) Summary and Conclusion of the sixty-fourth meeting of the Joint FAO/WHO Expert
Committee On Food Additives (JECFA), JECFA/64/SC; pp 1-18.
FAO/WHO (Food and Agriculture Organization/World Health Organization). (2009). The FAO/WHO acrylamide
in food network. Available online at http://www.foodrisk.org/acrylamide/
Favor, J; Mailing, H; Layton, D; et al. (1994) Genetic risk extrapolation from animal data to human disease: a
taskgroup report of the International Commission for Protection Against Environmental Mutagens and Carcinogens
(ICPEMC). Mutat Res 330:23-34.
Favor, J; Shelby, MD. (2005) Transmitted mutational events induced in mouse germ cells following acrylamide or
glycidamide exposure. Mutat Res. 580(1-2):21-30.
Feng, Y; Forgac, M. (1992) Cysteine 254 of the 73kDa A subunit is responsible for inhibition of the coated vesicle
(H+)-ATPase upon modification by sulfhydryl reagents. J Biol Chem 267:5817-5822.
Fennell, TR; Snyder, RW; Krol, WL; et al. (2003) Comparison of the hemoglobin adducts formed by administration
of N-methylolacrylamide and acrylamide to rats. Toxicol Sci 71:164-175.
Fennell TR; Sumner SC; Snyder RW; et al. (2005) Metabolism and hemoglobin adduct formation of acrylamide in
humans. Toxicol Sci 85(l):447-59.
Field, EA; Price, CJ; Sleet, RB; et al. (1990) Developmental toxicity evaluation of acrylamide in rats and mice.
Fundam Appl Toxicol 14:502-512.
281
-------
Frantz, SW; Dryzga, MD; Fresheur, NL; et al. (1995) In vitro/in vivo determination of cutaneous penetration of
residual acrylamide monomer from polyacrylamide water solutions. Dow Chemical Toxicology Research
Laboratory.
Friedman, MA; Dulak, LH; Stedham, MA. (1995) A lifetime oncogenicity study in rats with acrylamide. Fundam
Appl Toxicol 27:95-105.
Friedman, MA; Tyl, RW; Marr, MC; et al. (1999a) Effects of lactational administration of acrylamide on rat dams
and offspring. Reprod Toxicol 13:511-520.
Friedman, MA; Dulak, LH; Keefe, RT. (1999b) Effect of acrylamide on rat hormone levels in a 28-day drinking
water study. Unpublished study report.
Fuhr, U; Boettcher, MI; Kinzig-Schippers, M; et al. (2006) Toxicokinetics of acrylamide in humans after ingestion
of a defined dose in a test meal to improve risk assessment for acrylamide carcinogenicity. Cancer Epidemiol
Biomarkers Prev 15(2):266-271.
Fullerton, PM; Barnes, JM. (1966) Peripheral neuropathy in rats produced by acrylamide. Br J Ind Med
23:210-221.
Fullerton, PM. (1969) Electrophysiological and histological observations on peripheral nerves in acrylamide
poisoning in man. J Neurol Neurosurg Psychiat 32:186-192.
Gamboa da Costa, G; Churchwell, MI; Hamilton, LP; et al. (2003) DNA adduct formation from acrylamide via
conversion to glycidamide in adult and neonatal mice. Chem Res Toxicol 16(10): 1328-1337.
Garey, J; Paule, MG. (2007) Effects of chronic low-dose acrylamide exposure on progressive ratio performance in
adolescent rats. NeuroToxicology 28(5):998-1002.
Garland, TO; Patterson, MW. (1967) Six cases of acrylamide poisoning. Brit Med J (4): 134-138.
Generoso, WM; Sega, GA; Lockhart, AM; et al. (1996) Dominant lethal mutations, heritable translocations, and
unscheduled DNA synthesis induced in male mouse germ cells by glycidamide, a metabolite of acrylamide. Mutat
Res 371:175-183.
Ghanayem, BI; Witt, KL; El-Hadri, L; et al. (2005a) Comparison of germ cell mutagenicity in male CYP2El-null
and wild-type mice treated with acrylamide: evidence supporting a glycidamide-mediated effect. Biol Reprod
72(1):157-163.
Ghanayem, BI; Witt, K; Kissling, GE; et al. (2005b) Absence of acrylamide-induced genotoxicity in CYP2El-null
mice: Evidence consistent with a glycidamide-mediated effect. Mutat Res 578:284-297.
Gho, M; McDonald, K; Ganetzky, B; et al. (1992) Effects of kinesin mutations on neuronal functions. Science
258:313-316.
Gilbert, SG; Maurissen, JPJ. (1982) Assessment of the effects of acrylamide, methylmercury, and 2,5-hexanedione
on motor functions in mice. J Toxicol Environ Health 10:31-42.
Gjerleff, T; Elsborg, H; Bonde, JP. (2001) Svaerkronisk acrylamid-intoksikation [Severe chronic acrylamide
intoxication]. UgeskrLaeg 163:4204-4205.
Goodman, DG; Ward, JM; Squire, RA; et al. (1979) Neoplastic and nonneoplastic lesions in aging F344 rats.
Toxicol Appl Pharmacol 48:237-248.
Gould, RM; Brady, ST. (2004) Neuropathy: many paths lead to hereditary spastic paraplegia. Curr Biol
14:R903-R904.
282
-------
Granath, F; Ehrenberg, L; Tornqvist, M. (1992) Degree of alkylation of macromolecules in vivo from variable
exposure. Mutat. Res. 284, 297-306.
Gutierrez-Espeleta, GA; Hughes, LA; Piegorsch, WW; et al. (1992) Acrylamide: dermal exposure produces genetic
damage in male mouse germ cells. Fundam Appl Toxicol 18:189-192.
Haber, LT; Maier, A; Kroner, OL; et al. (2009) Evaluation of human relevance and mode of action for tunica
vaginalis mesotheliomas resulting from oral exposure to acrylamide. Regul Toxicl Pharmacol 53(2): 134-149.
Hagmar, L; Tornqvist, M; Nordander, C; et al. (2001) Health effects of occupational exposure to acrylamide using
hemoglobin adducts as biomarkers of internal dose. Scand J Work Environ Health 27(4):219-226.
Hagmar, L; Wirfalt, E; Paulsson, B; et al. (2005) Differences in hemoglobin adduct levels of acrylamide in the
general population with respect to dietary intake, smoking habits and gender. Mutat Res 580(1-2): 157-165.
Hanioka, N; Tanaka-Kagawa, T; Miyata, Y; et al. (2003) Functional characterization of three human cytochrome
p450 2E1 variants with amino acid substitutions. Xenobiotica 33(6):575-586.
Hansch, C; Leo, A; Hoekman, D. (1995) Exploring QSAR. hydrophobic, electronic, and steric constants. In: ACS
Professional Reference Book. Heller, SR, ed. Washington, DC: American Chemical Society, p. 6.
Harris, CH; Gulati, AK; Friedman, MA; et al. (1994) Toxic neurofilamentous axonopathies and fast axonal
transport. V. Reduced bidirectional vesical transport in cultured neurons by acrylamide and glycidamide. J Toxicol
Envion Health 42:343-356.
Harry, GJ. (1992) Acrylamide-induced alterations in axonal transport. Mol Neurobiol 6:203-216.
Hartmann, EC; Boettcher, MI; Schettgen, T; et al. (2008) Hemoglobin adducts and mercapturic acid excretion of
acrylamide and glycidamide in one study population. J Agric Food Chem. 56(15):6061-8.
Hashimoto, K; Aldridge, WN. (1970) Biochemical studies on acrylamide, a neurotoxic agent. Biochem Pharmacol
19:2591-2604.
Hashimoto, K; Sakamoto, J; Tanii, H. (1981) Neurotoxicity of acrylamide and related compounds and their effects
on male gonads in mice. Arch Toxicol 47:179-189.
Hashimoto, K; Tanii, H. (1985) Mutagenicity of acrylamide and its analogues in Salmonella typhimurium. Mutat
Res 158:129-133.
Hays, SM; Aylward, LL. (2008a) Using Biomonitoring Equivalents to interpret human biomonitoring data in a
public health risk context. J Appl Toxicol. 2008 Dec 29 [Epub ahead of print]
Hays, SM; Aylward, LL. (2008b) Biomonitoring Equivalents (BE) dossier for acrylamide (AA) (CAS No. 79-06-1).
Regul Toxicol Pharmacol. 2008 Aug;51(3 Suppl):S57-67.
Hazleton Laboratories. (1953) Inhalation toxicity supplement to reports dated May 2, 1951 and August 13, 1952.
Submitted under TSCA Section 8D. EPA Document No. 878211664. NTIS No. OTS206055.
Hazleton Laboratories. (1954) Chronic inhalation exposure-acrylamide. Submitted under TSCA Section 8D. EPA
Document No. 878211670. NTIS No. OTS206055.
He, F; Zhang, S; Wang, H; et al. (1989) Neurological and electroneuromyographic assessment of the adverse effects
of acrylamide on occupationally exposed workers. Scand J Work Environ Health 15:125-129.
Hersch, MI; McLeod, JG; Satchell, PM; et al. (1989) Breathing pattern, lung inflation reflex and airway tone in
acrylamide neuropathy. Respir Physiol 76:257-276.
283
-------
Heudorf, U; Hartmann, E; Angerer, J. (2009) Acrylamide in children - exposure assessment via urinary acrylamide
metabolites asbiomarkers. Int JHygEnvironHealth212(2):135-141.
Hinson, JA; Roberts, DW. (1992) Role of covalent and noncovalent interactions in cell toxicity: effects on proteins.
Ann Rev Pharmacol Toxicol 32:471-510.
Ho, WH; Wang, SM; Yin, HS. (2002) Acrylamide disturbs the subcellular distribution of GABAa receptor in brain
neurons. J Cellular Biochem 85:561-571.
Hogervorst, JG; Schouten, LJ; Konings, EJ; et al. (2007) A prospective study of dietary acrylamide intake and the
risk of endometrial, ovarian, and breast cancer. Cancer Epidemiol Biormark Prev 16(11):2304-2313.
Hogervorst, JG; Schouten, LJ; Konings, EJ; et al. (2008a) Dietary acrylamide and the risk of renal cell, bladder, and
prostate cancer. Am J Clin Nutr 87(5): 1428-1438.
Hogervorst, JG; Schouten, LJ; Konings, EJ; et al. (2008b) Dietary acrylamide intake is not associated with
gastrointestinal cancer risk. J Nutr 138(11):2229-2236.
Holland, N; Ahlborn, T; Turteltaub, K; et al. (1999) Acrylamide causes preimplantation abnormalities in embryos
and induces chromatinadducts in male germ cells of mice. Reprod Toxicol 13:167-178.
Hoorn, AJW; Custer, LL; Myhr, BC; et al. (1993) Detection of chemical mutagens using Muta® mouse: a
transgenic mouse model. Mutagenesis 8(1):7-10.
HSDB (Hazardous Substances Data Bank). (2005) HSDB: Acrylamide. The National Library of Medicine.
Available online at http://toxnet.nlm.nih.gOv/cgi-bin/sis/search/f7./temp/~KNYp4J:l.
Kurd, DD; Saxton, WM. (1996) Kinesin mutations cause motor neuron disease phenotypes by disrupting fast axonal
transport in drosophila. Genetics 144:1075-1085.
Husain, R; Dixit, R; Das, M; et al. (1987) Neurotoxicity of acrylamide in developing rat brain: changes in the levels
of brain biogenic amines and activities of monoamine oxidase and acetylcholine esterase. Ind Health 25:19-28.
IARC (International Agency for Research on Cancer). (1994a) Acrylamide. In: IARC monographs on some
industrial chemicals; summary of data reported and evaluation. Vol. 60:389-433.
IARC. (International Agency for Research on Cancer) (1994b) Ethylene oxide. In: IARC monographs on the
evaluation of carcinogenic risks to humans, some industrial chemicals. Vol. 60:73-159.
IARC. (International Agency for Research on Cancer). (1999) Acrylonitrile. In: IARC monographs on the
evaluation of carcinogenic risks to humans, Part One: Re-evaluation of some organic chemicals, hydrazine and
hydrogen peroxide. Vol. 71:43-108.
IARC (International Agency for Research on Cancer). (2007a) Ethylene oxide. In: IARC monographs on the
evaluation of carcinogenic risks to humans. Vol 97:185-309.
IARC (International Agency for Research on Cancer).(2007b) 1,3-Butadiene. In: IARC monographs on the
evaluation of carcinogenic risks to humans. Vol. 97:45-184.
latropoulos, MJ; Lebish, I; Wang, CX; et al. (1998) Microscopic evaluation of proliferative mesothelial lesions
diagnosed previously as mesotheliomas of the tunica vaginalis testis. Unpublished report to CYTEC Industries,
October 14, 1998.
ICPEMC (International Commission for Protection Against Environmental Mutagens and Carcinogens). (1993a)
Use of in vivo genetic toxicology data to construct human risk assessments, Final report submitted to Department of
Health, Canada, July, 1993, Contract No. 3138.
284
-------
ICPEMC (International Commission for Protection Against Environmental Mutagens and Carcinogens). (1993b)
Genetic risk extrapolation from animal data to human disease, Final report submitted to U.S. Environmental
Protection Agency, September, 1993, under interagency agreement number 1824-D094-A1 between the USEPA and
the Biomedical and Environmental Information Analysis Section of the Oak Ridge National Laboratory.
Igisu, H; Matsuoka, M. (2002) Acrylamide encephalopathy. J Occup Health 44:63-68.
Igisu, H; Goto, I; Kawamura, Y; et al. (1975) Acrylamide encephalneuropathy due to well water pollution. J Neurol
Neurosurg Psychiatry 38:581-584.
Ikeda, GJ; Miller, E; Sapienza, PP; et al. (1983) Distribution of [14C]-labeled acrylamide and betaine in foetuses of
rats, rabbits, beagle dogs and miniature pigs. Food Chem Toxicol 21:49-58.
Ikeda, GJ; Miller, E; Sapienza, PP; et al. (1985) Maternal-foetal distribution studies in late pregnancy. II.
Distribution of 1-[14C]-acrylamide in tissues of beagle dogs and miniature pigs. Food Chem Toxicol 23(8):757-
761.
IRRM (Institute for Reference Materials and Measurements).. (2004) Monitoring database on acrylamide levels in
food maintained by the Institute for Reference Materials and Measurements together with Health and Consumer
Protection DG. Available online at http://www.irmm.jrc.be/.
Jiang, L; Cao, J; An, Y; et al. (2007) Genotoxicity of acrylamide in human hepatoma G2 (HepG2) cells. Toxicol In
Vitro 21(8):1486-1492.
JIFSAN (Joint Institute for Food Safety and Applied Nutrition). (2002). Acrylamide in food workshop: scientific
issues, uncertainties, and research strategies. Available online at http://www.jifsan.umd.edu/acrylamide2002.htm.
JIFSAN (Joint Institute for Food Safety and Applied Nutrition). (2004). Proceedings and reports from the Joint
Institute for Food Safety and Applied Nutrition Acrylamide in food workshop: update - scientific issues,
uncertainties, and research strategies; April 13-15, 2004; Chicago, Illinois. Available online at
http://www.jifsan.umd.edu/acrylamide2004.htm.
Jin, L; Chico-Galdo, V; Massart, C; et al. (2008) Acrylamide does not induce tumorigenesis or major defects in
mice in vivo. JEndocrinol 198(2):301-307.
Johnson, KA; Gorzinski, SJ; Bodner, KM; et al. (1984) Acrylamide: A two-year drinking water chronic toxicity-
oncogenicity study in F344 rats. Dow Chemical, USA. Midland, MI. Submitted under TSCA. NTIS No. 0507273.
Johnson, KA; Beyer, JE; Bell, TJ; et al. (1985) Acrylamide: A two-year drinking water chronic toxicity-
oncongenicity study in F344 rats. Dow Chemical, USA. Midland, MI. Submitted under TSCA. EPA Document
No. 878216184. NTIS Document No. OTS0206849.
Johnson, KA; Gorzinski, SJ; Bodner, KM; et al. (1986) Chronic toxicity and oncogenicity study on acrylamide
incorporated in the drinking water of F344 rats. Toxicol Appl Pharmacol 85:154-168.
Johnsrud, EK; Koukouritaki, SB; Divakaran, K; et al. (2003) Human hepatic CYP2E1 expression during
development. J Pharmacol Exp Ther. 307(1):402-407. Erratum in: (2004) J Pharmacol Exp Ther 309(1):439.
Jung, R; Engelhart, G; Herbolt B; et al. (1992) Collaborative study of mutagenicity with Salmonella typhimurium
TA102. Mutat Res 278:265-270.
Kadry, AM; Friedman, MA; Abdel-Rahman, MS. (1999) Pharmacokinetics of acrylamide after oral administration
in male rats. Environ Toxicol Pharmacol 7:127-133.
Kaplan, ML; Murphy, SD; Gilles, FH. (1973) Modification of acrylamide neuropathy in rats by selected factors.
Toxicol Appl Pharmacol 24:564-579.
285
-------
Kaster, JK; Kamendulis, LM; Friedman, MA; et al. (1998) Syrian hamster embryo (SHE) cell transformation by
acrylamide and hormones. Toxicol Sci 42(1-S):76.
Kellert, M; Scholz, K; Wagner, S; et al. (2006) Quantitation of mercapturic acids from acrylamide and glycidamide
in human urine using a column switching tool with two trap columns and electrospray tandem mass spectrometry. J.
Chromatogr, A 1131:58.
Kesson, CM; Lawson, DH; Baird, AW. (1977) Acrylamide poisoning. Postgrad Med J 53:16-17.
Khan, MA; Davis, CA; Foley, GL; et al. (1999) Changes in thyroid gland morphology after acute acrylamide
exposure. Toxicol Sci 47:151-157.
Kirman, CR; Gargas, ML; Deskin, R; et al. (2003) A physiologically based pharmacokinetic model for acrylamide
and its metabolite, glycidamide, in the rat. J Toxicol Environ Health A 66(3):253-274.
Klaunig, J. (2000) Cellular effects of acrylamide. Unpublished data cited in Environ, 2002.
Klaunig, JE. (2008) Acrylamide carcinogenicity. J Agric Food Chem 56(15):5984-5988.
Klaunig, JE; Kamendulis, LM. (2005) Mechanisms of acrylamide induced rodent carcinogenesis. Adv Exp Med
Biol 561:49-62.
Kligerman, AD; Atwater, AL; Bryant, MF. (1991) Cytogenetic studies of ethyl acrylate using C57BL/6 mice.
Mutagenesis 6(2):137-141.
Knaap, AG; Kramers, PG; Voogd, CE; et al. (1988) Mutagenic activity of acrylamide in eukaryotic systems but not
in bacteria. Mutagenesis 3(3):263-268.
Ko, M-H; Chen, W-P; Lin-Shiau; et al. (1999) Age-dependent acrylamide neurotoxicity in mice: morphology,
physiology, and function. ExpNeurol 158:37-46.
Konings, EJM; Baars, AJ; Van Klaveren, JD; et al. (2003) Acrylamide exposure from foods of the Dutch population
and an assessment of the consequent risk. Food Chem Toxicol 41:1569-1576.
Kopp, EK; Dekant, W. (2009) Toxicokinetics of acrylamide in rats and humans following single oral administration
of low doses. Toxicol Appl Pharmacol. 235(2): 135-42.
Krebs, O; Favor, J. (1997) Somatic and germ cell mutagenesis in lambda lacZ transgenic mice treated with
acrylamide or ethylnitrosourea. Mutat Res 388:239-248.
Krishna, G; Theiss, JC. (1995) Concurrent analysis of cytogenetic damage in vivo: A multiple endpoint-multiple
tissue approach. Environ Mol Mutagen 25:314-320.
KS Crump Group, Inc. (1999a) Consideration of the potency classification of acrylamide based on the incidence of
tunica vaginalis mesotheliomas (TVMs) in male F344 rats. June 11, 1999. Unpublished report prepared for the
Acrylamide Monomer Producers Association, Frankfort, KY.
KS Crump Group, Inc. (1999b) Mechanism of acrylamide induction of benign mammary fibroadenomas in the
aging female F344 rat: relevance to human health risk assessment. December 14, 1999. Unpublished report
prepared for the Acrylamide Monomer Producers Association, Frankfort, KY.
Lahdetie, J; Suutari, A; Sjoblom, T. (1994) The spermatid micronucleus test with the dissection technique detects
the germ cell mutagenicity of acrylamide in rat meiotic cells. Mutat Res 309:255-262.
Lande, SS; Bosch, SJ; Howard, PH. (1979) Degradation and leaching of acrylamide in soil. J Environ Qua!
8(1):133-137.
286
-------
Lapadula, DM; Bowe, M; Carrington, CD; et al. (1989) In vitro binding of [14C]-acrylamide to neurofilament and
microtubule proteins of rats. Brain Res 481:157-161.
Larsson, SC; Hakansson, N; Akesson, A; et al. (2009a) Long-term dietary acrylamide intake and risk of endometrial
cancer in a prospective cohort of Swedish women. Int J Cancer 124(5): 1196-1199.
Larsson, SC; Akesson, A; Wolk, A. (2009b) Long-term dietary acrylamide intake and risk of epithelial ovarian
cancer in a prospective cohort of Swedish women. Cancer Epidemiol Biomark Prev 18(3):994-997.
Larsson, SC; Kesson, A; Bergkvist, L; et al. (2009c) Dietary acrylamide intake and risk of colorectal cancer in a
prospective cohort of men. Eur J Cancer 45(4):513-516.
Larsson, SC; Akesson, A; Wolk, A. (2009d) Long-term dietary acrylamide intake and breast cancer risk in a
prospective cohort of Swedish women. Am J Epidemiol 169(3):376-381.
Lehning, EJ; Persaud, A; Dyer, KR; et al. (1998) Biochemical and morphologic characterization of acrylamide
peripheral neuropathy. Toxicol Appl Pharmacol 151:211-221.
Lehning, EJ; Balaban, CD; Ross, JF; et al. (2002) Acrylamide neuropathy I. Spatiotemporal characteristics of nerve
cell damage in rat cerebellum. NeuroToxicology 23:397-414.
Lehning, EJ; Balaban, CD; Ross, JF; et al. (2003a) Acrylamide neuropathy II. Spatiotemporal characteristics of
nerve cell damage in brainstem and spinal cord. NeuroToxicology 24:109-123. [Note: This is a corrected version
from the version that appeared in NeuroToxicology (2002) 23:415-429, which had Figures 2-4 in black and white
instead of color. Otherwise the two versions are identical].
Lehning, EJ; Balaban, CD; Ross, JF; et al. (2003b) Acrylamide neuropathy III. Spatiotemporal characteristics of
nerve cell damage inforebrain. NeuroToxicology 24:1125-136.
Lewis, RJ, Sr, ed. (1997) Hawley's condensed chemical dictionary. 13th ed. New York, NY: John Wiley & Sons,
Inc., p. 17.
Lide, DR, ed. (2000) CRC handbook of chemistry and physics. 81st Ed. Boca Raton, FL: CRC Press LLC, pp.
3-287.
Lijinsky, W; Andrews, AW. (1980) Mutagenicity of vinyl compounds in Salmonella typhimurium. Teratogen
Carcinogen Mutagen 1:259-267.
Lipscomb, JC. (2004) Evaluating the relationship between variance in enzyme expression and toxicant
concentration in health risk assessment. Hum Ecol Risk Assess 10:39-55.
Lipscomb, JC; Teuschler, LK; Swartout, J; et al. (2003) The impact of cytochrome P450 2El-dependent metabolic
variance on a risk-relevant pharmacokinetic outcome in humans. Risk Anal 23(6): 1221-1238.
LoPachin, RM; Lehning, EJ. (1994) Acrylamide-induced distal axon degeneration: A proposed mechanism of
action. NeuroToxicology 15 (2): 247-260.
LoPachin, RM; Ross, JF; Lehning, EJ. (2002a) Nerve terminals as the primary site of acrylamide action: a
hypothesis. NeuroToxicology 23:43-59.
LoPachin, RM; Ross, JF; Reid, ML; et al. (2002b) Neurological evaluation of toxic axonopathies in rats: acrylamide
and 2,5-hexandione. NeuroToxicology 23:95-110.
LoPachin, RM; Balaban, CD; Ross, JR. (2003) Acrylamide axonopathy revisited. Toxicol Appl Pharmacol
188:135-153.
LoPachin, RM; Canady, RA. (2004) Acrylamide toxicities and food safety: Session IV. summary and research
needs. NeuroToxicology 25: 507-509.
287
-------
LoPachin, RM; Schwarcz, AI; Gaughan, R; et al. (2004) In vivo and in vitro effects of acrylamide on synaptosomal
neurotransmitter uptake and release. NeuroToxicology 25:349-363.
LoPachin, RM; DeCaprio, AP. (2005) Protein adduct formation as a molecular mechanism in neurotoxicity.
Toxicol. Sci86(2):214-225.
LoPachin, RM; Barber, DS. (2006) Synaptic cysteine sulfhydryl groups as targets of electrophilic neurotoxicants.
Toxicol Sci 94:240-255.
LoPachin, RM; Barber, DS; He, D; et al. (2006) Acrylamide inhibits dopamine uptake in rat striatial synaptic
vesicles. Toxicol Sci 89:224-234.
LoPachin, RM; Gavin, T; Geohagen, BC; et al. (2007a) Neurotoxic mechanisms of electrophilic type-2 alkenes: soft
soft interactions described by quantum mechanical parameters. Toxicol Sci 98(2):561-70.
Lopachin, RM; Barber, DS; Geohagen, BC; et al. (2007b) Structure-toxicity analysis of type-2 alkenes: in vitro
neurotoxicity. Toxicol Sci 95:136-146.
LoPachin, RM; Barber, DS; Gavin, T. (2008) Molecular mechanisms of the conjugated alpha, beta-unsaturated
carbonyl derivatives: relevance to neurotoxicity and neurodegenerative diseases. Toxicol Sci 104(2):235-249.
LoPachin, RM; Geohagen, BC; Gavin, T. (2009) Synaptosomal toxicity and nucleophilic targets of 4-hydroxy-2-
nonenal. Toxicol Sci 107(1): 171-181.
Lyon, M; Adler, I-D; Bridges, B; et al. (1983) International Commission for Protection against Environmental
Mutagens and Carcinogens (ICPEMC), Committee 4 Final Report, Estimation of genetic risks and increased
incidence of genetic disease due to environmental mutagens. Mutat Res 115:255-291.
Manjanatha, MG; Aidoo, A; Shelton, SD; et al. (2006) Genotoxicity of acrylamide and its metabolite glycidamide
administered in drinking water to male and female Big Blue mice. Environ Mol Mutagen 47:6-17.
Mapp, C; Mazzotta, M; Bartolucci, GB; et al. (1977) La neuropatia da acrilamide: prime osservazioni in Italia
[Nervous system disease caused by acrylamide: 1st cases in Italy]. Med Lav 68(1): 1-12.
Marchetti, F; Lowe, X; Bishop, J; et al. (1997) Induction of chromosomal aberration in mouse zygots by acrylamide
treatment of male germ cells and their correlation with dominant lethality and heritable translocations. Environ Mol
Mutagen 30:410-417.
Marchetti, F; Bishop, J; Lowe, X; et al. (2009) Chromosomal mosaicism in mouse two-cell embryos after paternal
exposure to acrylamide. Toxicol Sci 107(1): 194-205.
Marlowe, C; Clark, MJ; Mast, RW; et al. (1986) The distribution of [14C]-acrylamide in male and pregnant Swiss-
Webster mice studied by whole-body autoradiography. Toxicol Appl Pharmacol 86:457-465.
Marsh, GM; Lucas, LJ; Youk, AO; et al. (1999) Mortality patterns among workers exposed to acrylamide: 1994
follow up. Occup Environ Med 56:181-190.
Marsh, GM; Youk, AO; Buchanich, JM, et al. (2007) Mortality patterns among workers exposed to acrylamide:
updated follow-up. J Occup Environ Med 49(l):82-95.
Martins, C; Oliveira, NG; Pingarilho, M; et al. (2007) Cytogenic damage induced by acrylamide and glycidamide in
mammalian cells: correlation with specific glycidamide-DNA adducts. Toxicol Sci 95(2):383-390.
Marty, J-P; Vincent, C-M. (1998) In vitro percutaneous absorption of acrylamide across human skin. Paris, France,
Faculty of Pharmacy, Universite de Paris Sud.
Maurissen, JPJ; Weiss, B; Davis, HT. (1983) Somatosensory thresholds in monkeys exposed to acrylamide.
Toxicol Appl Pharmacol 71:266-279.
288
-------
Mayerhofer, A; Hammings, HC; Snyder, GL; et al. (1999) Functional dopamine-1 receptors and DARPP-32 are
expressed in human ovary and granulosa luteal cells in vitro. J Clin Endocrinol 84:257-264.
McCollister, DD; Oyen, F; Rowe, VK. (1964) Toxicology of acrylamide. Toxicol Appl Pharmacol 6:172-181.
McConnell, RF; Western, HH; Ulland, BM; et al. (1992) Proliferative lesions of the testes in rats with selected
examples from mice. In: Guides for Toxicologic Pathology, STP/ARP/AFIP. Washington, DC, pp. 1-32. (as cited
in latropolous et al., 1998).
Mei, N; Hu, J; Churchwell, MI; et al. (2008) Genotoxic effects of acrylamide and glycidamide in mouse lymphoma
cells. Food Chem Toxicol 46(2):628-636.
Michels, KB; Rosner, BA; Chumlea, WC; et al. (2006) Preschool diet and adult risk of breast cancer. Int J Cancer
118(3):749-754.
Miller, MJ; Carter, DE; Sipes, IG. (1982) Pharmacokinetics of acrylamide inFischer-334 rats. Toxicol Appl
Pharmacol 63:36-44.
Moore, MM; Amtower, A; Doerr, C; et al. (1987) Mutagenicity and clastogenicity of acrylamide inL5178Y mouse
lymphoma cells. Environ Mutagen 9:261-267.
Mori, H; Arakawa, S; Ohkawa, T; et al. (1994) The involvement of dopamine in the regulation of steroidogenesis in
rat ovarian cells. Horm Res 41:36-40.
Moser, VC; Anthony, DC; Sette, WF; etal. (1992) Comparison of subchronic neurotoxicity of 2-hydroxyethyl
acrylate and acrylamide in rats. Fundam Appl Toxicol 18:343-352.
Mottram, DS; Wedzicha, BL; Dodson, AT. (2002) Acrylamide is formed in the Maillard reaction. Nature 19:448-
449.
Mucci, LA; Dickman, PW; Steineck, G; et al. (2003) Dietary acrylamide and cancer of the large bowel, kidney, and
bladder: absence of an association in a population-based study in Sweden. Br J Cancer 88:84-89.
Mucci, LA; Lindblad; P; Steineck, G; et al. (2004) Dietary acrylamide and risk of renal cell cancer. Int J Cancer
109(5):774-776.
Mucci, LA; Sandin, S; Baiter, K; et al. (2005) Acrylamide intake and breast cancer risk in Swedish women. JAMA
293(11):1326-1327.
Mucci, LA; Adami, HO; Wolk, A. (2006) Prospective study of dietary acrylamide and risk of colorectal cancer
amongwomen. Int J Cancer 118(1):169-173.
Mtiller, W; Engelhart, G; Herbold, B; et al. (1993) Evaluation of mutagenicity testing with Salmonella typhimurium
TA102 in three different laboratories. Environ Health Perspect 101(suppl 3):33-36.
Mulloy, KB. (1996) Two case reports of neurological disease in coal mine preparation plant workers. Am J Ind
Med 30:56-61.
Myers, JE; Macun, I. (1991) Acrylamide neuropathy in a South African factory: an epidemiologic investigation.
Am J Ind Med 19:487-493.
Nagao, T. (1994) Developmental abnormalities due to exposure of mouse paternal germ cells, preimplantation
embryos, and organogenic embryos to acrylamide. Congen Anom 34:35-46.
NRC (National Research Council). (1983). Risk Assessment in the Federal Government: Managing the Process.
National Academy Press: Washington, D.C.
289
-------
NRC (National Research Council) (1994). Science and judgment in risk assessment. Washington, DC: National
Academy Press.
NTP (National Toxicology Program). (1993) Final report on the reproductive toxicity of acrylamide (ACRL) (CAS
no. 79-06-1) in CD-I Swiss mice. NTIS Technical Report.
NTP (National Toxicology Program). (1989) Toxicology and carcinogenesis studies of N-methylolacrylamide in
F344/N rats and B6C3F! mice (gavage studies). Research Triangle Park, NC: US Department of Health and
Human Services. NIH Publication No. 89-2807. NTP TR 352.
NTP (National Toxicology Program). (2005) 11th Report on carcinogens. Research Triangle Park, NC: U.S.
Department of Health and Human Services.
NTP/CERHR (National Toxicology Program Center for the Evaluation of Risks to Human Reproduction). (2004)
Expert panel report on reproductive and developmental toxicity of acrylamide. Available online at
http ://cerhr. niehs. nih. gov/.
Neuman, F. (1991) Early indicators for carcinogenesis in sex-hormone-sensitive organs. Mutat Res 248:341-356.
Olesen, PT; Olsen, A; Frandsen, H. (2008) Acrylamide exposure and incidence of breast cancer among
postmenopausal women in the Danish diet, cancer and health study. Int J Cancer 122(9):2094-2100.
01stern, HB; Paulsen, JE; Alexander, J. (2007) Effects of perinatal exposure to acrylamide and glycidamide on
intestinal tumroigenesis in Min/+mice and their wild-type litter mates. Anticancer Res 27(6B):3855-3864.
Pacchierotti, F; Tiveron, C; D'Archivio, M; et al. (1994) Acrylamide-induced chromosomal damage in male mouse
germ cells detected by cytogenetic analysis of one-cell zygotes. Mutat Res 309:273-284.
Padilla, S; Atkinson, MB; Breuer, AC. (1993) Direct measurement of fast axonal organelle transport in the sciatic
nerve of rats treated with acrylamide. J Toxicol Environ Health 39:429-445.
Park, J; Kamendulis, LM; Friedman, MA; et al. (2002) Acrylamide-induced cellular transformation. Toxicol Sci
65:177-183.
Paulsson, B; Grawe, J; Tornqvist, M. (2002) Hemoglobin adducts and micronucleus frequencies in mouse and rat
after acrylamide or TV-methylolacrylamide treatment. Mutat Res 516:101-111.
Paulsson, B; Kotova, N; Grawe, J; et al. (2003a) Induction of micronuclei in mouse and rat by glycidamide,
genotoxic metabolite of acrylamide. Mutat Res535(l):15-24.
Paulsson, B; Athanassiadis, I. Rydgerg, P; et al. (2003b) Hemoglobin adducts from glycidamide: acetonization of
hydrophilic groups for reproducible gas chromatography/tandem mass spectrometric analysis. Rapid Commun.
Mass Spectrom. 17: 1859-1865.
Paulsson, B; Rannug, A; Henderson, AP; et al. (2005) In vitro studies of the influence of glutathione transferases
and epoxide hydrolase on the detoxification of acrylamide and glycidamide in blood. Mutat Res. 580(l-2):53-9.
Pearson, RG and Songstad, J. (1967) Application of the principle of hard and soft acids and bases to organic
chemistry. J Am Chem Soc 89:1827-1836.
Pelucchi, C; Galeone, C; Levi, F; et al. (2006) Dietary acrylamide and human cancer. Int J Cancer 118(2):467-471.
Pelucchi, C; Galeone, C; Dal Maso, L; et al. (2007) Dietary acrylamide and renal cell cancer. Int J Cancer
120(6):1376-1377.
290
-------
Perrault, SD. (2003) Distinguishing between fertilization failure and early pregnancy loss when identifying male-
mediated adverse pregnancy outcomes. Adv Exp Med Biol 518:189-198.
Petersen, DW; Kleinow, KM; Kraska, RC; et al. (1985) Uptake, disposition and elimination of acrylamide in
rainbow trout. Toxicol Appl Pharmacol 80:58-65.
Pleasure, DE; Mischler, KD; Engel, WK. (1969) Axonal transport of proteins in experimental neuropathies.
Science 166:524-525.
Post, EJ; McLeod, JG. (1977) Acrylamide autonomic neuropathy in the cat. J Neurol Sci 33:353-374.
Prentice, DE; Meikle, AW. (1995) A review of drug-induced Ley dig cell hyperplasia and neoplasia in the rat and
some comparisons with man. Human Exper Toxicol 14:562-572.
Rafales, LS; Lasley, SM; Greenland, RM; et al. (1983) Effects of acrylamide on locomotion and central monoamine
function in the rat. Pharmacol Biochem Behavior 19:635-644.
Ramsey, J; Young, J; Gorzinski, S. (1984) Acrylamide: Toxicodynamics in rats. Midland, MI: Dow Chemical
USA.
Raymer, JH; Sparacino, CM; Velez, GR; et al. (1993) Determination of acrylamide in rat serum and sciatic nerve by
gas chromatography-electron-capture detection. J Chromatogr 619(2):223-234.
Reid, E. (2003) Science in motion: common molecular pathological themes emerge in the hereditary spastic
paraplegias. J Med Genet 40:81-86.
Reid, E; Kloos, M; Ashley-Koch, A; et al. (2002) A kinesin heavy chain (KIF5A) mutation in heredity spastic
paraplegia (SPG 10). Am JHumGenet71:1189-1194.
Rice, JM. (2005) The carcinogenicity of acrylamide. MutatRes 580(l-2):3-20.
Roach, JA; Andrzejewski, D; Gay, ML; et al. (2003) Rugged LC-MS/MS survey analysis for acrylamide in foods.
J Agric Food Chem 51:7547-7554.
Robinson, M; Bull, RJ; Knutsen, GL; et al. (1986) A combined carcinogen bioassay utilizing both the lung adenoma
and skin papilloma protocols. Environ Health Perspect 68:141-145.
Russell, LB; Hunsicker, PR; Cacheiro, NL; et al. (1991) Induction of specific-locus mutations in male germ cells of
the mouse by acrylamide monomer. Mutat Res 262:101-107.
Russo, A; Gabbani, G; Simoncini, B. (1994) Weak genotoxicity of acrylamide onpremeiotic and somatic cells of
the mouse. Mutat Res 309:263-272.
Rutledge, JC; Generoso, WM; Shourbaji, A; et al. (1992) Developmental anomalies derived from exposure of
zygotes and first-cleavage embryos to mutagens. Mutat Res 296:167-177.
Sabri, MI; Spencer, PS. (1990) Acrylamide impairs fast and slow axonal transport in rat optic system. Neurochem
Res 15:603-608.
Sakamoto, J; Hashimoto, K. (1986) Reproductive toxicity of acrylamide and related compounds in mice- effects on
fertility and sperm morphology. Arch Toxicol 59:201-205.
Sakamoto, J; Kurosaka, Y; Hashimoto, K. (1988) Histological changes of acrylamide-induced testicular lesions in
mice. Exp MolPathol 48:324-334.
Sankaranarayanan, K. (1982) Genetic effects of ionizing radiation in multicellular eukaryotes and the evaluation of
genetic radiation hazards in man. Amsterdam: Elsevier, pp .385.
291
-------
Satchell, PM; McLeod, JG. (1981) Megaoesophagus due to acrylamide neuropathy. J Neurol Neurosurg Psychiat
44:906-913.
Schettgen, T; Weiss, T; Drexler, H; et al. (2003). A first approach to estimate the internal exposure to acrylamide in
smoking and non-smoking adults from Germany. Int. J. Hyg. EnViron. Health 206, 9-14.
Schmid, TE; Wang, X; Adler, ID. (1999) Detection of aneuploidy by multicolor FISH in mouse sperm after in vivo
treatment with acrylamide, colchicine, diazepam, or thiabenzole. Mutagenesis 14:173-179.
Sega, GA; Generoso, EE. (1990) Measurement of DNA breakage in specific germ-cell stages of male mice exposed
to acrylamide, using an alkaline-elution procedure. Mutat Res 242:79-87.
Sega, GA; Valdivia Alcota, RP; Tancongco, CP; et al. (1989) Acrylamide binding to the DNA and protamine of
spermiogenic stages in the mouse and its relationship to genetic damage. Mutat Res 216:221-230.
Sega, GA; Generoso, EE; Brimer, PA. (1990) Acrylamide exposure induces a delayed unscheduled DNA synthesis
in germ cells of male mice that is correlated with the temporal pattern of adduct formation in testis DNA. Environ
MolMutagen 16:137-142.
Segerback, D; Calleman, CJ; Schroeder, JL; et al. (1995) Formation of N-7-(2-carbamoyl-2-hydroxyethyl) guanine
in DNA of the mouse and the rat following i.p. administration of [14C]-acrylamide. Carcinogenesis 16:1161-1165.
Shelby, MD; Cain, KT; Hughes, LA; et al. (1986). Dominant lethal effects of acrylamide in male mice. Mutat Res
173:35-40.
Shelby, MD; Cain, KT; Cornett, CV; et al. (1987) Acrylamide: Induct ion of heritable translocations in male mice.
Environ Mutagen 9:363-368.
Shipp, A; Lawrence, G; Gentry, R; et al. (2006) Acrylamide: review of toxicity data and dose-response analyses for
cancer and noncancer effects. Crit Rev Toxicol 36:481-608.
Shiraishi, Y. (1978) Chromosome aberrations induced by monomeric acrylamide in bone marrow and germ cells of
mice. Mutat Res 57:313-324.
Sickles, DW. (1989a) Toxic neurofilamentous axonopathies and fast anterograde axonal transport. I. The effects of
single doses of acrylamide on the rate and capacity of transport. Neurtoxicology 10:91-102.
Sickles, DW. (1989b) Toxic neurofilamentous axonopathies and fast anterograde axonal transport. II. The effects of
single doses of neurotoxic and non-neurotoxic diketones and B,B'-iminodipropionitrile (IDPN) on the rate and
capacity of transport. NeuroToxicology 10:103-112.
Sickles, DW. (1991) Toxic neurofilamentous axonopathies and fast anterograde axonal transport.III. Recovery from
single injections and multiple dosing effects of acrylamide and 2,5-hexanedione. Toxicol Appl Pharmacol
108:390-396.
Sickles, DW. (1992) Toxic neurofilamentous axonopathies and fast anterograde axonal transport. IV. In vitro
analysis of transport following acrylamide and 2,5-hexanedione. Toxicol Lett 61:199-204.
Sickles, DW; Brady, ST; Testino, A; et al. (1996) Direct effect of the neurotoxicant acrylamide on kinesin-based
microtubule motility. J Neurosci Res 46:7-17.
Sickles, DW; Stone, JD; Friedman, MA. (2002a) Fast axonal transport: A site of acrylamide neurotoxicity?
NeuroToxicology 23:223-251.
Sickles, DW; Stone, D; Friedman, MA. (2002b) Fast axonal transport: a site of acrylamide neurotoxicity: a rebuttal.
NeuroToxicology 23:265-270.
292
-------
Smith, BL. (1991) Fibroadenomas. In: Harris, JR; Hellman, S; Henderson, S; et al., eds. Breast diseases.
Philadelphia: JP Lippincott Company, pp. 34-37.
Smith CJ; Perfetti TA; Rumple MA; et al. (2000) "IARC group 2A Carcinogens" reported in cigarette mainstream
smoke. Food Chem Toxicol 38(4):371-383.
Smith, MK; Zenick, H; Preston, RJ; et al. (1986) Dominant lethal effects of subchronic acrylamide administration in
the male Long-Evans rat. Mutat Res 173:273-277.
SNFA (Swedish National Food Agency). (2002) Press release, 24 April 2002. Available online at
http://www.slv.se/engdefault.asp.
Sobel, W; Bond, GG; Parsons, TW; et al. (1986) Acrylamide cohort mortality study. Br J Ind Med 43:785-788.
Sobels, F.H. (1977) Some problems associated with the testing for environmental mutagens and a perspective for
studies in comparative mutagenesis. Mutat Res 46:245-260.
Sobels, F.H. (1982) The parallelogram: an indirect approach for the assessment of genetic risks from chemical
mutagens. in: Bora, KC; Douglas, GR: Nestman, ER, eds. Progress in mutation research. Vol. 3., pp. 323-327.
Sobels, F.H. (1989) Models and assumptions underlying genetic risk assessment, Mutat Res, 114:93-105.
Solleveld, HA; Haseman, JK; McConnell, EE. (1984) Natural history of body weight gain, survival, and neoplasia
in F344 rat. J Natl Cancer Inst 72:929-940.
Solomon, JJ; Fedyk, J; Mukai, F; et al. (1985) Direct alkylation of 2'-deoxynucleosides and DNA following in vitro
reaction with acrylamide. Cancer Res 45(8):3465-3470.
Spencer, PS; Schaumburg, HH. (1974) A review of acrylamide neurotoxicity part II. Experimental animal
neurotoxicity and pathological mechanisms. Can J Neuro Sci 1:152-169.
Spencer, PS; Schaumburg, HH. (1991) A review of acrylamide neurotoxicity: Part 2. Experimental animal
neurotoxicity and pathologic mechanisms. Can J Neurol Sci 152:69.
Spencer, PS; Schaumburg, HH. (1977) Ultrastructural studies of the dying-back process IV. Differential
vulnerability of PNS and CNS fibers in experimental central-peripheral distal axonopathies. J Neuropath Exp
Neurol 36:300-320.
Spencer, PS; Sabri, MI; Schaumburg, HH; et al. (1979) Does a defect of energy metabolism in the nerve fiber
underlie axonal degeneration in polyneuropathies? Ann Neurol 5:501-507.
Stadler, RH; Blank, I; Varga, N; et al. (2002) Acrylamide fromMaillard reaction products. Nature 419:449.
Stamler, JS; Lamas, S; Fang, C. (2001) Nitrosylation: the prototypic redox-based signaling mechanism. Cell
106:675-683.
Stone, JD; Peterson, AP; Eyer, J; et al. (1999) Axonal neurofilaments are nonessential elements of toxicant-induced
reductions in fast axonal transport: video-enhanced differential interference microscopy in peripheral nervous
system axons. Toxicol Appl Pharmacol 161:50-58.
Stone, JD; Peterson, AP; Eyer, J; et al. (2000) Neurofilaments are non-essential elements of toxicant-induced
reductions in fast axonal transport: Pulse labeling in CNS neurons. NeuroToxicology 21(4):447-458.
Sublet, VH; Zenick, H; Smith, MK; et al. (1989) Factors associated with reduced fertility and implantation rates in
females mated to acrylamide-treated rats. Toxicology 55:53-67.
293
-------
Sumner, SC; MacNeela, JP; Fennell, TR. (1992) Characterization and quantitation of urinary metabolites of 1,2,3-
[13C]-acrylamide in rats and mice using 13C nuclear magnetic resonance spectroscopy. Chem Res Toxicol 5:81-89.
Sumner, SC; Fennell, TR; Moore, TA; et al. (1999) Role of cytochrome P450 2E1 in the metabolism of acrylamide
and acrylonitrile in mice. Chem Res Toxicol 12:1110-1116.
Sumner, SC; Williams, CC; Snyder, RW; et al. (2003) Acrylamide: a comparison of metabolism and hemoglobin
adducts in rodents following dermal, i.p., oral, or inhalation exposure. Toxicol Sci 75(2):260-270.
Suzuki, K; Pfaff, L. (1973) Acrylamide neuropathy in rats. An electron microscopic study of degeneration and
regeneration. Acta Neuropath 24:197-213.
Svensson, K; Abramsson, L; Becker, W; et al. (2003) Dietary intake of acrylamide in Sweden. Food Chem Toxicol
41:1581-1586.
Swaen, GM; Haidar, S; Burns, CJ; et al. (2007) Mortality study update of acrylamide workers. Occup Environ
Med. 64(6): 3 96-401.
Takahashi, H; Ohara, T; Hashimoto, K. (1971) Electrophysiological study of nerve injuries in workers handling
acrylamide. Int Arch Arbeitsmed 28:1-11.
Tanigawa, H; Onodera, H; Maekawa, A. (1987) Spontaneous mesotheliomas in Fischer rats - A histological and
electron microscopic study. Toxicol Pathol 15:157-163.
Tareke, E; Rydberg, P; Karlsson, P; et al. (2000) Acrylamide: A cooking carcinogen? Chem Res Toxicol
13:517-522.
Tareke, E; Rydberg, P; Karlsson, P; et al. (2002) Analysis of acrylamide, a carcinogen formed in heated foodstuffs.
J Agric Food Chem 50:4998-5006.
Tareke, E; Twaddle, NC; McDaniel, LP; et al. (2006) Relationships between biomarkers of exposure and
toxicokinetics in Fischer 344 rats and B6C3F1 mice administered single doses of acrylamide and glycidamide and
multiple doses of acrylamide.Toxicol Appl Pharmacol. 217(l):63-75.
Tegeris Laboratories. (1989) Initial submission: A lifetime oncogenicity study with acrylamide in rats (final report)
with cover letter dated 043092. Submitted under TSCA Section 8ECP. EPA Document No. 88-920002367. Fiche
No. OTS536419. (Volumes VIII to XI of Appendix XIX, which contained individual animal pathology reports,
were not included in the TSCA submission/fiche. These volumes were received on CD from Dr. M. Friedman in
May, 2003.)
Thier, R; Lewalter, J; Selinski, S; et al. (2002) Possible impact of human CYP2E1 polymorphisms on the
metabolism of acrylonitrile. Toxicol Lett 128(l-3):249-255.
Tilson, HA. (1981) The neurotoxicity of acrylamide: An overview. Neurobehav Toxicol Teratol 3:445-461.
Tilson, H; Cabe, PA. (1979) The effects of acrylamide given acutely or in repeated doses on fore- and hindlimb
function of rats. Toxicol Appl Pharmacol 47:253-260.
Tong, GC; Cornwell, WK; Means, GE. (2004) Reactions of acrylamide with glutathione and serum albumin.
Toxicol Lett 147(2):127-31.
Tornqvist, M; Paulsson, B; Vikstrom, AC; and Granath, F. (2008) Approach for cancer risk estimation of
acrylamide in food on the basis of animal cancer tests and in vivo dosimetry. J Agric Food Chem. 56(15):6004-12.
Tripathy, NK; Patnaik, KK; Nabi, MJ. (1991) Acrylamide is genotoxic to the somatic and germ cells ofDrosophila
melanogaster. Mutat Res 259:21-27.
294
-------
Tsuda, H; Shimizu, CS; Taketomi, MK; et al. (1993) Acrylamide; induction of DNA damage, chromosomal
aberrations and cell transformation without gene mutations. Mutagenesis 8(l):23-29.
Turek, FW; Desjardins, C. (1979) Development of leydig cell tumors and onset of changes in the reproductive and
endocrine systems of aging F344 rats. J Natl Cancer Inst 63:969-975.
Twaddle, NC; McDaniel, LP; Gamboa da Costa, G; et al. (2004) Determination of acrylamide and glycidamide
serum toxicokinetics in B6C3FJ mice using LC-ES/MS/MS. Cancer Lett 207(1):9-17.
UNSCEAR (United Nations Scientific Committee on the Effects of Atomic Radiation). (1986) Genetic and somatic
effects of ionizing radiation, report to the General Assembly, United Nations, New York.
Uphouse, L; Russell, M. (1981) Rapid effects of acrylamide on spiroperidol and serotonin binding in neural tissue.
Neurobehav Toxicol Teratol 3:281-284.
Uphouse, LL; Nemeroff, CB; Mason, G; et al. (1982) Interactions between "handling" and acrylamide on endocrine
responses in rats. NeuroToxicology 3:121-125.
Urban, M; Kawadias, D; Riedel, K; et al. (2006) Urinary mercapturic acids and a hemoglobin adduct for the
dosimetry of acrylamide exposure in smokers and nonsmokers. Inhal Toxicol 18(10)831-839.
U.S. EPA (Environmental Protection Agency). (1986a) Guidelines for the health risk assessment of chemical
mixtures. Federal Register 51(185):34014-34025.
U.S. EPA (Environmental Protection Agency). (1986b) Guidelines for mutagenicity risk assessment. Federal
Register 51(185):34006-34012.
U.S. EPA. (Environmental Protection Agency) (1988) Recommendations for and documentation of biological
values for use in risk assessment. Springfield, VA: National Technical Information Service. EPA Document No.
600/6-87/008. NTIS Document No. PB88-179874/AS.
U.S. EPA. (Environmental Protection Agency) (1991) Guidelines for developmental toxicity risk assessment. Fed
Reg 56(234):63798-63826.
U.S. EPA. (Environmental Protection Agency) (1994a) Interim policy for particle size and limit concentration
issues in inhalation toxicology: notice of availability. Federal Register 59:531799.
U.S. EPA. (Environmental Protection Agency) (1994b) Methods for derivation of inhalation reference
concentrations and application of inhalation dosimetry. Springfield, VA: National Technical Information Service.
NTIS Document No. PB2000-500023. Available online at http//:www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (Environmental Protection Agency) (1995) Use of the benchmark dose approach in health risk
assessment. Springfield, VA: National Technical Information Service. EPA Document No. 630/R-94/007. NTIS
Document No. PB95-213765. Available online at http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (Environmental Protection Agency) (1996) Guidelines for reproductive toxicity risk assessment. Federal
Register 61(212):56274-56322.
U.S. EPA. (Environmental Protection Agency) (1998a) Guidelines for neurotoxicity risk assessment. Federal
Register 63(93):26926-26954.
U.S. EPA. (Environmental Protection Agency) (1998b) Science policy council handbook: peer review.
Washington, DC: Office of Science Policy, Office of Research and Development. Springfield, VA: National
Technical Information Service. EPA Document No. 100-B-98-001. NTIS Document No. PB98-140726. Available
online at http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (Environmental Protection Agency) (1998c) Assessment of thyroid follicular cell tumors. Risk
Assessment Forum. Office of Research and Development, Washington, DC. EPA/630/R-97/002.
295
-------
U.S. EPA. (Environmental Protection Agency) (2000a) Science policy council handbook: peer review. 2nd edition.
Prepared by the Office of Science Policy, Office of Research and Development, Washington, DC. EPA 100-B-OO-
001. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (Environmental Protection Agency) (2000b) Science policy council handbook: risk characterization.
Prepared by the Office of Science Policy, Office of Research and Development, Washington, DC. EPA 100-B-OO-
002. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Environmental Protection Agency) (2000c) Benchmark dose technical guidance document [external
review draft]. EPA/630/R-00/001. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (Environmental Protection Agency) (2000d) Supplementary guidance for conducting for health risk
assessment of chemical mixtures. Risk Assessment Forum, Washington, DC; EPA/630/R-00/002. Available online
at http://www.epa.gov/iris/backgr d.htm.
U.S. EPA. (Environmental Protection Agency)(2002) A review of the reference dose and reference concentration
processes. Risk Assessment Forum, Washington, DC; EPA/630/P-02/0002F. Available onlilne at
http://www.epa.gov/iris/backgr-d.htm.
*U.S. EPA. (Environmental Protection Agency) (2003) National primary drinking water regulations. Organic
chemicals. Available online at http://www.epa.gov/safewater/mcl.html. (accessed June 10, 2003).
U.S. EPA (Environmental Protection Agency)(2005a) Guidelines for carcinogen risk assessment. Risk Assessment
Forum, Washington, DC. EPA/630/P-03/001B. Available online at
http://cfpub.epa. gov/ncea/cfm/recordisplay.cfm?deid= 116283.
U.S. EPA (Environmental Protection Agency) (2005b) Supplemental guidance for assessing susceptibility from
early life exposure to carcinogens, risk assessment forum, Washington, DC. EPA/630/R-03/003F.. Available online
at: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=l 16283.
U.S. EPA (Environmental Protection Agency) (2008). Toxicological Review of Acrylamide (CAS No. 79-06-1) in
Support of Summary Information on the Integrated Risk Information System (IRIS) External Review Draft (PDF)
Available online at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=187729
U.S. FDA (Food and Drug Administration) (2006) The U.S. FDA 2006 exposure assessment for acrylamide (slide
presentation by Dr. Michael DiNovi). Available online at http://www.cfsan.fda.gov/~dms/acryexpo.html.
U.S. FDA (Food and Drug Administration) (2009) The U.S. FDA Center for Food Safety and Applied Nutrition
(CFSAN) site for information on acrylamide and FDA research. Available online at
http://www.cfsan.fda.gov/~lrd/pestadd.htmMacrylamide.
Valdivia, RP; Lafuente, NM; Katoh, M. (1989) Acrylamide-induced chromosome-type aberrations in spermiogenic
stages evaluated in the first cleavage metaphases in the mouse. Environ Mol Mutagen 14(Suppl 15):205.
Verlaan, M; Te Morsche, RH; Roelofs, HM; et al. (2004) Genetic polymorphisms in alcohol-metabolizing enzymes
and chronic pancreatitis. Alcohol Alcohol 39(l):20-24.
Verschueren, K. (2001) Handbook of environmental data on organic chemicals, 4th ed., Vol. 1-2. New York, NY:
John Wiley & Sons, Inc., p. 124.
Vesper, HW; Licea-Perez, H; Meyers, T; et al. (2005) Pilot study on the impact of potato chips consumption on
biomarkers of acrylamide exposure. AdV Exp. Med. Biol. 561:89-96.
Vesper, HW; Bernert, JT; Ospina, M; et al. (2007) Assessment of the relation between biomarkers for smoking and
biomarkers for acrylamide exposure in humans. Cancer Epidemiol Biomark Prev 16(11):2471-2478.
296
-------
Vesper, HW; Slimani, N; Hallmans, G; et al. (2008) Cross-sectional study on acrylamide hemoglobin adducts in
subpopulations from the European prospective investigation into cancer and nutrition (EPIC) study. J Agric Food
Chem50(15):6046-6053.
Von Tungeln, LS; Churchwell, MI; Doerge, DR; et al. (2009) DNA adduct formation and induction of micronuclei
and mutations in B6C3Fi/Tk mice treated neonatally with acrylamide or glycidamide. Int J Cancer 124:2006-2015.
Voogd, CE; van der Stel, JJ; Jacobs, JJ. (1981) The mutagenic action of aliphatic epoxides. Mutat Res 89:269-282.
Walden, R; Squibb, RE; Schiller, CM. (1981) Effects of prenatal and lactational exposure to acrylamide on the
development of intestinal enzymes in the rat. Toxicol Appl Pharmacol 58:363-369.
Walker, K; Hattis, D; Russ, A; et al. (2007) Approaches to acrylamide physiologically based toxicokinetic modeling
for exploring child-adult dosimetry differences. J Toxicol Environ Health A. 70(24):2033-55.
Walker, RA; O'Brien, ET: Epstein, DL; et al. (1997) N-Ethylmaleimide and ethacrynic acid inhibit kinesin binding
to microtubules in a motility assay. Cell Motil Cytoskeleton 37:289-299.
Walum, E; Flint, OP. (1993) Selective effects of acrylamide, metylene bisacrylamide, and haloperidol on neuronal
development in rat embryo midbrain micromass cultures. In Vitro Toxicol 6:125-134.
Warr, T; Parry, J; Callander, R; et al. (1990) Methyl vinyl sulphone: a new class of Michael-type genotoxin. Mutat
Res 245:191-199.
Waters MD, Nolan C. (1995) EC/US workshop report: assessment of genetic risks associated with exposure to
ethylene oxide, acrylamide, 1,3-butadiene and cyclophosphamide. Mutat Res 330(1-2): 1-11.
WHO (1985). Environmental Health Criteria No. 49, Acrylamide. Geneva. Available online at
http://www.inchem.org/documents/ehc/ehc/ehc49.htm.
Wilson, KM; Vesper, HW; Tocco, P; et al. (2008) Validation of a food frequency questionnaire measurement of
dietary acrylamide intake using hemoglobin adducts of acrylamide and glycidamide. Cancer Causes Control [epub
ahead of print].
Wilson, KM; Baiter, K; Adami, HO; et al. (2009a) Acrylamide exposure measured by food frequency questionnaire
and hemoglobin adduct levels and prostate cancer risk in the cancer of the prostate in Sweden Study. Int J Cancer
124(10):2384-2390.
Wilson, KM; Mucci, LA; Cho, E; et al. (2009b) Dietary acrylamide intake and risk of premenopausal breast cancer.
AmJEpidemiol 169:954-961.
Wise, LD; Gordon, LR; Soper, KA. (1995) Developmental neurotoxicity evaluation of acrylamide in Sprague-
Dawley rats. Neurotoxicol Teratol 17:189-198.
Working, PK; Bentley, KS; Hurtt, ME; et al. (1987a) Dominant lethal assay of acrylonitrile and acrylamide in the
male rat. Environ Mutagen Suppl 9(8): 115.
Working, PK; Bentley, KS; Hurtt, ME; et al. (1987b) Comparison of the dominant lethal effects of acrylonitrile and
acrylamide in the male F344 rat. Mutagenesis 2:215-220.
Xiao, Y; Tales, AD. (1994) Increased frequencies of micronuclei in early spermatids of rats following exposure of
young primary spermatocytes to acrylamide. Mutat Res 309:245-254.
Yamada, T; Nakamura, J; Murakami, M; et al. (1995) Effect of chronic L-DOPA administration on serum
luteinizing hormone levels in male rats. Toxicology 97:173-182.
Young, JF; Luecke, RH; Doerge, DR. (2007) Physiologically based pharmacokinetic/pharmacodynamic model for
acrylamide and its metabolites in mice, rats, and humans. Chem Res Toxicol. 20(3):388-399.
297
-------
Zeiger, E; Anderson, B; Haworth, S. (1987) Salmonella mutagenicity tests: III. Results from the testing of 255
chemicals. Environ Mutagen 9(suppl 9): 1-110.
Zeiger, E; Recio, L; Fennell, TR; et al. (2009) Investigation of the low-dose response in the in vivo induction of
micronuclei and adducts by acrylamide. Toxicol Sci 107(l):247-257.
Zenick, H; Itope, E; Smith, MK. (1986) Reproductive toxicity associated with acrylamide treatment in male and
female rats. J Toxicol Environ Health 17:457-472.
Zenick, H; Clegg, ED; Perreault, SD; et al. (1994) Assessment of male reproductive oxicity: a risk assessment
approach. In: Hayes AW, ed. Principles and methods of toxicology. Chapter 27. 3rd ed. New York: Raven Press,
pp. 947.
Zhu, YJ; Zeng, T; Zhu, YB; et al. (2008) Effects of acrylamide on the nervous tissue antioxidant and sciatic nerve
electrophysiology in the rat. Neurochem Res 33(11)2310-2317.
298
-------
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION
The Toxicological Review of Acrylamide (December 2007 external review draft) has
undergone formal external peer review performed by scientists in accordance with EPA guidance
on peer review (U.S. EPA, 2006a). In this case, the external reviewers were members of a U.S.
EPA Scientific Advisory Board (SAB) panel who were tasked with providing written answers to
general questions on the overall assessment and on chemical-specific questions in areas of
scientific controversy or uncertainty. The Panel deliberated on the charge questions during a March
10-11, 2008 face-to-face meeting and discussed its draft report in a subsequent conference call on
July 16, 2008. The final draft of the panel's report was released on December 19, 2008. A summary
of the significant comments made by the SAB reviewers and EPA's responses to these comments
follow. In many cases the comments of the individual reviewers have been synthesized and
paraphrased in the development of Appendix A. EPA also received scientific comments from the
public. These comments and EPA's respopnses are included in a separate section of this
appendix.
The efforts of the review panel members are gratefully acknowledged. The following is a
summary of the SAB reviewers' comments and EPA responses.
1. Comment: SAB reviewers noted that the determination of accurate benchmark doses
for AA-induced neurotoxicity in the principal studies for the RfD and RfC may be
compromised by a lack of functional testing of neurotoxicity and the use of a relatively
insensitive measure, peripheral axonopathy detected by light microsopy, as the primary
index of neurotoxicity. Of concern was the possibility that, in looking at axonal
degeneration, preceding terminal degeneration may have been missed, particularly at
lower doses.
Response: Additional text was added to Section 5.1.1 (Choice of Principal Study and
Critical Effect—with Rationale and Justification) to discuss this area of uncertainty in
selecting the critical effect and principal study for the derivation of the RfD and RfC.
2. Comment: SAB reviewers noted that future studies may demonstrate AA effects on
male reproductive function at lower exposure levels than those associated with
neurotoxicity in animal studies. Of particular concern to the reviewers were:
(1) observations that impaired male reproductive performance (e.g., male-mediated
implantation losses) occurred in laboratory animals at oral exposure levels (-3-13
mg/kg-day) that are only "somewhat above" the lowest chronic doses associated with
neurotoxicity (1-2 mg/kg-day); (2) in-depth animal studies of dose-response
relationships for heritable germ cell effects are not available; (3) possible associations
08/28/2009 A-l DRAFT-DO NOT CITE OR QUOTE
-------
between human exposure to AA and reproductive effects (especially heritable germ cell
effects and effects on sperm endpoints) have not been adequately studied; and (4)
growing evidence that cigarette smoking, a known source of AA exposure, is associated
with altered male reproductive tissue endpoints, including sperm concentration,
morphology, motility, and DNA fragmentation.
Response: Several changes were made in Chapter 5 to explicitly note the SAB's
concerns about heritable germ cell effects and to strongly encourage further research,
including text to clearly state that: (1) the lower end of the range of animal oral doses
associated with lethal germ cell effects, particularly male-mediated implantation losses, is
close to the lowest exposure levels associated with neurotoxicity in orally exposed
animals; (2) dose-response relationship for heritable germ cell effects in AA-exposed
animals are not well described, particularly at dose levels below 50 mg/kg-day; and
(3) possible associations between human exposure to AA and altered sperm
characteristics have not been adequately studied. In addition, text was added to Chapters
5 and 6 to note that any future studies of possible associations between AA exposure and
sperm characteristics should adjust for smoking history and alcohol consumption,
especially due to the growing evidence of associations between cigarette smoking and
altered sperm endpoints.
The database deficiency UF used in the derivation of the RfD and RfC (UFDB =1) was
not changed in response to these comments, because, as discussed in several places in
Chapter 5, the animal database for oral exposure to AA is robust. Although the human
data are limited, they clearly demonstrate neurotoxicity as the predominant observable
noncancer adverse effect. Although animal studies for inhalation exposures are limited,
kinetic studies in animals and humans indicate no portal of entry effects (i.e., lung or GI
effects that would be unique to that route of exposure). The database identifies nerve
degeneration as the critical effect from chronic oral exposure. There are unresolved
issues that warrant further research including the MOA of AA-induced neurotoxicity, the
potential for behavioral or functional adverse effects not detected in the assays to date,
and the uncertainty that heritable germ cell effects may occur at lower than previously
reported doses. These issues, however, do not warrant applying a UF for database
deficiencies.
3. Comment: SAB reviewers noted that the discussion of MOA(s) for neurotoxicity in two
sections of the 2007 document (Section 4.6.1, pp 123-124; and Section 4.7.3, pp 134-
136) was confusing and recommended their incorporation into a single section. In
addition, several deficiencies in the discussion were noted (inadequate discussion of the
current molecular understanding of the MO As, AA adduct chemistry pertinent to AA
MO As for neurotoxicity, and the respective roles of AA, the parent compound, and its
08/28/09 A-2 DRAFT-DO NOT CITE OR QUOTE
-------
epoxide metabolite, GA). Panel members provided more specific text to address these
deficiencies and offered the text to EPA for consideration in revising the document.
Response: In response to this comment, the two sections were consolidated into one
section (Section 4.7.3.1 in the current version) and the recommended text was
incorporated into the revised discussion of the MO As for neurotoxicity.
4. Comment: SAB reviewers commented that the discussion of heritable germ cell effects
studies in Section 4.4 should be modified to add synthesis, analysis, and scrutiny.
Recommended issues to be discussed further include considerable deficiencies in the
database (e.g., characterization of dose-response relationships for certain endpoints), the
significance of the effect, and MO As for heritable germ cell effects (including DNA
adduct formation leading to mutagenicity, as well as clastogenic and mitotic spindle
effect mechanisms).
Response: Section 4.4 (Heritable Germ Cell Effects) was rewritten to add a new section
"Synthesis, and evaluation of heritable germ cell effects'', as well as further discussion of
potential MO As for heritable germ cell effects.
5. Comment: SAB reviewers considered studies other than the 2-year rat study by Johnson
et al. (1986) that could serve as the basis of the RfD and RfC, noting that the critical
effect of axonal degeneration detected by light microscopy is relatively less sensitive than
neuronal changes visible under the electron microscope or functional/behavioral
alterations. The Panel suggested that EPA generate an RfD from the Calleman et al.
(1994) study of AA-exposed workers for the purpose of comparison with the RfD based
on the rat data from Johnson et al. (1986), despite the limitations of this human study
including the small sample size and restriction to young adult males, the bias toward the
null from the healthy worker effect, the effect of confounding exposures to other
neurotoxicants, and the short duration of exposure. The Panel saw this exercise as a type
of sensitivity analysis to help to determine whether the RfD and RfC based on the
relatively insensitive measure of neurotoxicity in the rat study appears to be adequately
health protective.
Response: As suggested, EPA conducted an alternate derivation of the RfC based upon
the Calleman et al. (1994) study of worker exposure to AA via the inhalation and dermal
routes. This analysis can be found in Appendix F, and the details of Calleman et al.
(1994) study are presented in Section 4.1. EPA notes that the human data from the
Calleman et al. (1994) study are very limited due to the small number of subjects (n=41),
the narrow representation of the general public (i.e., workers), and by a number of
potentially confounding factors including concurrent exposure to another neurotoxin
(acrylonitrile), aggregate exposure via both dermal and inhalation routes, a composite
08/28/09 A-3 DRAFT-DO NOT CITE OR QUOTE
-------
index of neurotoxicity, and control groups of varying size and composition. There are
also no other human inhalation toxicity studies to support or challenge the reproducibility
or validity of the Calleman et al. (1994) study results.
6. Comment: The SAB panel could not reach consensus about EPA's decision that a
database deficiency UF greater than 1 was not needed in the derivation of the RfD and
RfC. Some panel members agreed with EPA's position, whereas others argued that
deficiencies in the database could lead to lower values for the RfD and RfC when they
are filled, thereby warranting a database deficiency UF. Database deficiencies pointed
out included: (1) the absence of robust neurotoxicity evaluations including
histopathology and electron microscopy coupled with systemic evaluation of functional
or behavioral endpoints at multiple time points in adult animals and in animals exposed
during early development to determine whether critical lifestage differences exist in
susceptibility to AA neurotoxicity; (2) the chronic neurotoxicity of AA has only been
assessed in rats; and (3) heritable germ cell effects have not been fully characterized,
especially dose-response relationships.
Response: See reponse to Question 2. Although the database deficiency UF used in the
derivation of the RfD and RfC (UFoB =1) was not changed in response to these
comments, it is agreed that there are unresolved issues that warrant further research
including the MOA of AA-induced neurotoxicity, the potential for behavioral or
functional adverse effects not detected in the assays to date, and the uncertainty that
heritable germ cell effects may occur at lower than previously reported doses. Text
changes were made in several places in Sections 5.1, 5.2, and 5.3 to emphasize these
points.
7. Comment: The SAB panel recommended that the document should include some
discussion of the potential for cumulative effects from exposure to AA and other
type-2 alkenes, which can produce similar noncancer effects via common mechanisms of
action. Evaluating the cumulative effects of type-2 alkenes was noted to be particularly
germane since human exposure can be pervasive due to environmental pollution (e.g.,
acrolein, acrylonitrile), contamination of food (e.g., AA, methyl acrylate), and
endogenous generation (e.g., acrolein, 4-hydroxy-2-nonenal, as reviewed by LoPachin et
al., 2008b.
Response: Text was added to Section 6.1 regarding this issue.
8. Comment: The SAB panel noted that the recalibrated Kirman PBTK model was
superior to the Young et al. model, but requested that EPA provide further descriptions of
the model and its parameterization/development and consider further evaluation and
refinement of the model with recently available toxicokinetics data. To justify the latter
08/28/09 A-4 DRAFT-DO NOT CITE OR QUOTE
-------
actions, the Panel noted several discrepancies between the PBTK predicted and measured
dose metrics and that the use of other available kinetic data (e.g., Hartmann et al., 2008;
Vesper et al., 2008, 2006; Fennell et al., 2006) to refine the model may help to resolve
the apparent discrepancies.
Response: The recalibrated Kirman model used in the external review draft of the
Toxicological Review of Acrylamide (December 2007), has since been published by
Walker et al. (2007). EPA agrees that an update of the Walker et al. (2007) PBPK model
parameter values, as well as the model structure, are needed based upon more recent
kinetic data, and the need to represent first pass lung clearance for an inhalation
exposure. For the purposes of the current assessment, however, recent kinetic data from
Doerge et al. (2005 a,b,c) and Tareke et al. (2006) in conjunction with the human adduct
data from Fennel et al. (2005) are sufficient to conduct a direct extrapolation of the rat
dose-response POD to a human equivalent administered dose based on equivalent AUCs
in the blood for AA or GA. Because AA or GA AUC in the blood was also the dose
metric that would be simulated with a PBPK model, the equivalent AUC method is a
viable alternative in lieu of using the uncertainty factor for interpecies toxicokinetic
differences, and thus EPA will not delay the assessment pending update and peer review
of a revised acrylamide PBPK model. The equivalent AUC method also has fewer,
relatively well supported parameters, and thus has inherently less uncertainty than a
PBPK model. An acrylamide PBPK model, of course, has far greater applicability to
generate dose metrics other than the AUC in blood, to compare different dosing
regimens, to simulate a far greater range of data, and to conduct variability analyses.
Thus, EPA recognizes and supports the important use of PBPK models in risk
assessment, and encourages the research community to continue to develop and submit
peer reviewed acrylamide PBPK models for consideration. In future reviews or revisions
to this document, EPA will consider available acrylamide PBPK model results, and if
needed, revise accordingly the reference values derived in this current version of the
Toxicological Review of Acrylamide.
Comment: The SAB panel noted that hemoglobin adduct data and other data in several
recent publications (Hartmann et al., 2008; Vesper et al., 2008, 2006; Fennell et al.,
2005) provide a means of estimating Human Equivalent Concentration (HECs) by
alternative empirical approaches that might be compared with the predictions from the
PBTK models.
Response: In accordance with the SAB recommendation (and as discussed in the
response to Question 8), EPA extensively reviewed and evaluated the kinetic data
needed to derive an HEC based on observed levels of hemoglobin adducts and serum
AUCs for AA or GA following a specified administered dose. EPA concluded that this
method, is indeed, a viable and simple (i.e., few parameters) means to derive the HEC.
Thus the AUC method was used in this current version of the assessment, and text has
08/28/09 A-5 DRAFT-DO NOT CITE OR QUOTE
-------
been added in Sections 3.5 and 5.1.2. Appendix E provides additional details the data and
calculations used to develop the in vivo second order adduct formation rate constants,
and the AUCs normalized to administered dose.
10. Comment: The SAB panel noted there are recent data indicating human variability in
the metabolism and toxicokinetics of AA (e.g., Harmann et al., 2008; Heudorf et al.,
2008; Vesper et al., 2008, 2006; Fennell et al., 2006, 2005) and asked EPA to consider
how to incorporate this information into the PBTK model.
Response: EPA agrees that human variability in both internal dose and response to AA
and GA is an important consideration in risk assessment, and an updated and peer
reviewed PBPK model could account for altered disposition of AA or GA (e.g., due to
variability in metabolism). More data, however, are needed to identify both the critical
factors and level of variability at low dose exposures in the general human population.
Assessing the impact of altered internal levels of AA and GA on the response is also
challenging because both are toxins, thus variability in metabolism leading to decreased
GA and increased AA might change the "spectrum" of adverse effects.
11. Comment: The SAB panel agreed with the use of the PBTK models to conduct route-
to-route extrapolations for noncancer effects and cancer. The panel noted that the value
generated in the default approach to estimating a human equivalent dose was very similar
to the value derived using the PBTK model.
Response: In the external review draft of the Toxicological Review of Acrylamide
(December 2007), the default approach to generate an RfD uses a UFA-TK of 3, and
resulted in an RfD similar to the value derived from the results with the recalibrated
Kirman et al. PBPK model. This could possibly have been due to the dose metric used in
the model simulation, which was AA AUC in blood, and which apparently scaled
roughly comparably to the uncertainty value of 3. The default approach, however, was
grossly in error for the oral slope factor that was based on the GA AUC dose metric (as a
metabolite of AA), so for the cancer assessment, a PBPK model, or the currently used
equivalent AUC method are superior to the use of a UFA-TK of 3.
12. Comment: The SAB panel recommended including a table displaying relevant
outcomes from reliable and well-performed studies for the following categories of
noncancer effects: neurotoxicity in the adult and developing organism, reproductive
toxicity including heritable germ cell effects, developmental toxicity, and general
systemic toxicity following various durations of exposure, as appropriate.
Response: A data array figure (Figure 5-1) was added to Section 5.1.1 with reliable
NOAELs and LOAELs for the following categories of noncancer effects from oral
exposure studies: subchronic and chronic effects; reproductive effects (including
08/28/09 A-6 DRAFT-DO NOT CITE OR QUOTE
-------
testicular and sperm effects, male-mediated implantation losses, and female reproductive
performance effects); and developmental effects (including fetal effects in standard
developmental bioassays and neurological assessments in offspring exposed during
gestation and beyond). The text in Section 5.1 was modified to guide the reader through
the data in the figure in defense of selecting degenerative nerve changes as the critical
effect for the RfD and RfC. There are no studies of heritable germ cell effects in orally
exposed animals, but the text in Section 5.1 discusses the results of the i.p. and dermal
exposure studies on this endpoint.
13. Comment: The SAB panel agreed with the inclusion of Section 5.5, QuantitatingRisk
for Heritable Germ Cell Effects, in the document, but asked that: (1) the risk
extrapolation factors (REFs) be explained in more detail; (2) the basis of the asuumed
number of human loci capable of mutating to dominantly expressed disease alleles
(1,000) be clarified in the modified direct approach; and (3) how, in the doubling dose
approach, the four data sets, each of which used high AA dosing rates, could accurately
predict the number of new disease in the offspring at low doses.
Response: Additional clarifying text was added to Section 5.5 to provide more
information on the bases for the REFs; however, the basis of the assumed number of
1,000 mutable genes in the modified direct approach was not available. It is agreed that,
in the doubling dose approach, extrapolation from the high-dose studies to low-dose
human exposure scenarios is of highly uncertain accuracy. The following statement was
added to emphasize this uncertainty: "Nevertheless, the accuracy of extrapolation of
these high exposure rates to the expected human exposure scenarios presented in
Table 5-16 is another major uncertainty in the calculations." The Panel agreed with the
recommendations in Section 5.5 and elsewhere in the document for further research and
data to fill the critical data gaps: in the REFs, the quantitative relationship between
genetic alterations in germ cells and heritable disease, and the shape of the low-dose
relationship.
14. Comment: The SAB panel agreed that the rationale and justification for the "likely to
be carcinogenic to humans" hazard descriptor was clearly described and that the
conclusion was scientifically supportable. The panel suggested that the rationale and
justification could be further expanded by:
a) Noting that the NTP and IARC have placed AA in cancer classification
groups similar to EPA's "likely to be carcinogenic to humans" category;
b) Emphasizing that concordance between tumor sites in animals and humans is
not as important as the observed concordance that pertinent modes of action
(e.g., somatic cell mutagenicity) operate in cells of humans and animals;
c) Adding primary CNS tumors to the list of experimental tumors induced by
AA;
08/28/09 A-7 DRAFT-DO NOT CITE OR QUOTE
-------
d) Emphasizing that the spectrum of tumors seen in AA-exposed rats is
completely consistent with a DNA-reactive MO A, based on published data
about other substances that induce or initiate the same kinds of neoplasms;
and
e) Emphasizing that the demonstration of AA' s tumor initiation activity by
multiple routes of administration provides strong support that AA causes
cancer by a DNA-reactive MOA.
Response: Additional text was added to Section 4.8 in general accordance with the
Panel's recommendations. It should be noted that the Agency generally does not include
information regarding other agencies determinations, such as IARC, in Toxicological
Reviews.
15. Comment: The SAB panel noted that rationale and justification for the weight of
evidence for a mutagenic MOA for AA carcinogenicity was sound and clearly and
objectively presented. The SAB panel further noted that hormonal disruption MO As
proposed for AA are highly speculative and supported by, at most, limited evidence. The
Panel made several recommendations for improving the presentation as follows:
a) Expand the discussion of biological plausibility and coherence beyond DNA
adducts and expand the human relevance section;
b) Reconsider the statement regarding the lack of relationship of cytogenetic
damage to a mutagenic mode MOA because the literature is full of such
correlations;
c) Consider adding the results of the case-control study of post-menopausal
breast cancer by Olesen et al. (2008) (reporting an association between
AA-hemoglobin level and risk for breast cancer after adjustment for smoking
status) to the discussion;
d) Emphasize that observations in humans of GA-hemoglobin adducts and GA
urinary metabolites demonstrate that internal exposure to GA, the mutagenic
AA metabolite, occurs in the general population at low levels of AA exposure;
e) Add discussion that AA/GA is not unique among DNA-reactive epoxides
(e.g., glycidol, ethylene oxide) in displaying carcinogenic action in the
thyroid, peritesticular mesothelium and mammary tissue and
f) Add discussion that CNS tumors were observed in both chronic bioassays and
that this observation represents strong evidence for a DNA-damaging
mechanism;
g) Add discussion of observations that short-term exposure to high doses of AA
in male F344 rats found no evidence for hormonal dysregulation in the
hypothalamus-pituitary-thyroid axis, yet some studies report associations with
hormonal changes, low level AA exposure, and cancer; and
h) The lack of data to describe dose-response relationships for DNA adducts or
pertinent mutagenic events in animals exposed to low levels of AA.
08/28/09 A-8 DRAFT-DO NOT CITE OR QUOTE
-------
Response: Additional text was added to Section 4.8 in accordance with the Panel's
recommendations.
16. Comment: The SAB panel recommended that data from the two chronic bioassays in
F344 rats (Friedman et al., 1995; Johnson et al., 1986) be modeled for the purpose of
deriving oral slope factors, noting that they did not agree that the Friedman study was a
better basis for the oral slope factor, that both are reasonably strong studies, and that the
strengths and limitation of both studies should be discussed in greater depth.
Response: In accordance with the Panel's concerns, the tumor incidence data from the
Johnson et al. (1986) bioassay were analyzed and compared with the results of the
analysis of the Friedman et al. (1995) bioassay. The numerical value of the oral slope
factor was indeed changed (increased risk) based on using the summed risk for increased
incidence of tumors in the Johnson et al. male rats. Both studies were needed to support
the use of the Johnson et al. data, and are now considered to be co-principal studies. Text
in Section 5.4 was modified to compare the strengths and limitations of the two
bioassays, and to describe the change in the derivation of the oral slope factor and
inhalation unit risk. Appropriate additions were also made to Appendix D to provide
details of the analysis for the tumor incidence data from the Johnson et al. (1986).
17. Comment: The SAB panel noted that the cancer dose-response analysis did not include
a factor to scale for pharmacodynamic differences in potency between animals and
humans, that such a factor should be considered as per the EPA Cancer Guidelines, and
that the potential for human pharmacokinetic variability to influence the cancer potency
estimate should be discussed qualitatively and quantitatively.
Response: For a mutagenic carcinogen, as is the case here, the current EPA Cancer
Guidelines do not include an adjustment for pharmacodynamic differences in
extrapolating from animals to humans. Rather, the method used is a low dose linear
extrapolation from the BMDL as a point of departure. The text does discuss variability
(kinetic and dynamic) in the human population as a source of uncertainty.
18. Comment: The SAB panel agreed that the AUC for GA is the best choice for the
internal dose metric used in deriving the oral slope factor, but asked EPA to consider the
inclusion of additional human data on variability to form GA from AA.
Response: Additional references and text have been added in Sections 3.3 and 3.5 that
address variability in human metabolism of AA and GA, and that present epidemiology
data on differences in GA-Val adduct levels, and urinary metabolites.
08/28/09 A-9 DRAFT-DO NOT CITE OR QUOTE
-------
19. Comment: The SAB panel agreed that the recommendation to use the age-dependent
adjustment factors is well justified and transparently and objectively described. The
panel noted that using the PBTK model to evaluate the effect of lifestage on CYP2E1 and
glutathione levels on internal exposure to GA and that such analysis could be used to
develop chemical-specific adjustment factors for early life exposure.
Response: EPA has developed specific guidelines for age dependent adjustment factors.
There is also insufficient data to determine whether children would be more or less
susceptible to AA induced toxicity because both the parent and GA metabolite are toxic.
Differences in AA or GA metabolism, or other kinetic drivers for different age groups
may alter the internal disposition of AA or GA, but the effects on the resulting spectrum
of adverse effects are not known at present.
20. Comment: The SAB panel noted that the discussion of uncertainties in the cancer
assessment and toxicity values was good, but could be improved by expanding the
discussion of human variability (specifically how human polymorphisms, or age-related
changes, in CYP2E1 and glutathione transferase(s) may influence cancer risk) and of the
limitation of not having another rodent species.
Response: Additional data and discussion has been added to Section 3.3 and 3.5,
however, as in the response to comment 19, it is difficult to predict the ultimate impact
on the spectrum of adverse effect from altered disposition of AA and GA due to
polymorphisms or enzyme status.
08/28/09 A-10 DRAFT-DO NOT CITE OR QUOTE
-------
Public comments and disposition
General comments are summarized below with a EPA response given for each. Specific
comments related to each subject area are provided in a bulleted list with the reviewer attribution
noted.
1. General comment: A hormonal MOA for AA carcinogenesis is possible.
Response: EPA agrees that disruption of hormone levels or signaling is a possible,
although well less supported, MOA for AA carcinogenesis. The experimental data
supporting this MOA are discussed in Section 4.8.3.2 of the Toxicological Review.
2. General comment: Multiple MO As are likely to be responsible for AA-induced cancer.
Response: EPA agrees that a mixed MOA is possible, i.e., an increased mutagenic
burden in hormonally-sensitive tissues with or without disruption of the hormonal
pathways (see Section 4.8.3.3 of the Toxicological Review).
3. General comment: The hormonal MOA is the most plausible MOA for AA
carcinogenesis.
Response: EPA disagrees with this conclusion and considers the data for the hormonal
MOA to be limited or lacking (see Section 4.8.3 of the Toxicological Review). The SAB
Review Panel agreed with EPA and considered the hormone disruption MOA to be
highly speculative. In addition, the SAB Panel concluded that the existing short-term
mouse studies in SENCAR, ICR (skin), and A/J (lung) show no such selectivity of
carcinogenicity for hormonally regulated tissues. Also, the CNS tumors observed in both
chronic AA cancer bioassays were considered strong evidence for a DNA-damaging
mechanism. The SAB Panel cited a short-term, high dose study of AA in male F344 rats
that found essentially no evidence for hormonal dysregulation in the hypothalamus-
pituitary-thyroid axis based on measurements of gene expression, neurotransmitters,
hormones, and histopathology (Bowyer et al., 2008).
4. General comment: Evidence for a mutagenic MOA for cancer is weak. DNA adducts
are found in both target and nontarget tissues. Genotoxicity findings interpreted as
mutagenicity may actually represent chromosome deletion. Genotoxic effects such as
micronuclei formation may exhibit a nonlinear dose-response.
Response: EPA disagrees with the conclusion that the weight of evidence for a
mutagenic MOA is weak. Section 4.8.3 of the Toxicological Review illustrates that the
majority of the data support a mutagenic MOA for AA carcinogenicity. The SAB panel
agreed, indicating that a sound rationale and justification for a mutagenic MOA were
provided in the Toxicological Review. The Panel discussed the fact that AA/GA is not
unique among DNA-reactive epoxides for carcinogenic action in thyroid, peritesticular
mesothelium, and mammary tissue (e.g., glycidol, ethylene oxide). In addition, the SAB
panel cited new data further supporting a mutagenic MOA (i.e., recent studies showing
GA-hemoglobin adducts or GA urinary metabolites in humans suggesting internal
exposure at low environmental concentrations). The observation that DNA adducts are
08/28/09 A-11 DRAFT-DO NOT CITE OR QUOTE
-------
found in both target and nontarget tissue does not alter the conclusion that adduct
formation is likely related to cancer. Target organ responses may be related to
differences in DNA repair and organ susceptibility to cancer. Findings in the mouse
lymphoma and Big Blue mouse assays are considered relevant for the identification of
DNA reactive carcinogens. EPA Guidelines for Carcinogen Risk Assessment indicate
that linear low dose extrapolation should be used for agents that are DNA-reactive and
demonstrate mutagenic activity (U.S. EPA, 2005a).
5. General comment: Combination of tumor types is not appropriate for hazard
identification or dose-response assessment of AA.
Response: The EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a)
allow for the combination of tumor data using several possible options. These include
adding risk estimates derived from different tumor sites and representing the overall response
in each experiment by counting animals with any tumor showing a statistically significant
increase.
6. General comment: Mortality adjustment of tumor data was not necessary and the use of
a time-to-tumor model was not warranted.
Response: Although not strictly necessary, EPA performs a mortality adjustment of
tumor data, when the raw data are available, to assess the impact of survival across dose
groups on cancer incidence.
7. General comment: The development and validation of the PBTK model for AA is not
fully described.
Response: The recalibrated Kirman model used in the external review draft of the
Toxicological Review of Acrylamide (December 2007), has since been published by
Walker et al. (2007). EPA agrees that an update of the Walker et al. (2007) PBPK model
parameter values, as well as the model structure, are needed based upon more recent
kinetic data, and the need to represent first pass lung clearance for an inhalation
exposure. For the purposes of the current assessment, however, recent kinetic data from
Doerge et al. (2005 a,b,c) and Tareke et al. (2006) in conjunction with the human adduct
data from Fennel et al. (2005) are sufficient to conduct a direct extrapolation of the rat
dose-response POD to a human equivalent administered dose based on equivalent AUCs
in the blood for AA or GA. Because AA or GA AUC in the blood was also the dose
metric that would be simulated with a PBPK model, the equivalent AUC method is a
viable alternative in lieu of using the uncertainty factor for interpecies toxicokinetic
differences, and thus EPA was not subject to delay of the assessment pending update and
peer review of a revised acrylamide PBPK model. The equivalent AUC method also has
fewer, relatively well supported parameters, and thus has inherently less uncertainty than
a PBPK model. An acrylamide PBPK model, of course, has far greater applicability to
generate dose metrics other than the AUC in blood, to compare different dosing
regimens, to simulate a far greater range of data, and to conduct variability analyses.
Thus, EPA recognizes and supports the important use of PBPK models in risk
08/28/09 A-12 DRAFT-DO NOT CITE OR QUOTE
-------
assessment, and encourages the research community to continue to develop and submit
peer reviewed acrylamide PBPK models for consideration. In future reviews or revisions
to this document, EPA will consider available acrylamide PBPK model results, and if
needed, revise accordingly the reference values derived in this current version of the
Toxicological Review of Acrylamide.
08/28/09 A-13 DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX B. MUTAGENICITY TEST RESULTS
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
Dose/Concentration
HID or
LEDb
Result
Reference
Bacterial gene mutation assays
Reverse mutation
Fluctuation test
S. typhimurium TA1535,
TA1537, TA98, TA100
S. typhimurium TA1535, TA97,
TA98, TA100
S. typhimurium TA1535,
TA1537, TA98, TA100, TA102
S. typhimurium TA1535,
TA1537, TA98, TA100
Escherichia coli WP2 uvrA~
S. typhimurium TA1535
S. typhimurium TA1535,
TA1537, TA1538, TA98,
TA100
S. typhimurium TA1535,
TA1537, TA1538, TA98,
TA100
K. pneumoniae ur~ pro"
10-10,000 ug/plate
±S9 activation
100-10,000 ug/plate
±S9 activation
1-100 mg/plate
±S9 activation
0.5-50 mg/plate
±S9 activation
Up to 5 mg/plate
±S9 activation
Up to 1 mg/plate
±S9 activation
0.5-5,000 ug/plate
±S9 activation
2-10 mg/mL
100
10,000
100
50
5
1
5,000
10
Weakly positive in TA98,
TA100 only with activation;
others negative
Negative
Negative
Negative in both systems
Negative
Negative
Negative
Negative
Zeigeretal., 1987
Knaapetal., 1988
Tsudaetal., 1993
Mtilleretal., 1993;
Jungetal., 1992
Lijinsky and
Andrews, 1980
Hashimoto and Tanii,
1985
Knaapetal., 1988
Nonmammalian gene mutation assays in vivo
Sex-linked
recessive lethal
Somatic mutation,
recombination
D. melanogaster
D. melanogaster
D. melanogaster
D. melanogaster
40-50 mM
abdominal injection
0.24-5 mM
larvae feeding
1-1.5
larvae feeding
(unit unspecified)
1-1.5 mM
larvae feeding
50
1
1
1
Negative
Positive
Weakly positive
Positive
Knaapetal., 1988
Tripathy et al., 1991
Knaapetal., 1988
Batiste-Alentorn et al.,
1991
08/28/2009
B-l
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
D. melanogaster
Dose/Concentration
0.25-5 mM
larvae feeding
HID or
LEDb
1
Result
Positive
Reference
Tripathy et al., 1991
Mammalian gene mutation assays in vitro
Mouse lymphoma
L5178YTK+A,tk locus
Mouse lymphoma
L5178YTK+/',tk locus
Mouse lymphoma
L5178Y/TK+/-,tk locus
Mouse lymphoma
L5 178Y TK+/", tk and HPRT
loci
Mouse lymphoma
L5178Y TK+/~, HPRT locus
Chinese hamster V79H3 cells,
HPRT locus
Human promyelocytic leukemia
HL-60 and NB4 cells, HPRT
locus
lOmM
0-0.85 mg/mL
without activation
0-18 mM
no activation
0.5-7.5 mg/mL
with or without
metabolic activation
0. 1-0.5 mg/mL
with cocultivated
mammalian cells
1-7 mM
no activation
0-700 mg/L
no activation
10
0.5
12
0.3
7
700
Positive (more pronounced
without activation)
Positive
Positive
Equivocal, increases only at
cytotoxic concentrations
Positive
Negative
Positive
Barfknecht et al.,
1988
Moore etal., 1987
Meietal.,2008b
Knaapetal., 1988
Knaapetal., 1988
Tsudaetal., 1993
Ao etal., 2008
Mammalian gene mutation assays in vivo
Transgenic mouse
liver ell,
lymphocyte
HPRT
Transgenic mouse
lacZ
Mouse B6C3F!/Tk+/-, (M, F)
spleen lymphocytes
tk and HPRT loci
Mouse B6C3F!/Tk+/-, (M, F)
spleen lymphocytes
tk and HPRT loci
Big Blue Mouse
(M,F)
Muta® Mouse
0-0.70 mmol/kg
i.p. injection post-
natal days 1, 8, 15
0-0.70 mmol/kg
i.p. injection post-
natal days 1-8
100, 500 mg/L
AA or GA
Drinking water for
3-4 weeks
5 x 50 mg/kg-day
i.p. injection
0.70
0.14
100
(est. 19-25
mg/kg-
day)
5 x 50
Negative
Positive
Positive
Weakly positive, no
statistical analysis
Von Tungeln etal.,
2009
Von Tungeln etal.,
2009
Manjanathaet al.,
2006
Hoornetal., 1993
08/28/2009
B-2
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Mouse spot test
Morphological
specific locus
Test system"
Muta® Mouse
Mouse embryos
(T x HT)P!
Mouse (C3H/R1 x lOl/Rl^
(M)
Mouse (102/E1 x CSH/E^Fj
(M)
Dose/Concentration
50-100 mg/kg
i.p. injection
1 x 50 or 75 mg/kg
3 x 50 or 75 mg/kg
i.p. injection
5 x 50 mg/kg
i.p. injection
100-125 mg/kg
i.p. injection
HID or
LEDb
100
50
3 x 50
50
100
Result
Negative
Positive
Positive
Positive (postspermatogonia)
Positive (postspermatogonia;
spermatogonia)
Reference
Krebs and Favor,
1997
Neuhauser-Klaus and
Schmahl, 1989
Russell et al., 1991
Ehling and
Neuhauser-Klaus,
1992
Chromosomal alterations in mammalian cells in vitro
Chromosomal
aberrations
Cell division
aberration
Chromosome
enumeration
Polyploidy
Spindle
disturbances
Micronucleus
Chinese hamster cells
Chinese hamster cell line (V79)
Chinese hamster cell line (V79)
Mouse lymphoma
L5178YTK+/--3.7.2cells
Chinese hamster lung cell line
DON:Wg3h
Chinese hamster lung fibroblast
LUC2 p5
Chinese hamster lung fibroblast
LUC2 p5
Chinese hamster cell line (V79)
Chinese hamster cell line (V79)
Seminiferous tubular segments
(spermatids from SD rats)
Human hepatoma G2 cells
0.5-5 mM
no activation used
0.1-3mg/mL
± S9 activation
0-2,000 uM
no activation
0.65-0.85 mg/mL
without activation
0.2-1 mg/mL
0.01-1 mg/mL
0.0125-0.5 mg/mL
0.5-5 mM
0.01-1 mg/mL
5-50 ug/mL
0-2.5 mM
2
1
2,000
0.75
0.2
0.01
0.5
1
0.01
50
0.625
Positive
Positive, with or without
metabolic activation
Weakly positive
Positive
Positive
Positive
Positive
Positive
Positive
Negative
Positive
Tsudaetal., 1993
Knaapetal., 1988
Martins et al., 2007
Moore etal., 1987
Warretal., 1990
Warretal., 1990
Warretal., 1990
Tsudaetal., 1993
Adleretal., 1993
Lahdetieetal., 1994
Jiang etal., 2007
08/28/2009
B-3
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
Dose/Concentration
HID or
LEDb
Result
Reference
Chromosomal alterations in mammalian cells in vivo
Chromosomal
aberrations
Mouse (101/E1 x CSH/E^Fj
(bone marrow cells)
Mouse (ICE-SPF)
(bone marrow cells)
Mouse (ddY)
(bone marrow cells)
Mouse (ddY)
(bone marrow cells)
Rat
(bone marrow cells)
Mouse (C57BL/6J)
(spleen lymphocytes)
Mouse (C57BL/6)
(splenocytes)
Mouse (101/E1 x CSH/E^Fj
(spermatogonia)
Mouse (C57BL/6J)
(spermatogonia)
Mouse (102/E1 x CSH/E^Fj
(spermatogonia)
Mouse (102/E1 x CSH/E^Fj
(spermatocytes)
Mouse (CFj)
(first cleavage embryos)
Mouse (B6C3FJ) (M)
(first cleavage one-cell zygotes,
examined after mating)
50-150 mg/kg
i.p. injection
100 mg/kg
i.p. injection
100-200 mg/kg
i.p. injection
500 ppm in diet for
7 to 21 days
(78 mg/kg-day)
100 mg/kg
i.p. injection
50-125 mg/kg
i.p. injection
100 mg/kg
i.p. injection
50-150 mg/kg
i.p. injection
50-125 mg/kg
i.p. injection
5 x 50 mg/kg-day
i.p. injection
100 mg/kg
i.p. injection
150 mg/kg
i.p. injection
75 and 125 mg/kg
or 5 x 50 mg/kg-day
i.p. injection
50
100
200
78
100
125
100
150
125
5 x50
100
150
75
Positive
Positive
Negative
Negative
Negative
Negative
Negative
Negative
Negative
Negative
Positive
Positive in embryos from
which the males had mated
6-8 days following treatment
(early spermatozoa stage)
Positive
Adleretal., 1988
Cihak and
Vontorkova, 1988
Shiraishi, 1978
Shiraishi, 1978
Krishna and Theiss,
1995
Backer etal., 1989
Kligermanetal., 1991
Adleretal., 1988
Backer etal., 1989
Adler, 1990
Adler, 1990
Valdivia et al., 1989
Pacchierottietal.,
1994
08/28/2009
B-4
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Polyploidy or
aneuploid
Spindle
disturbances
Micronucleus
Test system"
Mouse (B6C3FO
(first cleavage zygotes,
examined after mating)
Mouse bone marrow cells
Mouse bone marrow cells
Mouse (102/E1 x C3H/E1)
bone marrow cells
Mouse (101/E1 x CSH/E^Fj
bone marrow cells (M,F)
Mouse (ICR-SPF)
bone marrow cells (M)
Mouse (ICR-SPF)
bone marrow cells (M)
Mouse (Swiss NIH)
bone marrow cells (M,F)
Mouse (ICR-SPF)
bone marrow cells (M,F)
Rat (Sprague-Dawley)
bone marrow cells (M)
Rat
bone marrow cells
Mouse (BALB/c)
reticulocytes
Mouse (CBA)
reticulocytes
Mouse (CBA)
reticulocytes
Dose/Concentration
50 mg/kg
i.p. injection (males)
for 5 days before
mating
100-200 mg/kg
i.p. injection
500 ppm in the diet
for 7-21 days
(78 mg/kg-day)
120 mg/kg
i.p. injection
50-125 mg/kg
i.p. injection
100 mg/kg
i.p. injection
25-100 mg/kg-day
for 2 days
i.p. injection
136 mg/kg
i.p. injection
42.5-100 mg/kg-day
(1,2, or 3 days)
i.p. injection
100 mg/kg
i.p. injection
100 mg/kg
i.p. injection
50-100 mg/kg
i.p. injection
25-50 mg/kg
i.p. injection
0.18,0.35,
0.70 mmol/kg; i.p.
injection
HID or
LEDb
50
100
78
120
50
100
25
136
M: 42.5
F:55
100
100
50
25
0.35
Result
Positive
Positive
Positive
Negative
Positive
Positive
Positive
Positive
Positive
Negative
Negative
Positive
Positive, but results were not
analyzed statistically
Positive, but results were not
analyzed statistically
Reference
Marchettietal., 1997
Shiraishi 1978
Shiraishi 1978
Adleretal., 1993
Adleretal., 1988
Cihak and
Vontorkova, 1988
Cihak and
Vontorkova, 1988
Knaapetal., 1988
Cihak and
Vontorkova, 1990
Paulsson et al., 2002
Krishna and Theiss,
1995
Russoetal., 1994
Paulsson et al., 2002
Paulsson et al., 2003a
08/28/2009
B-5
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
Mouse (B6C3FO
reticulocytes and
normochromatic erythrocytes
(M,F)
Mouse (B6C3FO
reticulocytes and
normochromatic erythrocytes
(M,F)
Mouse (B6C3FJ)
reticulocytes
Mouse (B6C3FO
normochromatic erythrocytes
Mouse (wild-type or CYP2E1-
null)
erythrocytes (F)
Rat (Sprague-Dawley)
reticulocytes
Mouse (C57BL/6J) (M)
spleen lymphocytes
Mouse (C57BL/6) (M)
splenocytes
Mouse (C57BL/6J)
spermatids
Mouse (BALB/c)
spermatids
Rat (Lewis)
spermatids
Dose/Concentration
0,0.14,
0.70 mmol/kg
i.p. injection
postnatal days 1, 8,
15
0,0.14,
0.70 mmol/kg
i.p. injection
postnatal days 1-8
0-24 mg/kg-day for
28 days
gavage
0-24 mg/kg-day for
28 days
gavage
0, 25, 50 mg/kg-day
i.p. injection for
5 days
0.70, 1.4 mmol/kg
i.p. injection
50-125 mg/kg
i.p. injection
100 mg/kg
i.p. injection
10-100 mg/kg
i.p. injection
50-100 mg/kg or
4 x 50 mg/kg-day
i.p. injection
50-100 mg/kg or
4 x 50 mg/kg-day
i.p. injection
HID or
LEDb
0.70
0.70
6
4
25
0.7
50
100
50
50
100
Result
Negative
Negative
Positive
Positive
Positive (wild-type mice
only)
Positive, but nonmonotonic,
probably due to toxicity at
high dose
Positive
Positive
Positive
Positive
Positive
Reference
VonTungelnetal.,
2009
VonTungelnetal.,
2009
Zeigeretal., 2009
Zeigeretal.,2009
Ghanayem et al.,
2005b
Paulsson et al., 2003a
Backer etal., 1989
Kligermanetal., 1991
Collins et al., 1992
Russoetal., 1994
Xiao and Tates, 1994
08/28/2009
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Synaptonemal
complex
aberrations
Synaptonemal
complex
irregularities
Heritable
translocations
Reciprocal
translocations
Test system"
Rat (Sprague-Dawley)
spermatids
Mouse (C57BL/J6) (M)
germ cells
Mouse (C57BL/J6) (M)
germ cells
Mouse (C3H x 10 1^ (M)
Mouse (C3H/E1) (M)
Mouse (C3H/E1) (M)
Mouse (C3H/E1) (M)
Dose/Concentration
50-100 mg/kg or
4 x 50 mg/kg-day
i.p. injection
50-150 mg/kg
i.p. injection
50-150 mg/kg
i.p. injection
5 x 40-50 mg/kg-
day
i.p. injection
50-100 mg/kg
i.p. injection
5 x 50 mg/kg-day
dermal
5 x 50 mg/kg-day
i.p. injection
HID or
LEDb
4 x50
150
50
40
50
50
50
Result
Positive
Negative
Weakly positive, asynapsis in
meiotic prophase
Positive
Positive
Positive
Positive
Reference
Lahdetieetal., 1994
Backer etal., 1989
Backer etal., 1989
Shelby etal., 1987
Adleretal., 1994
Adler etal., 2004
Adler, 1990
DNA damage and repair and DNA adduct formation
Spore rec assay
In vitro DNA
breakage (comet
assay)
In vivo DNA
breakage (comet
assay)
Oxidative DNA
damage
Bacillus subtilis
H17(rec+)andM45(rec")
Human hepatoma G2 cells
Mouse (C3H x C57BL/10)F!
(M)
Mouse Pzh:SFIS (M)
bone marrow, spleen, liver,
kidney, lungs, testes
Mouse (wild-type or CYP2E1-
null)
leukocytes, liver, lung (F)
Human hepatoma G2 cells
1-50 mg/disk
0-20 mM
25-125 mg/kg
i.p. injection
0-125 mg/kg
i.p.injection
0, 25, 50 mg/kg-day
i.p injection, for
5 days
0-20 mM
10
2.5
25
50
25
5
Positive
Positive
Positive
Positive
Positive (wild-type mice
only)
Positive
Tsudaetal., 1993
Jiang et al., 2007
Sega and Generoso,
1990
Dobrzynska, 2007
Ghanayem et al.,
2005b
Jiang etal., 2007
08/28/2009
B-7
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
In vitro UDS
In vivo/m vitro
UDS
In vivo UDS
In vitro DNA
adducts
In vivo DNA
adducts
Test system"
Rat primary hepatocytes
Rat (F344) (M)
primary hepatocytes
Human mammary epithelial
cells
Rat (F344)(M)
hepatocytes
Rat (F344)(M)
spermatocytes
Mouse (C3H x lO^Fj and
(C3H x BLIO^ (M)
germ cells
Chinese hamster cells (V79)
Mouse lymphoma cells
L5178Y/TK+/-
Big Blue mouse embryonic
fibroblasts (with lambda phage
ell transgene)
Human bronchial epithelial cells
(with lambda phage ell
transgene)
Mouse (C3H x ELIO^
testis
Mouse (C3H x ELIO^ (M)
liver
Rat (Sprague-Dawley)
liver, lung, kidney, brain, testis
Mouse (BALB/c)
liver, kidney, brain
Neonatal mouse (B6C3F!)
whole body
Dose/Concentration
5-20 mM
0.01-1 mM
1-10 mM
1 x 100 mg/kg
5 x 30 mg/kg-day
gavage
1 x 100 mg/kg
5 x 30 mg/kg-day
gavage
7.8-125 mg/kg
i.p. injection
0-2,000 uM
0-20 mM
0, 0.0032 mM,
0.320 mM, 16 mM
0, 0.320 mM,
3.2 mM
46 mg/kg
i.p. injection
46 mg/kg
i.p. injection
46 mg/kg
i.p. injection
53 mg/kg
i.p. injection
50 mg/kg
i.p. injection
HID or
LEDb
17.5
1
1
1 x 100
5x30
5x30
7.8
2,000
20
0.0032
0.320
46
46
46
53
50
Result
Weakly positive
Negative
Positive
Negative
Positive
Positive
Positive
Negative
Positive
Positive
Positive
Positive
Positive
Positive
Positive
Reference
Barfknecht et al.,
1988
Butterworthetal.,
1992
Butterworthetal.,
1992
Butterworthetal.,
1992
Butterworthetal.,
1992
Segaetal., 1990
Martins et al., 2007
Meietal.,2008b
Besaratinia and
Pfeifer, 2004
Besaratinia and
Pfeifer, 2004
Segaetal., 1990
Segaetal., 1990
Segerbacketal., 1995
Segerbacketal., 1995
Gamboa da Costa et
al., 2003
08/28/2009
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
Mouse (C3H/HeNMTV) (M)
and(C57Bl/CN)(F)
liver, lung, kidney
Mouse (C3H/HeNMTV) (M)
liver, lung
Mouse (B6C3FJ)
lung, liver, spleen, bone marrow
Mouse (B6C3FO
lung, liver, spleen, bone marrow
Mouse (B6C3FO (M)
liver
Mouse (B6C3FO (M, F)
liver, lung, kidney, leukocytes,
testis
Mouse (B6C3FJ) (M, F)
liver
Rat (F344) (M, F)
liver, brain, thyroid, leukocytes,
mammary gland, testis
Rat (F344) (M, F)
liver
Dose/Concentration
50 mg/kg
i.p. injection
0-50 mg/kg
i.p. injection
0,0.14,
0.70 mmol/kg
i.p. injection
postnatal days 1, 8,
15
0,0.14,
0.70 mmol/kg
i.p. injection post-
natal days 1-8
0-24 mg/kg-day for
28 days
Gavage
50 mg/kg
i.p.injection
1 mg/kg-day
drinking water
50 mg/kg
i.p.injection
1 mg/kg-day
drinking water
HID or
LEDb
50
1
0.14
0.14
0.125
50
1
50
1
Result
Positive
Positive
Positive
Positive
Positive
Positive
Positive
Positive
Positive
Reference
Gamboa da Costa et
al., 2003
Gamboa da Costa et
al., 2003
VonTungelnetal.,
2009
VonTungelnetal.,
2009
Zeigeretal., 2009
Doergeetal.,2005a
Doergeetal.,2005a
Doergeetal.,2005a
Doerge etal., 2005a
Sister chromatic! exchange
In vitro
Chinese hamster V79 cells
Chinese hamster V79 cells
0.1-lmg/mL
± S9 activation
0.5-2.5 mM
no activation used
0.3
1
Positive at 0.3 mg/mL
without S9 and 1.0 mg/mL
withS9
Positive
Knaapetal., 1988
Tsudaetal., 1993
08/28/2009
B-9
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
In vivo
Test system"
Chinese hamster V79 cells
Mouse (C57BL/6J) (M)
spleen lymphocytes
Mouse (C57BL/6) (M)
splenocytes
Mouse (BALB/c)
differentiating spermatogonia
Dose/Concentration
0-2,000 uM
no activation
50-125 mg/kg
i.p. injection
100 mg/kg
i.p. injection
50-100 mg/kg
HID or
LEDb
2,000
50
100
50
Result
Positive
Positive
Positive
Positive
Reference
Martins et al., 2007
Backer etal., 1989
Kligermanetal., 1991
Russoetal., 1994
Cell transformation
Mouse C3H/10T1/2 clone 8
cells
Mouse NIH/3T3 cells
Mouse C3H/10T1/2 cells
Mouse BALB/c 3T3 cells
Syrian hamster embryo cells
Syrian hamster embryo cells
25-200 ug/mL
2-200 ug/mL
0.01-0.3 mg/mL
0.5-2 mM
0. 1-0.7 mM
0.001-10 mM
50
0.0125
0.3
1
0.5
10
Positive
Positive
Negative
Positive
Positive
Negative
Banerjee and Segal,
1986
Banerjee and Segal,
1986
Abernethy and
Boreiko, 1987
Tsudaetal., 1993
Park et al., 2002
Kasteretal., 1998
Germ cell effects
Sperm head DNA
alkylation
Sperm head
protamine
alkylation
Sperm head
abnormalities
Sperm aneuploidy
Mouse (C3H x 10 1^
Mouse (C3H x lO^Fj
Mouse (ddY)
Mouse (102/ElxC3H/ElFO (M)
125 mg/kg
i.p. injection
125 mg/kg
i.p. injection
0.3-1.2 mM in
drinking water for
4 weeks
0, 60, 120 mg/kg
i.p. injection
125
125
1.2
120
Weakly positive
Positive
Positive
Negative
Sega etal., 1989
Sega etal., 1989
Sakamoto and
Hashimoto, 1986
Schmidetal., 1999
aM = male, F = female.
bHID, highest ineffective dose/concentration for negative tests; LED, lowest effective dose/concentration for positive tests.
08/28/2009
B-10
DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX C. DOSE-RESPONSE MODELING FOR DERIVING THE RfD
All available models in the EPA Benchmark Dose Software (BMDS version 1.3.2) were
fit to incidence data for microscopically detected degenerative nerve changes in male and female
F344 rats from the two 2-year drinking water studies (Friedman et al., 1995; Johnson et al.,
1986). The data that were modeled are shown below in Table C-l. The benchmark response
predicted to affect 5% of the population (BMR5) was selected for the point of departure. A BMR
of 5% extra risk was selected for the following reasons (1) this effect level is considered to be a
minimal biologically significant change given the critical effect of degenerative nerve changes;
(2) the BMDLs remained near the range of observation; and (3) the 5% extra risk level is
supportable given the relatively large number of animals used in the prinicipal studies.
Table C-l. Incidence data for degenerative changes detected by light
microscopy in nerves of male and female F344 rats exposed to acrylamide in
drinking water for 2 years
Reference
Johnson et al. , 1986
(incidence of rats with changes in tibial
nerves: see Table 4-9)
Males (moderate to severe)
Females (slight to moderate)
Friedman et al., 1995"
(incidence of rats with minimal to mild
changes in sciatic nerves: see
Table 4-12)
Males
Females
Dose (mg/kg-day)
0
9/60
3/60
30/83
7/37
0
—
-
29/88
2/43
0.01
6/60
7/60
—
_
0.1
12/60
5/60
21/65
_
0.5
13/60
7/60
13/38
_
1.0
—
-
—
2/20
2.0
16/60b
16/61C
26/49c
_
3.0
—
-
—
38/86c
"Two control groups were included in the study design to assess variability in background tumor responses.
bStatistically significant trend test.
Statistically significantly different from control incidences.
All models provided adequate fits to the data for changes in tibial nerves in male and
female rats in the Johnson et al. (1986) study, as assessed by a %2 goodness-of-fit test (see
Tables C-2 and C-3 and following plots [Figures C-l and C-2] of observed and predicted values
from the various models). The log-logistic model provided the best fit for the male rat data as
assessed by Akaike's Information Criterion (AIC) and was thus selected to estimate a benchmark
dose (BMD) from the Johnson et al. (1986) data. The probit model provided the best fit of the
female rat data. Table C-4 lists the predicted doses associated with 10, 5, and 1% extra risk for
nerve degeneration in female and male rats in the Johnson et al. (1986) study.
6/15/2009
C-l
DRAFT-DO NOT CITE OR QUOTE
-------
Table C-2. Predictions (mg/kg-day) from models for doses associated with a
10% extra risk for nerve degeneration in male rats exposed to acrylamide in
drinking water
Model
Log-logistic3
Gammab
Multistage0
Quanta! linear
Weibullb
Probit
Logistic
Quanta! quadratic
Log-probita
BMD (ED10)
1.22
1.28
1.28
1.28
1.28
1.45
1.48
1.75
1.72
BMDL
0.57
0.64
0.64
0.64
0.64
0.87
0.90
1.19
1.06
X2 p-value
0.49
0.48
0.48
0.48
0.48
0.45
0.44
0.34
0.33
AIC
288.59
288.65
288.65
288.65
288.65
288.85
288.88
289.57
289.67
aSlope restricted to >1.
bRestrictpower>l.
'Restrict betas >0, degree of polynomial = 4.
Source: Johnson etal. (1986).
Log-Logistic IVbdel with 0.95 Confidence Level
0.4
0.35
s °3
$0.25
IQ2
gO. 15
LL
0.1
0.05
Log-Logistic
BMDL
BMD
0
0.5
15:0906/062003
Source: Johnson et al. (1986).
1
dose
1.5
Figure C-l. Observed and predicted incidences for nerve changes in male
rats exposed to acrylamide in drinking water for 2 years.
6/15/2009
C-2
DRAFT-DO NOT CITE OR QUOTE
-------
Table C-3. Predictions (mg/kg-day) from models for doses associated with a
10% extra risk for nerve degeneration in female rats exposed to acrylamide
in drinking water
Model
Probit
Logistic
Quantal quadratic
Quanta! linear
Log-probita
Gammab
Multistage0
Weibullb
Log-logistic3
BMD (ED10)
.19
.24
.40
0.98
.31
.10
.19
.11
.10
BMDL (LED10)
0.88
0.93
1.07
0.59
0.91
0.60
0.60
0.60
0.54
X2/7-value
0.62
0.62
0.59
0.59
0.59
0.41
0.41
0.41
0.41
AIC
220.68
220.69
220.92
220.75
220.94
222.69
222.68
222.69
222.69
aSlope restricted to >1.
bRestrictpower>l.
'Restrict betas >0, degree of polynomial = 3.
Source: Johnson etal. (1986).
Probit Model with 0.95 Confidence Level
0.4
0.35
0.3
CD
|>0.25
^ 0.2
c
o
=8 0.15
& 0.1
0.05
0
nuu
<
c
L
^_
^ —
0
^^— -""
BMDL
5
^^^~^~
^^*~~^^
1
dose
^^^^
^^^^^
^^~~~^^
BMD
1.5 ;
_.
-
-
-^^^
_
-
-
-
-
2
19:0506/102003
Source: Johnson et al. (1986).
Figure C-2. Observed and predicted incidences for nerve changes in female
rats exposed to acrylamide in drinking water for 2 years.
6/15/2009
C-3
DRAFT-DO NOT CITE OR QUOTE
-------
Table C-4. Predictions (mg/kg-day) from best-fitting models for doses
associated with a 10, 5, and 1% extra risk for nerve degeneration in male and
female rats exposed to acrylamide in drinking water
Model
Male
Log-logistic
Female
Probit
BMD
(ED10)
1.22
1.19
BMDL
(LED10)
0.57
0.88
BMD
(ED5)
0.58
0.67
BMDL
(LED5)
0.27
0.49
BMD
(EDO
0.11
0.15
BMDL
(LEDi)
0.05
0.11
Source: Johnson etal. (1986).
Several models in the software provided adequate fits to the data for minimal to mild
changes in sciatic nerves in male and female rats in the Friedman et al. (1995) study, as assessed
by a x2 goodness-of-fit test (see Tables C-5 and C-6 and following plots [Figures C-3 and C-4]
of observed and predicted values from the best-fitting models). The quantal quadratic model
provided the best fit to the male rat data as assessed by AIC and was selected to estimate a BMD.
The BMD associated with a 10% extra risk for minimal to mild changes in sciatic nerves for
male rats was 1.1 mg/kg-day and its lower 95% confidence limit (BMDL) was 0.8 mg/kg-day.
Table C-7 lists the predicted doses associated with 10, 5, and 1% extra risk for sciatic nerve
changes in female and male rats in the Friedman et al. (1995) study.
6/15/2009
C-4
DRAFT-DO NOT CITE OR QUOTE
-------
Table C-5. Predictions (mg/kg-day) from models for doses associated with a
10% extra risk for sciatic nerve changes in male rats exposed to acrylamide
in drinking water
Model
Quantal quadratic
Logistic
Probit
Gamma3
Multistage13
Quantal linear
Weibulf
Log-logistic0
Log-probif
BMD (ED10)
1.11
0.73
0.73
1.30
1.39
0.65
1.38
BMDL (LED10)
0.82
0.46
0.45
0.37
0.37
0.35
0.13
X2/>-value
0.96
0.89
0.89
0.86
0.86
0.86
0.86
AIC
422.84
423.15
423.16
424.82
424.82
423.28
424.82
NAd
NA
"Restrict power >1.
bRestrict betas >0, degree of polynomial = 4.
°Slope restricted to >1.
dNA = failed to generate a model.
Source: Friedman etal. (1995).
Quantal Quadratic Model with 0.95 Confidence Level
0.7 Quantal Quadratic
0.6
o 0.4
'
0.2
BMDL I
BMP
0
0.5
15:3006/062003
Source: Friedman et al. (1995).
1
dose
1.5
Figure C-3. Observed and predicted incidences for nerve changes in male
rats exposed to acrylamide in drinking water for 2 years.
6/15/2009
C-5
DRAFT-DO NOT CITE OR QUOTE
-------
Table C-6. Predictions (mg/kg-day) from models for doses associated with a
10% extra risk for sciatic nerve changes in female rats exposed to acrylamide
in drinking water
Model
Gamma3
Multistage13
Quanta! quadratic
Probit
Logistic
Quanta! linear
Weibulf
Log-probif
Log-logistic0
BMD (ED10)
2.48
2.02
1.68
1.20
1.23
1.04
2.75
BMDL (LED10)
0.93
0.86
1.35
0.88
0.91
0.65
0.93
XV-value
0.25
0.22
0.18
0.11
0.11
0.09
0.09
AIC
224.85
225.12
225.69
226.92
226.85
227.46
226.85
NAd
NA
a= Restrict power >1.
b= Restrict betas >0, degree of polynomial = 4.
°= Slope restricted to >1.
dNA = failed to generate a model.
Source: Friedman etal. (1995).
Garrma Multi-Hit Model wth 0.95 Confidence Level
0.6(
0.5
0.4
0.3
0.2
0.1
0
Gfemma Multi-hit
BMDL
BIVD
0 0.5
18:5406/102003
Source: Friedman et al. (1995).
1.5
dose
2.5
Figure C-4. Observed and predicted incidences for nerve changes in female
rats exposed to acrylamide in drinking water for 2 years.
6/15/2009
C-6
DRAFT-DO NOT CITE OR QUOTE
-------
Table C-7. Predictions (mg/kg-day) from best-fitting models for doses
associated with 10, 5, and 1% extra risk for sciatic nerve changes in male and
female rats exposed to acrylamide in drinking water
Model
Male
Quantal quadratic
Female
Gamma3
BMD
(ED10)
1.11
2.48
BMDL
(LED10)
0.82
0.93
BMD
(ED5)
0.77
2.25
BMDL
(LED5)
0.57
0.46
BMD
(EDO
0.34
1.86
BMDL
(LEDi)
0.25
0.09
""Restrict power >1.
Source: Friedman etal. (1995).
6/15/2009
C-7
DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX D. DOSE-RESPONSE MODELING FOR CANCER
METHODS
Data: Tumor data from the 2-year bioassays with F344 rats (Friedman et al., 1995;
Johnson et al., 1986) were modeled to obtain potential points of departure for deriving an oral
slope factor and inhalation unit risk (Tables D-l and D-2).
Table D-l. Incidence of tumors with statistically significant increases in the
Friedman et al. (1995) bioassay with F344 rats exposed to acrylamide in
drinking water
Reference/tumor type
Friedman et al., 1995/males"
Follicular cell adenoma/carcinoma
Tunica vaginalis mesotheliomab
Friedman et al., 199 5 / females"
Follicular cell adenoma/carcinoma
Mammary malignant/benign0
Combined mammary or thyroid tumord
Dose (mg/kg-day)
0
3/100
4/102
1/50
7/46
8/46
0
2/102e
4/102
1/50
4/50
4/50f
0.1
12/203
9/204
-
0.5
5/101
8/102
-
1.0
-
10/100
21/94J
27/94gJ
2.0
17/751
13/751
-
3.0
-
23/100J
30/95J
48/95^
"Two control groups were included in the study design to assess variability in background tumor responses.
blncidences reported herein are those originally reported by Friedman et al. (1995) and not in the reevaluation study
by latropoulos et al. (1998).
Incidences of benign and adenocarcinoma were added herein, based on an assumption that rats assessed with
adenocarcinoma were not also assessed with benign mammary gland tumors.
dMammary tissue was not available for testing in four animals in one control group, six animals in the 1 mg/kg-day
dose group and five animals in the 3 mg/kg-day dose group; these animals were not counted for either tumor type,
and subtracted from the total number of animals in the group.
eThe data reported in Table 4 in Friedman et al. (1995) lists one follicular cell adenoma in the second control group,
however, the raw data obtained in the Tegeris Laboratories (1989) report (and used in the time-to-tumor analysis)
listed no follicular cell adenomas in this group. The corrected number for adenomas (zero) and the total number
(two) of combined adenomas and carcinomas in the second control group are used in the tables of this assessment.
fOne animal had both a mammary and a thyroid tumor; this animal was only counted once in the combined total.
gThree animals had both a mammary and a thyroid tumor; these animals were only counted once in the combined
total.
hFive animals had both a mammary and a thyroid tumor; these animals were only counted once in the combined
total.
'Statistically significant (p < 0.05).
JStatistically significant (p< 0.001).
Source: Friedman etal. (1995).
6/15/2009
D-l
DRAFT-DO NOT CITE OR QUOTE
-------
Table D-2. Incidences of tumors with statistically significant increases in the
Johnson et al. (1986) bioassay with F344 rats exposed to acrylamide in
drinking water
Tumor type
Males
Thyroid (follicular cell) adenoma (no carcionomas found)
Tunica vaginalis mesothelioma
Pheochromocytomas, benign (adrenal)
Females
Mammary gland adenocarcinoma
Mammary gland benign tumors (adenoma, fibroadenoma, or
fibroma)
Mammary malignant + benign
CNS tumors of glial origin
Thyroid (follicular cell) adenoma or adenocarcinoma
Oral cavity, squamous cell carcinoma
Oral Cavity squamous papilloma
Oral cavity malignant + benign
Uterus adenocarcinoma
Clitoral adenoma, benign
Pituitary gland adenoma
Dose (mg/kg-day)
0
1/60
3/60
3/60
2/60
10/60
12/60
1/60
1/58
0/60
0/60
0/60
1/60
0/2
25/59
0.01
0/58
0/60
7/59
1/60
11/60
12/60
2/59
0/59
0/60
3/60
3/60
2/60
1/3
30/60
0.1
2/59
7/60
7/60
1/60
9/60
10/60
1/60
1/59
0/60
2/60
2/60
1/60
3/4
32/60
0.5
1/59
ll/60a
5/60
2/58
19/58
21/58
1/60
1/58
2/60
1/60
3/60
0/59
2/4
27/60
2.0
7/59a
10/60a
10/60a
6/61
23/6 la
29/61
9/6 la
5/60a
1/61
7/6 la
8/60
5/60a
5/5a
32/60a
aSignificantly different from control, p < 0.05, after Mantel-Haenszel mortality adjustment.
blncidences of benign and malignant tumors in these sites (mammary gland or oral cavity) were added herein, based
on an assumption that rats assessed with malignant tumors were not also assessed with tumors.
Source: Johnson etal. (1986).
Adenoma and carcinoma incidences within each site were combined by counting animals
with either of the responses, under the assumption that the tumor types represent different
realizations along a continuum of effects resulting from the same mechanism, as recommended
by the cancer guidelines (U.S. EPA, 2005a).
Extrapolation Models: When there are no biologically based models suitable for
modeling the available data, EPA has generally used one dose-response model to promote
consistency across cancer assessments. The multistage model (and the related multistage-
Weibull model) has been used by EPA in the vast majority of quantitative cancer assessments
because it is thought to reflect the multistage carcinogenic process and it fits a broad array of
dose-response patterns. Occasionally the multistage model does not fit the available data, in
which case an alternative model should be considered. The related multistage-Weibull model
has been the preferred model when individual data are available for time-to-tumor modeling,
which considers more of the observed response than does the simpler dichotomous response
model.
6/15/2009
D-2
DRAFT-DO NOT CITE OR QUOTE
-------
The multistage model is given by:
P(d) = 1 exp[-(q0 + qid + q2d2 + . . .
where P(d) represents the lifetime risk (probability) of cancer at dose d, and qt (for i = 0, 1, ..., 6)
are parameters estimated in fitting the model. The multistage model in BMDS (U.S. EPA
Benchmark Dose Software, version 1.3.2) was used for all multistage model fits.
The multistage-Weibull model is given by:
P(d,t) = 1- exp[-(q0 - qid- q^d. . . - q^ ) ft -
where P(d) represents the lifetime risk (probability) of cancer at dose d, t is the time to
observation of the tumor, to is the time from initiation of the tumor to the time it is observed, and
j and qt (for i = 0, 1, ..., 6) are parameters estimated in fitting the model. Most often there are not
sufficient data to estimate t0, which would at least involve interim sacrifice data at multiple
intervals. Without data which help identify times of tumor initiation from the concurrent study
or other studies, to is set to 0. The model was fit using the licensed software, MULTI-WEIB (KS
Crump and Company, Ruston, LA).
RESULTS
Friedman et al. (1995) Female Rat Tumor Modeling. For mammary gland tumors
(benign or malignant), the two female control groups were combined for modeling, obtaining
incidences of 1 1/96, 21/94, and 30/95 for the 0, 1, and 3 mg/kg-day groups. A one-stage
multistage model provided an adequate fit (p = 0.47) (see Figure D-l). The POD was based on
10% extra risk because this was the lowest level of extra risk that is consistent with the lower
end of the observed data range. The BMDio was estimated to be 1.2 mg/kg-day, with a BMDLio
of 0.78 mg/kg-day. For linear low-dose extrapolation, the slope factor associated with this site is
0.17(0.78 mg/kg-day), or 0.13 (mg/kg-dayf1 (see Table D-3).
For thyroid follicular cell adenomas or carcinomas, the two female control groups were
combined for modeling, obtaining incidences of 2/100, 10/100, and 23/100 for the 0, 1, and
3 mg/kg-day groups. A one-stage multistage model provided an adequate fit (p = 0.90; see
Figure D-2). The POD was based on 10% extra risk which represented the lowest extra risk
consistent with the lower end of the observed data range. The BMDio was estimated to be
1.3 mg/kg-day, with a BMDLio of 0.94 mg/kg-day. For linear low-dose extrapolation, the slope
factor associated with this site is 0. 17(0.94 mg/kg-day), or 0. 1 1 (mg/kg-day)"1 (see Table D-3).
6/15/2009 D-3 DRAFT-DO NOT CITE OR QUOTE
-------
Table D-3. Risk estimate derived from separate and combined incidence of
mammary or thyroid tumors in female F344 rats exposed to acrylamide in
drinking water
Tumor site
Mammary
(benign and malignant)
Thyroid
(adenomas and carcinomas)
Mammary or thyroid tumors
(tumor-bearing animals)
BMDR
(mg/kg-day)
1.2
1.3
1.2
BMDLR
(mg/kg-day)
0.78
0.94
0.88
Rat Cancer Slope
Factor3
(mg/kg-day)"1
0.13
0.11
0.23
a Rat Cancer Slope Factor is the upper bound on lifetime extra risk, calculated using R/BMDLR, where R = 0.1 for
mammary tumors or for thyroid tumors and 0.2 for the combination mammary or thyroid tumors.
Source: Friedman etal. (1995).
Despite a few early mortalities, there were no statistically significant incidences of early
mortalities in female rats exposed to acrylamide. Consequently, it was judged that the
multistage-Weibull model would not provide an appreciably different estimate of risk compared
to the multistage model for either tumor site.
The rat cancer slope factors corresponding to mammary tumors and to follicular cell
thyroid tumors in female F344 rats were very similar, 0.13 vs. 0.11 (mg/kg-day)"1. Given that
there was more than one tumor site, basing the unit risk on one tumor site may underestimate the
carcinogenic potential of acrylamide.
The EPA cancer guidelines (U.S. EPA, 2005a) suggest two approaches for calculating
risks when there are multiple tumor sites in a data set to assess the total risk from multiple tumor
sites. The simpler approach suggested in the cancer guidelines would be to estimate cancer risk
from the combined incidence of tumor-bearing animals. EPA traditionally used this approach
until the NRC (1994) Science and Judgment document made a case that evaluating tumor-
bearing animals would tend to underestimate overall risk when tumor types occur in a
statistically independent manner. The NRC-recommended an approach that adds distributions of
the individual tumor incidence to obtain a distribution of the summed incidence for all tumor
types. Both approaches were considered for this assessment.
Following the combined incidence approach, the combined incidence of female rats
bearing thyroid or mammary tumors from exposure to acrylamide in the drinking water (Tegeris
Laboratories, 1989) were considered for dose-response modeling. The data that were modeled
are shown in Table D-l, with the control groups combined as above. A one-stage multistage
model provided an adequate fit (p = 0.85) (see Figure D-3). The POD was based on 20% extra
risk because this was the lowest level of extra risk that is consistent with the lower end of the
observed data range, yielding a BMD20 of 1.2 mg/kg-day and a BMDL20 of 0.88 mg/kg-day. For
linear low-dose extrapolation, the slope factor associated with this site is 0.2/(0.88 mg/kg-day),
6/15/2009
D-4
DRAFT-DO NOT CITE OR QUOTE
-------
or 0.23 (mg/kg-day) l, approximately twofold higher than either of the risks estimated from the
individual sites.
Following the other recommendation of the EPA cancer guidelines for summing risks
from multiple tumor sites (U.S. EPA, 2005a; NRC, 1994), etiologically different tumor types—
that is, tumors in different organs—are not combined across sites prior to modeling, to allow for
the possibility that different tumor types can have different dose-response relationships.
Consequently, the modeling carried out separately for the two tumor types was used as a basis
for estimating a statistically appropriate upper bound on total risk. Note that this estimate of
overall risk describes the risk of developing any combination of the tumor types considered, not
just the risk of developing both simultaneously. The estimate involved the following steps:
4. It was assumed that the tumor types associated with acrylamide exposure were
statistically independent—that is, that the occurrence of mammary tumors was not
dependent on whether there were thyroid follicular cell adenomas/carcinomas. This
assumption cannot currently be verified and if not correct could lead to an overestimate
of risk from summing across tumor sites. NRC (1994) argued that a general assumption
of statistical independence of tumor-type occurrences within animals was not likely to
introduce substantial error in assessing carcinogenic potency from rodent bioassay data.
5. The central tendency or maximum likelihood estimates of unit potency (i.e., risk per unit
of exposure) were estimated by R/BMDR, and the upper confidence limit on the unit risk
estimated by R/BMDLR.
6. The central tendency or maximum likelihood estimates of unit potency (i.e., risk per unit
of exposure), estimated by R/BMDR, were summed across the multiple sites for male or
female F344 rats.
7. An estimate of the 95% upper bound on the summed unit risk was calculated by
assuming a normal distribution for the individual risk estimates, and deriving the
variance of the risk estimate for each tumor site from its 95% upper confidence limit
(UCL), according to the formula
95% UCL = MLE + (1.645 x SD)
where 1.645 is the t-statistic corresponding to a one-sided 95% confidence interval and
>120 degrees of freedom, and the standard deviation (SD) is the square root of the
variance of the MLE. The variances were summed across tumor sites to obtain the
variance of the sum of the MLE. The 95% UCL on the sum of the individual MLEs was
calculated from the variance of the sum of the MLE.
6/15/2009 D-5 DRAFT-DO NOT CITE OR QUOTE
-------
Table D-4 lists the site-specific risk estimates derived via multistage model extrapolation
to low exposures and the summed risks for female rats. First note that the individual unit risks
are virtually the same as those estimated using the POD approach above. Specifically, the
model-extrapolated slope factor for mammary tumors is 0.14 (mg/kg-day )~* compared with
0.13 (mg/kg-day )~* using the POD approach (Table D-2), and both methods lead to the same
slope factor for thyroid tumors, 0.11 (mg/kg-day )~\
There is some potential for greater model uncertainty in the model-extrapolated estimates
because it is unknown whether the multistage model adequately characterizes the underlying
dose-response relationship in this low-exposure range; however, it appears to be minimal for
these data. Consequently, the multistage model extrapolations introduce little additional
uncertainty into summing risks across these tumor sites.
The resulting 95% UCL on the summed risk of mammary tumors or thyroid follicular cell
adenomas/carcinomas for female F344 rats was 0.21 (mg/kg-day)"1, and the summed central
tendency was 0.16 (mg/kg-day)"1, about a 1.3-fold difference (Table D-4). The estimated risk
for mammary tumors was more variable, contributing about 70% of the overall variability in the
summed risk. As was the case with the tumor-bearing approach, the summed upper bound risk is
nearly twofold higher than either of the individual risks. For these data, the two approaches
yield very similar results.
Table D-4. Risk estimates derived from separate and summed dose-response
modeling of mammary and thyroid tumors in female F344 rats exposed to
acrylamide in drinking water
Tumor site
Mammary
(benign and malignant)
Thyroid
(adenomas and carcinomas)
BMDR
(mg/kg-day)
1.2
1.3
BMDLR
(mg/kg-day)
0.78
0.94
Risk of either mammary or thyroid tumors
Central tendency
oral potency"
(mg/kg-day)"1
8.3 x 10~2
7.7 x 1Q-2
0.16
Upper bound on
lifetime extra risk
(mg/kg-day)"1
0.13
0.11
0.21b
aCentral tendency oral potency = R/BMDR, where R = .1. The combined central tendency risk is the sum of the
individual oral potencies.
bThe rat cancer slope factor for the combination of tumor sites is the 95% UCL on the sum of the central tendency
unit potencies, not the sum of the individual slope factors; see the preceding text for derivation. This rat cancer
slope factor should not be used with exposures greater than 3 mg/kg-day, because above this level the dose-response
relationship is likely to be nonlinear.
Data Source: Friedman et al. (1995).
Friedman et al. (1995) Male Rat Tumor Modeling. As was done with the female rat
control groups, the two male rat control groups were combined into one control group:
5/202 males had thyroid follicular cell adenomas or carcinomas, and 8/202 had tunica vaginalis
mesotheliomas.
6/15/2009
D-6
DRAFT-DO NOT CITE OR QUOTE
-------
Because male rats in the highest dose group in the Friedman et al. (1995) study showed
early mortalities, models that adjusted for early mortality were fit to the data for tunica vaginalis
mesotheliomas and thyroid follicular cell adenomas and carcinoma. Pathology reports for
individual rats in the study (Tegeris Laboratories, 1989) were examined to extract time-to-death
and tumor occurrence data for each animal. Outputs from the computer program follow.
For TVM, MULTI-WEIB provided a model fit with a one-degree polynomial. The dose
associated with 10% extra risk (EDi0) at 108 weeks (i.e., full lifetime) was 1.2 mg/kg-day, with a
lower 95% confidence limit (LEDio) of 0.75 mg/kg-day. For linear low-dose extrapolation, the
slope factor associated with this site, using the POD approach, is 0.17(0.75 mg/kg-day), or
0.13 (mg/kg-dayf1 (see Table D-5).
For thyroid follicular cell adenomas or carcinomas, MULTI-WEIB provided a model fit
with a one-degree polynomial. The dose associated with 10% extra risk (EDio) at 108 weeks
(i.e., full lifetime) was 0.71 mg/kg-day, with an LEDio of 0.45 mg/kg-day. For linear low-dose
extrapolation, the slope factor associated with this site, using the POD approach, is
0.1/(0.45 mg/kg-day), or 0.22 (mg/kg-dayf1 (see Table D-5).
Table D-5. Risk estimates for separate and combined incidence of TVMs or
thyroid tumors in male rats exposed to acrylamide in drinking water
Incidence modeled
TVM
Follicular cell thyroid tumors
TVM or thyroid tumors
BMDRa
(mg/kg-day)
1.2
0.71
0.70
BMDLRa
(mg/kg-day)
0.75
0.45
0.30
Rat Cancer Slope
Factor
(risk level/BMDL)
(mg/kg-day)"1
1.3 x 10"1
2.2 x 1Q-1
3.3 x 10"1
aR = 10% extra risk.
bTumor-bearing animal method: Individual rats that had more than one of the tumor types were counted only once
(see Table D-l for incidences). For the NRC (1994) approach, the slope factor was 0.34 (see discussion below).
Data Source: Friedman et al. (1995).
The first recommended method in the EPA cancer guidelines for assessing total risk from
multiple tumor sites (U.S. EPA, 2005a; NRC, 1994) does not combine data from etiologically
different tumor types prior to modeling to allow for the possibility that different tumor types can
have different dose-response relationships. Note that the multistage-Weibull model yielded
distinctly different values of j, the parameter that describes the relationship of incidence with
increasing age, for the two tumor sites. For TVM, j was 1, indicating no difference between the
groups regarding incidence increasing with increasing age. For thyroid tumors, y was 3.7,
indicating relatively greater tumor incidence with increasing exposure as age increases.
Consequently, keeping the dose-response assessments separate maintains a better
6/15/2009
D-7
DRAFT-DO NOT CITE OR QUOTE
-------
correspondence with the observed biological events. The risks from the individual sites were
summed using the statistical approach as described for female rats above.
Table D-6 lists the site-specific risk estimates derived via multistage-Weibull model
extrapolation to low exposures, and the summed risks. First note that these individual unit risks
are virtually the same as those estimated using the POD approach above. Specifically, the
model-extrapolated slope factor for TVM is 0.14 (mg/kg-day)"1 compared with 0.13 (mg/kg-
day)"1, using the POD approach (Table D-5), and the model-extrapolated factor for thyroid
tumors is 0.23 (mg/kg-day)"1 compared with 0.22 (mg/kg-day)"1, using the POD approach (Table
D-3). While there is some potential for greater model uncertainty in the model-extrapolated
estimates, because it is unknown whether the multistage model adequately characterizes the
underlying dose-response relationship in this low-exposure range, it appears to be minimal for
these data. Consequently, the multistage model extrapolations introduce little additional
uncertainty into summing risks across these tumor sites.
Table D-6. Risk estimates derived from modeling separate and summed
incidence of TVM and thyroid tumors in male F344 rats exposed to
acrylamide in drinking water
Tumor site
TVM
Thyroid
(adenomas and carcinomas)
BMDR
(mg/kg-day)
1.2
0.71
BMDLR
(mg/kg-day)
0.75
0.45
Risk of either TVM or thyroid tumors
Central tendency
oral potency"
(mg/kg-dayJT1
8.3 x 10~2
0.14
0.22
Rat Cancer Slope
Factor
(mg/kg-day)'1
0.13
0.22
0.32b
aCentral tendency oral potency = R/BMDR, where R = .1. The combined central tendency risk is the sum of the
individual oral potencies.
bThe rat cancer slope factor for the combination of tumor sites is the 95% upper bound on lifetime extra risk (UCL)
on the sum of the central tendency unit potencies, not the sum of the individual slope factors; see the preceding text
for derivation. This rat cancer slope factor should not be used with exposures greater than 2 mg/kg-day, because
above this level the dose-response relationship is likely to be nonlinear.
Source: Friedman etal. (1995).
The resulting 95% UCL on the summed risk of TVM or thyroid follicular cell
adenomas/carcinomas for male F344 rats was 0.32 (mg/kg-day)"1, and the summed central
tendency was 0.22 (mg/kg-day)"1, about a 1.4-fold difference (Table D-6). The estimated risk
for thyroid tumors was the more variable, contributing about 73% of the overall variability in the
summed risk. The upper bound on the summed risks is about 1.4-fold higher than the risk of
thyroid tumors alone, the higher of the two individual risks.
Based on the analyses discussed above, the recommended upper bound estimate on rat
extra cancer risk from continuous, lifetime oral exposure to acrylamide is 0.3 (mg/kg-day)"1,
6/15/2009
D-8
DRAFT-DO NOT CITE OR QUOTE
-------
rounding the summed risk for male rats above to one significant digit.10 The slope factor can be
used to estimate cancer risks from doses up to approximately 2.0 mg/kg-day due to the
approximate linear dose-response throughout the observable range. This slope factor should not
be used with exposures greater than 2.0 mg/kg-day, the highest exposure in the male rat
bioassay, because above this level the cancer dose-response relationships are not likely to
continue linearly, and there are no data to indicate where this nonlinearity would begin to occur.
As in most risk assessments, extrapolation of study data to estimate potential risks to
human populations from exposure to acrylamide has engendered some uncertainty in the results.
The uncertainty falls into two major categories: model uncertainty and parameter uncertainty.
Model uncertainty refers to a lack of knowledge needed to determine which is the correct
scientific theory on which to base a model, whereas parameter uncertainty refers to a lack of
knowledge about the values of a model's parameters (U.S. EPA, 2005a). In the absence of a
biologically based model, a multistage model was the preferred model because it has some
concordance with the multistage theory of carcinogenesis and serves as a benchmark for
comparison with other cancer dose-response analyses. That said, it is unknown how well this
model or the linear low-dose extrapolation predicts low-dose risks for acrylamide. Also, while
the female rats did not appear to have as strong a carcinogenic response as the male rats, it is not
known which gender is more relevant for extrapolation of risk to humans.
Parameter uncertainty can be assessed through confidence intervals and probabilistic
analysis. Each description of parameter uncertainty assumes that the underlying model and
associated assumptions are valid. Uncertainty in the animal dose-response data can be assessed
through the ratio of BMDs to their BMDLs. For the tumor sites evaluated here, the ratios were
below a factor of 2, which is typical in similarly designed bioassays.
Johnson et al. (1986) F344 Rat Tumor Modeling. Data for tumors with statistically
significant increases in the Johnson et al. (1986) drinking water bioassays were modeled to
derive potential PODs for an oral slope factor and inhalation unit risk. For males, the tumor
types were tunica vaginalis mesotheliomas, thyroid follicular cell (adenoma/carcinoma), and
adrenal pheochromocytomas. For females, the tumor types were mammary gland tumors
(malignant and benign combined), thyroid follicular cell (adenoma/carcinoma), CNS tumors of
glial origin, and oral cavity tumors (malignant and benign combined). The data for uterine
adenocarcinomas and pituitary gland adenomas were not analyzed because the statistical
10 For comparison, the tumor-bearing animal approach applied to the combined incidence of thyroid or
TVM tumors (see Table D-l for data) led to a multistage-Weibull model with a three-stage polynomial, and j = 5.4.
The dose associated with a 10% extra risk (ED10) at 108 weeks (i.e., full lifetime) was 0.70 mg/kg-day, with an
LED10 of 0.30 mg/kg-day (see the last output). For linear low-dose extrapolation, the slope factor associated with
this combination, using the point of departure approach, is 0.1/(0.30 mg/kg-day), or 0.33 per mg/kg-day, virtually
identical to that estimated above using a method consistent with the NRC (1994) recommendation.
6/15/2009 D-9 DRAFT-DO NOT CITE OR QUOTE
-------
significance of the elevated incidences in the high-dose group was only demonstrated after
Mantel-Haenszel mortality adjustment (Table D-2). The data for clitoral adenomas were not
analyzed because the number of tissues examined in each group was small (<5, Table D-2).
The tumor data for each sex and tumor site were fit with the multistage model to estimate
the BMD and the 95% lower confidence limit on the BMD, the BMDL. Because individual
animal data were not available for the time of death from the Johnson et al. (1986) bioassay, no
adjustments or special modeling was done for early mortalities.
The POD results for separate modeling of the female mammary, thyroid, CNS, and oral
cavity tumor incidence data are presented in Table D-7. In addition, rat cancer slope factors for
the summed risks for these tumor types were calculated using the method described above, and
are presented in Table D-7. Table D-8 shows the calculations for summing risks across tumor
sites in the female rats.
Table D-7. Risk estimates derived from separate incidence of mammary,
thyroid, CNS, or oral cavity tumors in female F344 rats exposed to
acrylamide in drinking water
Tumor site
Mammary
(benign and malignant)
Thyroid follicular cell
(adenomas and carcinomas)
CNS tumors of glial origin
Oral cavity (malignant and benign)
BMD10
(mg/kg-day)
0.44
2.93
1.80
1.80
BMDL10
(mg/kg-day)
0.30
1.47
1.03
0.99
Slope factor3
(mg/kg-day)"1
0.34
0.07
0.10
0.10
1 Rat Cancer Slope factor is the upper bound on lifetime extra risk, calculated using R/BMDLR, where R = 0.1.
Data Source: Johnson et al. (1986).
6/15/2009
D-10
DRAFT-DO NOT CITE OR QUOTE
-------
Table D-8. Calculation of summed risks for tumors at several sites in female F344
rats exposed to acrylamide in drinking water in the Johnson et al. (1986) bioassay
Tumor site
Mammary
Thyroid
BMR
0.1
0.1
OSFa
(central
tendency)
(mg/kg-day) *
2.3 x 10"1
3.4 x 10"2
OSFb
(upper bound)
(mg/kg-day)1
3.4 x 10"1
6.8 x 10"2
t-
statistic
1.645
1.645
Standard
deviation
6.79 x 10"2
2.06 x 10"2
a2
4.61 x 10"3
4.25 x 10"4
Cumulative Variance 5.04 x 10"3 (Eo2)
Cumulative standard deviation 7.10 x 10"2 (\/Eo2)
Sum of central tendency risks 2.6 x 10"1 (mg/kg-day)"1
Upper bound on cumulative risk 3.8 x 10"1 (mg/kg-day)"1
Mammary
Thyroid
CNS
0.1
0.1
0.1
2.3 x 10"1
3.4 x 10"2
5.6 x 10"2
3.4 x 10"1
6.8 x 10"2
9.7 x 10"2
1.645
1.645
1.645
6.79x 10"2
2.06 x 10"2
2.52 x 10"2
4.61 x 10"3
4.25 x 10"4
6.37 x 10"4
Cumulative Variance 5.67 x 10"3 (Eo2)
Cumulative standard deviation 7.53 x 10"2 (\/Eo2)
Sum of central tendency risks 3.2 x 10"1 (mg/kg-day)"1
Upper bound on cumulative risk 4.4 x 10"1 (mg/kg-day)"1
Mammary
Thyroid
CNS
Oral cavity
0.1
0.1
0.1
0.1
2.3 x 10"1
3.4 x 10"2
5.6 x 10"2
5.6 x 10"2
3.4 x 10"1
6.8 x 10"2
9.7 x 10"2
1.0 x 10"1
1.645
1.645
1.645
1.645
6.79 x 10"2
2.06 x 10"2
2.52 x 10"2
2.76 x 10"2
4.61 x 10"3
4.25 x 10"4
6.37 x 10"4
7.64 x 10"4
Cumulative Variance 6.44 x 10"3 (Eo2)
Cumulative standard deviation 8.02 x 10"2 (\/Eo2)
Sum of central tendency risks 3.7 x 10"1 (mg/kg-day)"1
Upper bound on cumulative risk 5.0 x 10"1 (mg/kg-day)"1
3Derived by dividing the BMR (0.1) by the BMD10.
b Derived by dividing the BMR (0.1) by the BMDL10
The POD results for separate modeling of the male tunica vaginalis, thyroid, and adrenal
tumor incidence data are presented in Table D-9. In addition, rat cancer slope factors for
summed risks for these tumor types were calculated, and are presented in Table D-9. Table D-10
shows the calculations for summing risks across tumor sites in the male rats.
6/15/2009
D-ll
DRAFT-DO NOT CITE OR QUOTE
-------
Table D-9. Risk estimates derived from separate incidence of TVM, thyroid
tumors in male F344 rats exposed to acrylamide in drinking water
Tumor site
TVMb
Thyroid
(adenomas and carcinomas)
Adrenal pheochromo-cytoma
BMD10
(mg/kg-day)
0.27
2.04
2.55
BMDLR
(mg/kg-day)
0.16
1.12
1.08
Rat Cancer
Slope Factor"
(mg/kg-day)"1
0.61
0.09
0.09
A Rat Cancer Slope Factor is the upper bound on lifetime extra risk, calculated using R/BMDLR, where R = 0.1.
b An adequate fit could not be achieved using the full dataset, however, dropping the highest dose did produce an
adequate fit to the data with thel degree polynomial model ( %2 goodness of fit value = 0.08).
Data Source: Johnson etal. (1986).
Table D-10. Calculation of summed risks for tumors at several sites in male
F344 rats exposed to acrylamide in drinking water in the Johnson et al.
(1986) bioassay
Tumor site
TVM
Thyroid
BMR
0.1
0.1
OSFa
(central
tendency)
(mg/kg-day) *
3.7 x 10"1
4.9 x 10"2
OSFb
(upper bound)
(mg/kg-day)1
6.1 x 10"1
8.9 x 10"2
t-
statistic
1.645
1.645
Standard
deviation
1.48 x 10"1
2.45 x 10"2
a2
2.18 x 10"2
5.99 x 10"4
Cumulative Variance 2.24 x 10"2 (Eo2)
Cumulative standard deviation 1.50 x 10"1 (\/Eo2)
Sum of central tendency risks 4.2 x 10"1 (mg/kg-day)"1
Upper bound on cumulative risk 6.7 x 10"1 (mg/kg-day)"1
TVM
Thyroid
Adrenal
0.1
0.1
0.1
3.7 x 10"1
4.9 x 10"2
3.9 x 10"2
6.1 x 10"1
8.9 x 10"2
9.3 x 10"1
1.645
1.645
1.645
1.48 x 10"1
2.45 x 10"2
3.24 x 10"2
2.18 x 10"2
5.99 x 10"4
1.05 x 10"3
Cumulative Variance 2.35 x 10"2 (Eo2)
Cumulative standard deviation 1.53 x 10"1 (\/Eo2)
Sum of central tendency risks 4.6 x 10"1 (mg/kg-day)"1
Upper bound on cumulative risk 7.1 x 10"1 (mg/kg-day)"1
"Derived by dividing the BMR (0.1) by the BMD10.
b Derived by dividing the BMR (0.1) by the BMDL1C
6/15/2009
D-12
DRAFT-DO NOT CITE OR QUOTE
-------
DATA PRINTOUTS FOR BMD MODELING FOR THE
FRIEDMAN ET AL. (1995) TUMOR DATA SETS
FEMALE RATS, MALIGNANT AND BENIGN MAMMARY TUMORS, ACRYLAMIDE
DATA SOURCE: Tegeris Laboratories, 1989
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: G:\_BMDS\PCE\ACRYLAMIDE_FRIEDMAN_F.(d)
Gnuplot Plotting File: G:\_BMDS\PCE\ACRYLAMIDE_FRIEDMAN_F.plt
Mon Jun 05 11:32:19 2006
BMDS MODEL RUN
The form of the probability function is:
The parameter betas are restricted to be positive
Dependent variable = mamm
Independent variable = mg kg d
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Default Initial Parameter Values
Background = 0.131573
Beta(l) = 0.0827018
Beta(2) = 0
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.71
Beta(l) -0.71 1
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0835445 -0.0391471 0.288342
0.0565531 -0.0221264 0.199558
NA
6/15/2009 D-13 DRAFT-DO NOT CITE OR QUOTE
-------
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -143.354 3
Fitted model -143.609 2 0.51136 1 0.4746
Reduced model -149.278 1 11.8483 2 0.002674
d.f. = 1
P-value = 0.4713
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.18762
BMDL = 0.776448
Specified effect = 0.0001
Risk Type = Extra risk
Confidence level = 0.95
Specified effect = le-005
Risk Type = Extra risk
Confidence level = 0.95
Specified effect = le-006
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.1272e-005
BMDL = 1.12717e-005
6/15/2009
D-14
DRAFT-DO NOT CITE OR QUOTE
-------
0.45
0.4
0.35
T3
3 0.3
o
CD
-------
FEMALE RATS, THYROID FOLLICULAR CELL ADENOMAS OR CARCINOMAS, ACRYLAMIDE
DATA SOURCE: Tegeris Laboratories, 1989
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: G:\_BMDS\PCE\ACRYLAMIDE_FRIEDMAN_F.(d)
Gnuplot Plotting File: G:\_BMDS\PCE\ACRYLAMIDE_FRIEDMAN_F.plt
Mon Jun 05 11:38:01 2006
The form of the probability function is:
Dependent variable = thyroid
Independent variable = mg kg d
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Default Initial Parameter Values
Background = 0.0220015
Beta(l) = 0.0800466
Beta(2) = 0
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.71
Beta(l) -0.71 1
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0851609 -0.14648 0.187344
0.0507808 -0.0182223 0.180835
NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
6/15/2009 D-16 DRAFT-DO NOT CITE OR QUOTE
-------
has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -96.2398 3
Fitted model -96.2474 2 0.0150352 1 0.9024
Reduced model -108.069 1 23.6586 2 <.0001
d.f. = 1
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.29585
BMDL = 0.941045
Specified effect = le-005
Risk Type = Extra risk
Confidence level = 0.95
Specified effect = 0.0001
Risk Type = Extra risk
Confidence level = 0.95
Specified effect = le-006
Risk Type = Extra risk
Confidence level = 0.95
6/15/2009
D-17
DRAFT-DO NOT CITE OR QUOTE
-------
0.35
0.3
0.25
T3
CD
I °-2
"o
0.1
0.05
0
Multistage Model with 0.95 Confidence Level
Multistage
0
BMDL
BMP
0.5
1
1.5
dose
2.5
11:38 06/05 2006
Source: Friedman et al. (1995).
Figure D-2: Observed and predicted incidences for thyroid tumors in female
rats exposed to acrylamide in drinking water for 2 years.
6/15/2009
D-18
DRAFT-DO NOT CITE OR QUOTE
-------
FEMALE RATS, MAMMARY OR THYROID FOLLICULAR CELL TUMORS, ACRYLAMIDE
DATA SOURCE: Tegeris Laboratories, 1989
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: G:\_BMDS\PCE\ACRYLAMIDE_FRIEDMAN_F.(d)
Gnuplot Plotting File: G:\_BMDS\PCE\ACRYLAMIDE_FRIEDMAN_F.plt
Wed Jun 14 12:51:00 2006
The form of the probability function is:
Dependent variable = com
Independent variable = mg kg d
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Default Initial Parameter Values
Background = 0.130609
Beta(l) = 0.188994
Beta(2) = 0
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.67
Beta(l) -0.67 1
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0836998 -0.0368541 0.291243
0.0612613 0.0721662 0.312306
NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
6/15/2009 D-19 DRAFT-DO NOT CITE OR QUOTE
-------
has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -158.381 3
Fitted model -158.4 2 0.0370709 1 0.8473
Reduced model -175.349 1 33.9343 2 <.0001
Chi~2 = 0.04 d.f. = 1 P-value = 0.8471
Benchmark Dose Computation
Specified effect = 0.2
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.16078
BMDL = 0.88194
6/15/2009 D-20 DRAFT-DO NOT CITE OR QUOTE
-------
0.6
0.5
T3
CD
I °'4
•2 0.3
o
2
LJ_
0.2
0.1
Multistage Model with 0.95 Confidence Level
Multistage
BMDL
BMD
0 0.5 1
12:51 06/142006
Source: Friedman et al. (1995).
1.5
dose
2.5
Figure D-3: Observed and predicted incidences for mammary or thyroid
tumors in female rats exposed to acrylamide in drinking water for 2 years.
6/15/2009
D-21
DRAFT-DO NOT CITE OR QUOTE
-------
MALE RAT, TUNICA VAGINALIS MESOTHELIOMA, ACRYLAMIDE WITH INDUCTION TIME
ESTIMATED (TIME UNIT = WEEKS)
DATA SOURCE: Tegeris Laboratories, 1989
DATE: 06-07-03 TIME: 18:11:17
MULTI-WEIB (MAR 1985)
(C) COPYRIGHT CLEMENT ASSOCIATES, INC. 1983-1987
K.S. CRUMPS COMPANY, INC.
1201 GAINES STREET
RUSTON, LA 71270
(318)255-4800
THE 36 OBSERVATIONS AT LEVEL 1 WITH A DOSE OF .000000
TUMOR TUMOR
TIME # INDICATOR TIME # INDICATOR
41.0
69.0
74.0
77.0
79.0
85.0
88.0
90.0
93.0
95.0
96.0
98.0
99.0
101.0
103.0
105.0
107.0
108.0
1
1
1
2
1
1
1
5
3
3
2
1
9
3
12
4
4
2
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
2
58.0
73.0
76.0
78.0
82.0
87.0
89.0
91.0
94.0
96.0
97.0
99.0
100.0
102.0
104.0
106.0
107.0
108.0
1
2
2
2
2
2
3
5
1
1
1
1
3
3
9
17
57
36
1
1
1
1
1
1
1
1
1
2
1
2
1
1
1
1
1
1
THE 46 OBSERVATIONS AT LEVEL 2 WITH A DOSE OF .100000
TUMOR TUMOR
TIME # INDICATOR TIME # INDICATOR
46.0
63.0
67.0
72.0
78.0
80.0
82.0
84.0
86.0
89.0
91.0
93.0
94.0
96.0
97.0
98.0
1 1
1 1
1 1
1 1
2 1
1 1
3 1
1 1
2 1
1 1
1 1
1 2
4 1
3 1
2 1
5 1
61.0
65.0
68.0
76.0
79.0
81.0
83.0
85.0
87.0
90.0
92.0
93.0
95.0
97.0
98.0
99.0
1
1
1
1
1
1
2
2
3
2
1
1
1
1
1
2
1
1
1
2
1
1
1
1
1
1
1
1
1
2
2
1
6/15/2009
D-22
DRAFT-DO NOT CITE OR QUOTE
-------
100.0 1 2
101.0 2 1
102.0 5 1
104.0 10 1
106.0 1 2
107.0 1 2
108.0 1 2
100.0 3 1
102.0 1 2
103.0 11 1
105.0 6 1
106.0 15 1
107.0 57 1
108.0 38 1
THE 26 OBSERVATIONS AT LEVEL
TUMOR
TIME # INDICATOR TIME
2.0 1 1
49.0 1 1
77.0 1 1
78.0 1 1
86.0 2 1
90.0 3 1
95.0 1 1
98.0 3 1
101.0 1 1
103.0 7 1
105.0 3 1
106.0 6 1
107.0 30 1
32.0 1 1
72.0 1 1
78.0 1 2
79.0 2 1
88.0 1 1
91.0 1 1
97.0 2 1
99.0 3 1
103.0 2 2
104.0 4 1
106.0 3 2
107.0 2 2
108.0 19 1
THE 39 OBSERVATIONS AT LEVEL
TUMOR
TIME # INDICATOR TIME #
51.0 1 1
58.0 1 1
67.0 1 2
72.0 1 1
76.0 1 1
78.0 1 2
80.0 1 2
82.0 1 2
86.0 1 2
92.0 1 1
93.0 1 1
95.0 1 1
98.0 1 2
100.0 3 1
102.0 1 1
103.0 5 1
104.0 5 1
105.0 1 1
107.0 2 2
108.0 2 1
55.0 1 1
61.0 1 1
67.0 1 1
74.0 1 1
77.0 1 1
79.0 2 1
81.0 1 1
83.0 2 1
88.0 4 1
93.0 1 2
95.0 1 2
97.0 1 1
98.0 1 1
101.0 2 1
103.0 1 2
104.0 1 2
105.0 1 2
106.0 6 1
107.0 14 1
3 WITH A DOSE OF .500000
TUMOR
# INDICATOR
4 WITH A DOSE OF 2.00000
TUMOR
INDICATOR
FORM OF PROBABILITY FUNCTION:
P(DOSE) = 1 - exp( (-00 - 01 * D - 02 * DA2 - 03 * DA3) * (T - TO)AJ)
THE MAXIMUM LIKELIHOOD ESTIMATION OF:
PROBABILITY FUNCTION COEFFICIENTS
6/15/2009
D-23
DRAFT-DO NOT CITE OR QUOTE
-------
Q(0)=.384153255996E-03
Q(1)=.812864704009E-03
Q( 2)= .000000000000
Q( 3)= .000000000000
TIME FUNCTION COEFFICIENTS
T0= .000000000000
J= 1.00000000000
THE MAXIMUM LIKELIHOOD IS -133.655450741
MAXIMUM LIKELIHOOD ESTIMATES OF EXTRA RISK
WEIBULL LOWER CONFIDENCE LIMITS ON DOSE FOR FIXED RISK
RISK MLE DOSE
CONFIDENCE
LOWER BOUND UPPER BOUND LIMIT
ON DOSE ON RISK INTERVAL TIME
.100000 1.20015 .747920 .155548 95.0%
1.000000E-03 1.139660E-02 7.102226E-03 1.604166E-03
1.000000E-06 1.139090E-05 7.116445E-06 1.600645E-06
108.000
95.0% 108.000
95.0% 108.000
WEIBULL UPPER CONFIDENCE LIMITS ON RISK FOR FIXED DOSE
DOSE MLE RISK
.500000 4.294526E-02
2.00000 .161029
CONFIDENCE
UPPER BOUND
ON RISK
LIMIT
INTERVAL
TIME
6.801233E-02
.252245
95.0%
108.000
95.0% 108.000
NORMAL COMPLETION!
6/15/2009
D-24
DRAFT-DO NOT CITE OR QUOTE
-------
MALE RAT, FOLLICULAR CELL ADENOMA AND CARCINOMA, ACRYLAMIDE WITH NO INDUCTION
TIME ESTIMATED
DATA SOURCE: Tegeris Laboratories, 1989
[NOTE FOR THE RECORD: When SRC examined the individual male rat pathology reports provided in
the Tegeris Laboratories 1989 Report (provided on CD by Marvin Friedman), 2 rats with follicular
cell adenomas (#138 and #175), and one rat with a follicular cell carcinoma (#182) were found in
Control Group 1. These numbers agree with the numbers reported in Table 4 of the Friedman et
al. (1995) report. Among the individual animal pathology reports for male rats in Control Group 2,
however, SRC found two male rats with follicular cell carcinomas (#'s 335 and 345), but no male
rats with follicular cell adenomas. This does not agree with Table 4 in Friedman et al. (1995),
which reported that Control Group 2 had 2 male rats with follicular cell carcinomas and one male
rat with a follicular cell adenoma. The dose-response analysis described in here in Appendix D for
the male rat follicular cell adenomas plus carcinomas used the Tegeris Laboratories 1989 report
numbers. In addition, based on SRC's examination of the individual animal pathology reports, the
total number of male rats assessed for thyroid histopathology in the two control groups was 202
(rather than the 204 male rats included in these control groups); 2 male rats in Control Group 1
did not have thyroid histopathology.]
DATE: 06-09-03 TIME: 19:38:24
MULTI-WEIB (MAR 1985)
(C) COPYRIGHT CLEMENT ASSOCIATES, INC. 1983-198
K.S. CRUMPS COMPANY, INC.
1201 GAINES STREET
RUSTON, LA 71270
(318)255-4800
THE 35 OBSERVATIONS AT LEVEL 1 WITH A DOSE OF .000000
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS INDICATOR
41.0
69.0
74.0
77.0
79.0
85.0
88.0
90.0
93.0
95.0
97.0
99.0
101.0
103.0
104.0
106.0
1 1
1 1
1 1
2 1
1 1
1 1
1 1
5 1
3 1
3 1
1 1
10 1
3 1
12 1
8 1
16 1
58.0
73.0
76.0
78.0
82.0
87.0
89.0
91.0
94.0
96.0
98.0
100.0
102.0
104.0
105.0
107.0
1
2
2
2
2
2
3
5
1
3
1
3
3
1
4
2
1
1
1
1
1
1
1
1
1
1
1
1
1
2
1
2
6/15/2009
D-25
DRAFT-DO NOT CITE OR QUOTE
-------
107.0 58 1 108.0 2 2
108.0 36 1
THE 44 OBSERVATIONS AT LEVEL 2 WITH A DOSE OF .100000
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS INDICATOR
46.0 1 1 61.0 1 1
63.0 1 1 65.0 1 1
67.0 1 1 68.0 1 1
72.0 1 1 76.0 1 1
78.0 2 1 79.0 1 1
80.0 1 1 81.0 1 1
82.0 3 1 83.0 2 1
84.0 1 1 85.0 2 1
86.0 2 1 87.0 3 1
89.0 1 1 90.0 2 1
91.0 1 1 92.0 1 1
93.0 2 1 94.0 4 1
95.0 1 1 96.0 3 1
97.0 3 1 98.0 6 1
99.0 2 1 100.0 1 2
100.0 3 1 101.0 2 1
102.0 6 1 103.0 1 2
103.0 10 1 104.0 1 2
104.0 9 1 105.0 6 1
106.0 1 2 106.0 15 1
107.0 4 2 107.0 54 1
108.0 4 2 108.0 35 1
THE 26 OBSERVATIONS AT LEVEL 3 WITH A DOSE OF
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS
2.0 1 1 32.0 1 1
49.0 1 1 72.0 1 1
77.0 1 1 78.0 1 2
78.0 1 1 79.0 2 1
86.0 1 2 86.0 1 1
88.0 1 1 90.0 3 1
91.0 1 1 95.0 1 1
97.0 2 1 98.0 3 1
99.0 31 101.0 1 1
103.0 9 1 104.0 4 1
105.0 3 1 106.0 9 1
107.0 2 2 107.0 30 1
108.0 1 2 108.0 18 1
THE 38 OBSERVATIONS AT LEVEL 4 WITH A DOSE OF '.
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS
.500000
INDICATOR
2.00000
INDICATOR
51.0 1 1 55.0 1 1
58.0 1 1 61.0 1 1
67.0 2 1 72.0 1 1
6/15/2009
D-26
DRAFT-DO NOT CITE OR QUOTE
-------
74.0
77.0
79.0
80.0
82.0
86.0
92.0
93.0
97.0
98.0
101.0
103.0
104.0
105.0
106.0
107.0
1
1
1
1
1
1
1
1
1
1
2
2
1
2
4
9
2
1
2
1
1
1
1
1
1
1
1
2
2
1
1
1
76.0
78.0
79.0
81.0
83.0
88.0
93.0
95.0
98.0
100.0
102.0
103.0
104.0
106.0
107.0
108.0
1
1
1
1
2
4
1
2
1
3
1
4
5
2
7
2
1
1
1
1
1
1
2
1
2
1
2
1
1
2
2
1
FORM OF PROBABILITY FUNCTION:
P(DOSE) = 1 - exp( (-00 - 01 * D ) * (T - TO )AJ)
THE MAXIMUM LIKELIHOOD ESTIMATION OF:
PROBABILITY FUNCTION COEFFICIENTS
Q(0)=.107582747873E-08
Q( 1)= .420830494317E-08
TIME FUNCTION COEFFICIENTS
T0= .000000000000
J= 3.71285084690
THE MAXIMUM LIKELIHOOD IS -127.749366108
MAXIMUM LIKELIHOOD ESTIMATES OF EXTRA RISK
WEIBULL LOWER CONFIDENCE LIMITS ON DOSE FOR FIXED RISK
CONFIDENCE
LOWER BOUND UPPER BOUND LIMIT
RISK MLEDOSE ON DOSE ON RISK INTERVAL TIME
.100000 .705946 .451674 .151830 95.0% 108.000
1.000000E-03 6.703644E-03 4.289084E-03 1.562515E-03 95.0% 108.000
1.000000E-06 6.700295E-06 4.308122E-06 1.555270E-06 95.0% 108.000
WEIBULL UPPER CONFIDENCE LIMITS ON RISK FOR FIXED DOSE
CONFIDENCE
UPPER BOUND LIMIT
DOSE MLERISK ON RISK INTERVAL TIME
.500000 7.190726E-02 .110089 95.0% 108.000
2.00000 .258066 .372827 95.0% 108.000
NORMAL COMPLETION!
6/15/2009 D-27 DRAFT-DO NOT CITE OR QUOTE
-------
Time-to-Tumor Model Results for the Combined Incidence of Thyroid Tumors or TVM in Male Rats
Exposed to Acrylamide in the Drinking Water
MULTI-WEIB (MAR 1985)
(C) COPYRIGHT CLEMENT ASSOCIATES, INC. 1983-1987
K.S. CRUMPS COMPANY, INC.
1201 GAINES STREET
RUSTON, LA 71270
(318)255-4800
THE 28 OBSERVATIONS AT LEVEL 1 WITH A DOSE OF 0.000000
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS INDICATOR
42.0
74.0
77.0
82.0
89.0
91.0
94.0
96.0
97.0
101.0
103.0
105.0
107.0
108.0
1 1
1 1
1 1
2 1
3 1
2 1
1 1
2 1
1 1
3 1
5 1
1 1
25 1
13 1
73.0
76.0
78.0
87.0
90.0
93.0
95.0
96.0
99.0
102.0
104.0
106.0
107.0
108.0
1
2
2
1
4
1
3
1
5
3
2
10
5
1
1
1
1
1
1
1
1
2
1
1
1
1
2
2
THE 25 OBSERVATIONS AT LEVEL 2 WITH A DOSE OF .000000
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS INDICATOR
58.0 1 1
73.0 1 1
79.0 1 1
87.0 1 1
90.0 1 1
93.0 1 1
98.0 1 1
99.0 1 2
103.0 7 1
104.0 1 2
106.0 7 1
107.0 1 2
108.0 2 2
69.0
77.0
85.0
88.0
91.0
94.0
99.0
100.0
104.0
105.0
107.0
108.0
1
1
1
1
3
1
4
3
6
3
32
22
1
1
1
1
1
1
1
1
1
1
1
1
THE 48 OBSERVATIONS AT LEVEL 3 WITH A DOSE OF .100000
6/15/2009
D-28
DRAFT-DO NOT CITE OR QUOTE
-------
TUMOR TUMOR
TIME # OF ANIMALS INDICATOR TIME # OF ANIMALS INDICATOR
46.0 1 1
63.0 1 1
67.0 1 1
72.0 1 1
78.0 2 1
80.0 1 1
82.0 3 1
84.0 1 1
86.0 2 1
89.0 1 1
91.0 1 1
93.0 1 1
94.0 4 1
96.0 3 1
97.0 1 2
98.0 1 2
100.0 2 1
101.0 2 1
102.0 1 2
103.0 1 2
104.0 1 2
106.0 14 1
107.0 53 1
108.0 33 1
61.0 1 1
65.0 1 1
68.0 1 1
76.0 1 2
79.0 1 1
81.0 1 1
83.0 2 1
85.0 2 1
87.0 3 1
90.0 2 1
92.0 1 1
93.0 1 2
95.0 1 1
97.0 3 1
98.0 4 1
99.0 2 1
100.0 2 2
102.0 5 1
103.0 10 1
104.0 9 1
105.0 6 1
106.0 2 2
107.0 5 2
108.0 6 2
THE 28 OBSERVATIONS AT LEVEL 4 WITH A DOSE OF
TUMOR
TIME # OF ANIMALS
2.0 1 1
49.0 1 1
77.0 1 1
78.0 1 3
86.0 1 1
88.0 1 1
92.0 1 1
97.0 2 1
99.0 3 1
103.0 6 1
104.0 4 1
106.0 6 1
107.0 29 1
108.0 18 1
TUMOR
INDICATOR TIME # OF ANIMALS
32.0 1 1
72.0 1 1
78.0 1 1
79.0 2 1
86.0 1 3
90.0 3 1
95.0 1 1
98.0 3 1
101.0 1 1
103.0 3 3
105.0 3 1
106.0 3 3
107.0 3 3
108.0 1 3
500000
INDICATOR
THE 40 OBSERVATIONS AT LEVEL 5 WITH A DOSE OF 2.00000
TUMOR
TIME # OF ANIMALS
TUMOR
INDICATOR TIME # OF ANIMALS
INDICATOR
6/15/2009
D-29
DRAFT-DO NOT CITE OR QUOTE
-------
51.0 1
58.0 1
67.0 1
72.0 1
76.0 1
78.0 1
79.0 1
81.0 1
83.0 2
88.0 4
93.0 1
95.0 1
97.0 1
100.0 3
102.0 1
103.0 2
104.0 2
105.0 1
106.0 2
107.0 8
1
1
1
1
1
2
2
1
1
1
1
1
1
1
2
2
2
2
2
2
55.0 1
61.0 1
67.0 1
74.0 1
77.0 1
79.0 1
80.0 1
82.0 1
86.0 1
92.0 1
93.0 1
95.0 1
98.0 2
101.0 2
103.0 4
104.0 4
105.0 1
106.0 4
107.0 8
108.0 2
1
1
2
2
1
1
2
2
2
1
2
2
2
1
1
1
1
1
1
1
FORM OF PROBABILITY FUNCTION:
P(DOSE) = 1 - exp( (-00 - 01 * D - 02 * DA2 - 03 * DA3) * (T - TO)AJ)
THE MAXIMUM LIKELIHOOD ESTIMATION OF:
PROBABILITY FUNCTION COEFFICIENTS
Q(0)=.106410244171E-11
Q(1)=.135503864185E-11
Q( 2)= .000000000000
0(3)=.499366216636E-12
TIME FUNCTION COEFFICIENTS
T0= .000000000000
J= 5.39821051674
THE MAXIMUM LIKELIHOOD IS -185.712125973
MAXIMUM LIKELIHOOD ESTIMATES OF EXTRA RISK
WEIBULL LOWER CONFIDENCE LIMITS ON DOSE FOR FIXED RISK
CONFIDENCE
LOWER BOUND UPPER BOUND LIMIT
6/15/2009
D-30
DRAFT-DO NOT CITE OR QUOTE
-------
RISK MLEDOSE ON DOSE ON RISK INTERVAL TIME
.100000 .695915 .304814 .213802 95.0% 108.000
5.000000E-02 .379173 .148395 .122838 95.0% 108.000
1.000000E-02 7.805583E-02 2.907622E-02 2.661966E-02 95.0% 108.000
1.000000E-03 7.787649E-03 2.894507E-03 2.688219E-03 95.0% 108.000
1.000000E-06 7.783932E-06 3.575852E-06 2.176804E-06 95.0% 108.000
WEIBULL UPPER CONFIDENCE LIMITS ON RISK FOR FIXED DOSE
CONFIDENCE
UPPER BOUND LIMIT
DOSE MLERISK ON RISK INTERVAL TIME
.100000 1.281155E-02 3.397492E-02 95.0% 108.000
.500000 6.774880E-02 .158717 95.0% 108.000
2.00000 .470433 .600987 95.0% 108.000
NORMAL COMPLETION!
6/15/2009
D-31
DRAFT-DO NOT CITE OR QUOTE
-------
DATA PRINTOUTS FOR BMD MODELING FOR THE
JOHNSON ET AL. (1986) TUMOR DATA SETS
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Mammary Tumors
(Malignant and Benign) in Female F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra
risk)
T3
£
o
c
o
'
o
(0
0.6
0.5
0.4
0.3
0.2
0.1
20:24 03/09 2009
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL BMD
0.5
1
dose
1.5
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C: \USEPA\IRIS\acrylamide\female\maittmary\lMulfeniMS_. (d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\mammary\lMulfemMS_.plt
Mon Mar 09 21:24:49 2009
BMDS Model Run
6/15/2009
D-32
DRAFT-DO NOT CITE OR QUOTE
-------
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.205539
Beta(l) = 0.21947
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.53
Beta(l) -0.53 1
Interval
Conf. Limit
Variable
Background
Beta(l)
Parameter Estimates
Estimate
0.196578
0.238242
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-167.257
-168.557
-177.557
341.114
# Param's
5
2
1
Deviance Test d.f.
2.60026
20.5999
P-value
0.4574
0.0003801
Goodness of Fit
Scaled
6/15/2009
D-33
DRAFT-DO NOT CITE OR QUOTE
-------
Dose
Est. Prob.
Expected
Observed
Size
ChiA2 =2.62
d.f. = 3
P-value = 0.4542
Residual
0.0000
0.0100
0.1000
0.5000
2.0000
0.1966
0.1985
0.2155
0.2868
0.5011
11.795
11.909
12.930
16.635
30.567
12.000
12.000
10.000
21.000
29.000
60
60
60
58
61
0.067
0.029
-0.920
1.267
-0.401
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0.1
Extra risk
0.95
0.442241
0.295052
0.778339
Taken together, (0.295052, 0.778339) is a 90
interval for the BMD
% two-sided confidence
Multistage Cancer Slope Factor =
0.338923
6/15/2009
D-34
DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Thyroid Follicular
Cell (Adenomas and Carcinomas) in Female F344 Rats Exposed to Acrylamide in the Drinking Water
(10% extra risk)
O
I
O
'•4-*
O
(0
0.2
0.15
0.1
0.05
20:22 03/09 2009
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMID
0.5
1.5
dose
2.5
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C:\USEPA\IRIS\acrylamide\female\FCAC\lMulfemMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\FCAC\lMulfeitiMS_.plt
Mon Mar 09 21:22:28 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-35
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0075858
Beta(l) = 0.0386267
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.54
Beta(l) -0.54 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.00893025
0.0359638
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-32.3827
-33.2785
-36.7233
70.5569
# Param's
5
2
1
Deviance Test d.f.
1.79148
8.68109
P-value
0.6168
0.06958
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0089
0.0093
0.0125
0.0266
0.0777
Expected
0.518
0.548
0.737
1.542
4.663
Observed
1.000
0.000
1.000
1.000
5.000
Size
58
59
59
58
60
Scaled
Residual
0.673
-0.744
0.309
-0.443
0.163
=1.32
d.f. = 3
P-value = 0.7236
6/15/2009
D-36
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 2.92963
BMDL = 1.46827
BMDU = 9.25522
Taken together, (1.46827, 9.25522) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.0681074
6/15/2009 D-37 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of CMS Tumors of
Glial Origin in Female F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra risk)
Multistage Cancer Model with 0.95 Confidence Level
T3
£
O
c
O
•*=
O
(0
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
0.5
1
dose
BMD
1.5
20:20 03/09 2009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C: \USEPA\IRIS\acrylamide\female\CNS\lMulfeitiMS_. (d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\CNS\lMulfemMS_.plt
Mon Mar 09 21:20:09 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-38
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
****
****
the web sight for model updates which will eventually
incorporate these convergence criterion. Default values used.
****
****
Default Initial Parameter Values
Background = 0.0124331
Beta(l) = 0.069725
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.52
Beta(l) -0.52 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.0178358
0.0585541
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-49.516
-50.7987
-56.5743
105.597
# Param's
5
2
1
Deviance Test d.f.
2.56534
14.1166
P-value
0.4636
0.006932
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0178
0.0184
0.0236
0.0462
0.1264
Expected
1.070
1.086
1.414
2.770
7.709
Observed
1.000
2.000
1.000
1.000
9.000
Size
60
59
60
60
61
Scaled
Residual
-0.068
0.885
-0.352
-1.089
0.497
=2.35
d.f. = 3
P-value = 0.5038
6/15/2009
D-39
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.79937
BMDL = 1.03024
BMDU = 4.00245
Taken together, (1.03024, 4.00245) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.0970644
6/15/2009 D-40 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Oral Cavity Tumors
(Malignant and Benign) in Female F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra
risk)
Multistage Cancer Model with 0.95 Confidence Level
T3
£
o
c
o
•*=
o
(0
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
20:27 03/09 2009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C:\USEPA\IRIS\acrylamide\female\oral\lMulfemMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\oral\lMulfeitiMS_.plt
Mon Mar 09 21:27:04 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-41
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0250102
Beta(l) = 0.0586007
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.55
Beta(l) -0.55 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.0249031
0.0585561
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-56.151
-58.2474
-62.4646
120.495
# Param's
5
2
1
Deviance Test d.f.
4.19284
12.6273
P-value
0.2414
0.01325
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0249
0.0255
0.0306
0.0530
0.1327
Expected
1.494
1.528
1.836
3.182
7.960
Observed
0.000
3.000
2.000
3.000
8.000
Size
60
60
60
60
60
Scaled
Residual
-1.238
1.206
0.123
-0.105
0.015
= 3.01
d.f. = 3
P-value = 0.3897
6/15/2009
D-42
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.79931
BMDL = 0.988695
BMDU = 4.74373
Taken together, (0.988695, 4.74373) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.101143
6/15/2009 D-43 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Uterus
Adenocarcinomas in Female F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra
risk)
Multistage Cancer Model with 0.95 Confidence Level
0.2
0.15
•
I
o
'
0.1
0.05
Multistage Cancer
Linear extrapolation
BMID
3.5
20:31 03/09 2009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: C:\USEPA\IRIS\acrylamide\female\uterus\lMulfemMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\uterus\lMulfeitiMS_.plt
Mon Mar 09 21:31:38 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-44
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
****
****
the web sight for model updates which will eventually
incorporate these convergence criterion. Default values used.
****
****
Default Initial Parameter Values
Background = 0.0136297
Beta(l) = 0.0329157
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.51
Beta(l) -0.51 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.0171456
0.0258801
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-36.1508
-38.4257
-40.3921
80.8514
# Param's
5
2
1
Deviance Test d.f.
P-value
4.54986
8.48273
0.2079
0.07541
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0171
0.0174
0.0197
0.0298
0.0667
Expected
1.029
1.044
1.181
1.757
4.003
Observed
1.000
2.000
1.000
0.000
5.000
Size
60
60
60
59
60
Scaled
Residual
-0.029
0.944
-0.168
-1.346
0.516
= 3.00
d.f. = 3
P-value = 0.3921
6/15/2009
D-45
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 4.07111
BMDL = 1.79347
BMDU = 27.8943
Taken together, (1.79347, 27.8943) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.0557578
6/15/2009 D-46 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Clitoral Adenomas
(Benign) in Female F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra risk)
o
I
o
'•4-*
o
(0
0.8
0.6
0.4
0.2
BMDL
BMD
08:0803/11 2009
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.5
1
dose
1.5
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C: \USEPA\IRIS\acrylamide\female\clitoral\lMulfeitiMS_. (d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\clitoral\lMulfeitiMS_.plt
Wed Mar 11 09:08:19 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-47
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0
Beta(l) = 5.10063e+019
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.47
Beta(l) -0.47 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.325271
1.62417
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-6.93147
-9.06517
-12.0285
22.1303
# Param's
5
2
1
Deviance Test d.f.
4.2674
10.194
P-value
0.234
0.03728
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.3253
0.3361
0.4264
0.7005
0.9738
Expected
0.651
1.008
1.706
2.802
4.869
Observed
0.000
1.000
3.000
2.000
5.000
Size
2
3
4
4
5
Scaled
Residual
-0.982
-0.010
1.309
-0.875
0.367
= 3.58
d.f. = 3
P-value = 0.3109
6/15/2009
D-48
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.0648702
BMDL = 0.022986
BMDU = 0.29871
Taken together, (0.022986, 0.29871) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 4.35048
6/15/2009 D-49 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Pituitary Gland
Adenomas in Female F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra risk)
o
I
o
'•4-*
o
(0
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.5
1
dose
1.5
20:29 03/09 2009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\female\pituitary\lMulfemMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\female\pituitary\lMulfemMS_.plt
Mon Mar 09 21:29:12 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-50
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.47609
Beta(l) = 0.0514418
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.55
Beta(l) -0.55 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.474642
0.0517505
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-205.995
-206.935
-207.169
417.87
# Param's
5
2
1
Deviance Test d.f.
1.88098
2.34909
P-value
0.5975
0.6718
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.4746
0.4749
0.4774
0.4881
0.5263
Expected
28.004
28.495
28.641
29.284
31.578
Observed
25.000
30.000
32.000
27.000
32.000
Size
59
60
60
60
60
Scaled
Residual
-0.783
0.389
0.868
-0.590
0.109
= l.i
d.f. = 3
P-value = 0.5981
6/15/2009
D-51
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 2.03593
BMDL = 0.559619
BMDU did not converge for BMR = 0.100000
BMDU calculation failed
BMDU = Inf
6/15/2009 D-52 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (all doses included)
(10% extra risk)
T3
0)
•5
I
C
o
•*=
o
(0
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.5
1
dose
1.5
13:1403/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:14:47 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-53
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0755454
Beta(l) = 0.0641515
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.61
Beta(l) -0.61 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.0626677
0.0881625
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-89.143
-96.1436
-99.7038
196.287
Param's
5
2
1
Deviance Test d.f.
14.0011
21.1215
P-value
0.002904
0.0002996
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0627
0.0635
0.0709
0.1031
0.2142
Expected
3.760
3.810
4.254
6.185
12.852
Observed
3.000
0.000
7.000
11.000
10.000
Size
60
60
60
60
60
Scaled
Residual
-0.405
-2.017
1.381
2.044
-0.897
= 11.12
d.f. = 3
P-value = 0.0111
6/15/2009
D-54
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.19507
BMDL = 0.660363
BMDU = 3.50031
Taken together, (0.660363, 3.50031) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.151432
6/15/2009 D-55 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 2-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (all doses included)
(10% extra risk)
T3
0)
•5
I
C
o
•*=
o
(0
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.5
1
dose
1.5
13:1503/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:15:41 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl-beta2*dose/x2) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-56
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0755454
Beta(l) = 0.0641515
Beta(2) = 0
by the user,
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
have been estimated at a boundary point, or have been specified
and do not appear in the correlation matrix )
Background Beta(l)
Background 1 -0.61
Beta(l) -0.61 1
Interval
Conf. Limit
Variable
Background
Beta(l)
Beta(2)
Estimate
0.0626677
0.0881624
0
Parameter Estimates
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-89.143
-96.1436
-99.7038
196.287
# Param' s
5
2
1
Deviance
14.0011
21.1215
Test d.f.
3
4
P-value
0.002904
0.0002996
Dose
Est. Prob.
Goodness of Fit
Expected Observed Size
Scaled
Residual
6/15/2009
D-57
DRAFT-DO NOT CITE OR QUOTE
-------
0.0000 0.0627 3.760 3.000 60 -0.405
0.0100 0.0635 3.810 0.000 60 -2.017
0.1000 0.0709 4.254 7.000 60 1.381
0.5000 0.1031 6.185 11.000 60 2.044
2.0000 0.2142 12.851 10.000 60 -0.897
= 11.12 d.f. = 3 P-value = 0.0111
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.19507
BMDL = 0.660363
BMDU = 3.50031
Taken together, (0.660363, 3.50031) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.151432
6/15/2009 D-58 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 3-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (all doses included)
(10% extra risk)
Multistage Cancer Model with 0.95 Confidence Level
T3
£
O
c
O
•*=
O
(0
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.5
1
dose
1.5
13:1603/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:16:36 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl-beta2*dose/x2-betaS*doseA3)
The parameter betas are restricted to be positive
6/15/2009
D-59
DRAFT-DO NOT CITE OR QUOTE
-------
Dependent variable = incidence
Independent variable = dose
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0755454
Beta(l) = 0.0641515
Beta(2) = 0
Beta(3) = 0
by the user,
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2) -Beta(3)
have been estimated at a boundary point, or have been specified
and do not appear in the correlation matrix )
Background Beta(l)
Background 1 -0.61
Beta(l) -0.61 1
Interval
Conf. Limit
Variable
Background
Beta(l)
Beta(2)
Beta(3)
Estimate
0.0626677
0.0881624
0
0
Parameter Estimates
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-89.143
-96.1436
-99.7038
196.287
# Param's
5
2
1
Deviance Test d.f.
14.0011
21.1215
P-value
0.002904
0.0002996
6/15/2009
D-60
DRAFT-DO NOT CITE OR QUOTE
-------
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0627
0.0635
0.0709
0.1031
0.2142
Expected
3.760
3.810
4.254
6.185
12.851
Observed
3.000
0.000
7.000
11.000
10.000
Size
60
60
60
60
60
Scaled
Residual
-0.405
-2.017
1.381
2.044
-0.897
ChiA2 = 11.12 d.f. = 3 P-value = 0.0111
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.19507
BMDL = 0.660363
BMDU = 3.50031
Taken together, (0.660363, 3.50031) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.151432
6/15/2009 D-61 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 4-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (all doses included)
(10% extra risk)
O
I
O
'•4-*
O
(0
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.5
1
dose
1.5
13:1703/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:17:33 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl-beta2*dose/x2-beta3*dose/x3-beta4*doseM)]
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
6/15/2009
D-62
DRAFT-DO NOT CITE OR QUOTE
-------
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 5
Total number of specified parameters = 0
Degree of polynomial = 4
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0755454
0.0641515
Beta(l) =
Beta(2) =
Beta(3) =
Beta(4) =
by the user,
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2) -Beta(3) -Beta(4)
have been estimated at a boundary point, or have been specified
and do not appear in the correlation matrix )
Background Beta(l)
Background 1 -0.61
Beta(l) -0.61 1
Interval
Conf. Limit
Variable
Background
Beta(l)
Beta(2)
Beta(3)
Beta(4)
Estimate
0.0626677
0.0881624
0
0
0
Parameter Estimates
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-89.143
-96.1436
-99.7038
196.287
# Param's
5
2
1
Deviance Test d.f.
14.0011
21.1215
P-value
0.002904
0.0002996
6/15/2009
D-63
DRAFT-DO NOT CITE OR QUOTE
-------
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0627
0.0635
0.0709
0.1031
0.2142
Expected
3.760
3.810
4.254
6.185
12.851
Observed
3.000
0.000
7.000
11.000
10.000
Size
60
60
60
60
60
Scaled
Residual
-0.405
-2.017
1.381
2.044
-0.897
ChiA2 = 11.12 d.f. = 3 P-value = 0.0111
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.19507
BMDL = 0.660363
BMDU = 3.50031
Taken together, (0.660363, 3.50031) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.151432
6/15/2009 D-64 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (Highest dose
dropped) (10% extra risk)
•
I
o
13
(0
0.3
0.25
0.2
0.15
0.1
0.05
13:21 03/092009
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.1
0.2
0.3
0.4
0.5
dose
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:21:13 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
6/15/2009
D-65
DRAFT-DO NOT CITE OR QUOTE
-------
Dependent variable = incidence
Independent variable = dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0423896
Beta(l) = 0.335431
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.6
Beta(l) -0.6 1
Interval
Conf. Limit
Variable
Background
Beta(l)
Parameter Estimates
Estimate
0.0340742
0.390914
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-62.1094
-65.5208
-71.2117
135.042
Param's
4
2
1
Deviance Test d.f.
6.8229
18.2046
P-value
0.03299
0.0003991
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
Est. Prob.
0.0341
0.0378
0.0711
0.2056
Expected
2.044
2.271
4.266
12.334
Observed
3.000
0.000
7.000
11.000
Size
60
60
60
60
Scaled
Residual
0. 680
-1.536
1.373
-0.426
6/15/2009
D-66
DRAFT-DO NOT CITE OR QUOTE
-------
ChiA2 = 4.89 d.f. = 2 P-value = 0.0867
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.269523
BMDL = 0.163397
BMDU = 0.578796
Taken together, (0.163397, 0.578796) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.612008
6/15/2009 D-67 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 2-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (Highest dose
dropped) (10% extra risk)
O
I
O
'•4-*
O
(0
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.1
0.2 0.3
dose
0.4
0.5
13:2203/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:22:01 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl-beta2*dose/x2) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
6/15/2009 D-68
DRAFT-DO NOT CITE OR QUOTE
-------
Independent variable = dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0423896
Beta(l) = 0.335431
Beta(2) = 0
by the user,
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
have been estimated at a boundary point, or have been specified
and do not appear in the correlation matrix )
Background Beta(l)
Background 1 -0.6
Beta(l) -0.6 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Beta(2)
Estimate
0.0340742
0.390914
0
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-62.1094
-65.5208
-71.2117
135.042
# Param's
4
2
1
Deviance Test d.f.
6.8229
18.2046
P-value
0.03299
0.0003991
Dose
Est. Prob.
Goodness of Fit
Expected Observed Size
Scaled
Residual
6/15/2009
D-69
DRAFT-DO NOT CITE OR QUOTE
-------
0.0000
0.0100
0.1000
0.5000
0.0341
0.0378
0.0711
0.2056
2.044
2.271
4.266
12.334
3.000
0.000
7.000
11.000
60
60
60
60
0. 680
-1.536
1.373
-0.426
= 4.89 d.f. = 2 P-value = 0.0867
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.269523
BMDL = 0.163397
BMDU = 0.578796
Taken together, (0.163397, 0.578796) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.612008
6/15/2009 D-70 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 3-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (Highest dose
dropped) (10% extra risk)
O
I
O
'•4-*
O
(0
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.1
0.2 0.3
dose
0.4
0.5
13:2203/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:22:39 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl-beta2*dose/x2-beta3*dose/x3)]
The parameter betas are restricted to be positive
Dependent variable = incidence
6/15/2009 D-71
DRAFT-DO NOT CITE OR QUOTE
-------
Independent variable = dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0423896
Beta(l) = 0.335431
Beta(2) = 0
Beta(3) = 0
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2) -Beta(3)
have been estimated at a boundary point, or have been specified
by the user,
and do not appear in the correlation matrix )
Background Beta(l)
Background 1 -0.6
Beta(l) -0.6 1
Parameter Estimates
95.0% Wald Confidence
Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper
Conf. Limit
Background 0.0340742 * *
*
Beta(l) 0.390914 * *
*
Beta(2) 0 * *
*
Beta(3) 0 * *
*
Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -62.1094 4
Fitted model -65.5208 2 6.8229 2 0.03299
Reduced model -71.2117 1 18.2046 3 0.0003991
AIC: 135.042
6/15/2009 D-72 DRAFT-DO NOT CITE OR QUOTE
-------
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
Est. Prob.
0.0341
0.0378
0.0711
0.2056
Expected
2.044
2.271
4.266
12.334
Observed
3.000
0.000
7.000
11.000
Size
60
60
60
60
Scaled
Residual
0.680
-1.536
1.373
-0.426
ChiA2 = 4.89 d.f. = 2 P-value = 0.0867
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.269523
BMDL = 0.163397
BMDU = 0.578796
Taken together, (0.163397, 0.578796) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.612008
6/15/2009 D-73 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (Two highest dose
dropped) (10% extra risk)
Multistage Cancer Model with 0.95 Confidence Level
•
I
o
13
(0
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMD
0.12
13:2603/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:26:35 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
6/15/2009
D-74
DRAFT-DO NOT CITE OR QUOTE
-------
Dependent variable = incidence
Independent variable = dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.022083
Beta(l) = 0.985041
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.59
Beta(l) -0.59 1
Interval
Conf. Limit
Parameter Estimates
Variable
Background
Beta(l)
Estimate
0.0263158
0.854491
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-33.5247
-36.1941
-38.6206
76.3881
# Param's
3
2
1
Deviance Test d.f.
5.33864
10.1918
P-value
0.02086
0.006122
Dose
0.0000
0.0100
0.1000
Est. Prob.
0.0263
0.0346
0.1061
Goodness of Fit
Expected Observed Size
1.579
2.076
6.364
3.000
0.000
7.000
60
60
60
Scaled
Residual
1.146
-1.466
0.267
= 3.54
d.f. = 1
P-value = 0.0601
6/15/2009
D-75
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.123302
BMDL = 0.0608676
BMDU = 0.519103
Taken together, (0.0608676, 0.519103) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 1.64291
6/15/2009 D-76 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 2-Degree Polynomial Model Results for the Incidence of Tunica Vaginalis
Mesothelioma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (Two highest dose
dropped) (10% extra risk)
Multistage Cancer Model with 0.95 Confidence Level
O
I
O
"•4-*
O
(0
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMD
0.02
0.04
0.06
0.08
0.1
dose
13:2703/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\tunicavaginalismesothelioma\lMulmalMS_.plt
Mon Mar 09 14:27:23 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose/xl-beta2*dose/x2) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
6/15/2009 D-77
DRAFT-DO NOT CITE OR QUOTE
-------
Independent variable = dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.0249685
Beta(l) = 0
Beta(2) = 9.85045
by the user,
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(l)
have been estimated at a boundary point, or have been specified
and do not appear in the correlation matrix )
Background Beta(2)
Background 1 -0.56
Beta(2) -0.56 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Beta(2)
Estimate
0.0251256
0
9.72992
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood)
-33.5247
-35.7015
-38.6206
75.4029
# Param's
3
2
1
Deviance Test d.f.
4.35343
10.1918
P-value
0.03693
0.006122
Dose
Est. Prob.
Goodness of Fit
Expected Observed Size
Scaled
Residual
6/15/2009
D-78
DRAFT-DO NOT CITE OR QUOTE
-------
0.0000 0.0251 1.508 3.000 60 1.231
0.0100 0.0261 1.564 0.000 60 -1.267
0.1000 0.1155 6.931 7.000 60 0.028
=3.12 d.f. = 1 P-value = 0.0772
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.10406
BMDL = 0.0688637
BMDU = 0.324378
Taken together, (0.0688637, 0.324378) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 1.45214
6/15/2009 D-79 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Thyroid Follicular
Cell Adenoma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra risk)
Multistage Cancer Model with 0.95 Confidence Level
T3
£
o
c
o
•*=
o
(0
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMID
0.5
1
dose
1.5
13:1003/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\follicularcelladenoma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\follicularcelladenoma\lMulmalMS_.plt
Mon Mar 09 14:10:33 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
The parameter betas are restricted to be positive
Dependent variable = incidence
6/15/2009 D-80
DRAFT-DO NOT CITE OR QUOTE
-------
Independent variable = dose
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.00953434
Beta(l) = 0.0562379
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.54
Beta(l) -0.54 1
Parameter Estimates
Interval
Conf. Limit
Variable
Background
Beta(l)
Estimate
0.0117608
0.0517152
Std. Err.
95.0% Wald Confidence
Lower Conf. Limit Upper
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood) #
-40.
-41.
3781
9922
-46.9722
87.9844
Param's
5
2
1
Deviance Test d.f.
3.22812
13.1881
P-value
0.3578
0.01039
Goodness of Fit
Dose
0.0000
0.0100
0.1000
0.5000
2.0000
Est. Prob.
0.0118
0.0123
0.0169
0.0370
0.1089
Expected
0.706
0.712
0.995
2.182
6.423
Observed
1.000
0.000
2.000
1.000
7.000
Size
60
58
59
59
59
Scaled
Residual
0.352
-0.849
1.017
-0.816
0.241
=2.60
d.f. = 3
P-value = 0.4572
6/15/2009
D-81
DRAFT-DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 2.03732
BMDL = 1.11679
BMDU = 4.98748
Taken together, (1.11679, 4.98748) is a 90 % two-sided confidence
interval for the BMD
Multistage Cancer Slope Factor = 0.0895426
6/15/2009 D-82 DRAFT-DO NOT CITE OR QUOTE
-------
BMD Cancer Multistage 1-Degree Polynomial Model Results for the Incidence of Adrenal
Pheochromocytoma in Male F344 Rats Exposed to Acrylamide in the Drinking Water (10% extra
risk)
Multistage Cancer Model with 0.95 Confidence Level
•
I
C
o
13
ro
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BM1D
0.5
1.5
2.5
dose
13:31 03/092009
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
C:\USEPA\IRIS\acrylamide\males\adrenalpheochromocytoma\lMulmalMS_.(d)
Gnuplot Plotting File:
C:\USEPA\IRIS\acrylamide\males\adrenalpheochromocytoma\lMulmalMS_.plt
Mon Mar 09 14:31:23 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl) ]
6/15/2009
D-83
DRAFT-DO NOT CITE OR QUOTE
-------
The parameter betas are restricted to be positive
Dependent variable = incidence
Independent variable = dose
Total number of observations = 5
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: 2.22045e-016
Parameter Convergence has been set to: 1.49012e-008
**** We are sorry but Relative Function and Parameter Convergence ****
**** are currently unavailable in this model. Please keep checking ****
**** the web sight for model updates which will eventually ****
**** incorporate these convergence criterion. Default values used. ****
Default Initial Parameter Values
Background = 0.087802
Beta(l) = 0.0427132
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.55
Beta(l) -0.55 1
Parameter Estimates
95.0% Wald Confidence
Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper
Conf. Limit
Background 0.0879228 * *
*
Beta(l) 0.0413823 * *
*
Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -99.2572 5
Fitted model -100.594 2 2.67284 3 0.4449
Reduced model -101.734 1 4.95289 4 0.2922
AIC: 205.187
Dose
0.0000
0.0100
0.1000
Est. Prob.
0.0879
0.0883
0.0917
Goodness of Fit
Expected Observed Size
5.275
5.210
5.501
3.000
7.000
7.000
60
59
60
Scaled
Residual
-1.037
0.821
0.670
6/15/2009 D-84 DRAFT-DO NOT CITE OR QUOTE
-------
0.5000 0.1066 6.396 5.000 60 -0.584
2.0000 0.1604 9.622 10.000 60 0.133
ChiA2 = 2.56 d.f. = 3 P-value = 0.4647
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 2.54603
BMDL = 1.0776
BMDU did not converge for BMR = 0.100000
BMDU calculation failed
BMDU = 1.57654e+068
Multistage Cancer Slope Factor = 0.0927984
6/15/2009 D-85 DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX E. DERIVATION OF IN VIVO SECOND ORDER RATE CONSTANTS
AND THE ADDUCT FORMATION SIMULATION MODEL
BACKGROUND
Two methods have been used to construct blood AUC's for AA and GA following
administration: 1) calculation directly from measured blood AA and GA time course data, or 2)
by inference from concentrations of AA-valine or GA-valine hemoglobin adducts measured in
red blood cells (Fennell et al., 2005). Blood AA and GA AUC's are calculated according to
Fennell et al. (2005) by dividing the measured adduct levels by the second order rate constant for
the formation of the adduct:
n 1 RHb
Dose = — x
k Hb
where k, the second order reaction rate constant, expressed in
units of 1/g globin/h, [RHb] is the adduct concentration, and
[Hb] is the concentration of hemoglobin.
This hemoglobin adduct formation rate constant has typically been derived in vitro using
red blood cell hemoglobin rather than whole blood (Fennell et al., 2003, 2005; Tareke et al.,
2006; Tornqvist et al. 2008). The accuracy of AUC's calculated using the in vitro derived rate
constant depends directly upon the accuracy of this estimate for the actual formation rates in
vivo. The results are also sensitive to processes involved in the "loss or elimination" of adducts
over time—red blood cell turnover, chemical loss of adduct, and body weight dependent
increases in blood volume. Equations to account for these losses in adducts over time are
available (Fennel et al., 2005; Walker et al., 2007), however, for exposures of 1 day or less, the
influence of these processes on adduct levels and AUC's estimated from them can be ignored
(Fennel et al., 2005; Walker et al., 2007).
Until recently there were insufficient data from which to derive either animal or human in
vivo hemoglobin adduct formation or elimination rates. Human data remains unavailable at this
time, however, Tareke et al. (2006) reported AA-Valine and GA-Valine hemoglobin adduct
levels in rats and mice exposed to single doses of AA or GA where serum level time course data
had also been collected (Doerge et al., 2005 b, c). Tareke et al. (2006) also measured AA-Valine
and GA-Valine hemoglobin adduct levels in rats exposed to AA in drinking water for up to 42
days, as well as the loss rate of both AA and GA adducts at the end of drinking water exposure.
Together, these data are sufficient to derive in vivo adduct formation and loss rates for the AA-
6/12/09 E-l DRAFT-DO NOT CITE OR QUOTE
-------
Valine and GA-Valine adducts that can be used to improve the estimates of blood AA and GA
AUCs for a drinking water exposure when only data on hemoglobin adduct levels are available.
DEVELOPMENT OF IN VIVO ADDUCT FORMATION RATE CONSTANTS
The derivation of in vivo adduct formation rates requires three critical types of data from
a single study: 1) the administered dose, 2) time course serum levels, and 3) time course adduct
levels (including sufficient post dosing sample times to determine elimination rates) if longer
than one day of exposure.
In vivo adduct formation rate constants for rats
The only studies meeting the above data requirement to derive rat in vivo rate constants
are those from Doerge et al. (2005c) and Tareke et al. (2006). EPA has derived in vivo gender
specific adduct formation rates based upon these single dose in vivo studies in male and female
F344 rats. Table E-l below list the raw serum AUCs that were available in numerical tables from
Doerge et al. (2005c) and the levels of hemoglobin that were taken from bar chart compilations
in Tareke et al. (2006).
Table E-l. Serum AUC from Doerge et al. (2005c) and hemoglobin adduct levels
from Tareke et al (2008) for a 0.1 mg/kg single dose of AA in male and female F344
rats.
Type of
adduct from
AA dosing
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
AA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
GA-Val
sex-
route
M-control
M-Diet
M-
gavage
M-IV
F-control
F-Diet
F-gavage
F-IV
M-control
M-IV
M-
gavage
M-Diet
F-control
F-IV
F-Diet
F-gavage
Hb adduct level
(fmole/mg
globin)
9
19.5
20
46.5
12
23
29
49.5
32.5
36
64
98.5
45
48.5
102
131
AUC
(uM-hr)
0
1.8
2.4
4.1
0
1.5
4.5
4.6
0
0.58
1.3
1.9
0
0.6
1.5
4.4
A linear regression of the hemoglobin adduct levels against the AUCs resulted in the
following correlations coefficients (r), equations, and slopes.
6/12/09
E-2
DRAFT-DO NOT CITE OR QUOTE
-------
Regression of hemoglobin adduct levels to ADC to derive in vivo second order rate constants for
adduct formation (i.e., the slope of the regression line)
y- Ratio of
Intercep Slope GA-
Gender Adduct Slope t r2 Val / AA-Val r
Male
Female
Male
Female
AA-Val
AA-Val
GA-Val
GA-Val
8.92
5.90
Average = 7.4
35.33
19.98
Average = 27.7
5.24
12.73
24.36
49.16
0.94
0.85
0.96
0.93
0.97
0.92
3.96 0.98
3.38 0.96
Example equation:
AA-Val (fmoles/mg globin) for males = 8.9 AA AUC (uM-h) + 5.2
GA-Val (fmoles/mg globin) for females = 20.0 AA AUC (uM-h) + 49.2
The slopes from these linear regressions (8.9 for male F344 rats, 5.9 for female F344 rats,
and a 7.4 average value) represent the in vivo second order rate constants for AA-Val in units of
1/g globin/h (multiplied by 106 ). 7.4 as the average of both genders (in units of [1/g globin/h]
multiplied by 106 in Table 5-5 for ease of presentation). The GA-Val formation rate constants
are 35.3 for males, 20.0 for females, and a 27.7 average for both genders.
Tareke et al. (2006) do not report gender specific slopes, but they do report a slope of 7.5 for
AA-Val (both genders), and 35 for GA-Val (both genders). The higher GA-Val slope for both
genders from Tareke et al. (2006) of 35 compared with the average from EPA's derivation (27.7)
may have resulted from including the serum and adduct data from administered doses of GA in
their analysis, data that were not included in EPA's regression analysis which only used the data
from AA dosing. EPA chose to derive the gender specific rates to more accurately estimate the
internal rat AUC from the POD because the measured AUC data in Doerge et al. (2005c) from
single doses of AA indicate that males rats have a lower serum GA-AUC than females from a
given dose of AA. The estimated AUCs per mg AA/kg bw reported in the Tornqvist et al. (2008)
drinking water study also confirm this gender difference in rats (see Tables 5-6 and 5-7). The
higher overall values of AUC reported by Tornqvist et al. (2008) may result from their use of in
vitro derived formation rate constant to estimate AUC based on adduct data compared with the
direct serum AUC measurements reported by Doerge et al. (2005c).
6/12/09 E-3 DRAFT-DO NOT CITE OR QUOTE
-------
In vivo adduct formation rate constants for humans
No human in vivo data are available for all three of the variables needed to
directly develop human in vivo adduct formation rate constants. The best surrogate for a human
in vivo rate constant in the derivation of reference values is the in vivo rate constants that Tareke
et al. (2006) derived from a linear regression of the all the available in vivo adduct level versus
AUC data for male and female mice and rats from the single dose studies of Doerge et el. (2005
b,c) and Tareke et al. (2006). Below is the scatter plot and results of the regression reproduced
from Tareke et al. (2006).
2 3
AUC (uM x h)
Fig. 4. Correlation of hemoglobin adducls for acrylarnide and glycidarnide with
Ihe respectiv e serum AUC in F344 rats and B6C3F ] mice exposed to sin gle dose
gavage administration of acrylarnide (Q.I mg^g hw) or an equimolar gavage
dose of glycidarnidc. Individual data points shown represent group mean
hemoglobin adducls and AUC values for male and female mice and rats.
Source: Tareke et al. (2006)
The regression results for AA-Val are:
AA-Val (fmoles/mg globin) = 7.5 AA AUC (uM-h) + 8
(r2 = 0.88, p<0.001)
andforGA-Val:
GA-Val (fmoles/mg globin) = 32.5 GA AUC (uM-h) + 41.5 (r2 = 0.83, p<0.001)
6/12/09
E-4
DRAFT-DO NOT CITE OR QUOTE
-------
The resulting in vivo adduct formation rate constants of 7.5 x 10"6 AA-Val 1/g globin/h,
and 32.5 x 10"6 GA-Val 1/g globin/h were applied to the Fennell 2005 data to derive an AA or
GA AUC/AA mg kg bw/d factor then used to convert the rat POD-AUC to a human
administered dose.
These in vivo adduct rate constants are not gender specific. Gender specific rates were
not considered to be critical in the derivation of the HEC for humans based on epidemiology
results from Hartmann et al. (2009) who did not observe a gender-related difference in internal
exposure and metabolism of AA in a study of a nonsmoking general population especially
designed for an even distribution of age and gender.
6/12/09 E-5 DRAFT-DO NOT CITE OR QUOTE
-------
DEVELOPMENT OF THE ADDUCT FORMATION MODEL TO ESTIMATE THE RAT
AUCS AT THE POD
Summary
To derive the rat AA-AUC or GA-AUC per dose of AA administered in a drinking water
study, EPA developed a simple model to fit the reported levels of AA-Val and GA-Val (Tareke
et al., 2006) that resulted from the Doerge et al. (2000a) rat drinking water exposure. The model
was parameterized with the gender specific rat in vivo adduct formation rate constants derived
by EPA (discussed above), and the adduct elimination rate constants reported by Tareke et al.
(2006). The model simulations resulted in the following AUC conversion factors:
AA AUC in |iM-hr per mg AA/kg bw = 22 (for males) and 48 (for females)
GA AUC in |iM-hr per mg AA/kg bw =15 (for males) and 48 (for females)
These values were used to estimate the rat internal AA-AUC that would result from the
BMDL (i.e., as the the POD from the Friedman et al. (1995) study data) for neurotoxicity used as
the basis for the RfD, and for the GA-AUC from the BMDL (POD from the Johnson et al. (1986)
data) for increased risk of tumors used to derive the oral slope factor.
The Adduct Formation Model
Tereke et al. (2006) reported the AA-Valine and GA-Valine hemoglobin adduct level
time course in male and female rats exposed chronically by drinking water to AA. Tereke also
measured the loss rate of both AA and GA adducts after the end of AA exposure. Separately,
Doerge et al (Doerge, da Costa et al. 2005) reported blood AA and GA concentrations at several
time points over the duration of the 48-day study. The data from both studies was obtained by
digitization from original graphs using Data Thief software.
One can either directly estimate the AUC fro a given DW dose if there is sufficient blood
level time course data. Alternately, one can derive blood AA and GA AUC from time course
data on hemoglobin adduct levels and in vivo formation and loss rates for the AA-Valine and
GA-Valine adducts. The authors of the two studies, did not report when blood samples for
adducts or AA and GA measurements were taken. Although the experimental blood AA and GA
data appear to represent a steady state, the clearance of both AA and GA are high enough that
oral gavage, drinking water or other repeat dose studies result in blood AA and GA levels that
fluctuate significantly throughout the day. Thus, the timing of the blood samples taken for
analysis of AA and GA will have a significant impact on any estimate of the AUC. If blood
sampling was conducted during an active period of drinking, blood concentrations and presumed
6/12/09 E-6 DRAFT-DO NOT CITE OR QUOTE
-------
exposure would be high, the inverse being true if sampling was conducted just after a
sleeping/non drinking period. Because the adduct half-life is between 10 and 13 days (Doerge,
da Costa et al. 2005; Tareke, Twaddle et al. 2006) adduct measurements are not subject to this
uncertainty.
The blood AA and GA concentration data reported by Doerge and Tereke at "various"
but otherwise unspecified times, were therefore assumed not to reliable measures of overall
blood AA or GA exposure. The alternate method of using adduct levels and formation and loss
rate constants was therefore used to estimate the AUCs from a drinking water exposure, and to
develop conversion factors to estimate the AUC from the rat study PODs.
The rates used to parameterize the model were the gender specific rat in vivo adduct
formation rate constants derived by EPA (discussed above), and the adduct elimination rate
constants reported by Tareke et al. (2006).
Adduct Model Equations
A simulation model of the formation and removal rates of AA-valine and GA-valine
hemoglobin adducts as a function of hemoglobin and AA or GA concentrations was developed
in ACSLXtreme. The amount of adduct (AADDUCT, umoles) at any time post exposure was
calculated as the integral of the balance between the formation and removal rates
ff
AADDUCT = KHGBl x RBCHGB x VRBC x BLCONC - KHGBD x AADDUCT
4 (Eq.l)
Where KHGB1 (L/d-gram hemoglobin) is the second order rate constant for the formation of the
adduct, RBCHGB (grams hemoglobin/L in RBC's) is the concentration of hemoglobin in red
blood cells, VRBC is the volume of red blood cells (L), BLCONC is the concentration of AA or
GA in the blood (uM), and KHGBD (/day) is the first order rate constant for loss of AA-Valine
or GA-Valine adducts. A unit analysis was conducted to verify the code, which was found to be
accurate. Several other parameters are used in the model to derive values used in Eq.l. Body
weight (0.25 kg), hematocrit (0.45), and fraction of body weight that is blood (0.06) were taken
from Walker et al. (2007)
An ACSLXtreme table function was adapted for use with single blood concentrations
representative of daily blood AA and GA exposures expected over the course of each day (see
Section 3.1).
Parameter estimation tools in ACSLXtreme were used to optimize the value of the AA or
GA blood concentration. Scripts (.m files) were written for each simulation, as for all
optimizations, and can be found in the model workspace.
6/12/09 E-7 DRAFT-DO NOT CITE OR QUOTE
-------
Modeling Approach
The in vivo gender specific adduct formation rates used in the model were derived from
the serum and adult data reported in Doerge et al. (2005c) and Tareke et al. (2006) from single
dose in vivo studies in male and female F344 rats. The in vivo rate constants for loss of AA and
GA hemoglobin adducts were taken directly from Tereke (Tareke, Twaddle et al. 2006).
The adduct model was used to derive the AA and GA AUCs from the Doerge et al.
(2005a) drinking water study data by estimating the daily average blood concentration that
would be needed to optimally fit the AA-Val and GA-Val adduct levels measured in the study
while holding the adduct formation and elimination rates constant. The Nelder-Meade parameter
optimization routine in ACSLX was used to optimize the average blood concentration against
the adduct data.
Results
Optimized values of the AA-valine and GA-valine adduct formation and loss rates are
presented in Table 1. In all cases, the there was excellent correspondence between measured and
modeled AA-valine and GA-valine adduct levels using the fixed rate constants (Example in
Figure 1).
Table E-2: Fitted AUC's and Rates Based on the Use of Rate Constants Derived
from In Vivo Data
Compound
Aery 1 amide
Glycidamide
Gender
Female
Male
Female
Male
In Vivo
Formation Rate (L/hr-g
HG)
5.9 xlO'6
6.4 xlO'6
2.0 xlO'5
3.5xlO'5
Elimination
Rate
1.3 xlO'3
1.3 xlO'3
2.6 xlO'3
2.2 xlO'3
Predicted AUC
(uM-d)
75.96
43.97
96.55
30.95
HG, hemoglobin
6/12/09
DRAFT-DO NOT CITE OR QUOTE
-------
Simulated and Observed Valine AA Adduct Levels in
Male F344 Rats Following Drinking Water Eposure to AA
- Simulated Valine Adducts
I Measured Valine Adducts
Time (days)
Figure 1: Example of the fit to adduct data.
Uncertainties and Data Gaps
The most important uncertainty in this analysis is the use of in vivo formation rates derived
from studies that single dose, rather than drinking water studies. Any errors in the values of these
in vivo rate constants would be proportional to the difference between the assumed and actual
blood AA and GA AUCs /from the drinking water exposure. This error could be reduced if blood
sampling times and drinking behavior (time, volume) were known for the Doerge et al. (2005a)
drinking water study.
REFERENCES
Doerge, D. R., G. G. da Costa, et al. (2005a). "DNA adducts derived from administration of
acrylamide and glycidamide to mice and rats." MutatRes 580(1-2): 131-41.
Doerge, DR; Young, JF; McDaniel, LP; et al. (2005b) Toxicokinetics of acrylamide and
glycidamide in B6C3F1 mice. Toxicol Appl Pharmacol 202(3):258-267.
Doerge, DR; Young, JF; McDaniel, LP; et al. (2005c) Toxicokinetics of acrylamide and
glycidamide in Fischer 344 rats. Toxicol Appl Pharmacol 208:199-209.
Fennell TR; Sumner SC; Snyder RW; et al. (2005) Metabolism and hemoglobin adduct
formation of acrylamide in humans. Toxicol Sci 85(l):447-59.
6/12/09
E-9
DRAFT-DO NOT CITE OR QUOTE
-------
Tareke, E., N. C. Twaddle, et al. (2006). "Relationships between biomarkers of exposure and
toxicokinetics in Fischer 344 rats and B6C3F1 mice administered single doses of acrylamide and
glycidamide and multiple doses of acrylamide." Toxicol Appl Pharmacol 217(1): 63-75.
Tornqvist, M; Paulsson, B; Vikstrom, AC; and Granath, F. (2008) Approach for cancer risk
estimation of acrylamide in food on the basis of animal cancer tests and in vivo dosimetry. J
Agric Food Chem. 56(15):6004-12.
Walker, K., D. Hattis, et al. (2007). "Approaches to acrylamide physiologically based
toxicokinetic modeling for exploring child-adult dosimetry differences." J Toxicol Environ
Health A 70(24): 203 3-5 5.
6/12/09 E-10 DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX F. ALTERNATE RFC BASED ON HUMAN EPIDEMIOLOGY DATA
An RfC can be derived from the limited Calleman et al. (1994) data for comparison
purposes, and to highlight the uncertainties in the results due to these data to prompt
improvements in the design of future studies.
Briefly, Calleman et al. (1994) performed a cross-sectional analysis of hemoglobin
adduct formation and neurological effects in a group of 41 factory workers (34 males and 7
females, aged 18 to 42 years) who were exposed to AA (and acrylonitrile, from which AA is
formed) for 1 month to 11.5 years (mean 3 years) during the production of AA in a factory in
China. AA mean exposure concentrations, measured during the summer of 1991, were 1.07 and
3.27 mg/m3 in the synthesis and polymerization rooms, respectively. Exposure concentrations
measured during the time of collection of biomarker data (September 1991) were lower,
averaging 0.61 and 0.58 mg/m3 in the synthesis and polymerization rooms, respectively. The
exposed group included 13 synthesis workers, 12 polymerization workers, 5 packaging workers,
and 6 ambulatory workers, classified according to their primary work location. The remaining
four workers were either exposed for less than 6 months (two subjects) or had not been exposed
to AA during the 4 months preceding the study. Blood sampling and medical and neurological
examinations were performed approximately 1 hour after a work shift. The beginning of a work
shift marked the beginning of 24-hour urine sampling. For vibration sensitivity testing, a
referent group consisted of 105 unexposed healthy adults (51 males and 54 females aged 20-60
years). A historical control of 80 persons was used as referent for electroneuromyography tests.
A group of 10 nonexposed male workers from the same city as the exposed group was used as a
referent group for biomarkers of exposure and signs and symptoms of neurotoxicity.
A neurotoxicity index, with a maximal score of 50, was used to express severity of
peripheral neuropathy (Table F-l); the information used to derive the score was collected by
questionnaire. The prevalence of specific symptoms was also assessed individually. Biomarkers
of exposure to AA that were reported in the study included free AA in plasma, mercapturic acids
in urine, and the hemoglobin adduct formed by the reaction of AA with the N-terminal valine of
hemoglobin (AAVal).
F-1 DRAFT-DO NOT CITE OR QUOTE
-------
Table F-l. Scoring system for the neurotoxicity index
Endpoint
Numbness of extremities
Cramping pain
Loss of position sensation
Loss of pain sensation
Loss of touch sensation
Loss of vibration sensation0
According to tuning fork
Vibration threshold in big toe
Vibration threshold in index finger
Clumsiness of hands
Difficulty grasping
Unsteady gait
Decrease or loss of ankle reflexes
Muscular atrophy
Electroneuromyographic abnormalities'1
Maximum total score
Points3
1
1
2
0, 1,2, or3b
0, I,2,or3b
1
0,1, or 2
0,1, or 2
4
4
4
3 or 5
6
0.5 per abnormality (maximum 6)
50
aPoints were intended to reflect weight given to these observations by a clinical physician diagnosing a peripheral
neuropathy.
bWorkers who had lost their pain or touch sensation were assigned 1 to 3 points depending on the extent of loss:
fingers, hands, or forearms.
°The ratio between the vibration threshold of an individual and that of the corresponding control group with regard
to age was used for scoring vibration sensitivity using the Vibratron instrument. One point was given if this ratio
was 1.5-2.5 for fingers or 1.5-4.0 for toes and 2 points if it was 2.5-5.0 for fingers or 4.0-8.0 for toes.
Abnormalities consisted of measured alterations in electrical activity of selected muscles and nerves.
Source: Callemanetal. (1994).
Group mean biomarker levels and neurotoxicity indices are presented in Table 5-10 for
controls and the work locations of packaging, polymerization, ambulatory, and synthesis. The
average neurotoxicity index scores, as well as the averages of the hemoglobin adduct levels of
AA, decreased with physical distance from the synthesis room where the monomer itself was
handled. This relationship was not reflected by measured free plasma AA, urinary mercapturic
acid, or hemoglobin adduct levels of acrylonitrile or by results of hand or foot vibration
sensitivity measurements or estimates of accumulated in vivo doses of AA. Statistically
significant correlations were reported between each of the biomarkers of exposure and the
calculated neurotoxicity indices, with the exception of free plasma AA concentrations.
F-2
DRAFT-DO NOT CITE OR QUOTE
-------
Table F-2. Group means ± SD of biomarkers in different categories of
workers
Controls
Packaging
Polymerization
Ambulatory
Synthesis
Free AAa
(umol/L)
0.92
2.2
1.3
2.0
1.8 ±0.8
Merc. ac.b
(umol/24
hours)
3± 1.8
93 ±72
58 ±75
53 ±35
64 ±46
AAValc
(nmol/g)
0.0 ±0.0
3. 9 ±2.5
7.7 ±3.4
9.5 ±7.3
13.4 ±9.8
ANVald
(nmol/g)
0.23 ±0.18
19.1 ±5.7
19.1 ±12.9
16.3 ±3.7
19.5 ±7.6
AccD.4/
(mM/hour)
0.0 ±0.0
8.1 ±6.6
27.0 ±23.9
37.6 ±21.9
68.3 ±64.2
NInf
0.0 ±0.0
8.9 ±9.1
10.0 ±5.8
11. 3 ±9.8
19.2 ±10.6
Tree plasma AA.
bUrinary mercapturic acid.
'Hemoglobin adduct between N-terminal valine and AA.
dHemoglobin adduct between N-terminal valine and acrylonitrile.
Predicted cumulative in vivo AA dose (based on rates of AA-hemoglobin adduct formation in human globin
hydrolysates and mean AA exposure concentrations measured in areas of polymerization and synthesis by station
sampling) (see Section 3.1 and Bergmark et al. [1993] for additional information).
fNeurotoxicity index.
Source: Calleman et al. (1994).
The data for AA-Val neurotoxicity index (NIn) in Table F-2 are amenable to benchmark
dose analysis as presented in Figure F-l.
Pdynorrial IVbdel wth 0.95 Confidence Le\el
8.
C/5
CD
I
25
20
15
10
Fblynorrial
BVCL
BVD
6 8
dose
10
12
14
Figure F-l. Benchmark Dose Analysis for Calleman et al. (1994) data
Using a benchmark dose response level of 1 standard deviation (i.e., NIn = 6.9), from
overall mean response (NIn = 9.9), the BMD is 6.1 nmol AA-Val/g globin and the BMDL (as a
the POD) is 3.5 nmol AA-Val/g globin. Under the assumption that this BMDL represents a
F-3
DRAFT-DO NOT CITE OR QUOTE
-------
steady state level, the daily increment in adducts can be calculated using the equations presented
in Fennell et al. (2005) as follows:
2
Daily adduct increment = x steady state adduct level
RBClifespan
^ , ,, 2 3.5nmolAA-Val 0.058 nmol AA -Val
Daily adduct increment = x =
120 g globin g globin
To convert this daily increment in AA-Val to a daily intake of AA (in mg AA/kg bw), the
following equation is used for the value of 74.7 nmol AA-Val/g globin/mmol AA/kg bw reported
in Fennell et al. (2005), and a factor of 71.08 mg AA / mMoles AA to convert the daily intale of
A A to mg/kg bw:
,,,,„, 74.7 nmol AAVal 71.OS mMAA
Daily intake of AA = Daily increment of AA Val -=- —— x
' g globin mMAA
mM AA I kg bw
^ , T , , .. 0.05833 nmolAA- Val 74.7nmol AAVal 71.OS mMAA
Daily intake of AA = -T- —— x
g globin g globin mMAA
mM AA I kg bw
_ 0.056mgAA
kgbw
The air concentration that would provide a 70 kg person who breathes 20 m3 of air a
0.056 mg AA/kg bw daily intake is 0.19 mg/ m3.
Air ConcentrationBMDL_DajlIntake = 0.056mg/kg-day xlOkg+ —^j =0.19mg/m3
20m
This POD for a continuous inhalation exposure of 0.19 mg/m3 is divided by a total UF of
30: 10 for consideration of intraspecies variation (UFH: human variability), and 3 for database
uncertainty.
Total UF = 100
= 1 (UFA) x 10 (UFH ) x 10 (UFS) x 1 (UFD)
An UF 1 was selected for interspecies extrapolation because this RFC is based on human
data.
An UF of 10 was used to account for interindividual variability in toxicokinetics and
toxicodynamics to protect potentially sensitive populations and lifestages (UFH).
F-4 DRAFT-DO NOT CITE OR QUOTE
-------
An UF of 10 was used to extrapolate from a subchronic exposure duration to a chronic
exposure duration (UFS) because the point of departure was derived from a subchronic exposure
to humans (i.e., the UFS = 10).
An UF to account for database deficiency is not necessary for this derivation (i.e., UFo =
1) because the toxicity database for laboratory animals repeatedly exposed to AA is robust and
contains two 2-year carcinogenicity/toxicology drinking water studies in F344 rats and numerous
shorter-term oral toxicity studies in animals; two two-generation reproductive toxicity studies,
one in F344 rats and one in CD-I mice; several single-generation reproductive toxicity studies
involving prolonged prebreeding drinking water exposure of Long-Evans rats and ddY mice; and
several developmental toxicity studies involving gestational exposure of Sprague-Dawley and
Wistar rats and CD-I mice. The database identifies nerve degeneration as the critical effect from
chronic oral exposure. There are unresolved issues that warrant further research, including the
MOA of AA neurotoxicity, the potential for behavioral or functional adverse effects not detected
in the assays to date, and the uncertainty that heritable germ cell effects may occur at lower than
previously reported doses. These issues, however, do not warrant applying a UF for database
deficiencies.
The RfC for AA based on the Calleman et al. (1994) human data is calculated as follows:
RfC = Air Concentration BMDL-Daily intake ^UF
= 0.19mg/m3-100
= 0.002 mg/m3 (rounded to one significant digit)
Limitations in the Calleman et al (1994) data, and uncertainties in the results
The human data from the Calleman et al. (1994) study (described in detail in Section 4.1)
are very limited due to a small number of subjects (n=41), the narrow representation of the
general public (i.e., workers), and by a number of potentially confounding factors including
concurrent exposure to another neurotoxin (acrylonitrile), aggregate exposure via both dermal
and inhalation routes, a composite index of neurotoxicity, and control groups of different size
and composition. There are also no other human inhalation toxicity studies to support or
challenge the reproducibility or validity of the Calleman et al. (1994) study results.
F-5 DRAFT-DO NOT CITE OR QUOTE
------- |