DRAFT - DO NOT CITE OR QUOTE
EPA/635/R-07/009A
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)
December 2007
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 information is distributed solely for the purpose of pre-dissemination peer review
under applicable information quality guidelines. It has not been formally disseminated by EPA.
It does not represent and should not be construed to represent any Agency determination or
policy. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
DRAFT-DO NOT CITE OR QUOTE
-------
CONTENTS —TOXICOLOGICAL REVIEW OF ACRYLAMIDE
(CAS No. 79-06 1)
LIST OF TABLES v
LIST OF FIGURES x
ABBREVIATIONS AND ACRONYMS xi
FOREWORD xiv
AUTHORS, CONTRIBUTORS, AND REVIEWERS xv
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 11
3.1. ABSORPTION 11
3.2. DISTRIBUTION 18
3.3. METABOLISM 21
3.4. ELIMINATION 33
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS 37
4. HAZARD IDENTIFICATION 44
4.1. STUDIES IN HUMANS—EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS 44
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIO ASSAYS IN
ANIMALS—ORAL AND INHALATION 66
4.2.1. Oral Exposure 66
4.2.1.1. Subchronic Studies 66
4.2.1.2. Chronic Studies 71
4.2.2. Inhalation Exposure 82
4.2.2.1. Subchronic Studies 83
4.2.2.2. Chronic Studies 83
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION .. 83
4.3.1. Reproductive Toxicity Studies 83
4.3.2. Developmental Toxicity Studies 102
4.4 HERITABLE GERM CELL STUDIES 112
4.5. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES 119
4.5.1. Neurotoxicity Studies 119
4.5.2. Other Cancer Studies 120
4.6. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION 123
4.6.1. Neurotoxicity Studies 123
4.6.2. Genotoxicity Studies 124
4.7. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS 129
4.7.1. Oral 129
4.7.2. Inhalation 134
4.7.3. Mode-of-Action Information 134
4.8. EVALUATION OF CARCINOGENICITY 138
4.8.1. Summary of Overall Weight of Evidence 138
4.8.2. Synthesis of Human, Animal, and Other Supporting Evidence 139
ii DRAFT-DO NOT CITE OR QUOTE
-------
4.8.3. Mode of Action for Carcinogenicity 143
4.8.3.1. Hypothesized Mode of Action—Mutagenicity 143
4.8.3.2. Alternative Mode of Action—Disruption of Hormone Levels or
Signaling 153
4.8.3.3. Conclusion About the Mode of Action 164
4.9. SUSCEPTIBLE POPULATIONS 164
4.9.1. Possible Childhood Susceptibility 164
4.9.2. Possible Gender Differences 166
4.9.3. Other 167
5. DOSE-RESPONSE ASSESSMENTS 168
5.1. ORAL REFERENCE DOSE 168
5.1.1. Choice of Principal Study and Critical Effect—with Rationale and
Justification 168
5.1.2. Methods of Analysis—Including Models (PBTK, HMD, etc.) 170
5.1.3. RfD Derivation—Including Application of Uncertainty Factors 172
5.1.4. Previous RfD Assessment 175
5.2. INHALATION REFERENCE CONCENTRATION (RfC) 175
5.2.1. Choice of Principal Study and Critical Effect—with Rationale and
Justification 175
5.2.2. Methods of Analysis—Including Model (PBTK, BMD, etc.) 176
5.2.3. RfC Derivation—Including Application of Uncertainty Factors 177
5.2.4. Previous RfC Assessment 179
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE AND INHALATION
REFERENCE CONCENTRATION 179
5.3.1 Areas of Uncertainty 180
5.3.2 Uncertainty Factors in Deriving the RfD and RfC 184
5.4. CANCER ASSESSMENT 190
5.4.1. Choice of Study/Data—with Rationale and Justification 190
5.4.2. Dose-Response Data 191
5.4.3. Dose Adjustments and Extrapolation Method(s) 192
5.4.4. Human Equivalent Concentration Using the PBTK Model 195
5.4.5. Oral Slope Factor and Inhalation Unit Risk 196
5.4.5.1. Oral Slope Factor 196
5.4.5.2. Inhalation Unit Risk 197
5.4.6 Application of Age-Dependent Adjustment Factors 199
5.4.7. Uncertainties in Cancer Risk Values 202
5.4.7.1. Areas of Uncertainty 202
5.4.8. Previous Cancer Assessment 212
5.5. QUANTITATING RISK FOR HERITABLE GERM CELL EFFECTS 213
5.5.1. Quantitative Approaches 213
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD 223
AND DOSE RESPONSE 223
6.1. HUMAN HAZARD POTENTIAL 223
6.2. DOSE RESPONSE 226
6.2.1. Noncancer/Oral 226
6.2.2. Noncancer/Inhalation 227
6.2.3. Cancer/Oral 228
6.2.4. Cancer/Inhalation 229
iii DRAFT-DO NOT CITE OR QUOTE
-------
7. REFERENCES 231
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
1
APPENDIX F. YOUNG ET AL (2007) PBTK/TD MODEL SUPPORTING
DOCUMENTATION 1
iv 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 22
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 mice26
Table 4-1. Observed deaths and SMRs for selected causes by follow up period for all workers
(compared with the general US population) 49
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) 50
Table 4-3. Neurological symptoms self-reported by acrylamide workers and nonexposed
workers 57
Table 4-4. Scoring system for the neurotoxicity index 59
Table 4-5. Group means ± SD of biomarkers in different categories of workers 61
Table 4-6. Correlation coefficients (linear regression) for relationships between biomarkers and
neurotoxicity index 62
Table 4-7. Incidences of symptoms in 210 tunnel workers classified into exposure groups based
on levels of hemoglobin adducts of acrylamide 64
Table 4-8. Light and electron microscopic data for left sciatic nerves from rats exposed to
acrylamide in drinking water for 90 days 70
Table 4-9. Light microscopic data for tibial nerves from F344 rats exposed to acrylamide in
drinking water for 2 years 74
Table 4-10. Incidences of selected tumors in male and female F344 rats exposed to acrylamide
in drinking water for 2 years 75
Table 4-11. Dosing parameters of groups of rats given acrylamide in drinking water for 106-108
weeks in the carcinogenicity study 76
Table 4-12. Light microscopic data for sciatic nerves from F344 rats exposed to acrylamide in
drinking water for 2 years 78
Table 4-13. Incidences of tumors in male F344 rats exposed to acrylamide in drinking water for
2 years 79
v 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 80
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 82
Table 4-16. Changes in reproductive parameters in F344 rats exposed to acrylamide in drinking
water for two generations 87
Table 4-17. Results of the dominant lethal mutation assay in F344 rats 88
Table 4-18. Results of dominant lethality testing in male Swiss CD-I mice exposed to
acrylamide in the drinking water 90
Table 4-19. Effects of acrylamide in drinking water on grip strength of mice 92
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 93
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 97
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 100
Table 4-23. Maternal and fetal effects in Sprague-Dawley rats and CD-I mice following oral
(gavage) administration of acrylamide to pregnant dams 104
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 Ill
Table 4-25. Frequency of translocation carriers in offspring derived from males exposed to
acrylamide or glycidamide 114
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 114
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 115
Table 4-29. Acrylamide initiation of squamous cell carcinomas or papillomas in female
SENCARmice 120
Table 4-30. Acrylamide initiation of skin tumor masses > 1mm in female SENCAR mice.... 121
Table 4-31. Noncancer effects in animals repeatedly exposed to acrylamide by the oral route 130
Table 4-32. Neurological effects following exposure to acrylamide in species other than the rat
and mouse 132
Table 4-33. Incidence of tumors with statistically significant increases in both 2-year bioassays
withF344 rats exposed to acrylamide in drinking water 141
Table 4-34. Circulating thyroid hormone levels in F344 rats following exposure to acrylamide in
drinking water for 14 or 28 days 162
vi DRAFT-DO NOT CITE OR QUOTE
-------
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 163
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 .... 171
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 172
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 172
Table 5-4. PBTK model simulation results for HEC based on the rat neurotoxicity BMD 173
Table 5-5. PBTK model simulation results for HEC based on the rat neurotoxicity BMD 177
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 183
Table 5-7. Summary of uncertainty in the acrylamide noncancer risk assessment 187
Table 5-8. Incidence of tumors with statistically significant increases in a 2-year bioassay with
F344 rats exposed to acrylamide in drinking water 192
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
194
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 . 195
Table 5-11. PBTK model simulation results for HEC based on male rat carcinogenicity data 196
Table 5-12. PBTK model simulation results for HEC to derive the inhalation unit risk based on
male rat oral exposure cancer data 199
Table 5-13. Summary of uncertainty in the acrylamide cancer risk assessment 208
Table 5-14. Heritable genetic risk estimates for humans exposed to acrylamide 222
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
vii 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 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 3
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) 3
Table E-2: Data used to recalibrate the Kirman et al. (2003) model parameters 7
Table E-3: AUC Predictions from the Original Kirman Model versus AUCs Derived from
Hemoglobin Adduct Data 11
Table E-4:Recalibrated PBTK Model Parameter Values for the Rat 12
Table E-5: Results of the Recalibratedl Kirman et al (2003) Model versus Urinary Metabolite
Data and AUCs Derived from Hemoglobin Adduct Data 14
Table E-6: Estimated Internal AUC Acrylamide and Glycidamide Doses Produced by Various
Drinking Water Intakes 16
Table E-7: Available Data for Calibration of the Human PBTK model 17
Table E-8. Parameters for the Human (male) Acrylamide PBTK Model 21
Table E-9: Human PBTK Model Predictions versus AUCs and Urinary Metabolites 22
Table E-10: Estimated AUCs in Humans for Acrylamide and Glycidamide from a Drinking
Water Exposure 23
viii DRAFT-DO NOT CITE OR QUOTE
-------
Table E-l 1: Estimated AUCs in Humans for Acrylamide and Glycidamide from An Inhalation
Exposure 23
Table F-l: Data Generated at NCTR on AA and GA in rats and mice 2
Table F-2: Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration
to Rats [Mean± Standard Deviation (Range)] 3
Table F-3: Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration
toMicea 4
Table F-5:. Pharmacokinetic Parameters from AA Administration to Human Volunteers 5
ix DRAFT-DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
Figure 2-1. Chemical structure of acrylamide (AA) with carbon numbers indicated 3
Figure 3-1. Metabolic scheme for acrylamide (AA) and its metabolite glycidamide (GA) 25
Figure 3-2. Hemoglobin andDNA adducts of acrylamide and glycidamide 32
Figure 3-2. Schematic of theKirman et al. PBTK Model for Acrylamide 43
Figure 3-3. Schematic of the Young et al. PBTK Model for Acrylamide 43
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 D-l. Observed and predicted incidences for mammary gland tumors in female rats
exposed to acrylamide in drinking water for 2 years 12
Figure D-2: Observed and predicted incidences for thyroid tumors in female rats exposed to
acrylamide in drinking water for 2 years 15
Figure D-3: Observed and predicted incidences for mammary or thyroid tumors in female rats
exposed to acrylamide in drinking water for 2 years 18
DRAFT-DO NOT CITE OR QUOTE
-------
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
xi DRAFT-DO NOT CITE OR QUOTE
-------
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
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
LD50 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
xn
DRAFT-DO NOT CITE OR QUOTE
-------
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
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
xiii 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).
xiv 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, D.C.
AUTHORS (CONTRACT)
David W. Wohlers, Ph.D.
Peter R. McClure, Ph.D., DABT
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Contract Number: GSF0019L
PBTKModeling
Dale Hattis, Ph.D.
Center for Technology, Environment, and Development
Clark University
950 Main Street
Worcester, MA
Katherine Walker, Sc.D.
P.O. Box 6308
Lincoln, MA
xv DRAFT-DO NOT CITE OR QUOTE
-------
REVIEWERS
This document and the accompanying IRIS Summary has been peer reviewed by EPA
scientists and independent scientists external to EPA. Comments from all peer reviewers were
evaluated carefully and considered by the Agency during the fmalization of this assessment.
INTERNAL EPA REVIEWERS
Ila Cote
Office of Research and Development
National Center for Environmental Assessment
Kevin Crofton
Office of Research and Development
National Health and Environmental Effects Laboratory
Sally Darney
Office of Research and Development
National Health and Environmental Effects Laboratory
Kerry Dearfield
Office of Research and Development
Office of The Science Advisor
[Currently with the US Department of Agriculture, Food Safety and Inspection Service]
Lynn Flowers
Office of Research and Development
National Center for Environmental Assessment
Gary Foureman
Office of Research and Development
National Center for Environmental Assessment
Angela Howard
Office of Research and Development
National Center for Environmental Assessment
Gene Hsu
Office of Research and Development
National Center for Environmental Assessment
[Currently with Merck & Co Inc, West Point, PA]
John Vandenberg
Office of Research and Development
National Center for Environmental Assessment
xvi 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 acrylamide.
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 (possibly threshold)
mode of action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with
uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m3) is
analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for both the respiratory system (portal-of-entry) and for
effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
acute (24 hours), short-term (>24 hours up to 30 days), and sub chronic (>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 an upper bound on the
estimate of risk per mg/kg-day of oral exposure. Similarly, a "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 (1983). EPA Guidelines and Risk Assessment Forum Technical 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), Guidelines for Developmental Toxicity Risk
Assessment (U.S. EPA, 1991), Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA,
1996b), Guidelines for Neurotoxicity Risk Assessment (U.S. EPA, 1998a), Guidelines for
Carcinogen Risk Assessment (U.S. EPA, 2005a), Supplemental Guidance for Assessing
1 DRAFT-DO NOT CITE OR QUOTE
-------
Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA, 2005b), 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 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: Peer Review
(U.S. EPA, 1998b, 2000a, 2005c), Science Policy Council Handbook: Risk Characterization
(U.S. EPA, 2000b), Benchmark Dose Technical Guidance Document (U.S. EPA, 2000c),
Supplementary Guidance for Conducting Health Risk Assessment of Chemical Mixtures (U.S.
EPA, 2000d), and A Review of the Reference Dose and Reference Concentration Processes (U.S.
EPA, 2002).
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 through August 2007 is included in
the assessment.
Estimates of risk for acrylamide derived by other organizations are compiled by the
National Libraries of Medicine and can be found on the TOXNET webpage at
http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen7iter . Additionally, the Joint FAO/WHO Expert
Committee on Food Additives (JECFA) information on acrylamide risk and toxicity is available
at: http://www.who.int/foodsafetv/chem/chemicals/acrylamide/en/.
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).
O
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 factor:
79-06-1 (Verschueren, 2001)
71.08 (Verschueren, 2001)
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)
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 AA will exist
solely as a vapor in the ambient atmosphere. Vapor-phase AA will be degraded in the
4 DRAFT-DO NOT CITE OR QUOTE
-------
atmosphere by reaction with photochemically produced hydroxyl radicals; the half-life for this
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 fmgerling
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 non-smokers (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
5 DRAFT-DO NOT CITE OR QUOTE
-------
research in many European countries and the United States determined that AA is formed
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, 2006a) 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, 2006a).
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).
DRAFT-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
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
2762.0
116.0
432.0
1057.0
1243.0
151.0
457.0
909.0
1925.0
539.0
13.0
5399.0
St. Dev.
36.6
427.9
27.7
107.9
249.1
331.1
43.4
143.0
243.6
411.7
106.3
2.4
1018.8
"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).
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
Min
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
1321.8
Max
2228.0
3770.0
2779.0
3324.0
7834.0
2838.0
910.0
3044.0
846.0
975.0
2955.0
"Grated 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 3442 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
FAO/WHO (2000)
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)
NFCS, Netherlands
SNFA, Sweden (2002)
SNT, Norway (2003)
Daily intake jig/kg-day
Mean (age group)
0.3-0.8
0.2-0.4
1.1(15-18)
0.28 (16-57)
0.5 (>15)
1.4 (2-14)
0.7
0.43 (>2)
1.06 (2-5)
0.40 (>2)
1.07 (2-5)
0.48 (1-97)
1.04 (1-6)
0.71 (7-18)
0.45 (18-74)
0.49 (males)
0.46 (females)
0.36 (9, boys)
0.32 (9, girls)
0.52 (13, boys)
0.49 (13, girls)
0.53 (16-30, males)
0.50 (16-30, females)
Percentilea'b
3.4a
i.r
2.9a
0.92b
2.31b
0.95a
2.33b
0.6a
1.1"
0.9a
1.03
1.01b
0.86b
0.72b
0.61b
1.35b
1.2b
Source
http://www.who.int/foodsafety/publications/chem/en/acrylamide full, pdf
http://europa.eu.int/comm/food/fs/sc/scf/outl3 1 en.pdf
http://www .bfr.bund.de/cm/208/ Abschaetzung_der_Acrylamid_Aufnahme_durch_
hochbelastete Nahrungsmittel in Deutschland Studie.pdf
http://www.bag.admin.ch/verbrau/aktuell/d/DDS%20acrylamide%20preliminary%
20communication.pdf
http://www.afssa.fr/ftp/afssa/basedoc/acrylpoint2sansannex.pdf
http ://www. cfsan.fda.gov/~dms/acry expo . html
http://www.cfsan.fda.gov/~dms/acryexpo.html
http://www.cfsan.fda.gov/~dms/acryexpo.html
Konigs et al. (2003)
Svensson et al. (2003)
Dybing and Sanner (2003)
a = 95th percentile.
b = 90th percentile.
Source: For all exposures estimates from 2002-2004 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
Diet: general population
2- to 5-year-olds
Drinking water
Personal care products
Cigarette smoking
Occupational exposures
Mean or median3
0.43
1.06
No data
-0.5
0.67 (from cigarette data)
2.6 (from adduct data)b
1.4-18
90th percentile or upper boundary3
0.92
2.31
0.01
1.1 (female)
1.3
~6
43 (based on PELe)
Totals (adults)
General population
Nonsmokers
Smokers
Occupational exposure"1
Nonsmokers
Smokers
0.98C
0.85 (from adduct data)
1.7 (from cigarette data)
3.6 (from adduct data)
2.4-19
3. 1-20 (cigarette data)
5-22 (adduct data)
2.0
3.2
45-52
45
46
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 1000.
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/kgbw/day and 43 ug/kgbw/day, respectively.
Source: NTP/CERHR (2004).
10
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.
Glycidamide, 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., 2007). 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 acrylamide [AA] and glycidamide [GA] hemoglobin
adducts, and glycidamide 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 acrylamide exposure by using the results of the
toxicokinetic study described above in 24 volunteer adult males. The estimated average daily
background exposure to acrylamide was 1.26 jig/kg-day based on the subject's preexposure
background acrylamide-hemoglobin-terminal-valine adduct levels (AAVal) (averaging about
80 fmol/mg globin). In an occupational 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
11 DRAFT-DO NOT CITE OR QUOTE
-------
20-80 fmol AAVal/mg globin), Fennell et al. (2005) estimated a daily acrylamide intake of
0.31-1.26 ng/kg-day. For a 70 kg adult this translates into a total daily intake of 22-88 jig of
acrylamide. 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 acrylamide 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 acrylamide (or acrylonitrile). Blood samples from 5 of the 41
exposed workers were also analyzed for hemoglobin adducts of glycidamide (a principal
metabolite of acrylamide 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 5 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 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
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
12 DRAFT-DO NOT CITE OR QUOTE
-------
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
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 with N-terminal valine residues in human hemoglobin is 4.4 x 10 6 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
-------
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 glycidamide-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). 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 glycidamide, and the
respective metabolite of glycidamide, N-acetyl-S-(2-hydroxy-2-carbamoylethyl)cysteine
(GAMA), were quantified 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 glycidamide 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.
14 DRAFT-DO NOT CITE OR QUOTE
-------
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 h. Urinary AA mercapturic acid metabolites were considerably reduced after 48 h of
fasting, with levels even well below the median level in non-smokers. 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 h dietary recall as 21 (13-178) |j,g for non-smokers and 26 (12-67) |j,g for smokers. The
median dietary exposure to acrylamide was estimated to be 0.47 (range 0.17-1.16) |j,g /kg body
weight per day. The median (range) total excretion of acrylamide in urine during 24 h was 16 (7-
47) |j,g acrylamide for non-smokers and 74 (38-106) |j,g 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 B6C3F1 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 post
dosing, 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 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).
15 DRAFT-DO NOT CITE OR QUOTE
-------
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 oral 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 B6C3F1 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
exhaled air; and 56% remained in the body. In mice, 51% was recovered in urine, feces, and
nose-tube and cage washes; less than 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
2 If reference minute ventilation rates for rats (0.7 cmVmin-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:
16 DRAFT-DO NOT CITE OR QUOTE
-------
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 post application, 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
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
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
3(imol/mmol) = 6.5 jimol/rat-exposure period.
17 DRAFT-DO NOT CITE OR QUOTE
(211 grams/rat) x (m3/106cm ) x (mmol/71.08 mg) x (103jimol/mmol) = 6.5 jimol/rat-exposure period.
-------
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
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 5 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
18 DRAFT-DO NOT CITE OR QUOTE
-------
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 ng/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 B6C3F1 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
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 B6C3F1 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.
19 DRAFT-DO NOT CITE OR QUOTE
-------
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 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 of from 10-
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
20 DRAFT-DO NOT CITE OR QUOTE
-------
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.
Glycidamide (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
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]).
The Fennell et al. (2005) study also provided data on the amount of hemoglobin adducts
derived 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.
21 DRAFT-DO NOT CITE OR QUOTE
-------
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
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.
Fennell et al. (2005) also derived second order rate constants (see Table 3-1) for the
reaction of acrylamide or glycidamide with the N-terminal valine residue of hemoglobin in rats
and humans. These rate constants are of particular use for physiologically based toxicokinetic
(PBTK) models of acrylamide and glycidamide (see Section 3.5).
Table 3-1. Second order rate constants for reaction of acrylamide or
glycidamide with the N-terminal valine residue of hemoglobin
Rat
Human
Rat/human
AAVal L/g Hba/hour
3.82 x 10^
4.27 x 10^
0.89
GAVal L/g Hba/hour
4.96 x 1(T6
6.72 x 1(T6
0.73
AAVal/GAVal
0.77
0.64
aHb = hemoglobin.
Source: Fennell et al. (2005).
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
22 DRAFT-DO NOT CITE OR QUOTE
-------
acrylamide (d(3)-AA), representing 13 |ig/kg body weight, in drinking water. Urine samples
before dosing and within 46 h after the dose were analysed for d(3)-AAMA and d(3)-GAMA by
LC-ESI-MS/MS. Total recovery in urine after 24 h 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 glycidamide 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. Conjugation with glutathione exceeded
the formation of the reactive metabolite glycidamide. The data suggests an at least 2-fold and 4-
fold lower relative internal exposure for glycidamide from dietary acrylamide in humans
compared with rats or mice, respectively.
Previous studies in people who are occupationally exposed or who smoke established that
AA and GA form hemoglobin adducts (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 glycidamide in humans
(Bergmark et al., 1993). 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). Other related compounds like acrylonitrile and N-methylolacrylamide (NMA) also form
hemoglobin adducts. These confounders should be discussed if they are potentially present in
studies that use AA hemoglobin adducts as the basis for estimating exposure to acrylamide. This
is further discussed at the end of this section under the heading "Potential confounders for the
hemoglobin adduct biomarker of acrylamide exposure."
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 has also been reported (Bergmark
23 DRAFT-DO NOT CITE OR QUOTE
-------
et al., 1991). The hemoglobin binding index of AA to cysteine was found to be 6400 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 1820 pmol/g Hb/|imol GA/kg (Bergmark et al.,
1991). 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).
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 B6C3F1 mice
exposed to oral doses of 50 mg/kg [l,2,3-13C]-labeled acrylamide (Sumner et al., 1992).
24 DRAFT-DO NOT CITE OR QUOTE
-------
acrylamide Hb adducts
Hb-
O
H9C=C—C—NhL
2 H 2
acrylamide
GSH
GS-CH2-CH2-CONH2
I
CYP2E1
glycidamide Hb adducts
^— Hb
O
/ \
H9C C—CONH,
2 H 2
glycidamide
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
GSH
CH2OH
GS—C—CONH9
H 2
• DMA 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—CONH9
H 2
N-acetyl-S-(1-carbamoyl-2-hydroxyethyl)cysteine
Figure 3-1. Metabolic scheme for acrylamide (AA) and its metabolite glycidamide (GA).
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).
25
DRAFT-DO NOT CITE OR QUOTE
-------
Table 3-2 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 3-2. 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 mice
Metabolite21
From AA precursor
N-acetyl-S-(3 -amino-3 -oxopropyl)cy steine
Glycidamide
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
% 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 acrylamide to the
epoxide derivative, glycidamide. Gylcidamide (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. Glycidamide may
also undergo hydrolysis, perhaps catalyzed by epoxide hydrolases (Sumner et al., 1992, 1999),
leading to the formation of 2,3-dihydroxypropionamide and 2,3-dihydroxypropionic acid.
Glycidamide 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
acrylamide to glycidamide is about twofold 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). Glycidamide 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).
26
DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide, 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 glycidamide (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
glycidamide 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 CYP2E1-
null mice compensated for the CYP2E1 deficiency by metabolizing more of the administered
AA via direct conjugation with GSH.
Information on human specificity of CYP2E1 to AA oxidation is lacking. It is noted,
however, that 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 to 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, post hoc).
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 compared with older infants,
children, and adults. However, actual differences will depend upon the delivery rate and
substrate concentration relative to the value of the Michaelis-Menten constant (Km) for
CYP2E1, which will determine the total amount metabolized (or parent compound cleared)
(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.
27 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide 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 acrylamide 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 [l,2,3-13C]-labeled acrylamide. In other experiments, no exhaled 14CC>2 was
detected following oral administration of 50 mg/kg [l-14C]-labeled acrylamide to male Sprague-
Dawley rats (Kadry et al., 1999), whereas 3-4% of i.v. injected [l,3-14C]-acrylamide (2 or 100
mg/kg) was detected as 14CC>2 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 acrylamide, 14CC>2 in exhaled breath accounted for 1.8 ± 0.2% of radioactivity
recovered in exhaled air, urine, feces, and tissues (Sumner et al., 2003). Similarly, 14CC>2 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 B6C3F1 mice and F344 rats, respectively, following nose-
only inhalation exposure to 3 ppm of a mixture of [l,2,3-13C]-acrylamide and [2,3-14C]-
acrylamide (Sumner et al., 2003).
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 acrylamide conjugated to
GSH vs. the percentage of AA converted to GA. Following i.p. or gavage administration of
50 mg/kg [l,2,3-13C]-acrylamide 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. In contrast, metabolites associated with direct conjugation of AA with glutathione
accounted for 64 ± 2.4% of metabolites in urine collected for 24 hours following 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. 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. The percentages of urinary metabolites associated with GA
formation following i.p. and gavage exposure were reported to be statistically significantly
smaller than the value for inhalation exposure. These results indicate that a smaller percentage
of AA is converted to GA following oral or intraparenteral exposure compared with inhalation
exposure. The biological significance of this small difference is uncertain. These results in
28 DRAFT-DO NOT CITE OR QUOTE
-------
F344 rats are in contrast to the increased percentage of GA formation observed in mice from an
oral gavage or dietary exposure vs. an i.v. exposure (Doerge et al., 2005b).
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 acrylamide by gavage (24 hours after the
last dose of a drinking water solution of 20 mg/kg-day nonlabeled acrylamide for 34 days) or by
a single i.p. 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 (tl/2) 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 the longer oral regimen (30%) compared with the 11-day i.p. regimen (8%), although
the GA AUCs were similar (822 units for 20 mg/kg-day subchronic oral vs. 730 units for 50
mg/kg-day 11-day i.p.) and no correlation was found to the different enzyme activities involved
in GA formation (CYP2E1) or metabolism (epoxide hydrolase). Concentrations of AA-
hemoglobin adducts with the longer oral regimen were about 30% less than concentrations from
the i.p. regimen, and concentrations of GA-hemoglobin adducts were about twofold higher than
those from the i.p. regimen. 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
29 DRAFT-DO NOT CITE OR QUOTE
-------
was concluded that the mechanism of axonal degeneration did not appear to involve route- or
dose-rate differences in metabolism or disposition of aery 1 amide.
Doerge et al. (2005b) compared the toxicokinetics of AA and GA in serum and tissues of
male and female B6C3F1 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, likely due to a first-pass effect, but possibly the result of some other
kinetic change. In comparing the results of this study with previous studies by this laboratory at
a 500-fold higher concentration (Twaddle et al., 2004), an increase in 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 B6C3F1 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 B6C3F1 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 B6C3F1 mice.
Since rats and mice had a comparable GA elimination half-life, this approximately five-fold
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 B6C3F1 mice and F344 rats at 50 mg aery 1 amide/kg, or an equimolar dose of GA
(61 mg/kg). They report GA-derived DNA adducts of adenine and guanine formed in all tissues
30 DRAFT-DO NOT CITE OR QUOTE
-------
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.
Acrylamide and glycidamide 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 acrylamide 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 1Q~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 glycidamide,
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 glycidamide-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 et al., 2003). Using liquid
chromatography with tandem mass spectrometry and isotope dilution, Gamboa 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 2000 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
31 DRAFT-DO NOT CITE OR QUOTE
-------
AA treatment in adult mice showed a supralinear dose-response relationship, consistent with
saturation of oxidative metabolism at higher doses.
o
II
H9C=C—C—NhL
2 H 2
acrylamide
Hb-Val-N-CH2-CH2-CONH2
or
Hb-Cys-S-CH2CH2CONH2
Acylamide Hb Adducts
HB-Cys-SH
O
Hb-Val-NH,
\
HC - C— CONH2
H
glycidamide
GSH
Hb-Val-N-CH2-CH-CONH2
OH
Hb-Cys-S-CH2-CH-CONH2
Ix
Glycidamide Hb Adducts
r
DNA-guanine-N7-CH2-CH-CONH2
Glycidamide DMA Adduct
Figure 3-2. Hemoglobin and DNA adducts of acrylamide and glycidamide.
Sources: Dearfield et al. (1995); Bergmark et al. (1993, 1991).
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 acrylamide and,
like acrylamide, is used in the production of grouting agents. Acrylonitrile can be used as a
precursor in one method to manufacture acrylamide, and is also formed when acrylamide 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 acrylamide 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 acrylamide, 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).
32
DRAFT-DO NOT CITE OR QUOTE
-------
Paulsson et al. (2002) measured hemoglobin adducts (and micronucleus [MN]
frequencies) in mice and rats after acrylamide 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 and one group after 48 hours for the hemoglobin adduct analysis. The identical (N-[2-
carbamoylethyljvaline) adduct and the respective epoxide metabolite (N-[2-carbamoyl-2-
hydroxyethyljvaline) 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 acrylamide 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
acrylamide 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
acrylamide 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 glycidamide
(GA) i.e. N-acetyl-S-(2-carbamoylethyl)-L-cysteine (AAMA) andN-(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 non-smokers (n=16) with median levels
33 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide exposure. The level of AAMA in
the occupationally non-exposed collective (n=29) ranged from 3 to 338 |ig/L, the level of
GAMA from
-------
for non-smokers and 74 (38-106) |j,g acrylamide for smokers. Median intakes (range) of AA
were estimated based on 24 h dietary recall as 21 (13-178) |j,g for non-smokers and 26 (12-67)
|j,g for smokers. The median dietary exposure to acrylamide was estimated to be 0.47 (range
0.17-1.16) |j,g /kg body weight per day. 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 data
Results from animal studies indicate that urinary excretion of metabolites is the principal
route of elimination of absorbed acrylamide, 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 acrylamide 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 acrylamide
were placed in metabolism cages designed to trap expired air (Miller et al., 1982). In contrast,
3 5 DRAFT-DO NOT CITE OR QUOTE
-------
studies with radiolabel in the carbon-1 position suggest that exhalation of CO2 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 acrylamide (Hashimoto and
Aldridge, 1970) and about 4% of an injected dose of 2 mg/kg [l,3-14C]-labeled acrylamide (Dow
Chemical Co., 1984) were 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
acrylamide to male Sprague-Dawley rats (Kadry et al., 1999).
In studies with male F344 rats given single i.v. doses of [l,3-14C]-labeled acrylamide,
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% CO2; 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 acrylamide 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 acrylamide, 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., 2005 a,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
acrylamide 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.
36 DRAFT-DO NOT CITE OR QUOTE
-------
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, p<0.001; GA vs. GAMA,
r2 =0.81, p<0.001) and mice (AA vs. AAMA, r2 =0.86, p<0.001; GA vs. GAMA, r2 =0.57,
p<0.001) . 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,
p<0.001; AUC-AA vs. AAMA, r2 =0.83, p<0.001) and mice (AUC-AA vs. AA, r2 =0.41,
p<0.011; AUC-AA vs. AAMA, r2 =0.41, p<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. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS
Two physiologically based toxicokinetic (PBTK) models for acrylamide were available
from the literature (Kirman et al., 2003; Young et al., 2007). Kirman et al. (2003) developed a
PBTK model that simulates the disposition of AA and its epoxide metabolite, GA, in the rat
based upon published kinetic data from the 1980s and early 1990s (Miller et al., 1982; Ramsey et
al., 1984; Raymer et al., 1993; Sumner et al., 1992). The Kirman et al. model parameters were
recalibrated by EPA using more recent kinetic and hemoglobin adduct data in rats and mice
(Sumner et al., 2003, Fennell et al., 2005; Doerge et al., 2005b,c) and human kinetic and
hemoglobin adduct data (Fennell et al., 2005, Boettcher et al., 2005). The recalibrated Kirman et
al. PBTK model was used in deriving oral and inhalation toxicity values for acylamide's
potential noncancer and cancer effects (see Section 5).
Young et al. (2007) developed a PBTK/TD (toxicodynamic) model that simulates AA
and GA kinetics in mice, rats, and humans, and adds representation of GA-DNA adduct
37 DRAFT-DO NOT CITE OR QUOTE
-------
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., 2005 a,b,c) and
from the literature (Barber et al., 2001; Sumner et al., 1992, 2003; 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). For the Young et al. (2007) model to be used for specific
application in deriving a toxicity value, additional work is needed to determine which parameter
values would be the most appropriate to use.
The following discussion provides a general description of each model, and is
accompanied by Appendix E and Appendix F that contain tables of text with 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 3-2. 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.),
oral 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 and N-
acetyl-S-(carbamoyl-2-hydroxyethyl)cysteine; or 2) with further saturable metabolism by
epoxide hydrolase to yield 2,3-dihydroxypropionamide. 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.
3 8 DRAFT-DO NOT CITE OR QUOTE
-------
The model parameters values and sources in the original Kirman model are presented in
Appendix E, Table E-l. The 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
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.
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, mice, and humans have subsequently
been published (Boettcher et al., 2005; Doerge et al., 2005a,b,c; Fennell et al., 2005; Sumner et
al., 2003), and were used by EPA to recalibrate the Kirman et al. (2003) and to improve the
utility of the Kirmal et al. model for use in deriving toxicity values, specifically to estimate
human equivalent concentrations based on test animal dose-response data, and to conduct a
route-to-route extrapolation of the dose-response relationship from an oral-to-an inhalation
exposure.
In calibrating a PBTK model, hemoglobin adduct data are considered a better surrogate
of serum levels of AA and GA than urinary metabolite data because the formation of hemoglobin
adducts is a direct function of the blood concentration of the reactive agents over 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 if 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
39 DRAFT-DO NOT CITE OR QUOTE
-------
than 100%. Hemoglobin adduct levels, however, provide a direct measure of the total amount of
parent acrylamide and glycidamide 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 acrylamide-induced
toxicity.
The parameters that were recalibrated in the Kirman et al. model included tissue/blood
partition coefficients for both AA and GA, and the balance between P450 metabolism,
gluthathione conjugation, and other non-P450 metabolism. The values for the partition
coefficients in the Kirman et al. (2003) model were first re-estimated using a methodology
(detailed in Appendix E) based on octanol/water partition coefficients and supported by an
analysis of the observed versus predicted ratios of 50 chemicals. Additional refinements in the
partition coefficients were based on more recent measurements of the volume of distribution of
AA and GA in mice and rats (Doerge et al., 2005b,c).
EPA also evaluated an alternative pathway for AA and GA clearance by GSH based on
results from Tong et al. (2004) that suggest that the reaction between AA and glutathione may be
largely direct, and not requiring catalysis by glutathione transferase enzymes such as
glutathione-S-transferase. A variation of the rat PBTK model was developed to represent non-
enzymatic binding of AA and GA to glutathione throughout the body. The value of the second
order reaction rate constant for AA-glutathione binding was based on Tong et al.'s (2004)
results: 0.021/second-mol/L glutathione x 60 seconds/minute x 60 minutes/hour = 75.6/hour-
mol/L glutathione or 0.0756/hour-mmol/L glutathione. To implement this model, a similar
reaction rate for GA was derived from the GA to AA reaction rate ratios in the liver based on
previous simulations. There is also the possibility that the overall binding of AA and GA to
glutathione includes a mix of enzyme-mediated and direct binding, for example, with the AA
reaction being solely or primarily direct, and the GA reaction being catalyzed by a glutathione-S-
transferase. A mixed pathway model, however, was not evaluated pending further data to better
resolve this issue.
To estimate the human equivalent concentrations in the derivation of the noncancer and
cancer toxicity values, the Kirman et al. PBTK model parameters values were adjusted to
simulate an adult human male. The partition coefficients used in the rat model were re-estimated
based on human partition coefficient data for other compounds, and information on the
octanol/water partition coefficients of acrylamide and the other compounds. The metabolic
parameters were then recalibrated to human acrylamide and glycidamide hemoglobin adduct data
40 DRAFT-DO NOT CITE OR QUOTE
-------
collected at reasonably low external exposure levels in a limited number of adult male subjects
given a clinical oral exposure (Fennell et al., 2005).
The model parameters values in EPA's recalibrated rat model (and how they were
estimated) are presented in Appendix E, Table E-2. The fits to the data used to calibrate the
model are also presented in Appendix E, Table E-3 and E-4 and are discussed in greater detail in
Appendix E. The model parameters values in EPA's human model (and how they were
estimated) are presented in Appendix E, Table E-5.
The specific applications and results of the recalibrated Kirman et al. PBTK model in the
derivation of toxicity values are discussed in detail in Section 5 of this Toxicological Review.
Young et al. (2007) PBTK/TD Model
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 acrylamide, 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
3-3). 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 are listed in Appendix F, Table F-2. They
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 2.8mM 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 hr) (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
41 DRAFT-DO NOT CITE OR QUOTE
-------
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. (1992, 2003) and simulated
with the model. The Raymer et al. (1993) data were also combined with the urinary elimination
data of Sumner et al. (1992, 2003) 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 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 (see Appendix F, Table F-l for the results of fitting to the rat data).
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 six fold 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. model (2007) to be applicable for use in the
development of toxicity values for acrylamide, 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.
42 DRAFT-DO NOT CITE OR QUOTE
-------
Acrylamide
Glycidamide
1 F
O
cq
5
!»m,;
* Lung '
A
! — ,
s
— l>; Tissues — r *"
o
CD
i ' §
! ! i S
^ Liver ! n >
1 F r
,,->' v > !
i I
Lung
T
'11' » — IN Tissues
1 I
~ !
1 I
•MT
^^ 1— ~-~y I i Vi^r
.... ' .-*
' 1
— A
1
£
O
5
L^
^""'
* 7' Y
Figure 3-2. Schematic of the Kirman et al. PBTK Model for Acrylamide
(Source: Kirman et al. [2003])
1r pi it
til SI It
I'W'K-I
UTN Uwi;
lAA)
. >ro
. loc n
AJduuis
lun nvsr
PBPK-3
Acryluntide mciaholhc
(AA CIS)
v
Ui'iiic
DNA
(jhxitlarniiie
\
iirine
PBPK-4
nietahvi«1ilcsj
(CiA-(iS) s
Figure 3-3. Schematic of the Young et al. PBTK Model for Acrylamide
(Source: Young et al. [2007])
43
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, 2006b), 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, 2006b; FAO/WHO, 2005; JIFSAN, 2005) 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 acrylamide 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. Acrylamide, 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 acrylamide 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 acrylamide 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
[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. Subjects
44 DRAFT-DO NOT CITE OR QUOTE
-------
were administered a 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 human cancer study data are limited, although the recent discovery of AA in foods
has prompted a number of studies to evaluate a potential association between dietary acrylamide
intake and cancer. To date, no association has been established between increased levels of
acrylamide in the diet and increased risk for a variety of cancer types. Most of the available
epidemiology studies on increased risk of cancer from AA in food have been conducted by
Mucci and colleagues, including three case-control studies for increased risk of cancers of the
large bowel, bladder, kidneys, renal cell, or breast (Mucci et al., 2005, 2004, 2003) and one
prospective study for colorectal cancers (Mucci et al., 2006). In another large case-control
study, Pelucchi et al. (2006) evaluated the relation between dietary AA intake and cancers of the
45 DRAFT-DO NOT CITE OR QUOTE
-------
oral cavity and pharynx, esophagus, large bowel, rectum, larynx, breast, ovary, and prostate.
None of these studies report a significant increased incidence of cancer associated with increased
intake of AA in food at the levels of intake observed.
One study (of lesser utility) examined the incidence of breast cancer in later life relative
to diet composition during the 3- to 5-year-old early development stage (Michels et al., 2006).
The authors report an increased risk related to early consumption of French fries, a food
considered to have relatively high levels of AA. This limited study, however, makes no mention
of AA, and the study methods are lacking.
Two cohort mortality studies evaluated increased risk for cancer in AA workers (Marsh
et al., 1999; Collins et al., 1989; Sobel et al., 1986,).
No human studies were identified that assessed the potential for adverse reproductive or
developmental effects from exposure to acrylamide.
An important factor in evaluating epidemiology studies that relate dietary intake to
effects concerns the characterization of the variability in acrylamide 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
46 DRAFT-DO NOT CITE OR QUOTE
-------
Collins et al. (1989) conducted a cohort mortality study of all male workers (8854, of
which 2293 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 [7153 workers]; and Kalamazoo, MI [60 workers]) and one in the
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. Acrylamide 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, central nervous
system, 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, central nervous
system, or lymphopoietic system) with increasing level of exposure to AA.
In the latest update report for the three facilities in the United States, Marsh et al. (1999)
reported that, for the 1925-1994 study period (during which 3282 deaths occurred among the
8508 people included in the cohort), excess and deficit overall mortality risks were found for
several tissue sites of interest; however, none of these were statistically significant or associated
with exposure to AA. Table 4-1 lists all of the observed deaths and SMRs for selected causes.
Cited SMRs of interest included: brain and other central nervous system (SMR 0.65, 95% CI
0.36-1.09); thyroid gland (SMR 2.11, 95% CI 0.44-6.17); testis and other male genital organs
(SMR 0.28, 95% CI 0.01-1.59); and cancer of the respiratory system (SMR 1.10, 95% CI 0.99-
1.22). Table 4-2 lists the SMRs from observed deaths for selected cancer sites (esophagus,
rectum, pancreas, and kidney) for all U.S. workers who died between 1950 and 1994, according
to the following exposure parameters and categories: duration of employment (categories of
<1 year, 1-14 years, and >15 years), time since first employment ( <20, 20-29, and >30 years),
duration of exposure (unexposed, 0.001-4.999, 5-19, and >20 years), cumulative exposure
(O.001, 0.001-0.029, 0.03-0.29, and >0.30 mg/m3-years), and estimated mean exposure
47 DRAFT-DO NOT CITE OR QUOTE
-------
concentrations (unexposed, 0.001-0.019, 0.02-0.29, and >0.3 mg/m3). In these exploratory
exposure-response analyses of esophageal, rectal, pancreatic, and kidney cancers a statistically
significantly elevated SMR was found for pancreatic cancer in workers with the highest
cumulative AA exposure category (>0.30 mg/m3-years), but the risks for the four exposure
categories did not increase monotonically from the lowest to highest category.
SMRs for the lowest to highest cumulative exposure categories (mg/m3-years, SMR,
followed by 95% CI, and number of pancreatic cancer deaths) were as follows: <0.001, 0.80
(0.54-1.14, 30 deaths); 0.001-0.029, 2.77 (0.57-8.09, 3 deaths); 0.03-0.29, 0.73, (0.09-2.64, 2
deaths); and >0.30, 2.26 (1.03-4.29, 9 deaths). In contrast, risk of pancreatic cancer showed a
monotonic increase with exposure-duration category, but none of the SMRs were statistically
significantly elevated in any of the four exposure categories.
SMRs for the lowest to highest duration categories (years, SMR, followed by 95% CI,
and number of pancreatic cancer deaths) were as follows: unexposed, 0.80 (0.54-1.14,
30 deaths); 0.001-0.4999, 1.46 (0.47-3.41, 5 deaths); 0.5-19, 1.79 (0.58-4.17, 5 deaths); and
>20, 2.42 (0.66-6.19, 4 deaths). A relative risk modeling analysis showed patterns of relative
risks that were similar to those observed in the exploratory exposure-response analysis of SMRs.
Marsh et al. (1999) concluded that their study provides "little evidence for a causal relation
between exposure to AA and mortality from any cancer sites." 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 37% of the cohort had died through 1994). 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, important information will be gathered.
48 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 US 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 central nervous system (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 (390-458)
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
1925-1983
Obs SMR 95% CI
2167 0.91b 0.87-0.95
496 1.06 0.96-1.15
13 0.83 0.44-1.42
141 1.07 0.90-1.26
16 1.15 0.66-1.87
35 1.34 0.94-1.87
38 0.94 0.67-1.29
16 1.20 0.69-1.95
5 0.51 0.16-1.20
27 1.09 0.72-1.59
202 1.25b 1.08-1.44
8 1.10 0.48-2.18
194 1.27b 1.10-1.46
2 0.88 0.11-3.18
4 0.48 0.13-1.23
29 0.96 0.64-1.38
0 - 0.00-1.23
13 1.06 0.56-1.81
12 1.06 0.55-1.86
5 0.36C 0.12-0.85
2 2.32 0.28-8.37
39 0.88 0.62-1.20
6 0.70 0.26-1.53
8 1.39 0.60-2.74
14 0.78 0.43-1.31
11 0.92 0.46-1.66
8 1.24 0.54-2.44
26 0.77 0.50-1.12
1019 0.90b 0.85-0.96
105 0.75b 0.62-0.91
68 1.08 0.84-1.37
199 0.70b 0.61-0.81
101
8508
228,816
1984-1994
Obs SMR 95% CI
1115 0.76b 0.72-0.81
357 0.89C 0.80-0.99
8 0.96 0.41-1.89
85 0.89 0.71-1.10
15 1.30 0.73-2.14
12 0.95 0.49-1.66
28 0.78 0.52-1.13
8 1.26 0.55-2.49
5 0.58 0.19-1.35
17 0.91 0.53-1.46
139 0.94 0.79-1.11
6 1.25 0.46-2.71
133 0.94 0.78-1.11
0 - 0.00-6.19
6 0.89 0.33-1.93
38 0.82 0.58-1.13
1 1.92 0.05-10.70
14 1.38 0.75-2.31
10 1.11 0.53-2.04
9 1.15 0.53-2.19
1 1.80 0.04-10.01
21 0.60C 0.37-0.92
0 - 0.00-2.35
0 - 0.00-4.04
9 0.68 0.31-1.29
11 0.65 0.32-1.16
2 0.60 0.07-2.15
15 0.53b 0.30-0.87
434 0.61b 0.56-0.67
74 0.53b 0.42-0.67
12 0.54C 0.28-0.94
43 0.65b 0.47-0.87
70
5942
58,916
1925-1994
Obs SMR 95% CI
3282 0.85b 0.82-0.88
853 0.98 0.92-1.05
21 0.88 0.54-1.34
226 0.99 0.87-1.13
31 1.22 0.83-1.73
47 1.22 0.89-1.62
66 0.87 0.67-1.10
24 1.22 0.78-1.82
10 0.55 0.26-1.00
44 1.01 0.74-1.36
341 1.10 0.99-1.22
14 1.16 0.63-1.95
327 1.11 0.99-1.24
2 0.70 0.08-2.52
10 0.66 0.31-1.22
67 0.88 0.68-1.11
1 0.28 0.01-1.59
27 1.20 0.79-1.75
22 1.08 0.68-1.64
14 0.65 0.36-1.09
3 2.11 0.44-6.17
60 0.76C 0.58-0.97
6 0.59 0.22-1.29
8 1.20 0.52-2.37
23 0.74 0.47-1.10
22 0.76 0.48-1.16
10 1.02 0.49-1.87
41 0.66b 0.47-0.89
1453 0.79b 0.75-0.83
179 0.64b 0.55-0.74
80 0.94 0.74-1.17
242 0.69b 0.61-0.79
171
8508
287,731
49
DRAFT-DO NOT CITE OR QUOTE
-------
aMonson life table program ICD-8 categories, labels and codes for U.S. plants for 1925-89; corresponding rates for 1990-1994 from the mortality and
population data system (MPDS)maintained at the University of Pittsburgh.
V<0.05.
Source: Marsh etal. (1999).
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)
Duration of employment (years)
<1
1-14
>15
Time since first employment (years)
<20
20-29
>30
Duration of exposure (years)
Unexposed
0.001^.999
5-19
>20
Cumulative exposure (mg/m3 -years)
0.001
0.001-0.029
0.03-0.29
>0.30
Mean intensity of exposure (mg/m3)
Unexposed
0.001-0.019
0.02-0.29
>0.30
Esophagus
Obs SMR 95% CI
12 0.84 0.43-1.46
9 0.95 0.43-1.80
10 1.60 0.77-2.94
3 0.69 0.14-2.01
6 0.80 0.29-1.73
22 1.21 0.76-1.83
24 0.96 0.61-1.42
4 1.63 0.45^.18
3 1.80 0.37-5.26
0 - 0.00-4.10
24 0.96 0.61-1.42
2 2.58 0.31-9.30
3 1.70 0.35-4.97
2 0.82 0.10-2.98
24 0.96 0.61-1.42
2 1.37 0.17^.95
3 1.53 0.32-4.47
2 1.26 0.15-4.53
Rectum
Obs SMR 95% CI
5 0.51 0.16-1.18
12 1.37 0.71-2.39
5 0.86 0.28-2.02
3 0.71 0.15-2.07
5 0.85 0.28-1.98
14 0.98 0.53-1.64
17 0.82 0.48-1.32
3 1.92 0.40-5.61
1 0.71 0.02-3.96
1 1.20 0.03-6.70
17 0.82 0.48-1.32
1 2.31 0.06-12.9
2 1.73 0.21-6.23
2 0.92 0.11-3.31
17 0.82 0.48-1.32
2 2.12 0.26-7.65
0 - 0.00-2.03
3 2.89 0.60-8.43
Pancreas
Obs SMR 95% CI
17 0.87 0.51-1.39
15 0.95 0.53-1.57
12 1.19 0.61-2.08
4 0.66 0.18-1.68
11 1.08 0.54-1.92
29 1.00 0.67-1.44
30 0.80 0.54-1.14
5 1.46 0.47-3.41
5 1.79 0.58-4.17
4 2.42 0.66-6.19
30 0.80 0.54-1.14
3 2.77 0.57-8.09
2 0.73 0.09-2.64
9 2.26a 1.03-4.29
30 0.80 0.54-1.14
4 1.69 0.46-4.32
5 1.50 0.49-3.49
5 2.31 0.75-5.40
Kidney
Obs SMR 95% CI
8 0.79 0.34-1.55
7 0.86 0.35-1.78
7 1.36 0.55-2.79
2 0.58 0.01-2.09
3 0.54 0.11-1.58
17 1.18 0.69-1.89
16 0.84 0.48-1.37
2 0.99 0.12-3.59
3 1.88 0.39-5.48
1 1.18 0.03-6.56
16 0.84 0.48-1.37
1 1.45 0.04-8.08
2 1.17 0.14-4.23
3 1.49 0.31-4.35
16 0.84 0.48-1.37
1 0.66 0.02-3.66
3 1.71 0.35-5.01
2 1.68 0.20-6.08
ap < 0.05.
Source: Marsh etal. (1999).
50
DRAFT-DO NOT CITE OR QUOTE
-------
Sobel et al. (1986) conducted a mortality study on a cohort of 371 workers assigned to
acrylamide 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 central nervous system, 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 acrylamide
workers) and expand the cohort to include employees hired since 1979. A total of 696
acrylamide workers were followed from 1955 through 2001 to ascertain the long-term health
effects of occupational exposure to acrylamide among production and polymerization workers
and the cause of death. Exposure to acrylamide 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 acrylamide workers died (n = 141) compared to an expected number of
172.1 (SMR 81.9, 95% CI 69.0 to 96.6). No cause-specific SMR for any of the investigated
types of cancer was exposure related. Similar to the earlier Marsh et al. (1999) results, the
authors report more total pancreatic cancer deaths (n=5) than expected (n=2.3) (SMR 222.2,
95% CI 72.1 to 518.5), however, 3 of the 5 were in the low dose group, with no apparent dose-
response relationship with acrylamide exposure, and thus questionable support for an acrylamide
related carcinogenicity. Although these studies provide no good evidence of a cancer risk from
occupational exposure to acrylamide at production facilities, additional studies are needed to
further evaluate the potential carcinogenicity in humans from exposure to acrylamide.
51 DRAFT-DO NOT CITE OR QUOTE
-------
Case-control studies
No statistically significant associations were found between high consumption of foods
with high (300-1200 |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
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
52 DRAFT-DO NOT CITE OR QUOTE
-------
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; p = 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. Acrylamide intake was determined from food frequency
questionnaires 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 acrylamide intake among the
participants was estimated at 25.9 |ig/day, with less than 1.5% of the women consuming more
than 1 jig/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, 1772 controls), esophagus (395 cases, 1066 controls),
large bowel (1394 cases of colon cancer, 886 cases of rectal cancer, 4765 controls), larynx (527
cases, 1297 controls), breast (2900 cases, 3122 controls), ovary (1031 cases, 2411 controls), and
prostate (1294 cases, 1451 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 =
53 DRAFT-DO NOT CITE OR QUOTE
-------
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.
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 1569 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 1921-1965. 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 1433 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 1569 controls,
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%
54 DRAFT-DO NOT CITE OR QUOTE
-------
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 30-50 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 Pels
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 1929-1950, and the questionnaire
was administered in 1997. The mothers in this validation study ranged in age from 60-93 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
Pels 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.
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; p 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;/? = 0.77). Intake
of specific food items with elevated AA (e.g., coffee, crispbreads, and fried potato products) was
not associated with cancer risk.
55 DRAFT-DO NOT CITE OR QUOTE
-------
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
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 chi-square 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.
56 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
> < 0.05.
V < 0.01 (chi-square 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 of
57
DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide (and acrylonitrile, from which acrylamide
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
glycidamide adducts of hemoglobin in AA-exposed workers indicating that the transformation of
AA to GA occurs in humans, and by Deng et al. (1993). Acrylamide 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
58 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, I,2,or3b
0, 1,2, or3b
1
0, I,or2
0,1, or 2
4
4
4
3 or 5
6
0.5 per abnormality (maximum 6)
50
"Points 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 chi-square test to analyze symptoms and clinical signs
and the Student's t-test to assess ENMG parameters. Variance analysis and the Q-tesi 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 years of age, 31-40 years of age, and >40 years of age). Comparisons of
vibration threshold between AA-exposed workers and referents within these age groupings
59 DRAFT-DO NOT CITE OR QUOTE
-------
showed a significant increase in the exposed workers. Comparison of the results of ENMG
measurements between the exposed workers and the referent group revealed a 10 to 20%
decrease in conduction velocity in the peroneal and sural nerves and 25 to 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.
60 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
(jimol/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 acrylamide.
bUrinary mercapturic acid.
'Hemoglobin adduct between N-terminal valine and acrylamide.
dHemoglobin adduct between N-terminal valine and acrylonitrile.
Predicted cumulative in vivo acrylamide dose (based on rates of acrylamide-hemoglobin adduct formation in
human globin hydroly sates and mean acrylamide 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 acrylamide 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.
61
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
AAVaT
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
<0.001
0.001
O.001
Tree plasma acrylamide.
bUrinary mercapturic acid.
'Hemoglobin adduct between N-terminal valine and acrylamide.
dHemoglobin adduct between N-terminal valine and acrylonitrile.
Predicted cumulative in vivo acrylamide dose (based on rates of acrylamide-hemoglobin adduct formation in
human globin hydroly sates and mean acrylamide 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 acrylamide 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
62 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.
63 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/7-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.
64
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 to 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 and central nervous systems (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
65 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 acrylamide
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
66 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,
67 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
68 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.
69 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
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
Dose
0.2
3
350
27
17
0
(mg/kg-day)
1
o
5
453
30
78
0
44/350 108/453
0/10
0/10
0/10
0/10
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
o
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
70 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.
71 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
72 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.
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
73 DRAFT-DO NOT CITE OR QUOTE
-------
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.
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
5
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
5
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
"The 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).
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 carcinogenicity 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), central nervous system (CNS) tumors of glial origin, thyroid
74
DRAFT-DO NOT CITE OR QUOTE
-------
(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 male and female F344 rats
exposed to acrylamide in drinking water for 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
ll/60a
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
10/60a
6/60
10/60a
6/61
23/6 la
9/6 la
5/60a
1/61
7/6 la
5/60a
5/5a
32/60a
aSignificantly different from control, p < 0.05, after Mantel-Haenszel mortality adjustment.
Source: Johnson etal. (1986).
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 acrylamide 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
75 DRAFT-DO NOT CITE OR QUOTE
-------
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.
An additional group of 25 rats/sex was observed during the course of this study for signs
of viral infections (to address concerns about sialodacryoadenitis virus infection in the first
bioassay). Control rats were divided into two separate groups to more accurately assess the
variability of low-incidence background tumors.
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
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.
76 DRAFT-DO NOT CITE OR QUOTE
-------
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
those of control groups starting at week 8 and were significantly decreased from week 40
(398 grams vs. 408 grams in controls, approximately 2.5% lower) to study end (375 grams vs.
412 grams 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
77 DRAFT-DO NOT CITE OR QUOTE
-------
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.
At the level of behavioral and clinical observation performed in this bioassay protocol, no
clinical signs of neurotoxicity were reported in any treated rats. 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
aNumber 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
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.
78
DRAFT-DO NOT CITE OR QUOTE
-------
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
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).
79
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
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
"Does 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.
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
80 DRAFT-DO NOT CITE OR QUOTE
-------
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 Leydig 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
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 Leydig cell neoplasia. In contrast, rats
reevaluated as having focal mesothelial hyperplasia or benign mesothelioma showed either no
Leydig cell neoplasia or 25 or 50% of the testes occupied by Leydig cell neoplasia. The
comparison suggests that the extent of Leydig cell neoplasia and the development of malignant
mesotheliomas may have been linked.
81 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).
Yeydig 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
82 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 acrylamide 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 acrylamide 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
83 DRAFT-DO NOT CITE OR QUOTE
-------
the 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 anti-fertilization effect may be due to nongenetic actions, the doses
needed to elicit the anti-fertilization effects were generally higher than that needed to elicit the
postimplantation genetic effects, and thus the anti-fertilization 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., 1987; 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 etal., 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.
84 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
85 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 3 high-dose females and 2
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.
Acrylamide 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.
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.
86 DRAFT-DO NOT CITE OR QUOTE
-------
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 (mg/kg-day)
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
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
"Significantly (p < 0.05) different from control value. Values are group means ± SD.
Source: Tyl et al. (2000a).
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.
In summary, the two-generation reproductive toxicity/dominant lethal mutation study
with F344 rats exposed to AA in drinking water (Tyl et al., 2000a) identified 5.0 mg/kg-day as
the LOAEL and 2.0 mg/kg-day as the NOAEL for effects on reproduction in the FO and Fl
generations (decreased implantations/dam and decreased number of live pups/litter). The same
NOAEL and LOAEL were identified for dominant lethal mutation effects (decreased live
implants/litter and increased pre- and postimplantation loss) when exposed males were bred with
nonexposed females. Decreased body weights (4-6% compared with control values) were
observed most consistently in Fl males at doses >2.0 mg/kg-day (not at 0.5 mg/kg-day), and
signs of neurotoxicity (increased incidences of foot splay) were observed in FO males in the low-
and high-dose groups (0.5 and 5.0 mg/kg-day) and in FO females in the 2.0 and 5.0 mg/kg-day
groups.
87
DRAFT-DO NOT CITE OR QUOTE
-------
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 ± 2.6s
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.
ep < 0.05.
f/?<0.01.
sp< 0.001.
Source: Tyl et al. (2000a).
Chopin 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
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 1 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.
88
DRAFT-DO NOT CITE OR QUOTE
-------
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 1 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.
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.
Acrylamide 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.
89 DRAFT-DO NOT CITE OR QUOTE
-------
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.
Acrylamide 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 grams vs. 37.7 ± 0.9 grams 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.
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
"Mean ± 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
90
DRAFT-DO NOT CITE OR QUOTE
-------
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. Acrylamide 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 ±11.1 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
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. Acrylamide 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
91 DRAFT-DO NOT CITE OR QUOTE
-------
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
Fore limb
Hindlimb
Females
Fore limb
Hindlimb
Fl absolute grip strength (grams)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).
Significantly 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 acrylamide
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
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
92 DRAFT-DO NOT CITE OR QUOTE
-------
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 chi-square statistics. Effects on pre- and postimplantation loss were analyzed
using Kruskal-Wallis ANOVA with Mann-Whitney U-test for post hoc 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).
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
"(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. (1987) observed reversible male-mediated reproductive effects (dominant lethal
93
DRAFT-DO NOT CITE OR QUOTE
-------
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 oral 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.
Glycidamide as the putative toxin for dominant lethal effects
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 glycidamide. 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
94 DRAFT-DO NOT CITE OR QUOTE
-------
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 B6C3F1 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
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
95 DRAFT-DO NOT CITE OR QUOTE
-------
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
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
3 Calculated from graphically presented data on body weight and water consumption.
96 DRAFT-DO NOT CITE OR QUOTE
-------
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.
Table 4-21. Results of sperm analysis
fertility testing (following 10 weeks of
to acrylamide in the drinking water
(baseline and week 9) and male
treatment) of Long-Evans rats exposed
Parameter
Sperm count (^ 106)
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
d
95 ±1.8
d
111±20
146 ± 49
15/15
5/15 (33%)f
31.7 ± 3. 8f
Tour 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
{p
-------
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
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
98 DRAFT-DO NOT CITE OR QUOTE
-------
effects (decreased pup body weight), this study supports a LOAEL of 50 ppm (8.8 mg/kg-day)
and aNOAEL 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
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.
The study design of the assessment of reproductive effects in treated females was similar
to that of the treated males, but treatment was limited to a single exposure level (24 females
administered AA at a level of 1.2 mM in the drinking water, resulting in an AA dose of
approximately 18.7 mg/kg-day that was based on body weight and water consumption data
provided by the authors). The only effect noted in this part of the study was a slight but
statistically significant increase in the number of resorptions/dam (1.9 ± 1.5, n = 24) as compared
with controls (0.2 ± 0.3, n = 18) at day 13 of gestation.
99 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
o o
J.J
9.0
13.3
16.3
Fertility rate"
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
(grams)
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
o o
J.J
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
aNumber 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 acrylamide (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 oftesticular 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 and 65% of the
100
DRAFT-DO NOT CITE OR QUOTE
-------
60-day-old mice. In the prepubertal mice, body weight was significantly decreased at 1 and
5 days after dosing with 150 mg/kg acrylamide. 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 acrylamide. 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 acrylamide in either age group. Histologic
abnormalities in the testes of prepubertal animals treated with 150 mg/kg acrylamide 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. (1987) 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 oral 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 60
101 DRAFT-DO NOT CITE OR QUOTE
-------
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 (e.g., decreased auditory startle response) in the offspring of
female Sprague-Dawley rats exposed to 5 and 15 mg/kg-day, respectively, on GDs 6-10 (Wise
et al., 1995). 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 studies are summarized in Table 4-31, and discussed below, except
for the Zenick et al. (1986) study, which has been discussed previously in Section 4.3.1. It is
102 DRAFT-DO NOT CITE OR QUOTE
-------
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 oral 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). Acrylamide 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.
103 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.
0Significantly 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 wasp < 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
104 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.
105 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/?
< 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
106 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
oral 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, chi-square test, and a test
for statistical outliers. Differences in results between treated and control groups were considered
significant at the level ofp < 0.05.
107 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 grams, whereas a net mean weight gain of 47 grams 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.
Other developmental toxicity studies
Genotoxic effects observed in the germ cells of mice following i.p. injection of
125 mg/kg acrylamide 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
108 DRAFT-DO NOT CITE OR QUOTE
-------
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
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 grams/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 grams/rat/day feed
containing 400 ppm AA represents a daily dose of 16 mg/kg-day). Fetal weights were reduced
by AA treatment (acrylamide 2.4 ± 0.05 grams, control 3.2 ± 0.05 grams, mean ± SEM). (The/>-
value reported by the authors using the Student t-test was >0.2; however, the t-test performed by
CERHR gave/? < 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.
109 DRAFT-DO NOT CITE OR QUOTE
-------
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 from GDs 0 to 19. Acrylamide intake
was estimated 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
110 DRAFT-DO NOT CITE OR QUOTE
-------
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.
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
-
-
-
I
t
t
t
60
4
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 (GAB A) receptors developmentally. In
111
DRAFT-DO NOT CITE OR QUOTE
-------
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
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
Qualitative characterization
Two recent reviews of studies in mice for heritable germ cell effects from exposure to
acrylamide (Favor and Shelby, 2005; NTP/CERHR, 2004) have both concluded that AA induces
transmissible genetic damage in male germ cells of mice in the form of reciprocal translocations
and gene mutations. The studies consisted of five heritable translocation studies (Adler et al.,
2004, 1994, 1990; Generoso et al., 1996; Shelby et al., 1987) and two specific mouse locus
assays (Ehling and Neuhauser-Klaus, 1992; Russell et al., 1991). 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.
Favor and Shelby (2005) summarized their conclusions as follows: (la) 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. They further note that,
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. Further research is needed to
resolve the conflicting results between two specific-locus mutation experiments with respect to
the stem cell spermatogonial effects.
The NTP/CERHR panel also noted that AA-induced transmissible genetic damage can
lead to genetic disorders and infertility in subsequent generations, but these risks were not
included in the expert panel's quantitative evaluation of LOAELs for risk to the general
population because of the lack of testing at dose levels below where reproductive and
112 DRAFT-DO NOT CITE OR QUOTE
-------
developmental toxic effects were observed. The panel did hypothesize that, considering the
magnitude of the response detected for heritable translocations at the lowest dose level tested
(40 mg/kg-day x 5 days or 50 mg/kg as a single dose), it is likely that such effects would occur
at doses lower than these.
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.
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 17/8095 in the historical control. All ten of the semi-sterile
males sampled from the 5 x 50 treatment for cytogenetic analysis of spermatocytes had
translocations.
113 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"
—
—
7-16
7-16
3.5-7.5
7-10
7-10
7-11
36-42
1.5-8.5
F! progeny tested
Males
ll,292d
9,890e'f
362 f
367f
669
162
125
57
556
258
Females
48
449
217
Translocation carriers0
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
Generosoetal. (1996)
Adler et al. (2002)
Adleretal. (1994)
Adler etal. (1994)
Generosoetal. (1996)
Shelby etal. (1987)
Shelby etal. (1987)
Adler (1990)
Adler (1990)
Adler et al. (2004)
a5 x 40 and 5 x50 represent 40 or 50 mg acrylamide/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. (1994, 2004).
fBoth male and female Fj 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
1506
5077
5191
5312
5353
6419
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.
Sources: Data from Russell et al. (1990); table from Favor and Shelby (2005).
114
DRAFT-DO NOT CITE OR QUOTE
-------
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
1362
2226
2421
2453
2574
2925
23,489
771
1924
1948
2419
2598
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).
115
DRAFT-DO NOT CITE OR QUOTE
-------
Adler et al. (1990) administered acrylamide 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), 1005 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 acrylamide 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, 5/9890.
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 C3H/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 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.
(1990, 1994, 2004) 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
116 DRAFT-DO NOT CITE OR QUOTE
-------
plus an inversion on a third chromosome, and 1 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 glycidamide 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). Acrylamide 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 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 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
117 DRAFT-DO NOT CITE OR QUOTE
-------
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
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.
The two specific locus mutation studies provide evidence for specific locus mutations in
rodent spermatid and spermatozoal stage germ cells. Resolution of the conflicting results
between these two studies with respect to induced mutations in stem cell spermatogonia will
require further research. This research is needed because mutations in stem cell spermatogonia
118 DRAFT-DO NOT CITE OR QUOTE
-------
represent a more persistent and serious lesion that will be transmitted to all descendent
spermatids, in contrast with the later stage mutations that can be inherited only from fertilization
by that directly affected spermatid. There are also implications for determining risk from
exposure and the dose duration for a reference value since, 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).
Acrylamide 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
examination. All other signs of neurotoxicity (impaired mobility and righting reflex, decreased
119 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide 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). Acrylamide was dissolved in distilled water for oral and
injection routes and in ethanol for dermal applications. Acrylamide 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-
29.
Table 4-29. 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.
120 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 4-30.
Table 4-30. Acrylamide initiation of skin tumor masses > 1mm 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: Bull etal. (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 a 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 a 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-
121 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.
Acrylamide 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 a 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. Acrylamide 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 three 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
122 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.
4.6. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION
4.6.1. Neurotoxicity Studies
Several groups (Padilla et al., 1993; Harry, 1992; Sickles, 1991) have put forth the theory
that AA slows fast axonal transport within axons and that this produces the subsequent effects on
the nerve terminal and the axon. Sickles (1991) examined the rate and quantity of protein
transport in rat sciatic nerve following a single injection of 50 mg/kg AA and found an
immediate reduction of 48% in the quantity of protein transported down the rat sciatic nerve.
Transport remained depressed for 16 hours, and recovery occurred from 16 to 24 hours, reaching
control levels at 24 hour postinjection. Harry (1992) dosed adult male rats with 40 mg/kg i.p. for
nine consecutive days and found a 75% decrease in the amount of accumulated material by 24
hours. Padilla et al. (1993) treated male Long-Evans rats with a dose of 50 mg/kg AA i.p. twice
a day for 5 weeks and observed that fast axonal transport in the sciatic nerve of these animals
was decreased by 6.5-9.4%.
Other researchers have developed a hypothesis that nerve terminals are the primary site
of AA-induced neurotoxicity and that axonal degeneration is secondary to nerve terminal
changes (Lehning et al., 2003a, 2002, 1998; Lopachin et al., 2002a). Groups of adult male
Sprague-Dawley rats were exposed to either i.p. injection of 50 mg/kg-day of AA for up to 11
days or via drinking water at 28 mM AA (estimated oral dose of 21 mg/kg-day) for 38 to 49
days. Gait abnormalities were observed in both treatment groups but appeared earlier at the
higher dose rate. Using the de Olmos silver stain method (de Olmos et al., 1994), the authors
observed nerve terminal degeneration in both treatment groups and proposed that nerve terminal
degeneration precedes axonal degeneration, which occurred primarily at the subchronic, albeit,
lower dose rate. This same pattern of lesion was observed in the cerebellum, the brainstem, the
spinal cord, and the forebrain (Lehning et al., 2003a,b).
There remains uncertainty as to which is the most sensitive lesion and what is the
sequence of events in the mode of action leading to clinical signs of neurotoxicity. Sickles et al.
123 DRAFT-DO NOT CITE OR QUOTE
-------
(2002a) suggest that, because the most distal sites (i.e., nerve terminals) are likely to be the most
vulnerable to alterations in fast axonal transport, they should not be considered as a separate site.
The primary target currently under investigation for reduction in fast axon transport is the motor
protein, kinesin (Sickles et al., 2002b). The primary target currently under investigation for
nerve terminal damage is synaptosomal proteins involved in the presynaptic transmission
(LoPachin et al., 2004). Both kinds of proteins contain sulfhydryl groups as sites for AA
binding, and in both modes of action there is additional research needed to show a direct
relationship between the protein functional inhibition and the observed functional deficits
(LoPachin and Canady, 2004; Sickles et al., 2002a).
4.6.2. Genotoxicity Studies
Appendix B (Table B-l) summarizes results of numerous published mutagenicity tests
for acrylamide 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; 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, 1990; Generoso et al., 1996; 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 4.3.3.
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-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
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).
124 DRAFT-DO NOT CITE OR QUOTE
-------
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 10"6 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 KT6 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 acrylamide (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 micrenucleated 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
negative (Krishna and Theiss, 1995; Shiraishi, 1978) results. 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
125 DRAFT-DO NOT CITE OR QUOTE
-------
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.
Acrylamide-induced increases in micronuclei were seen in bone marrow cells,
reticulocytes, spleen lymphocytes, and splenocytes of mice and spermatids of rats and mice
(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).
Acrylamide 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 to 1 mg/mL (Adler et al.,
1993; Tsuda et al., 1993; Warr et al., 1990; Knaap et al., 1988; Moore et al., 1987). 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).
Earlier evidence for AA-induced DNA damage and repair included positive results in a
spore rec assay (Tsuda et al., 1993), DNA breakage in mice following i.p. injection of AA doses
>25 mg/kg (Sega and Generoso, 1990), 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).
Acrylamide tested positive for sister chromatid exchange in mammalian cells both
in vitro (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 (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).
126 DRAFT-DO NOT CITE OR QUOTE
-------
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 (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 in Chinese hamster V79H3 cells at the highest concentration
tested (7 mM) without activation (Tsuda et al., 1993).
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 notK. pneumoniae (Voogd et al., 1981).
Glycidamide 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). Glycidamide (125 mg/kg by i.p. injection) induced dominant lethal
mutations in male JH mice mated with nonexposed female SB mice (Generoso et al., 1996).
Glycidamide 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., 2003). 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 of DNA adducts
Glycidamide forms DNA adducts in mice and rats (see Figure 3-2) (Doerge et al., 2005a;
Gamboa 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
acrylamide (Segerback et al., 1995; Sega et al., 1990).
127 DRAFT-DO NOT CITE OR QUOTE
-------
Doerge et al. (2005a) measured DNA adducts following a single i.p. administration of
AA and GA to adult B6C3F1 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.
Acrylamide 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. Acrylamide and glycidamide 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. Acrylamide-DNA adduct formation was
saturable, whereas the formation of most GA-DNA adducts was dose-dependent. Glycidamide
treatment dose-dependently increased the frequency of ell mutations relative to control treatment
(P<.001). Glycidamide 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=.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<.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.
128 DRAFT-DO NOT CITE OR QUOTE
-------
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 grams 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 4-31). The studies in Table 4-31 provided the
information needed to characterize the dose-response relationships for noncancer effects.
Examination of NOAELs and LOAELs for the various effects noted in Table 4-31
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 4-31). Although studies selected for
inclusion in Table 4-31 only examined rats and mice, Table 4-32 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.
129 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-31. Noncancer effects in animals repeatedly exposed to acrylamide
by the oral route
Reference/species
Bureketal., 1980
F344 rat, M&F
Johnsonet al., 1986
F344 rat, M&F
Friedman et al.,
1995
F344 rat, M&F
Tyletal, 2000a
F344 rat, M&F
Chapinetal., 1995
CD-I mouse, M&F
Zenicketal., 1986
Long-Evans rat, M
Zenicketal., 1986
Long-Evans rat, F
Exposure
conditions
(mg/kg-day)
0,0.05,0.2, 1,5,
or 20
90 days in DW
0,0.01,0.1,0.5,
or 2.0
2 years in DW
0,0.1,0.5, or 2.0
(M)
0, 1.0, or 3.0 (F)
2 years in DW
0,0.5, 2.0, or 5.0
2 generations in D W
0,0.8, 3.1, or 7.5
2 generations in D W
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
NOAEL
LOAEL
(mg/kg-day)
0.2
1
5
5
5
0.5
2
0.5
0.5
2
0.5(M)
l.O(F)
2.0(M)
3.0(F)
0.5(M)
l.O(F)
0.5
2.0(M)
3.0(F)
2.0
ND
ND
5.0(F)
0.5(M)
3.1
7.5
3.1
7.5
3.1(F)
ND
4.6
5.1
5.1
14.6
1
5
20
20
20
2
ND
2
2
ND
2.0(M)
3.0(F)
ND
ND
2.0(M)
3.0(F)
2.0
ND
ND
5.0
5.0(M)
0.5(M)
ND(F)
2.0(M)
7.5
ND
7.5
ND
7.5(F)
7.9
7.9
8.8
8.8
ND
Effect
Degenerative nerve changes (EM)
Degenerative nerve changes (LM)
Hindlimb foot splay
Decreased body weight (8-20%)
Atrophy of testes & skeletal muscle
Degenerative nerve changes (LM)
Hindlimb foot splay
Decreased body weight (<5%, M only)
Early mortality after 24 weeks
Other nonneoplastic lesions
Degenerative nerve changes (LM)
Hindlimb foot splay
Decreased body weight (8-9%)
Early mortality after 60 weeks
Other nonneoplastic lesions
MM implantation losses (FO&F1)
Degenerative nerve changes (LM)
Hindlimb foot splay (FO M only)
Body weight effects
Decreased body weight (4-6%)
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)
Smith etal., 1986 0, 1.5, 2.8, or 5.8 1.5 2.8 MM postimplantation losses
Long-Evans rat, M 80 days in DW; 5.8 ND Peripheral nerve changes (LM)
mated w/ 5.8 ND Hindlimb foot splay
nonexposed F
Sakamoto and 0,3.3,9.0,13.3, 9.0 13.3 MM decreased fetuses/dam
Hashimoto, 1986 or 16.3 13.3 16.3 Slight hindlimb weakness
ddY mouse, M 4 weeks in DW; 13.3 16.3 Decreased sperm counts, abnormal sperm
mated w/ morphology
nonexposed F
Sakamoto and 0, 18.7 18.7 ND Female reproductive performance
Hashimoto, 1986 4 weeks in DW;
ddY mouse, F mated w/ ND 18.7 Slight hindlimb weakness
nonexposed M
130
DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-31. Noncancer effects in animals repeatedly exposed to acrylamide
by the oral route
Reference/species
Field etal, 1990
Sprague-Dawley
rat, F
Field etal., 1990
CD-I mouse, F
Wise etal., 1995
Sprague-Dawley
rat, F
Friedman et al.,
1999a
Wistar rat, F
Exposure
conditions
(mg/kg-day)
0, 2.5, 7.5, or 15
GD 6-20 by gavage
0,3, 15, or 45
GD 6-17 by gavage
0,5, 10, 15, or 20
GD 6-10 by gavage
0, 25 (maternal
doses) PND 0-21
by gavage
NOAEL
LOAEL
(mg/kg-day)
7.5
15
15
15
45
15
10
10
ND
10
ND
25
25
15
ND
ND
45
ND
45
15
15
5
15
25
ND
ND
Effect
Decreased maternal weight gain
Fetal malformations or variations
Hindlimb splay, maternal
Decreased maternal weight gain
Fetal malformations or variations
Hindlimb splay, maternal
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
DW = drinking water LM = light microscopy ND = not determined
EM = electron microscopy LOAEL = lowest-observed-adverse-effect level
F = female M = male NOAEL = no-adverse-effect level
GD = gestation days MM = male-mediated PND = postnatal days
131
DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-32. 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)
Eskin et al., 1985
Monkeys (macaque)
Maurissen et al., 1983
Monkeys (pigtail)
McCollisteretal., 1964
Monkeys (5.1 kg)
Gilbert and Maurissen, 1982
Mice (Balb/c)
Hashimoto et al., 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)
acrylamide in drinking water
for 12 days (total estimated
dose 310 mg/kg)
1/5 to l/2oftheLD50(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.
24 1 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
4-31 were not oriented towards neurobehavioral endpoints and did not evaluate gait
abnormalities. Instead, hindlimb foot splay, a gross characteristic sign of AA-induced peripheral
neuropathy, was measured in several of the studies cited in Table 4-31. Changes in foot splay
have been observed in most studies at oral dose levels above the lowest less-than-lifetime and
chronic doses associated with histologic signs of peripheral nerve damage (1-3 mg/kg-day), with
the exception of one study that 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 4-31). This observation does not appear consistent with
other observations, including the absence of hindlimb foot splay in Fl generation rats in the
same study exposed to doses as high as 5 mg/kg-day (Tyl et al., 2000a) and 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). Neurobehavioral studies with protocols
132 DRAFT-DO NOT CITE OR QUOTE
-------
and endpoints that are suitable for quantifying the dose-response are a research need, and efforts
are ongoing to measure more sensitive neurobehavioral responses to AA.
Acrylamide 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 4-31). 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 4-31). 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, 1990;
Generoso et al., 1996; Ehling and Neuhauser-Klaus, 1992; Russell et al., 1991; 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 4-31). 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 4-31 suggests that the onset of
adverse effects for male reproductive toxicity results from lower levels of AA exposure (2.8-
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 neurobehavioral changes (e.g., 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) (Table 4-31). 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 4-31). 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
133 DRAFT-DO NOT CITE OR QUOTE
-------
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
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
Since experimental acrylamide neuropathy was first reported (Hazleton Laboratories,
1953), acrylamide has been extensively studied in efforts to understand its toxicological
134 DRAFT-DO NOT CITE OR QUOTE
-------
properties and mode of action (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 acrylamide (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).
After four decades of research, the cellular and molecular site and the mode of action for
AA-induced neurotoxicity remain unresolved, although compromise of fast axonal transport is
the central theme for several of the hypotheses (Sickles et al., 2002b). With respect to
neurofilaments and fast axon transport, a series of studies on normal and transgenic mice
(lacking neurofilaments) demonstrated that the presence or absence of neurofilaments did not
alter AA's reduction of fast axonal transport in central (optic) or peripheral (sciatic) nerves and
that the reduction mode of action is independent of toxicant-induced modifications or
accumulations of neurofilaments (Stone et al., 2001, 2000, 1999). Sickles et al. (2002b) further
advanced the hypothesis for reduction in fast transport as the MOA in a review of the literature
and the identification of the motor protein, kinesin, as a primary target. Kinesin has a reduced
affinity in vitro for microtubules following preincubation with varying concentrations of AA
(Sickles et al., 1996). Sickles et al. (2002b) proposed that kinesin and microtubules are
covalently modified by AA leading to a reduction in their mutual affinity and, subsequently, the
level of fast axonal transport.
Lehning et al. (1998) reported that relatively low dose oral subchronic (26-45 days)
exposure to AA (e.g., 20 mg/kg-day from drinking water) induced axonal degeneration (tibial
nerve) while shorter-term (11 day) i.p. doses at higher levels (50 mg/kg-day) did not, yet both
dosing regimens resulted in moderate levels of behavioral neurotoxicity. No apparent
differences were observed in a comparison of AA metabolism and toxicokinetics by the above
135 DRAFT-DO NOT CITE OR QUOTE
-------
dosing regimens that provided an explanation for the occurrence of degeneration with the
subchronic but not acute dosing regimen (Barber et al., 2001). These results support a
hypothesis that the MOA for axon degeneration is conditional on dose rate and an
epiphenomenon (i.e., a secondary phenomenon accompanying, and resulting from, the primary
MOA for functional neurotoxicity). LoPachin et al. (2002a) proposed that this primary MOA
involves the nerve terminal region and that the primary sites are synaptosomal proteins involved
in presynaptic transmission (LoPachin et al., 2004). LoPachin et al. (2004) hypothesized that
AA alters neurotransmitter release by interacting with sulfhydryl groups on synaptosomal
proteins.
Although there is controversy over which events are primary and which are
epiphenomena, both proposed MO As agree that binding of AA (or metabolites) to sulfhydryl
rich proteins is a key event. Both research teams also recognize the need for additional research
to determine relative binding rates and to demonstrate a direct relationship between protein
inhibition (kinesin or synaptosomal or other proteins) and observed functional deficits (LoPachin
and Canady, 2004; Sickles et al., 2002a).
The neurotoxic potential of the AA metabolite, glycidamide, also needs further
evaluation. Costa et al. (1992) administered either AA or GA to groups of male Sprague-Dawley
rats by i.p. injection at doses of 25 or 50 mg/kg-day (AA) or 50 or 100 mg/kg-day (GA) for 8
days and assessed hindlimb foot splay and rotarod performance. In the AA-treated rats,
performance was significantly reduced on both rotarod and foot splay tests, relative to controls.
Glycidamide-treated rats exhibited decreased rotarod performance but no evidence of hindlimb
foot splay. The lack of this effect suggests that AA-induced peripheral neuropathy may not
involve GA. Abou-Donia et al. (1993), however, report that GA is capable of inducing
peripheral neuropathy in male Sprague-Dawley rats (six/group) given i.p. injections of 50
mg/kg-day AA or GA for up to 13 days. Acrylamide-exposed rats developed hindlimb weakness
and altered gait by day 5, with hindlimb paralysis occurring in all rats by day 13. Glycidamide-
exposed rats developed severe ataxia by day 11, and hindlimb paralysis developed in all treated
rats by day 13. The apparent inconsistency between the two reports concerning the ability of GA
to induce peripheral neuropathy remains unresolved. Some of the inter- and intraspecies
differences in dose-response may result from differences in AA toxicokinetics (as discussed
previously in the section on species differences in AA and GA metabolism). For example, it was
observed that mice had approximately a five-fold increase in internal exposure to GA compared
with rats due to an increased rate of GA formation in the mouse (Twaddle et al., 2004; Barber et
al., 2001; Miller et al., 1982).
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
136 DRAFT-DO NOT CITE OR QUOTE
-------
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; Tyl et al., 2000b; Sega et al., 1989; Adler et al., 2000).
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
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.
137 DRAFT-DO NOT CITE OR QUOTE
-------
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.
The available human studies on potential AA carcinogenicity are for either dietary
exposures—four case-control studies (Mucci et al., 2005, 2004, 2003; Pelucchi et al., 2006) and
one prospective study (Mucci et al., 2006) or occupational exposures from inhalation and/or
dermal exposure—three cohort mortality studies (Marsh et al., 1999; Collins et al., 1989; Sobel
et al., 1986). These studies are judged as providing limited data. 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 the most
comprehensive study, an increased risk for pancreatic cancer was reported in a subgroup with the
highest cumulative AA exposure. In the four recent case-control studies and one prospective
study, 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 cavity, pharynx, esophagus, larynx, ovary, or prostate cancer. These
studies evaluated Swedish or Italian populations, not U.S. populations or U.S. diets (i.e., no
studies on U.S. populations have been reported to date). 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.
There is one case-control study that evaluated whether diet during preschool age (3-5-
year-olds) affected a woman's risk of breast cancer later in life and reported a slightly increased
OR associated with consumption of French fries, but there is considerable uncertainty in the
138 DRAFT-DO NOT CITE OR QUOTE
-------
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 (Michels et al., 2006).
There are no animal data on the carcinogenicity of chronic inhalation exposure to AA or
human data exclusively for an inhalation exposure. 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.
The majority of the data support a mutagenic MOA for AA carcinogenicity. Acrylamide
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 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. 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 an MOA are limited or lacking.
4.8.2. Synthesis of Human, Animal, and Other Supporting Evidence
Cohort mortality studies of acrylamide workers at several locations in the United States
and the Netherlands (Marsh et al., 1999; Collins et al., 1989) and a location in Michigan (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. Four case-control
studies (Mucci et al., 2005, 2004, 2003; Pelucchi et al., 2006) and one prospective study (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 cavity, pharyngeal, esophageal, laryngeal, ovarian, or
prostate cancers. These studies evaluated Swedish or Italian populations, not U.S. populations or
U.S. diets (i.e., no studies on U.S. populations have been reported to date). 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 an
increased OR for female breast cancer in later life (OR = 1.27; 95% CI = 1.12-1.44) from
increased consumption of French fries in preschool diet, but the results are of questionable use in
this assessment due to uncertainties in the accuracy of the diet recall (from an estimated 40-60
years ago) and lack of information on AA content in the foods (Michels et al., 2006).
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,
139 DRAFT-DO NOT CITE OR QUOTE
-------
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.
Since the relationship between increased exposure and increased risk for pancreatic
cancer is tenuous, and has only been observed in one epidemiologic study, the available human
evidence supporting AA carcinogenicity is considered to be limited to inadequate. The
epidemiology study results, and the limitations of the studies to detect increased cancer mortality
risks, however, are limited to discount potential AA carcinogenicity in humans, as suggested by
the positive animal study carcinogenicity results. Limitations in the epidemiology studies
include small cohort size and limited follow-up period (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., 1999); 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). No studies on U.S. populations have been reported to-date.
The only study that showed a positive association had questionable data on diet composition and
AA content in the diet (Michels et al., 2006).
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 4-33.
140 DRAFT-DO NOT CITE OR QUOTE
-------
Table 4-33. Incidence of tumors with statistically significant increases in
both 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
Johnson et al., 1986; females
Follicular cell adenoma/carcinoma
Mammary adenocarcinoma
Mammary benign
Mammary benign + malignant3
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
1/58
2/60
10/60
12/60
3/100
4/102
1/50
7/46
0
-
-
2/102d
4/102
1/50
4/50
0.01
0/58
0/60
0/59
1/60
11/60
12/60
-
-
0.1
2/59
7/60
1/59
1/60
9/60
10/60
12/203
9/204
-
0.5
1/59
ll/60e
1/58
2/58
19/58
21/58
5/101
8/102
-
1.0
-
-
-
10/100
21/94e
2.0
7/59e
10/60e
5/60e
6/61
23/6 le
29/61
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 significant.
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
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) and, with the exception of the CNS tumors, are not
considered to add weight to the evidence for acrylamide carcinogenicity in animals. With
141 DRAFT-DO NOT CITE OR QUOTE
-------
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). CNS tumors are therefore considered to be one of
the tumor types replicated in the Friedman et al. (1995) study, 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.
Results from the mouse skin tumor initiation assays add considerable weight to the
evidence for acrylamide 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 mode of action. These findings are consistent with the positive
findings for AA and GA genotoxicity in numerous tests.
Other evidence of the carcinogenicity of acrylamide 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 a week for 8 weeks (Bull et al., 1984a) and in
Swiss-ICR mice 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, acrylamide 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
142 DRAFT-DO NOT CITE OR QUOTE
-------
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, the epoxide metabolite of AA, glycidamide, 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 toK. pneumoniae (Voogd et al., 1981).
Glycidamide 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). Glycidamide (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). Glycidamide 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 mode of action (MO A) 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 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 acrylamide-induced
carcinogenicity (including Besaratinia and Pfeifer, 2007; Besaratinia and Pfeifer, 2005; Schmid
143 DRAFT-DO NOT CITE OR QUOTE
-------
et al., 1999; Dearfleld et al., 1995; Segerback et al., 1995; Moore et al., 1987). Acrylamide 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:
• Acrylamide 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.
• 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.
• Glycidamide is mutagenic in short-term bacterial assays.
Glycidamide is mutagenic in male and female mouse somatic cells following oral
exposure and in male mouse germ cells (heritable translocations) following
intraparenteral exposure.
• Acrylamide induces heritable translocations in male mouse germ cells following
intraparenteral or dermal administration, and specific locus mutations in male germ cells
following intraparenteral 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 to 13.3 mg/kg-day range, which is near the range of
chronic dose levels associated with carcinogenic effects in rats (0.5 to 3 mg/kg-day).
144 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide and
mutagenic activity that could lead to the formation of tumors
There is ample evidence in the literature for the ability of acrylamide and glycidamide
(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).
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 acrylamide's
carcinogenic effects consists of positive findings of stable mutations in viable somatic cells.
145 DRAFT-DO NOT CITE OR QUOTE
-------
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 and 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 Salmonella 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).
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.
146 DRAFT-DO NOT CITE OR QUOTE
-------
Acrylamide and glycidamide 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
acrylamide 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. Acrylamide 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 acrylamide concentrations.
Segerback et al. (1995) reported much higher rates of DNA-adduct formation from
aery 1 ami de-generated glycidamide than from acrylamide 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 glycidamide 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.
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. Acrylamide and glycidamide 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. Acrylamide-DNA adduct formation was
saturable, whereas the formation of most GA-DNA adducts was dose-dependent for all doses
tested. Glycidamide 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=.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<.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
147 DRAFT-DO NOT CITE OR QUOTE
-------
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 B6C3F1 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 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 B6C3F1 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 (i.e., 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
148 DRAFT-DO NOT CITE OR QUOTE
-------
chromatid and chromosome breaks and rearrangements. Thus, the positive results in these
assays, although relevant for heritable mutations can not be definitively attributable to GA
related DNA mutations or AA related chromosomal alterations.
Acrylamide and 15 of its analogues have been tested for mutagenicity in five TA strains
of Salmonella typhimurium (Hashimoto and Tanii, 1985). Acrylamide 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 glycidamide (the other two were N,N-diglycidyl
acrylamide 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 acrylamide, via conversion
to its active epoxide metabolite, glycidamide, 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 acrylamide-induced
mutagenicity (via glycidamide) 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 non-target tissues 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. Acrylamide 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
149 DRAFT-DO NOT CITE OR QUOTE
-------
in an enhancement of transformation. Acrylamide 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 concordance and a temporal sequence of events is
generally limited. Specifically, the only data of positive mutagenicity in the dose range of the
chronic rat bioassays (0.5 to 3 mg/kg-day) to demonstrate dose-response concordance is the
increase in N7-GA-guanine adducts to steady state levels in female rat livers after approximately
14 days of repeat dosing of approximately 1 mg AA/kg bw/day in the drinking water. In male rat
livers, the adduct levels at this same regimen were consistently lower than in females, and the
increase to a maximum at approximately 14 days was followed by an apparent slow decline in
adduct levels (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.
Acrylamide'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 (three times a
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
150 DRAFT-DO NOT CITE OR QUOTE
-------
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
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
151 DRAFT-DO NOT CITE OR QUOTE
-------
(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 demonstrates 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.
Biological plausibility and coherence
DNA adducts and mutations in genes have been implicated in the carcinogenic effects of
a variety of chemicals and drugs (polycyclic aromatic hydrocarbons, vinyl chloride, benzene,
tamoxifen). Thus, it is biologically plausible that the formation of DNA adducts is a causal
event in the carcinogenicity of AA. However, the fact that adducts are 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, quantitating the cancer risk from DNA adducts requires an
ability to identify the critical adducts based on the nature of the chemical and the variety and
quantity of adducts formed, DNA repair rates, the proliferation rate of the target cells needed to
"fix" the adducts into mutations, and the mutagenic potency of those fixed adducts in critical
genes. Highly sensitive methods are also needed to correlate DNA adducts in target organs with
carcinogenic response to firmly establish both dose-response concordance and temporal
sequence.
Human relevance
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
measuring mutagenicity directly in human cells (human bronchial epithelial cells) (Besaratinia
and Pfeifer, 2004).
152 DRAFT-DO NOT CITE OR QUOTE
-------
Conclusion
There is evidence from a variety of studies of glycidamide'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 acrylamide-induced tumors in hormonally sensitive tissues (mammary gland
and thyroid) or in a tissue adjacent to hormonally sensitive tissue (tunica vaginalis, the scrotal
sac mesothelium) (Shipp et al., 2006; Environ, 2002; KS Crump Group, Inc., 1999a,b).
Although this is a possible MOA, 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
153 DRAFT-DO NOT CITE OR QUOTE
-------
proliferation in thyroid follicular cells and eventual progression to follicular cell adenomas
(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. The relevance
of the carcinogenicity of chemicals that induce Ley dig 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, short-term (2-7 days) exposure of female F344 rats to
AA caused follicular cell morphometric changes (decreased colloid area and increased cell
height) without significantly changing circulating levels of thyroid hormones or thyroid
stimulating hormone (TSH). Other 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. 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 Leydig cell membranes, leading to decreases in testosterone
production; (3) there is a subsequent compensatory increase in serum LH to maintain
testosterone at normal levels (Clegg et al., 1997; Cook et al., 1999; Prentice and Meikle, 1995);
154 DRAFT-DO NOT CITE OR QUOTE
-------
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 MO A 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
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 a 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 to 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.
Acrylamide 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
155 DRAFT-DO NOT CITE OR QUOTE
-------
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 post dosing, 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), GAB A 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
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 et al., 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).
Acrylamide-exposed rats showed increased psychomotor stimulation from amphetamine,
compared with controls, that was associated with short-term elevations of 5-hydroxyindoleacetic
156 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide 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 acrylamide-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
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 Leydig 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).
157 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide-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 Ley dig 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
that the extent of Leydig 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 Leydig cell
hyperplasia and adenomas involve elevation of serum LH and/or a change in Leydig 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 acrylamide 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 Leydig
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
158 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide 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 MO A 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
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.
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.
159 DRAFT-DO NOT CITE OR QUOTE
-------
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
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 (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
160 DRAFT-DO NOT CITE OR QUOTE
-------
Both of the available chronic exposure studies 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).
There is no clear 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 T4,
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 4-34) (Friedman et al., 1999b). A significant decrease in T3 and
T4 in 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.
161 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
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 4-35. 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.
Klaunig and Kamendulis (2005) later 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
in nontarget tissue (liver), which supports a mutagenic MOA.
162
DRAFT-DO NOT CITE OR QUOTE
-------
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
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)
aReported 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).
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 turnorigenicity 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.
163
DRAFT-DO NOT CITE OR QUOTE
-------
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 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
acrylamide-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 three
times a 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
164 DRAFT-DO NOT CITE OR QUOTE
-------
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
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 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 2000 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 seven-fold 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 MOA, 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
165 DRAFT-DO NOT CITE OR QUOTE
-------
in older infants (>90 days old), children, and young adults up to 18 years old were relatively
similar. A four-fold 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
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 acrylamide in
humans.
Acrylamide-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 4-31). 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).
Acrylamide-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.,
166 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide on other potentially susceptible
populations.
Genetic polymorphisms in the AA metabolizing 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 A A (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). There is currently no quantitative estimate of differences in parent AA or GA tissue or
blood levels that might result from CYP2E1 polymorphisms at high or low levels of AA
exposure. It is also noted that, since both the parent AA and the metabolite GA have adverse
effects, different catalytic activities of CYP2E1 may result in different spectra of adverse effects.
167 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 4.5.1, 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 Table 4-31): 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.
Acrylamide 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. However, heritable adverse effects were not observed for
the endpoints measured in the two-generation studies at the lower doses tested in these studies
(Tyl et al., 2000a; Chapin et al., 1995). Further evaluation of the linearity or nonlinearity of the
dose-response curve for these adverse heritable risks is a critical database need (see discussion
on database uncertainty in Section 5.1.3).
The most sensitive effects noted in animals are degenerative peripheral nerve changes
and male-mediated implantation losses (i.e., male-mediated dominant lethal mutations). The
lowest observed exposure levels associated with peripheral nerve changes are (with NOAELs
noted in parentheses): 1 mg/kg-day (0.2 mg/kg-day NOAEL) in male F344 rats exposed for 90
days (Burek et al., 1980) and 2 mg/kg-day (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 (Tyl et al., 2000a); 7.5 mg/kg-day (3.1 mg/kg-day NOAEL) in FO and Fl
CD-I mice (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). 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 4-31).
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
168 DRAFT-DO NOT CITE OR QUOTE
-------
mg/kg-day) associated with microscopically detected degenerative peripheral nerve changes
(Table 4-31). As discussed in Section 4.5.1, 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, including the absence of hindlimb splaying in Fl rats
exposed to doses as high as 5 mg/kg-day in the same study and in rats exposed to doses as high
as 2-3 mg/kg-day for 2 years and 5 mg/kg-day for 90 days (Table 4-31). Thus, microscopically
detected degenerative peripheral nerve changes appear to be the most sensitive effect from oral
exposure and are selected as the critical effect for deriving the RfD.
The two chronic 2-year drinking water studies (Friedman et al., 1995; Johnson et al.,
1986) are 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 better
candidates to derive the chronic RfD than the subchronic study (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
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 semiquantitative 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,
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;
169 DRAFT-DO NOT CITE OR QUOTE
-------
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). Since benchmark analysis was not possible on this data set,
the subchronic results are viewed as supporting the findings from the chronic studies. The two
chronic studies presented better quantified 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.
5.1.2. Methods of Analysis—Including Models (PBTK, BMD, etc.)
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 5-1. The benchmark response (BMR)
predicted to affect 5% of the population, BMRs, was selected for the point of departure (POD).
A BMR lower than a 10% extra risk was selected for the following reasons: (1) the 95% lower
bound of the benchmark dose (BMD), BMDLs, 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 which states that "while it is important to always report EDi0s [effective doses at 10%
extra risk] and LEDi0s [95% lower bounds of the EDi0s] for comparison purposes, the actual
'benchmark dose' used as a POD may correspond to response levels below (or sometimes above)
10%, although for convenience standard levels of 1%, 5%, or 10% have typically been used"
(U.S. EPA, 1995).
The PBTK model used in this derivation is described in Section 3.5, and a table of the
physiological and chemical parameters used in the model simulations is in Appendix E.
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.
170 DRAFT-DO NOT CITE OR QUOTE
-------
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
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., 1995d
(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
3Reported 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 chi-
square 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
5-2 (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 BMDs 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 BMDs is 0.58 mg/kg-day, and the BMDLs 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.
171
DRAFT-DO NOT CITE OR QUOTE
-------
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
Model
Male
Log-logistic
Female
Probit
BMD10
(ED10)
1.22
1.19
BMDL10
(LED10)
0.57
0.88
BMD5
(ED5)
0.58
0.67
BMDL5
(LED5)
0.27
0.49
BMDi
(EDO
0.11
0.15
BMDLi
(LEDO
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 of male and female rats in the Friedman et al. (1995) study, as assessed
by a chi-square 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 5-3
(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 5-3. 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
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
"Restrict power >1.
Source: Friedman etal. (1995).
5.1.3. RfD Derivation—Including Application of Uncertainty Factors
The male rats appeared to be slightly more sensitive than the female rats in the Johnson et
al. (1986) and Friedman et al. (1995) studies, as reflected by slightly higher BMDLs for female
rats (0.49 and 0.46 mg/kg-day, respectively) than for male rats (0.27 and 0.57 mg/kg-day,
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.
The recalibrated Kirman et al. (2003) PBTK model (discussed in Section 3.5) was used to
estimate the internal dose in a rat that would result from an external exposure to the BMDL5 of
172 DRAFT-DO NOT CITE OR QUOTE
-------
0.27 mg/kg-day. Both PBTK model variations were used: (1) enzyme-catalyzed glutathione
binding of AA and GA and (2) passive glutathione binding of AA and GA. AUC in blood for
the parent compound, AA, was considered the most appropriate choice for the internal dose
metric for neurotoxicity. Table 5-4 presents the model simulation results of the AUC for the
male and female BMDs and BMDLs, the human equivalent daily intake that would produce that
same AUC in humans from a drinking water exposure, and the drinking water concentration,
assuming a 70 kg person who drinks 2 L/day. For the drinking water simulations in humans, the
PBTK model used an ingestion pattern that divided the total daily intake among five daily water
bolus ingestions, three of which were at meal times (8 am, 1 pm, and 6 pm), each amounting to
25% of the total daily intake, and two that were in-between meals (10:30 am and 3:30 pm), each
amounting to 12.5% of the total daily intake.
Table 5-4. PBTK model simulation results for HEC based on the rat
neurotoxicity BMP
BMD analysis results (mg/kg-day)
PBTK model 1 — enzyme catalyzed glutathione binding
AUC for AA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable
AA AUC in blood, based on a drinking water exposure
Drinking water concentration that would result in the
HECa daily intake (mg/L) (70 kg person, 2 L/day)
PBTK model 2 — passive glutathione binding
AUC for AA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable
AA AUC in blood, based on a drinking water exposure
Drinking water concentration that would result in the
HEC daily intake (mg/L) (70 kg person, 2 L/day)
Female rats
BMD5
0.67
45.0
0.188
6.6
46.1
0.193
6.8
BMDL5
0.49
32.9
0.137
4.8
33.7
0.141
4.9
Male Rats
BMDs
0.58
38.9
0.163
5.7
39.9
0.167
5.9
BMDLs
0.27
18.1
0.076
2.7
18.5
0.078
2.7
aHEC = human equivalent concentration.
Source: Johnson etal. (1986).
The human equivalent concentration (HEC)-adjusted POD from the PBTK model results
based on the lowest of the BMDLs from the Johnson et al. (1986) study of 0.27 mg/kg-day for a
5% extra risk in males for mild-to-moderate lesions is 0.076 mg/kg-day for the model with
enzyme catalyzed glutathione binding and 0.078 mg/kg-day for the model with passive
glutathione binding. There is little difference between the two models results, and the slightly
lower value of 0.076 mg/kg-day is chosen for the HEC-adjusted POD in light of the uncertainty
as to which type of binding occurs in humans.
The HECpBTK model adjusted POD was divided by a total uncertainty factor (UF) of 30: 3
for extrapolation for interspecies toxicodynamic differences (UFA-TD: animal to human) and 10
for consideration of intraspecies variation (UFn: human variability).
173
DRAFT-DO NOT CITE OR QUOTE
-------
Total UF =30
= 3 (UFA.TD) x 1 (UFA-TK) x 10 (UFH) x 1 (UFS) 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 the toxicodynamics of orally administered AA. The lack of a mechanistic basis or
any quantitative information on toxicodynamic differences between rats and humans supports
not reducing the UFA_TD from 3. The PBTK model simulations are used to account for
intraspecies toxicokinetic differences, and thus the UFA.TK = 1 instead of the default value of
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-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 not
reduced.
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 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
MO A 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.
174 DRAFT-DO NOT CITE OR QUOTE
-------
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. The magnitude of response at low doses, and the
shape of the low dose-response curve for potentially serious heritable germ cell effects is also a
research need. Some of these data needs are currently being addressed.
The RfD for acrylamide was calculated as follows:
RfD = HECpBTK model
= 0.076 mg/kg-day •*- 30
= 0.003 mg/kg-day.
5.1.4. Previous RfD Assessment
This RfD replaces the previous RfD for acrylamide 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 1000: 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 (Johnson et al., 1986; Friedman et al., 1995), as well as
current methodology for characterizing the dose-response curve and for determining the POD
(i.e., the BMDL).
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
175 DRAFT-DO NOT CITE OR QUOTE
-------
al., 2001). These studies, however, do not provide reliable 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).
The PBTK model was used for a route-to-route extrapolation by simulating the internal
AUC for acrylamide in the blood that results from an oral exposure at the BMDLs for male rats
and then simulating the daily inhaled intake level that would be needed to produce a comparable
AUC in humans (Table 5-5). The level of AA in the air for such an inhalation exposure is based
on a 70 kg person who breathes 20 mVday. The benchmark response (BMR) predicted to affect
5% of the population, BMR5, was selected for the point of departure (POD). A BMR lower than
a 10% extra risk was selected for the following reasons: (1) the 95% lower bound of the
benchmark dose (BMD), BMDLs, 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 BMDLs is consistent with the technical guidance for BMD analysis (U.S. EPA,
1995).
Justification for deriving an RfC from the oral RfD comes from: (1) considerable
evidence from occupational experience involving dermal and inhalation exposure that AA-
induced peripheral neuropathy (including development of the types of degenerative lesions
observed in nerves of rats exposed via drinking water) is a well-established human health hazard;
(2) evidence that tissue distribution in rats is similar following i.v., i.p., oral, dermal, and
inhalation exposure to AA (Sumner et al., 2003; Kadry et al., 1999; Dow Chemical Co., 1984;
Miller et al., 1982; Hashimoto and Aldridge, 1970); (3) evidence that the elimination kinetics of
radioactivity from oral or i.v. administration of radiolabeled AA in rats was similar (Miller et al.,
1982); and (4) lack of support for portal of entry effects. As a caveat, the RfC would not account
for dermal absorption from AA vapor deposited on the skin, although it is not known how much
vapor would be absorbed dermally since the human dermal exposure studies applied AA in a
solution to the skin, then evaporated the liquid and covered the dried residue with gauze.
5.2.2. Methods of Analysis—Including Model (PBTK, BMD, etc.)
See Section 5.1 for derivation of the chronic oral RfD for acrylamide, Section 3.5 for a
discussion of the PBTK model, and Appendix E for a table of the physiological and chemical
parameters used in the model simulations.
176 DRAFT-DO NOT CITE OR QUOTE
-------
5.2.3. RfC Derivation—Including Application of Uncertainty Factors
The BMDLs for degenerative nerve lesions in male rats exposed to acrylamide in
drinking water for 2 years is taken as the POD for deriving the RfC. The BMDL5 from males
and females is listed in Table 5-5, along with model simulation results for the internal dose
metric (AUC of AA in blood), the human equivalent inhalation daily intake required to produce
that same AUC value, and the air concentration that would provide a 70 kg person who breathes
20 m3 of air that amount of daily exposure. For the inhalation exposure, the PBTK model
simulated a continuous 24-hour inhalation exposure (i.e., no variation in the amount of
intake/unit time at any time throughout the day).
Table 5-5. PBTK model simulation results for HEC based on the rat
neurotoxicity BMP
BMD analysis results (mg/kg-day)
PBTK Model 1 — enzyme catalyzed glutathione binding
AUC for AA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable
AA AUC in blood, based on continuous air exposure
Air concentration that would result in the HEC daily
intake (mg/m3) (70 kg person, 20 nrVday)
PBTK Model 2 — passive glutathione binding
AUC for AA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable
AA AUC in blood based, on continuous air exposure
Air concentration that would result in the HEC daily
intake (mg/m3) (70 kg person, 20 nrVday)
Female rats
BMDS
0.67
45.0
0.180
0.63
46.1
0.188
0.66
BMDLS
0.49
32.9
0.132
0.46
33.7
0.138
0.48
Male rats
BMD5
0.58
38.9
0.156
0.55
39.9
0.163
0.57
BMDLS
0.27
18.1
0.073
0.25
18.5
0.076
0.27
Source: Johnson etal. (1986).
The HEC-adjusted POD from the PBTK model results is based on the lowest of the
Johnson et al. (1986) study BMDLs of 0.27 mg/kg-day for a 5% extra risk in males for mild-to-
moderate lesions. This HEC is 0.073 mg/kg-day for the model with enzyme catalyzed
glutathione binding, and 0.076 mg/kg-day for the model with passive glutathione binding. There
is little difference between the two models results, and the lowest value of 0.073 mg/kg-day is
chosen for the HEC adjusted POD in light of the uncertainty as to which type of binding occurs
in humans. In a 70 kg person inhaling 20 m3/day air, the air concentration needed to result in the
HEC adjusted POD would be 0.25 mg/m3.
The HECpBTK model-adjusted POD was divided by a total UF of 30: 3 for extrapolation for
interspecies toxicodynamic differences (UFA-To: animal to human) and 10 for consideration of
intraspecies variation (UFH: human variability).
177
DRAFT-DO NOT CITE OR QUOTE
-------
Total UF = 30
= 3 (UFA.TD) x 1 (UFA.TK) x 10 (UFH ) x 1 (UFS) 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 not reducing the UFA-TD from 3. The PBTK model simulations are used to account
for intraspecies toxicokinetic differences, and thus the UFA.TK = 1 instead of the default value of
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 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 not reduced.
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).
An UF to account for database deficiency is not necessary for this derivation (i.e., UFD =
1) because a PBTK model was used to conduct the route-to-route extrapolation from 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
MO A 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.
178 DRAFT-DO NOT CITE OR QUOTE
-------
The RfC for acrylamide is calculated as follows:
RfC = HECpBTK model ^ UF
= 0.25 mg/m3 - 30
= 0.008 mg/m3
5.2.4. Previous RfC Assessment
The previous IRIS assessment did not derive a RfC.
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
acrylamide. 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), and 4) the uncertainties in the
structure and parameter values of the PBTK model relative to its use in deriving the toxicity
values.
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) and to ensure that the risk to
chemicals and stressors are not underestimated. 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. Default uncertainty factors are used in the derivation of the
RfD and RfC to adjust the POD downward (i.e., to lower the acceptable exposure level) and thus
reduce the potential risk of adverse effects to protect the public health.
The specific uncertainty factors used in deriving the acrylamide RfD and RfC were
previously discussed in sections 5.1.3 and 5.2.3, respectively. A PBTK model 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
179 DRAFT-DO NOT CITE OR QUOTE
-------
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 5-7.
5.3.1 Areas of Uncertainty
Completeness of the Database
The human data for potential non-cancer adverse effects from exposure to acrylamide 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 acrylamide 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 acrylamide 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 acrylamide 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 acrylamide (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.
180 DRAFT-DO NOT CITE OR QUOTE
-------
In summary, there is a substantial animal database to assess the noncancer effects of
acrylamide. 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.
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. Heritable germ cell effects (e.g.,
translocations, dominant lethals) have also been reported in animal studies, and are a potentially
more serious adverse event than neurotoxicity because these are effects that can occur not only
in the exposed individual, but also their offspring and subsequent generations. Heritable germ
cells effects, however, have only been observed at relatively high levels of acrylamide exposure
in animal studies (orders of magnitude higher than the levels where neurotoxicity has been
observed). Data are not available to determine if these effects would be seen at lower exposure
levels in test animals, or whether they appear in humans at any level of exposure. To resolve this
uncertainty as to the possibility that heritable germ cell effects are the most critical low dose
effect will require more data.
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 acrylamide.
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 damage, the values of the RfD and RfC could potentially be lower.
Dose-Response Modeling and Determination of the Point of Departure
Benchmark dose (BMD) modeling was used to estimate the point of departure (POD) for
the acrylamide 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.
181 DRAFT-DO NOT CITE OR QUOTE
-------
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 benchmark response (BMR) predicted to affect 5% of the population, BMRs was
selected for the point of departure (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 acrylamide, 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 5-6, all of the final POD
estimates are within 2-fold of each other, supporting a relatively high degree of confidence that
the estimated BMDLs 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 acrylamide 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.
182 DRAFT-DO NOT CITE OR QUOTE
-------
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.
BMD
Model (EDS)
BMDL
(LED5)
Johnson etal. (1986)
Male
Log-logistic 0.58
Female
Probit 0.67
0.27
0.49
Friedman et al. (1995)
Male
Quantal quadratic 0.77
Female
Gamma3 2.25
0.57
0.46
aRestrictpower>l
Adequacy of the PBTK Model for Use in Deriving the RfD and the RfC
EPA recalibrated and parameterized a PBTK model for acrylamide (originally developed
and published by Kirman et al., 2003) against more recent data (Boettcher et al., 2005; Doerge et
al., 2005a,b,c; Fennell et al., 2005; Sumner et al., 2003) for use in deriving toxicity values. The
recalibrated AA PBTK model was tested against the new kinetic and hemoglobin binding data in
rats, mice, and humans. The model was then used to estimate the oral human equivalent
concentration (i.e., extrapolate the animal dose-response relationship to humans) to derive the
RfD, and to conduct a route-to-route extrapolation (oral to inhalation) to estimate the inhalation
HEC to derive the RfC.
There is always some degree of uncertainty in a PBTK model's structure and estimates of
internal dose, because PBTK models are simplified mathematical representations of very
complex organisms. Within the context of the alternative of using default factors, however,
PBTK models are increasingly offering a more scientifically supportable means to: 1)
extrapolate risk as observed in studies on test animal to potential risks to humans, 2) account for
the most sensitive human subpopulations, or 3) conduct route-to-route extrapolations. For
acrylamide, a published and tested PBTK model was available, and in its recalibrated form,
provided acceptable fits to recent and relevant data. The resulting fits to the available data were
adequate to support the model's use to estimate the oral human equivalent concentration to
derive the RfD, and to reduce uncertainty in that estimate compared to use of the default
uncertainty factor for interspecies toxicokinetic differences. The AA PBTK model was also used
to estimate the inhalation HEC based on the oral dose-response data to derive an RfC, a value
183 DRAFT-DO NOT CITE OR QUOTE
-------
that would otherwise not have been possible since there are no credible default methods for
conducting a route-to-route extrapolation. It is unknown how inaccuracies in the AA PBTK
model structure or parameter values would impact the assessment; further refinements in the
model or new data could increase or decrease the estimate of risks of neurotoxicity in humans.
5.3.2 Uncertainty Factors in Deriving the RfD and RfC
Uncertainty in the Completeness of the Database
As discussed above, the animal database is robust and complete by IRIS assessment
standards. There is a possibility that heritable germ cell effects or functional neurotoxicity might
be more sensitive endpoints, but more data are needed to resolve this issue, and meanwhile the
standard suite of animal toxicity studies are available to support the derivation of toxicity values.
Although the human data are limited, the predominantly neurological endpoints that have been
reported are similar to those observed in the animal studies. Thus no additional reduction in the
toxicity value is needed to account for uncertainty in the completeness of the database (i.e., UFD
= 1).
Uncertainty in the Animal to Human Extrapolation
The accuracy of extrapolating the dose-response relationship as observed in animals to
the dose-response that will occur in humans is a source of uncertainty. This extrapolation is
based on species differences in toxicokinetics and toxicodynamics. A PBTK model is available
that accounts for the differences between rat and human toxicokinetics following exposure to
acrylamide in deriving the RfD. The impact of the uncertainties in the model for each
application are discussed below.
PBTK model estimate of the oral human equivalent exposure (HEC)
The acrylamide PBTK model simulated the toxicokinetics of acrylamide in the body of
the Fisher 344 rat used in the chronic bioassays or in humans (an average 70 kg male). The
model was exercised to estimate the external exposure in each species that would be needed to
produce the same internal level of exposure at a target site or at some surrogate site. The PBTK
model thus accounts for toxicokinetic differences in extrapolating the external dose at the POD
for the rat to an equivalent external dose for humans based on the resulting internal dose that
would be expected to result in similar level of response. For acrylamide, the internal dose metric
used to estimate this human equivalent concentration (HEC) was the average concentration of
acrylamide in the blood over a 24 hour period (or the area under the time-concentration curve for
a 24 hour period). Since acrylamide rapidly distributes throughout the body, the acrylamide
AUC in blood is considered to be a good surrogate for levels at the putative target site (brain or
nerve tissue). The acrylamide PBTK model is based on sufficient animal data to support this use,
however, the human data are limited contributing to some uncertainty as to the accuracy of the
overall PBTK model's simulation results. It is unknown what the direction would be on risks to
184 DRAFT-DO NOT CITE OR QUOTE
-------
humans from an inaccurate PBTK model estimate of the HEC; the actual risk to humans could be
higher or lower. Additional human and animal kinetic data are being developed in research
centers around the world to further support the AA PBTK model and increase the confidence in
it's predictive capability.
PBTK model estimate of the inhalation HEC (route-to-route extrapolation to derive the
RfC)
Unlike for interspecies extrapolation, there are no credible default approaches to conduct
a route-to-route extrapolation. The AA PBTK model does so by estimating the daily inhaled
intake level (mg/m3) that would be needed to produce a human AUC for acrylamide in the blood
comparable to the level estimated from an oral exposure at the POD. That daily inhaled intake
level is then further adjusted with the default factors to account for uncertainty in interspecies
toxicodynamic differences and human variability to derive the RfC.
A route-to-route extrapolation of the RfC from the oral dose-response data is justified
based upon: (1) considerable evidence from occupational experience demonstrating that dermal
and inhalation exposure results in AA-induced peripheral neuropathy (including development of
the types of degenerative lesions observed in nerves of rats exposed via drinking water); (2)
evidence that tissue distribution in rats is similar following i.v., i.p., oral, dermal, and inhalation
exposure to AA (Sumner et al., 2003; Kadry et al., 1999; Dow Chemical Co., 1984; Miller et al.,
1982; Hashimoto and Aldridge, 1970); (3) evidence that the elimination kinetics of radioactivity
from either oral or i.v. administration of radiolabeled AA in rats are similar (Miller et al., 1982);
and (4) lack of support for portal of entry effects. Since there are no credible default methods to
estimate a safe daily inhaled intake level in the absence of inhalation study data, the level of
uncertainty in the RfC based on the AA PBTK model must be compared to the complete
uncertainty of having no RfC. As with uncertainty in the accuracy of the model in the derivation
of the RfD, further refinements in the model or new data could | or | the estimate of risks to
humans.
Use of default factors for the interspecies differences in toxicodynamics in conjunction
with the PBTK model
The AA PBTK model replaced the default factor for interspecies toxicokinetic
differences of 3 (UFA-TK = 3.16 without the model; 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
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.
185 DRAFT-DO NOT CITE OR QUOTE
-------
without the model (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 the same value 0.003 mg/kg/day as the RfD
derived with the PBTK model. One interpretation of this similarity is not that the PBTK model is
deficient, rather that the interspecies differences for parent acrylamide toxicokinetics might scale
roughly to the ratio of body weight to the 3/4 power which for extrapolating between an average
rat (350 grams) and human (70kg) is approximately a 3 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
acrylamide-specific data on human variation, a default UFn 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 acrylamide 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).
8 The RfD using the default approach is 0.003 mg/kg/day. RfDdefeultapproach = 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.
186 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-7. 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 acrylamide.
The available acrylamide database is
sufficiently robust and adequate to
identify commonly known endpoints
for adverse effects, and to not warrant a
UFD>1.
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. The animal database for repeated oral
exposures is robust and evaluates a wide spectrum of adverse
effects. Although animal studies for inhalation exposures are
limited, kinetic studies in animals and humans indicate no
critical route specific endpoints.
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,
and further research is warranted to evaluate the potential for
these effects at lower doses. Reproductive effects have been
observed in animals, but at NOAEL 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
acrylamide dose-response data and
provided valid estimates of the POD.
A number of BMD models provided reasonable fits to the
acrylamide 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.
187
DRAFT-DO NOT CITE OR QUOTE
-------
Consideration/
Approach
Impact on noncancer risk
estimate
Decision
Justification
PBTK model use in
estimating 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 PBTK model was used to estimate
the human equivalent dose used in the
derivation of the RfD.
The acrylamide PBTK is a published and peer reviewed model
that has been further recalibrated and tested against more recent
data. It provides a more scientifically supportable estimate of
the HEC compared to the use of the default factor. The model's
functionality and fits to the data were sufficiently good to
support its use in the assessment. Additonal data sets are
anticpated in the future that can be used to further test the model
and reduce uncertainty in the model results.
PBTK model use in
estimate 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 f
or I the estimated risks to
humans.
PBTK model was used in a route-to-
route extrapolation to derive an RfC.
Justification for deriving an RfC from the oral RfD is based on
human and animal kinetic data that support a model based route-
to-route extrapolation and the assumptions of comparable
responses for comparable internal levels of exposure reagardless
of the route of exposure.
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
PBTK model derived HEC in the
derivation of the RfD and RfC.
The default uncertainty factor for toxicodynamic differences
was used in the absence of an adequately developed and tested
PBTD model, or other chemical or species specific data to
support a more informed extrapolation. In keeping with the
EPA's goal of protecting public health and the environment, the
default factor for intraspecies differences in toxicodynamics is
used to ensure that the risk to chemicals and stressors are not
underestimated.
188
DRAFT-DO NOT CITE OR QUOTE
-------
Consideration/
Approach
Impact on noncancer risk
estimate
Decision
Justification
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. As above, the
default factor for intrahuman variability is used to ensure that
the risk to chemicals and stressors are not underestimated.
189
DRAFT-DO NOT CITE OR QUOTE
-------
5.4. CANCER ASSESSMENT
5.4.1. Choice of Study/Data—with Rationale and Justification
As summarized in Section 4.6.1, acrylamide 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, and CNS tumors (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). Evidence from available human studies is judged to be
limited to inadequate. No statistically significant increased risks for cancer-related deaths were
found in two cohort mortality studies of AA, with the exception of the finding in one study of an
increased risk of pancreatic cancer in a subgroup of workers with the highest cumulative AA
exposure (Marsh et al., 1999; Collins et al., 1989). Most of the available epidemiology studies
on increased risk of cancer from AA in food have been conducted by Mucci and colleagues
including three case-control studies for increased risk of cancers of the large bowel, bladder,
kidneys, renal cell or breast (Mucci et al., 2003, 2004, 2005), and one prospective study for
colorectal cancers (Mucci et al., 2006). In another large case-control study, Pelucchi et al.
(2006) evaluated the relation between dietary AA intake and cancers of the oral cavity and
pharynx, esophagus, large bowel, rectum, larynx, breast, ovary and prostate. None of these
studies report a significant increased incidence of cancer associated with increased intake of AA
in food at the levels of intake observed.
The mechanisms by which AA induces cancer in animals are not fully understood,
however, the weight of the scientific evidence strongly supports a mutagenic mode of action
(MOA), as discussed in Section 4.7.3.1. 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 (see Section 4.7.3.2); however, the available evidence is limited
or nonexistent, and is not informative concerning the relevance of these tumors (or the other
tumors observed in animals) to humans9. Therefore, the cancer dose-response relationships for
9 As discussed in detail in Section 4.7.3, the evidence that acrylamide-induced mesotheliomas in male
Fischer 344 rats may not be relevant to humans includes observations that acrylamide caused decreased circulating
levels of prolactin in male Fischer 344 rats (presumably through dopamine agonist activity at the D2 dopamine
receptor); that chemicals that induce Ley dig 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
190 DRAFT-DO NOT CITE OR QUOTE
-------
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 Johnson et al. (1986) cancer rat bioassay reported increased
tumor incidence at other sites in females (CNS, oral cavity, pituitary gland); however, this 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 did not reproduce the CNS or
oral tumors and was designed to address some of the deficiencies in the Johnson et al. (1986)
study. These improvements included different dose spacing to support better characterization of
the dose-response relationship, and a substantially larger control group (n=204) and 0.1 mg/kg-
day male rat group (n = 204) to increase the statistical power in the study to detect significantly
increased tumor incidence. The resulting data further supported selection of the tumor types that
were replicated in the second study (thyroid, mammary, and testicular tumors) for development
of the oral slope factor.
5.4.2. Dose-Response Data
Incidences of tumors with statistically significant increases in both of the 2-year
bioassays with F344 rats exposed to AA in drinking water are shown in Table 5-8. As discussed
in the previous section, incidence data for thyroid tumors in male and female rats, tunica
vaginalis mesotheliomas in male rats, and mammary gland tumors in female rats from the
Friedman et al. (1995) bioassay were chosen for oral slope factor development to support
reliable characterizations of dose-response relationships.
mesotheliomas in control and acrylamide-exposed male Fischer 344 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.
191 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-8. Incidence of tumors with statistically significant increases in a
2-year bioassay with F344 rats exposed to acrylamide in drinking water
Reference/tumor type
Friedman et al., 199 5 /males"
Follicular cell adenoma/carcinoma
Tunica vaginalis mesotheliomab
Friedman et al., 199 5 /females"
Follicular cell adenoma/carcinoma
Mammary malignant/benign
Dose (mg/kg-day)
0
3/100
4/102
1/50
7/46
0
2/102c
4/102
1/50
4/50
0.01
-
—
0.1
12/203
9/204
—
0.5
5/101
8/102
—
1.0
-
10/100
21/946
2.0
17/75d
13/75d
—
3.0
-
23/1006
30/956
aTwo 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 those reported in the
reevaluation study by latropoulos et al. (1998).
°The 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.
dStatisically significant (p < 0.05).
eStatisically significant (p < 0.001).
Source: Friedman etal. (1995).
5.4.3. Dose Adjustments and Extrapolation Method(s)
The current EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005) 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, 2005). In the case of acrylamide, 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 acrylamide exposure.
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
192
DRAFT-DO NOT CITE OR QUOTE
-------
[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-WEffi (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 5-9. In addition, the results for considering female rats
with either tumor are also presented in Table 5-9. 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 )~\ 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, 2005) 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
that evaluating tumor-bearing animals would tend to underestimate overall risk when tumor
types occur in a statistically independent manner. The NRC-recommended approach involves
adding 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 point of departure for the combined incidence
approach was based on 20% extra risk, because 20% was the lowest extra risk consistent with
the lower end of the observed data range. The BMD20 is 1.2 mg/kg-day, and the BMDL20 is 0.88
mg/kg-day. For linear low-dose extrapolation, the rat slope factor associated with this combined
193 DRAFT-DO NOT CITE OR QUOTE
-------
risk is 0.2/0.88 (mg/kg-day) l, or 0.23 (mg/kg-day) l, 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.
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
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
Rat Slope factor
[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 the NRC (1994) approach, the rat slope factor was 0.24 (see Appendix D).
Data source: Friedman etal. (1995).
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 5-10. 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.
194
DRAFT-DO NOT CITE OR QUOTE
-------
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
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
Rat Slope factor
[risk level/BMDL]
(mg/kg-day)"1
1.3 x 10"1
2.2 x 10"1
3.3 x lO"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 rat slope factor was 0.34 (see Appendix D).
Data source: Friedman etal. (1995).
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,
0.22 (mg/kg-day)"1 and 2.5-fold higher than for testicular tumors, 0.13 (mg/kg-day)"1 (see
Appendix D for more details).
Based on the analyses discussed above, the recommended upper bound estimate on
human extra cancer risk from continuous, lifetime oral exposure to AA (see Sections 5.3.4 and
5.3.5 for derivation) should be based on a rat slope factor 0.3 (mg/kg-day)"1 as derived from the
male rat data for the risk of TVM or thyroid tumors. The human equivalent slope factor (derived
below) 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.
5.4.4. Human Equivalent Concentration Using the PBTK Model
The recalibrated Kirman et al. (2003) PBTK model (discussed in Section 3.5) was used to
simulate the rat internal exposure from the external BMDLs, then to simulate the external HEC
that would be needed to result in that same internal exposure. Both PBTK model variations were
used: enzyme-catalyzed glutathione binding of AA and GA and passive glutathione binding of
AA and GA. The most appropriate measure of internal dose was considered to be the AUC in
blood for the GA metabolite. The choice of AUC as the dose metric is based on a mutagenic
MO A for a GA-induced carcinogenic endpoint and the assumption that total amount of this
active metabolite would more strongly correlate to incidence of tumors than other dose metric
195
DRAFT-DO NOT CITE OR QUOTE
-------
choices such as peak concentration, time above peak, or average concentration. The choice of
AUC in the blood compartment is based on the available blood level (or hemoglobin adduct)
data that support model calibration and simulation of the blood compartment compared to the
relatively sparse data for tissue levels of glycidamide or the most tumor susceptible tissues in
humans.
Table 5-11 presents the model simulation results of the GA AUC for the male BMDio
and BMDLio and the human equivalent daily oral intake that would produce that same AUC in
humans from a drinking water exposure and the drinking water concentration, assuming a 70 kg
person who drinks 2 L/day. For the drinking water simulations in humans, the PBTK model
used an ingestion pattern that divided the total daily intake among five daily water bolus
ingestions, three of which were at meal times (8 am, 1 pm, and 6 pm), each amounting to 25% of
the total daily intake, and two that were in between meals (10:30 am and 3:30 pm), each
amounting to 12.5% of the total daily intake.
Table 5-11. PBTK model simulation results for HEC based on male rat
carcinogenicity data
BMD analysis results (mg/kg-day)a
PBTK Model 1 — enzyme catalyzed glutathione binding
AUC for GA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable GA
AUC in blood, based on a drinking water exposure
Drinking water concentration that would result in the HEC daily
intake (mg/L) (70 kg person, 2 L/day)
PBTK Model 2 — passive glutathione binding
AUC for GA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable GA
AUC in blood, based on a drinking water exposure
Drinking water concentration that would result in the HEC daily
intake (mg/L) (70 kg person, 2 L/day)
BMD10
0.7
35.3
0.52
18.3
36.0
0.53
18.5
BMDL10
0.3
15.1
0.22
7.8
15.4
0.23
7.9
aSee Tables D-4 and D-5 for BMD10 and BMDL10 derivations for male TVM or thyroid tumors.
Data source: Friedman etal. (1995).
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
196
DRAFT-DO NOT CITE OR QUOTE
-------
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 values used to derive an HEC POD as the basis for cancer risks in humans orally
exposed to acrylamide, are the male rat BMDio of 0.7 mg/kg-day and BMDLio of 0.3 mg/kg-day
for the combined risk of male rats bearing TVM or thyroid tumors. The PBTK model was used
to extrapolate to the HEC-BMDio of 0.52 mg/kg-day for the model with enzyme-catalyzed
glutathione binding and 0.53 mg/kg-day for the model with passive glutathione binding. The
PBTK model yielded an HEC-BMDLi0 of 0.22 mg/kg-day for the model with enzyme-catalyzed
glutathione binding and 0.23 mg/kg-day for the model with passive glutathione binding. There
is little difference between results from the two variations of the PBTK model, and, in light of
the uncertainty as to which type of binding occurs in humans, the lower value of 0.22 mg/kg-day
is used as the POD. The human oral slope factor is derived by linear extrapolation from the POD
to the origin, corrected for background. The slope of the linear extrapolation from the HEC PBTK
modei-BMDio (a central estimate) is 0.19 (mg/kg-day)"1 (response rate/BMDio; 0.1/[0.52 mg/kg-
day] = 0.19 [mg/kg-day]"1) and the oral slope factor derived from the linear extrapolation from
the HEC PBTK modei-BMDLio is 0.45 (mg/kg-day)"1 (response rate/BMDLi0; 0.1/0.22 [mg/kg-day]
= 0.45 [mg/kg-day]"1).
With rounding to one significant figure, the human oral slope factor based on the
HECpBTKmodei-BMDLio is 0.5 [mg/kg-day]'1.
The human slope factor for acrylamide should not be used with exposures exceeding the
POD (LEDio), because above this level the fitted dose-response model better characterizes what
is known about the carcinogenicity of acrylamide. Additionally, 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].
5.4.5.2. Inhalation Unit Risk
A PBTK model is used to conduct a route-to-route extrapolation from the acrylamide oral
cancer bioassay results to an equivalent inhalation exposure, i.e., inhalation exposures that result
in comparable internal doses as those obtained from an administered oral dose. Support for use
of PBTK model in the derivation of an inhalation unit risk value based on oral data comes from:
(1) evidence that tissue distribution in rats is similar following i.v., i.p., oral, dermal, and
197 DRAFT-DO NOT CITE OR QUOTE
-------
inhalation exposure to AA (Sumner et al., 2003; Kadry et al., 1999; Dow Chemical Co., 1984;
Miller et al., 1982; Hashimoto and Aldridge, 1970); (2) the widespread distribution of AA and
the formation of GA adducts in diverse tissue throughout the body (Doerge et al., 2005a); (3)
evidence that the elimination kinetics of radioactivity from oral or i.v. administration of
radiolabeled AA in rats was similar (Miller et al., 1982); and (4) lack of support for portal-of-
entry effects. The PBTK model also accounts for first-pass metabolism, one of the main factors
that can lead to different distributional profiles between an oral and an inhalation exposure.
The inhalation unit risk for AA is based on adult exposures and is derived from the HEC
of the rat bioassay BMDLio, which is the 95% lower bound on the exposure associated with a
10% extra cancer risk, by dividing the risk (as a fraction) by the BMDLio. 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 MOA.
The PBTK model is used to derive an inhalation HEC-BMDio and HEC -BMDLio based
on comparable levels achieved of the AUC for GA in blood to those from the oral exposure
BMDio of 0.7 mg/kg-day and the BMDLio of 0.3 mg/kg-day in male F344 rats for combined
incidence of animals bearing tunica vaginalis mesotheliomas or thyroid tumors (Friedman et al.,
1995). The PBTK model predicted an oral HECpBTKmodei-BMDio of 0.53 mg/kg-day and an
HEC-BMDLio of 0.22 mg/kg-day. The air concentrations required to achieve these daily intake
levels, assuming a continuous 24-hour inhalation exposure for a 70 kg person who breathes 20
m3/day air are 1.84 mg/m3 for the HECpBTKmodei-BMDio and 0.79 mg/m3 for the HEC PBTK modei-
BMDLio (see Table 5-12). The slope of the linear extrapolation from the HEC -BMDio (a central
estimate) of the equivalent air concentration as a POD is 5 x io~5 (jig/m3)"1 and the inhalation
unit risk based on the HEC -BMDLio equivalent air concentration as a POD is 1.3 x 10~4
HECpBTKmodei-BMDLio equivalent air concentration, lower 95% bound on exposure at
10% extra risk — 0.79 mg/m3
HECpBTKmodei-BMDio equivalent air concentration, central estimate of exposure at 10%
extra risk — 1.84 mg/m3
Inhalation unit risk based on the HEC PBTK modei-BMDLio
= 0.1/0.79 mg/m3= 1.3 x 10~4 (jig/m3)"1;
With rounding to one significant figure, the inhalation unit risk is 1 x 10~4 (ug/m3)"1.
198 DRAFT-DO NOT CITE OR QUOTE
-------
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].
This derivation of the inhalation unit risk is not suitable to derive MLEs or 95% UCLEs
for lifetime inhalation exposure concentrations that would be associated with various levels of
risks. Appendix D provides an alternate derivation for MLEs and 95% UCLEs with similar
results.
Table 5-12. PBTK model simulation results for HEC to derive the inhalation
unit risk based on male rat oral exposure cancer data
BMD10 analysis results (mg/kg-day)a
PBTK Model 1 — enzyme catalyzed glutathione binding
AUC for GA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable GA
AUC in blood, based on a continuous air exposure
Air concentration that would result in the HEC daily intake
(mg/m3) (70 kg person, 20 nrVday)
PBTK Model 2 — passive glutathione binding
AUC for GA in rat blood (uM-hour)
Human equivalent intake (mg/kg-day) for a comparable GA
AUC in blood, based on a continuous air exposure
Air concentration that would result in the HEC daily intake
(mg/m3) (70 kg person, 20 nrVday)
BMD10
0.7
35.3
0.53
1.84
36.0
0.54
1.89
BMDL10
0.3
15.1
0.22
0.79
15.4
0.23
0.81
"See Tables D-4 and D-5 for BMD and BMDL10 derivations.
Data source: Friedman etal. (1995).
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 3.4.1), 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, in accordance with the
Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
(U.S. EPA, 2005b). The oral slope factor of 0.5 per mg/kg-day and the inhalation unit risk of 1 x
10"4 per ng/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.
199
DRAFT-DO NOT CITE OR QUOTE
-------
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. These ADAFs and their age groups were obtained from the 2005
Supplemental Guidance., and they may be revised over time. 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 |o,g/kg-day.
70-year exposure to 1 |o,g/kg-day AA from ages 0-70:
Age group
0 - < 2 years
2-<16
years
> 16 years
ADAF
10
3
1
unit risk
(per ng/kg-
day)
0.5 x 10'3
0.5 x 10'3
0.5 x 10'3
exposure
concentration
(Hg/kg-day)
1
1
1
duration
adjustment
2 years/70
years
14 years/70
years
54 years/70
years
partial risk
1.4xlO'4
3.0 xlO'4
3.9 x 10'4
Totalrisk=8x 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 (0.5 x 10"3) x 1 x 2/70 = 1.4 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 |o,g /kg-day is equivalent to a lifetime
unit risk estimate of 8 x 10"4 per ug/kg-day, adjusted for early-life susceptibility, assuming a 70-
year lifetime and constant exposure across age groups.
200
DRAFT-DO NOT CITE OR QUOTE
-------
If calculating the cancer risk for a 30-year exposure to a constant AA exposure level of 1
|o,g /kg-day from ages 0-30, the duration adjustments would be 2/70, 14/70, and 14/70, and the
partial risks would be 1.4 x 10"4, 3.0 x 10"4, and 1.0 x 10"4, resulting in a total risk estimate of 5 x
io-4.
If calculating the cancer risk for a 30-year exposure to a constant AA exposure level of 1
Hg /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 2.1 x IO"4, resulting in a total risk estimate of 2 x IO"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 ng/m3 AA from ages 0-70:
Age group
0 - < 2 years
2-<16
years
> 16 years
ADAF
10
3
1
unit risk
(per ug/m3)
1 x 10'4
1 x 10'4
1 x 10'4
exposure
concentration
(ug/m3)
1
1
1
duration
adjustment
2 years/70
years
14 years/70
years
54 years/70
years
partial risk
2.9 xlO'5
6.0 xlO'5
7.7 xlO'5
Totalrisk = 2x IO"4
Note that the partial risk for each age group is the product of the values in columns 2-5
[e.g., 10 x (IxlO"4) x 1 x 2/70 = 2.9 x IO"5], and the total risk is the sum of the partial risks. This
70-year risk estimate for a constant exposure of 1 |o,g /m3 is equivalent to a lifetime unit risk
estimate of 2 x IO"4 per ug/m3, 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/m3 from ages 0-30, the duration adjustments would be 2/70, 14/70, and 14/70, and the partial
risks would be 2.9 x IO"5, 6.0 x IO"5, and 2.0 x IO"5, resulting in a total risk estimate of 1 x IO"4.
201
DRAFT-DO NOT CITE OR QUOTE
-------
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 4.3 x 10"5, resulting in a total risk estimate of 4 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 carcinogen!city 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 acrylamide, specifically the cancer oral slope factor (CSF)
and the inhalation unit risk (IUR). These uncertainties arise either from incomplete knowledge
about acrylamide'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 structure and parameter values of the PBTK model relative to
its use in deriving the oral slope factor and the inhalation unit risk (IUR), and 5) 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 5-13.
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
acrylamide 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
202 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide induced cancer in humans.
Selection ofBioassay(s), Tumor Types, and Relevance to Humans (Le., 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 acrylamide
exposure via the drinking water, both in the F344 rats (Friedman et al., 1995; Johnson et al.,
1986). The Friedman et al. (1995) study addressed some of the deficiencies in the Johnson et al.
(1986) study by improving the dose spacing to better characterize the dose-response relationship,
and by substantially increasing the size of the control group (n=204) and a 0.1 mg/kg-day male
rat group (n = 204) to increase the statistical power to detect increased incidence of tumors.
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 acrylamide and glycidamide
(U.S. FDA [2006b]). The impact of these new data could be either to increase or decrease the
estimate of risks of acrylamide 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. These were the tumor types used in the derivation of the CSF and IUR.
Uncertainty in the selection of tumor type arises because Johnson et al. (1986) reported a variety
of other tumor types in females (CNS, oral cavity, pituitary gland). Although 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 including, the CNS tumors are of concern considering
acrylamide's known neurotoxicity. Rice (2005) has raised an 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 acrylamide induced cancer in humans.
203 DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide 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 acrylamide, 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. 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-Weibull modeling of
the Friedman et al. (1995) male rat data yielded a combined incidence BMDio of 0.7 mg/kg-day,
and BMDLio of 0.3 mg/kg-day, an approximately 2.3 fold difference. The multistage modeling
204 DRAFT-DO NOT CITE OR QUOTE
-------
of the female rat data yielded a combined incidence BMD20 of 1.2 mg/kg-day, and BMDL20 of
0.88 mg/kg-day, an approximately 1.4 fold difference. 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 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. These relatively similar results from the female or
male rat data for different approaches to calculating total risk 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 of the PBTK Model for Use in Deriving the CSF andlUR
EPA recalibrated and parameterized a PBTK model for acrylamide (originally developed
and published by Kirman et al., 2003) against more recent data (Boettcher et al., 2005; Doerge et
al., 2005a,b,c; Fennell et al., 2005; Sumner et al., 2003) for use in deriving toxicity values. The
recalibrated AA PBTK model was tested against the new kinetic and hemoglobin binding data in
rats, mice, and humans. The model was then used to estimate the oral human equivalent
concentration (i.e., extrapolate the animal dose-response relationship to humans) to derive the
cancer slope factor, and to conduct a route-to-route extrapolation (oral to inhalation) to derive
the inhalation FIEC to derive the inhalation unit risk.
There is always some degree of uncertainty in a PBTK model's structure and estimates of
internal dose, because PBTK models are simplified mathematical representations of very
complex organisms. Within the context of the alternative of using default factors, however,
PBTK models are increasingly offering a more scientifically supportable means to: 1)
extrapolate risk as observed in studies on test animal to potential risks to humans, 2) account for
the most sensitive human subpopulations, or 3) conduct route-to-route extrapolations. For
acrylamide, a published and tested PBTK model was available, and in its recalibrated form,
provided acceptable fits to recent and relevant data. The resulting fits to the available data were
adequate to support the model's use in the assessment. The model was especially important to
205 DRAFT-DO NOT CITE OR QUOTE
-------
estimating the oral human equivalent concentration in the CSF derivation because the putative
toxin was the acrylamide metabolite, glycidamide, the internal levels of which the PBTK model
did account for, while the default uncertainty factor for interspecies toxicokinetic differences
would not. The AA PBTK model was also used to derive an inhalation HEC based on the oral
dose-response data, a value that would also otherwise not have been possible since there are no
credible default methods for conducting a route-to-route extrapolation.
To estimate the oral exposure HEC, the acrylamide PBTK model first simulated the
toxicokinetics of acrylamide and its metabolite, glycidamide, in the Fisher 344 rat used in the
chronic bioassays and then in an average male human (i.e., a 70 kg male). The model was
exercised to estimate the external exposure in each species that would be needed to produce the
same internal level of exposure at a target site or at some surrogate site. The PBTK model thus
accounted for toxicokinetic differences in extrapolating the external dose at the POD for the rat
to an equivalent external dose for humans based on the resulting internal dose that would be
expected to result in a similar level of response. For acrylamide, the internal dose metric used to
estimate the oral human equivalent concentration (HEC) was the average concentration of
glycidamide in the blood over a 24 hour period (or the area under the time-concentration curve
for a 24 hour period). Glycidamide is the putative toxin for carcinogenic effects.
The AA PBTK model was also used to derive an inhalation HEC based on the oral dose-
response data, a value that would otherwise not have been possible since there were no available
inhalation data to directly derive an IUR, and no credible alternate method for conducting a
route-to-route extrapolation. The AA PBTK model does so by estimating the daily inhaled
intake level (mg/m3) that would be needed to produce a human AUC for glycidamide in the
blood comparable to the level estimated from an oral exposure at the POD. That daily inhaled
intake level is then used as the POD to derive the IUR. The use of the AA PBTK model to
estimate the inhalation HEC in the derivation of the IUR based on the oral dose-response data is
justified because the model accounts for differences in acrylamide internal disposition following
exposure from the oral or the inhalation route of exposure.
Overall, it is unknown how inaccuracies in the AA PBTK model structure or parameter
values would impact the risks; further refinements in the model or new data could increase or
decrease the estimate of risks of cancer in humans
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
206 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 acrylamide metabolism nor human
variability in response to acrylamide has been well characterized. Factors that could contribute
to a range of human response to acrylamide 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 acrylamide 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
begin in less than 2-year-olds (U.S. EPA, 2005b, also see Section 5.4.6).
207 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-13. 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
acrylamide 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 2007 provide
limited to inadequate support for definitive statements. Animal
bioassays, however, clearly demonstrate multi-site
carcinogenicity, and provide good support for acrylamide 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
acrylamide 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 acrylamide and glycidamide (U.S.
FDA [2006b]).
208
DRAFT-DO NOT CITE OR QUOTE
-------
Consideration/
Approach
Impact on cancer risk
estimate
Decision
Justification
Selection of tumor
types, and relevance to
humans
A different selectionof
tumor types from the
Johnson et al (1986) study
could t or I the estimated
risks of acrylamide 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 and mammary gland 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
acrylamide related
cancer in humans.
A BMD analysis was used to fit to
the acrylamide 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. EPA cancer guidelines (U.S.
EPA, 2005a) was followed to calculate risks for multiple tumor
sites. The relatively similar results from the female or male rat
data for different approaches to calculating total risk increased
confidence in the results.
209
DRAFT-DO NOT CITE OR QUOTE
-------
Consideration/
Approach
Impact on cancer risk
estimate
Decision
Justification
UseofaPBTKmodel
in the derivation of the
CSF and the IUR
Alternative methods could
t or I the estimate of risks
to humans.
A PBTK model for acrylamide was
used to estimate the oral human
equivalent concentration in the
derivation of the CSF, and the
inhalation HEC in the derivation of
the IUR
The AA PBTK model was especially important to estimating the
oral human equivalent concentration in the CSF derivation
because the putative toxin was the acrylamide metabolite,
glycidamide. The default uncertainty factor for interspecies
toxicokinetic differences would not account for differences in the
internal levels of GA, while PBTK model could and did. The AA
PBTK model was also used to derive an inhalation HEC based on
the oral dose-response data, a value that would otherwise not
have been possible since there were no available inhalation data to
directly derive an IUR, and no credible alternate method for
conducting a route-to-route extrapolation.
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.
210
DRAFT-DO NOT CITE OR QUOTE
-------
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 acrylamide
metabolism nor human variability in response to acrylamide 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).
211
DRAFT-DO NOT CITE OR QUOTE
-------
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
at multiple sites in both sexes of rats and carcinogenic effects in a series of one-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 0.45 [mg/kg-day]"1 (rounded to 0.5 [mg/kg-day]"1) is based on different cancer
incidence data (combined incidence of thyroid tumors and tunica vaginalis mesotheliomas in
male rats), linear extrapolation from a point of departure determined by a benchmark dose
analysis, and the use of a PBTK model to estimate the human equivalent internal levels of GA,
the AA metabolite, as the dose metric. Glycidamide is considered to be the putative toxin for the
mutagenic MOA leading to carcinnogenicity, and thus a better internal dose metric to correlate
to response than the internal (or external) level of AA.
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 1.3 x 1CT4 (jig/m3)"1
is based on EPA's subsequent methodology for inhalation dosimetry (U.S. EPA, 1994, 1989), an
improved understanding of the toxicokinetics of aery 1 amide, and the use of a PBTK model to
conduct a route-to-route extrapolation (oral to inhalation) of the dose-response relationship
derived from the oral data, and to derive a human equivalent dose based on the internal level of
glycidamide.
212 DRAFT-DO NOT CITE OR QUOTE
-------
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
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). Acrylamide 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.
213 DRAFT-DO NOT CITE OR QUOTE
-------
Parallelogram Approach
The parallelogram approach was originally formulated by F.H. Sobels (1977, 1982, 1989)
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 5-1. 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).
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 5-1: 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 5-2. The first modification (Figure
5-2a) 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 5-2a. The EC/US EPA workshop particpants who evaluated this case study concluded,
however, that the modified parallelogram approach in Figure 5-2a was not relevant for AA,
214
DRAFT-DO NOT CITE OR QUOTE
-------
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
a)
In Vitro
Specific DNA adducts
Measured Mutations
Somatic In Vivo
Specific DNA adducts
Measured Mutations
Germ (mouse) in vivo
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 5-2: Two modifications in the parallelogram approach for estimating risk of heritable
germ cell effects from exposure to AA.
the parallelogram approach in Figure 5-2a. 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-2007.
A second parallelogram approach shown in Figure 5-2b 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
215
DRAFT-DO NOT CITE OR QUOTE
-------
DNA adduct formation. As with the approach in Figure 5-2a, however, the types of data needed
to implement the approach in Figure 5-2b are not available for AA. Specifically, the only
information on AA's effects in human somatic cells, is hemoglobin adduction (Bergmark et al.,
1993; Fennell et al, 2005; Boettcher et al., 2005), and GA induced unscheduled DNA synthesis
in human epithelial cells in vitro (Butterworth et al., 1992). Other deficiencies in the AA
database that preclude implementation of the Figure 5-2b 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 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 5-2b) 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 acrylamide, and to extrapolate risk from rodent (e.g.,
mice) experimental models to humans (ICPEMC, 1993a):
216 DRAFT-DO NOT CITE OR QUOTE
-------
Parameter
Locus specificity
DNA repair variability
Metabolic variability
Dose rate variability
Exposure route
Germ cell stage specificity
Dose-response kinetics
Overall REF
REF*
2
0.1
1
1
1
1
1
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.
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 (Generoso et al., 1996; Adler et al., 2000) compared to
AA, and is also the primary inducer of DNA-adducts in somatic cells (Besaratinia and Pfeifer,
2005). The acrylamide 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 1000 for dominant
single gene diseases, and 10 for dominant chromosomal diseases (i.e., this assumption represents
217
DRAFT-DO NOT CITE OR QUOTE
-------
another source of uncertainty). 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
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) derive a doubling dose in mice based on four data sets (Ehling and
Neuhauser-Klaus, 1992; Shelby et al., 1987; Adler et al., 1994; Adler, 1990) using the following
equation:
218 DRAFT-DO NOT CITE OR QUOTE
-------
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 7248,413)] / 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 non-linear 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
on a human background translocation frequency of 1.9 per 1000 newborns (Lyon et al., 1983).
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 5-14. In these
derivations, N is set at one million (1 x 106), 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 (Ehling and Neuhauser-Klaus, 1992; Shelby et al., 1987; Adler et al., 1994; Adler,
1990). 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 /106 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
219 DRAFT-DO NOT CITE OR QUOTE
-------
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 5-14 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/CERHR, 2004), charged with evaluating the evidence for acrylamide's adverse
reproductive and developmental effects, reviewed the Dearfield et al. (1995) quantitation of
heritable germ cell 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.
220 DRAFT-DO NOT CITE OR QUOTE
-------
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.
221 DRAFT-DO NOT CITE OR QUOTE
-------
Table 5-14. Heritable genetic risk estimates for humans exposed to acrylamide
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
Number of induced genetic diseases/106 offspring
Ingestion
1.3 x l(T5
mg/kg-
day
7.3 x 10~5
4.3 x 10~3
3.0 x 1Q-2
3.1 x 10~2
1.7 x 10~3
2.7 x 10~3
1.4 x 10~2
2.3 x 10~2
2.2 x 10~4
Inhalation
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
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
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
Dermal
0.016
mg/kg-day
grout
worker
0.09
5.3
3.7
38.2
2.0
3.3
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
227
2.2
"Ehling and Neuhauser-Klaus (1992).
bShelbyetal. (1987).
cAdleretal. (1994).
dAdler(1990).
Source: Dearfieldetal. (1995).
222
DRAFT-DO NOT CITE OR QUOTE
-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD
AND DOSE RESPONSE
6.1. HUMAN HAZARD POTENTIAL
Acrylamide (CASRN 79-06-1) has the chemical formula CsHsNO (structural formula
CH2=CH-CONH2) and a molecular weight of 71.08. Acrylamide (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. Acrylamide forms during
the high-temperature heating of starchy foods. Acrylamide 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.
223
-------
Impaired male reproductive performance (i.e., male-mediated implantation losses) has
been observed in laboratory animals orally exposed to AA, but the lowest dose levels associated
with this effect (-3-13 mg/kg-day) are generally higher than the lowest doses associated with
degenerative nerve changes. To date, associations between human exposure to AA and
reproductive effects have not been reported.
Two recent reviews of studies in mice for heritable germ cell effects from exposure to
AA have both concluded 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.
In accordance with the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a),
acrylamide is "likely to be carcinogenic to humans" based on findings of: (1) increased
incidences of thyroid tumors in male and female rats, scrotal sac mesotheliomas in male rats, and
mammary gland tumors in female rats in two drinking water bioassays; (2) AA initiation of skin
tumors following oral, i.p., or dermal exposure to AA and tumor promotion by TPA in two
strains of mice; and (3) increased incidence of lung adenomas in another mouse strain following
i.p. injection of AA. Evidence from available human studies is judged to be limited to
inadequate. No statistically significant increased risks for cancer-related deaths were found in
either of two cohort mortality studies of AA workers with the exception that, in an exploratory
dose-response analysis of the most comprehensive study, an increased risk for pancreatic cancer
was found in a subgroup with the highest cumulative AA exposure. In one case-control study,
no statistically significant associations were found between increased risks for large bowel,
kidney, or bladder cancer and frequent consumption of foods containing high or moderate levels
of AA.
224
-------
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 glycidamide 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 US 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 B6C3F1 mice. The proposed schedule for completion of these studies is as
follows: in-life phase complete by August 2007; pathology complete February 2008; Pathology
Working Group review by May 2008 (data will then become available on the NTP website);
statistics complete by August 2008; BSI Report preparation by May 2009; NTP Technical
Report Subcommittee approval 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
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
225
-------
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, to reduce the uncertainty in the
interspecies extrapolation factors 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.
Estimates of Risks from Other Organizations
Estimates of risk for acrylamide derived by other organizations are compiled by the
National Libraries of Medicine and can be found on the TOXNET webpage at
http://toxnet.nlm.nih.gov/cgi-bin/sis/htmlgen7iter . Additionally, the Joint FAO/WHO Expert
Committee on Food Additives (JECFA) information on acrylamide risk and toxicity is available
at: http://www.who.int/foodsafetv/chem/chemicals/acrylamide/en/.
6.2. DOSE RESPONSE
6.2.1. Noncancer/Oral
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
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
226
-------
selected as the POD for deriving the RfD. A PBTK model was used to derive an HEC based on
an internal dose metric of total AUC for AA in the blood and for a simulated drinking water
exposure. The HEC was 0.076 mg/kg-day, and 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 0.003 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 and complete by IRIS assessment standards. 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 acrylamide 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 sensitie endpoint. There might, however, be behavioral or functional
effects that were not evaluated in these bioassays that could have lower NOAELs than the
histological effects used to derive the RfD. There is also uncertainty as to low-does reponse
relationship for heritable germ cell effects. These two issues lower the confidence in the overall
RfD to medium to high. Some of these data needs are being addressed in on-going studies
sponsored by the NTP.
6.2.2. Noncancer/Inhalation
An inhalation RfC for acrylamide was derived by application of a PBTK model to
extrapolate the internal dose metric (AUC for AA in the blood) from an oral exposure in rat to an
oral exposure in humans and then to an equivalent inhalation exposure in humans, assuming
continuous inhalation exposure over 24 hours. 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 insufficient to describe dose-response
relationships for inhalation exposure. Justification for deriving an RfC from the oral RfD comes
from: (1) considerable evidence from occupational experience involving dermal and inhalation
exposure that AA-induced peripheral neuropathy (including development of the types of
degenerative lesions observed in nerves of rats exposed via drinking water) is a well-established
human health hazard; (2) evidence that tissue distribution in rats is similar following i.v., i.p.,
oral, dermal, and inhalation exposure to AA (Sumner et al., 2003; Kadry et al., 1999; Dow
Chemical Co., 1984; Miller et al., 1982; Hashimoto and Aldridge, 1970); (3) evidence that the
elimination kinetics of radioactivity from oral or i.v. administration of radiolabeled AA in rats
was similar (Miller et al., 1982); and (4) lack of support for portal-of-entry effects.
227
-------
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 0.25 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 0.008 mg/m3.
Since the RfC is based on a route to route extrapolation of the oral exposure data using
the same PBTK model to develop an inhalation HEC, the overall confidence in the RfC is similar
to that for the RfD, namely medium to high. The AA PBTK model provided a scientifically
supportable estimate of the inhalation HEC as the basis for the RfC, a reference value that would
otherwise have been unobtainable in the absence of adequate inhalation bioassay data.
6.2.3. Cancer/Oral
Two methods (incidence of tumor-bearing animals and summed risk) were considered for
estimating human cancer risk from male F344 rats bearing scrotal sac mesotheliomas or thyroid
follicular cell tumors (adenoma and carcinoma); or females bearing mammary gland tumors
(malignant and benign) or thyroid follicular cell tumors (adenoma and carcinoma), from a 2-year
drinking water rat bioassay (Friedman et al., 1995) in male. Both approaches yielded a similar
result for risks from multiple tumor sites when rounded to one significant digit. A linear
extrapolation to the origin, corrected for background, from a BMDL (as a POD) was used to
derive the oral slope factor. Support for a linear extrapolation comes from evidence of a
mutagenic MOA for AA, including observations of: (1) strong evidence of mutagenicity in 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 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 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 to rule out human relevance.
The POD for the derivation of the oral slope factor for cancer risks in humans exposed to
AA was the lowest BMDLio for male rats bearing TVM or thyroid tumors which was converted
to the human equivalent BMDLio by using a PBTK model. The human equivalent internal dose
is based on a metric of total AUC of GA in the blood. Glycidamide has been shown to be the
primary reactive mutagenic agent, and the total amount in blood is the most appropriate and
228
-------
supportable dose metric to use as a correlate to increased incidence of tumors. The resulting
HEC-BMDLio (0.22 mg/kg-day) was used to derive a human oral slope factor of 0.5 (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
evaluate differences in susceptibility, increased early-life susceptibility is assumed and the age-
dependent adjustment factors (ADAFs) should be applied to the slope factor, 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 study and medium confidence in the database. The principal 2-year study
(Friedman et al., 1995) corroborates earlier tumor incidence data (Johnson et al., 1986) for most,
but not all tumor types, and was a larger and better designed study. There remain, however,
uncertainties concerning the differences between the two study tumor types and incidence data,
in particular for the CNS tumors, and some controversy over the histopathological interpretation
of the male tunica vaginalis mesotheliomas. The database is also incomplete with only one
animal species tested, and little human data to support acrylamide's carcinogenic potential in
humans. At this time, the preponderance of evidence supports a mutagenic MOA with
insufficient evidence to confidently rule out the human relevance of the acrylamide-induced
tumors observed in the F344 rat bioassays (thyroid, mammary gland, and scrotal sac). Although
an alternate nonmutagenic MOA has been proposed involving hormonal pathway disruption for
tumors specific to Fischer 344 rats, supporting data are limited or nonexistent. Additional oral
cancer bioassay data and research into acrylamide's MOA(s) are needed to resolve these issues
and data needs.
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 available that simulates both oral and inhalation exposures and provides
the means to conduct a route-to-route extrapolation of the dose-response relationship from the
oral exposure data to what one might expect from an inhalation exposure in humans. The basis
of the extrapolation is an exposure (oral or inhalation) that yields a comparable internal dose
using the internal dose metric of area under the time-concentration curve for GA in blood. The
AA metabolite, GA, is considered to be the putative mutagen and most directly related to AA's
carcinogenicity.
229
-------
Support for the extrapolation of an inhalation unit risk from the oral data for AA comes
from: (1) evidence that tissue distribution in rats is similar following i.v., i.p., oral, dermal, and
inhalation exposure to AA (Sumner et al., 2003; Kadry et al., 1999; Dow Chemical Co., 1984;
Miller et al., 1982; Hashimoto and Aldridge, 1970); (2) the widespread distribution of AA and
the formation of GA adducts in diverse tissue throughout the body (Doerge et al., 2005a); (3)
evidence that the elimination kinetics of radioactivity from oral or i.v. administration of
radiolabeled AA in rats was similar (Miller et al., 1982); and (4) lack of support for portal-of-
entry effects. The AA PBTK model was especially important in estimating the internal levels of
GA, and in accounting for first-pass metabolism, one of the main factors that could lead to
different distributional profiles between an oral and an inhalation exposure.
An inhalation unit risk, calculated from data for adult exposures, is derived from the
BMDLio, the 95% lower bound on the exposure associated with an 10% extra cancer risk, by
dividing the risk (as a fraction) by the BMDLio. 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 are based on PBTK model simulations of the HEC for
intake from a continuous inhalation exposure that are comparable (based on an internal dose
metric of GA AUC in blood) to the oral exposure BMDio of 0.7 mg/kg-day and the BMDLio of
0.3 mg/kg-day. The PBTK model-derived oral exposure HEC-BMDLio is 0.22 mg/kg-day. The
air concentrations required to achieve these intake levels, assuming a continuous 24-hour
inhalation exposure for a 70 kg person who breathes 20 m3/day air are 1.84 mg/m3 for the HEC-
BMDio and 0.79 mg/m3 for the HEC-BMDLio. The resulting inhalation unit risk based on the
HEC-BMDLio air concentration as the POD is 1 x 10~4 (jig/m3)"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, 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 similar to that for the the oral slope
factor (see discussion in the previous section) since it is based upon the same study (medium
confidence) and database (medium confidence), and the same AA PBTK model to estimate an
inhalation HEC.
230
-------
7. REFERENCES
Abdelmagid, HM; Tabatabai, MA. (1982) Decomposition of acrylamide in soils. J Environ Qua! 11(4):701-704.
Abernethy, DJ; Boreiko, CJ. (1987) Acrylonitrile and acrylamide fail to transform C3H/10T1/2 cells. Environ
Mutagen9(suppl8):2.
Abou-Donia, MB; Ibrahim, SM; Corcoran, JJ; et al. (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; Moses, M, eds. Cold
Spring Harbor Laboratory Press: Cold Spring Harbor, NY. 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. (1990) Clastogenic effects of acrylamide in different germ-cell stages of male mice, in: J. Allen, B.
Bridges, M. Lyon, M. Moses and L. Russell (Eds.l, Biology of Mammalian Germ Cell Mutagenesis, Banbuty
Report Vol. 34, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp. 115-131.
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. (198 la) The effects of acrylamide treatment upon the dopamine receptor.
Toxicol Appl Pharmacol 58:89-99.
Agrawal, AK; Seth, PK; Squibb, RE; et al. (1981b) Neurotransmitter receptors in brain regions of acrylamide-
treatedrats. I: Effects of a single exposure to acrylamide. Pharmacol Biochem Behavior 14:527-531.
Albert Einstein College of Medicine. (1980) A Fetal Toxicity Study of Acrylamide in Rats. Wayne, NJ: 1980.
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.
231
-------
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.
Amador, A; Steger, RW; Bartke, A; et al. (1985) Testicular LH receptors during aging in F344 rats. J Androl 6:61-
64.
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):703-7.
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 J Ind 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. EnvironMol 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):241-8.
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; Ehrich, MF; et al. (2001) Metabolism, toxicokinetics and hemoglobin adduct formation in
rats following subacute and subchronic acrylamide dosing. NeuroToxicology 22:341-353.
Barfknecht, TR; Mecca, DJ; Naismith, RW. (1988) The genotoxic activity of acrylamide. EnvironMol Mutagen
ll(supplll):9.
Bartke, A; Sweeney, CA; Johnson, L; et al. (1985) Hyperprolactenemia inhibits development of Ley dig 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 ofDrosophila 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): 1023-9.
Besaratinia, A; Pfeifer, GP. (2005) DNA adduction and mutagenic properties of acrylamide. Mutat Res 580(1-
2):31-40.
Besaratinia, A; Pfeifer, GP. (2007) A review of mechanisms of acrylamide carcinogenicity. Carcinogenesis
(3):519-28.
Bio/Dynamics-Inc. (1979) A Fetal Toxicity Study of Acrylamide in Rats. EPA/OTS; Doc #878211679;
NTIS/OTS0206055
232
-------
Bishop, JB; Morris, RW; Seely, JC; et al. (1997) Alterations in the reproductive patterns of female mice exposed to
xenobiotics. Fund Appl Toxicol 40:191-204.
Bjellaas, T; Stolen, LH; Haugen, M; et al. (2007) Urinary acrylamide metabolites as biomarkers for short-term
dietary exposure to acrylamide. Food Chem Toxicol. 45(6): 1020-6.
Bjorge, C; Brunborg, G; Wiger, R; et al. (1996) A comparative study of chemically induced DNA damage in
isolated human and rat testicular cells. Reprod Toxicol 10:509-19.
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): 167-76.
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.
Bonassi, S; Lando, C; Ceppi, M; et al. (2004) No association between increased levels of high-frequency sister
chromatid exchange cells (HFCs) and the risk of cancer in healthy individuals.Environ Mol Mutagen. 43(2): 134-6.
Bondy, SC; Tilson, HA; Agrawal, AK. (1981) Neurotransmitter receptors in brain regions of acrylamide-treated
rats. II. Effects of extended exposure to acrylamide. Pharmacol Biochem Behavior 14:533-537.
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 144 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.
Calleman, CJ. (1996) The metabolism and pharmacokinetics of acrylamide: implications for mechanisms of
toxicity and human risk estimation. Drug Metab Rev 28(4) :527-590.
Calleman, CJ; Bergmark, E; Costa, LG. (1990) Acrylamide is metabolized to glycidamide in the rat: Evidence from
hemoglobin adduct formation. Chem Res Toxicol 3:406-412.
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.
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 on line at http://www.gpoaccess.gov/cfr/index.html
233
-------
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.
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.
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.
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. Mutat Res 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.
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.
de Olmos, JS; Beltramino, CA; de Olmos de Lorenzo, S. (1994) Use of an amino-cupric-silver technique for the
detection of early and semiacute neuronal degeneration caused by neurotoxicants, hypoxia and physical trauma.
Neurotox Teratol 16(6):545-561.
234
-------
Diembeck, W; Busing, 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 onbiogenic 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 Metabol Dispos Biol Fate Chem 10:196-197.
Dobrzynska, M; Lenarczyk, M; Gajewski, AK. (1990) Induction of dominant lethal mutations by combined X-ray -
acrylamide treatment in male mice. Mutat Res 232:209-15.
Doerge, DR; da Costa, GG; McDaniel, LP; et al. (2005a) DNA adducts derived from administration of acrylamide
and glycidamide to mice and rats. Mutat Res 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):258-67.
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(1) mice administered a single dose of acrylamide. Toxicol Lett. 169(l):34-42.
Donovan, JW; Pearson, T. (1987) Ingestion of acrylamide with severe encephalopathy, neurotoxicity and
hepatotoxicity. Vet Hum Toxicol 29:462.
Dow Chemical Co. (1984) Acrylamide: toxicodynamics in rats. Unpublished study submitted to EPA, Office of
Toxic Substances by Dow Chemical Co. OTS Fiche 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 in foods. Toxicological Sciences 75:7-15.
EC/US EPA Workshop (1995) Human Genetic Risks from Exposure to Chemicals, Focusing on the Feasibility of
the Parallelogram Approach. Edited by M.D. Waters and C. Nolan. Mutation Research [Fundamental of Molecular
Mechniasmof Mutagenesis] 330(1-2): 1-233.
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.
235
-------
Eskin, TA; Lapham, LW; Maurissen, JPJ; et al. (1985) Acrylamide effects on the macaque visual system. Invest
Ophthalmol Vis Sci 26:317-329.
FAO/WHO. (2007). The FAO/WHO Acrylamide in Food Network. Available from http://www.acrylamide-
food.org/.
Favor, J; Shelby, MD. (2005) Transmitted mutational events induced in mouse germ cells following acrylamide or
glycidamide exposure. MutatRes. 580(1-2):21-30.
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.
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):266-71.
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): 1328-37.
Garland, TO; Patterson, MW. (1967) Six cases of acrylamide poisoning. Brit Med J (4): 134-138.
Gassner, P; Adler, ID. (1996) Induction of hypoploidy and cell cycle delay by acrylamide in somatic and germinal
cells of male mice. MutatRes 367:195-202.
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(l):157-63.
236
-------
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.
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.
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.
Hagmar, L; Bonsai, S; Stromberg, U; et al. (1998) Chromosomal aberrations in lymphocytes predict human cancer:
a report from the European Study Group on Cytogenetic Biomarkers and Health (ESCH). Cancer Res. 58(18):4117-
21.
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): 157-65.
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):575-86.
Hansch, C; Leo, A; Hoekman, D. (1995) Exploring QSAR. Hydrophobic, electronic, and steric constants. In: ACS
Professional Reference Book. Heller, SR, ed. Amer Chem Soc, Washington, DC. 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 Tox and
Envion Health 42:343-356.
Harry, GJ. (1992) Acrylamide-induced alterations in axonal transport. Mol Neurobiol 6:203-216.
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.
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.
237
-------
Ho, WH; Wang, SM; Yin, HS. (2002) Acrylamide disturbs the subcellular distribution of GAB Aa receptor in brain
neurons. J Cellular Biochem 85:561-571.
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 13:167-78.
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.
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. Industrial Health
25:19-28.
IARC (International Agency for Research on Cancer). (1994a) Acrylamide. In: IARC Monographs Some Industrial
Chemicals; Summary of Data Reported and Evaluation. 60:389.
IARC. (1994b) Ethylene oxide, IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Some
Industrial Chemicals. 60:73-159.
IARC. (1999) Acrylonitrile, IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Part One: Re-
Evaluation of some Organic Chemicals, Hydrazine and Hydrogen Peroxide. 71:43-108.
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.
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.
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.
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. (2004) Monitoring database on acrylamide levels in food maintained by thelnstitute for Reference Materials
and Measurements (IRMM), together with Health and Consumer Protection DG. Available from:
http://www.irmm.jrc.be/.
JIFSAN. (2002). Acrylamide in Food Workshop: Scientific Issues, Uncertainties, and Research Strategies.
Available from: http://www.jifsan.umd.edu/acrylamide2002.htm.
238
-------
JIFSAN. (2004). Proceedings and Reports from the Joint Institute for Food Safety and Applied Nutrition (JIFSAN)
2004 Acrylamide in Food Workshop: Update - Scientific Issues, Uncertainties, and Research Strategies; April 13-
15, 2004; Chicago, Illinois. Available from: http://www.jifsan.umd.edu/acrylamide2004.htm.
JIFSAN. (2005). The Food Safety Risk Analysis Clearinghouse for Acrylamide. Operated by the Joint Institute for
Food Safety and Applied Nutrition (JIFSAN). Available from: http://www.foodrisk.org/acrylamide.cfm.
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. OTS 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. OTS 0206849.
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(l):402-7. 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.
Kaster, JK; Kamendulis, LM; Friedman, MA; et al. (1998) Syrian hamster embryo (SHE) cell transformation by
acrylamide and hormones. Toxicol Sci 42(1-S):A375.
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):253-74.
Klaunig, J. (2000) Cellular effects of acrylamide. Unpublished data cited in Environ, 2002.
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 and Chemical Toxicology 41:1569-1576.
239
-------
Krebs, O; Favor, J. (1997) Somatic and germ cell mutagenesis in lambda lacZ transgenic mice treated with
acrylamide or ethylnitrosourea. MutatRes 388:239-248.
Krishna, G; Theiss, JC. (1995) Concurrent analysis of cytogenetic damage in vivo: A multiple endpoint-multiple
tissue approach. EnvironMol 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 Qual
8(1):133-137.
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.
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 inbrainstem 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. John Wiley & Sons, Inc.: New York,
NY. p. 17.
Lide, DR, ed. (2000) CRC Handbook of Chemistry and Physics. 81st Edition. CRC Press LLC: Boca Raton, FL.
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):1221-38.
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.
240
-------
LoPachin, RM; Ross, JF; Reid, ML; Das, S; Masukhani, S; and Lehning, EJ (2002b) Neurological evaluation of
toxic axonopathies in rats: Acrylamide and 2,5-hexandione. NeuroToxicology 23: 95-110.
LoPachin, RM; Canady RA. (2004) Acrylamide Toxicities and Food Safety: Session IV Summary and Research
Needs. NeuroToxicology 25: 507-509.
LoPachin, RM, Schwarcz AI, Gaughan CL, Mansukhani S and Das S. (2004) In vivo and in vitro effects of
acrylamide on synaptosomal neurotransmitter uptake and release. NeuroToxicology 25:349-363.
LoPachin, RM and DeCaprio, AP. (2005) Protein adduct formation as a molecular mechanism in neurotoxicity.
Tox. Sci. 86(2): 214-225.
LoPachin, RM; Gavin, T; Geohagen, BC; Das, S. (2007) Neurotoxic mechanisms of electrophilic type-2 alkenes:
soft soft interactions described by quantum mechanical parameters.Toxicol Sci. 98(2):561-70.
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, MutationRes. 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.
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.
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.
Mayerhofer, A; Hemmings, 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. (Cited in
Iatropolous;etal., 1998).
Michels, KB; Rosner, BA; Chumlea, WC; et al. (2006) Preschool diet and adult risk of breast cancer. Int J Cancer
118(3):749-54.
Miller, MJ; Carter, DE; Sipes, IG. (1982) Pharmacokinetics of acrylamide inFischer-334 rats. Toxicol Appl
Pharmacol 63:36-44.
241
-------
Moore, MM; Amtower, A; Doerr, C; et al. (1987) Mutagenicity and clastogenicity of acrylamide in L5178Y 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. HormRes 41:36-40.
Moser, VC; Anthony, DC; Sette, WF; et al. (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):774-6.
Mucci; LA; Sandin, S; Baiter, K; et al. (2005) Acrylamide intake and breast cancer risk in Swedish women. JAMA
293(11): 1326-7.
Mucci, LA; Adami, HO; Wolk, A. (2006) Prospective study of dietary acrylamide and risk of colorectal cancer
among women. Int J Cancer 118(1): 169-73.
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.
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/CERHR (2004) National Toxicology Program / Center for the Evaluation of Risks to Human Reproduction
Expert Panel Report on Reproductive and Developmental Toxicity of Acrylamide. http://cerhr.niehs.nih.gov/.
Neuhauser-Klaus, A; Schmahl, W. (1989) Mutagenic and teratogenic effects of acrylamide in the mammalian spot
test. Mutat Res 226:157-162.
Neuman, F. (1991) Early indicators for carcinogenesis in sex-hormone-sensitive organs. Mutat Res 248:341-356.
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. Journal of Toxicology and Environmental Health 39:429-445.
242
-------
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 jV-methylolacrylamide treatment. MutatRes 516:101-111.
Paulsson, B; Kotova, N; Grawe, J; et al. (2003) Induction of micronuclei in mouse and rat by glycidamide,
genotoxic metabolite of acrylamide. Mutat Res 535(1): 15-24.
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):467-71.
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):223-34.
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 Agricul 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. MutatRes 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.
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.
243
-------
Sakamoto, J; Kurosaka, Y; Hashimoto, K. (1988) Histological changes of acrylamide-induced testicular lesions in
mice. Exp Mol Pathol 48:324-34.
Sankaranarayanan, K. (1982) Genetic Effects of Ionizing Radiation in Multicellular Eukaryotes and the Evaluation
of Genetic Radiation Hazards in Man, Elsevier, Amsterdam, 385 pp.
Satchell, PM; McLeod, JG. (1981) Megaoesophagus due to acrylamide neuropathy. J Neurol Neurosurg Psychiat
44:906-913.
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
Mol Mutagen 16:137-142.
Segerback, D; Calleman, CJ; Schroeder, JL; etal. (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: Induction of heritable translocations in male mice.
Environ Mutagen 9:363-368.
Shipp, A; Lawrence, G; Gentry, R; McDonald, T; Bartow, H; Bounds, J; Macdonald, N; Clewell, H; Allen, B; Van
Landingham, C. (2006) Acrylamide: review of toxicity data and dose-response analyses for cancer and noncancer
effects. CritRev Toxicol (6-7):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. (1991) Toxic neurofilamentous axonopathies and fast anterograde axonal transport.III. Recovery
from single injections and multiple dosing effects of acrylamide and 2,5-hexanedione. Tox and Applied Pharm
108:390-396.
Sickles, DW; Pearson, JK; Beall, A; et al. (1994) Toxic axonal degeneration occurs independent of neurofilament
accumulation. JNeurosciRes 39(3):347-354.
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, D; Friedman, MA. (2002a) Fast axonal transport: a site of acrylamide neurotoxicity: a rebuttal.
NeuroToxicology 23:265-270.
Sickles, DW; Stone, JD; Friedman, MA. (2002b) Fast axonal transport: A site of acrylamide neurotoxicity?
NeuroToxicology 23:223-251.
Smith, BL. (1991) Fibroadenomas. In: Breast Diseases. Harris, JR; Hellman, S: Henderson, 1C; et al., eds. JP
Lippincott Company: Philadelphia. pp34-37.
244
-------
Smith CJ; Perfetti TA; Rumple MA; et al. (2000) "IARC group 2A Carcinogens" reported in cigarette mainstream
smoke. Food ChemToxicol 38(4):371-83.
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 from
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, Mutation Res., 46, 245-260.
Sobels, F.H. (1982) The parallelogram: an indirect approach for the assessment of genetic risks from chemical
mutagens, in: K.C. Bora, G.R. Douglas andE.R. Nestman (Eds.l, Progress in Mutation Research, Vol. 3, pp. 323-
327.
Sobels, F.H. (1989) Models and assumptions underlying genetic risk assessment, Mutation 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):3465-70.
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. (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 from Maillard reaction products. Nature 419:449.
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 Pharm 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.
Stone, JD; Peterson, AP; Eyer, J; et al. (2001) Neurofilaments are nonessential to the pathogenesis of toxicant-
induced axonal degeneration. J Neurosci 21(7):2278-2287.
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.
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.
245
-------
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):260-70.
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 and Chemical
Toxicology 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? ChemRes 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.
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 aerylonitrile. Toxicol Lett 128(l-3):249-55.
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.
Titenko-Holland, N; Ahlborn, T; Lowe, X; et al. (1998) Micronuclei and developmental abnormalities in 4-day
mouse embryos after paternal treatment with acrylamide. Environ Mol Mutag 31:206-217.
Tong, GC; Cornwell, WK; Means, GE. (2004) Reactions of acrylamide with glutathione and serum albumin.
Toxicol Lett 147(2): 127-31.
Tripathy, NK; Patnaik, KK; Nabi, MJ. (1991) Acrylamide is genotoxic to the somatic and germ cells ofDrosophila
melanogaster. Mutat Res 259:21-27.
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.
Tyl, RW; Friedman, MA. (2003) Effects of acrylamide on rodent reproductive performance. Reprod Toxicol 17:1-
13.
246
-------
Tyl, RW; Friedman, MA; Losco, PE; et al. (2000a) Rat two-generation reproduction and dominant lethal study of
acrylamide in drinking water. Reprod Toxicol 14:385-401.
Tyl, RW; Marr, MC; Myers, CB; et al. (2000b) Relationship between acrylamide reproductive and neurotoxicity in
male rats. Reprod Toxicol 14:147-157.
Twaddle, NC; McDaniel, LP; Gamboa da Costa, G; et al. (2004) Determination of acrylamide and glycidamide
serum toxicokinetics inB6C3Fl 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.
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 (U.S. Environmental Protection Agency). (1986b) Guidelines for mutagenicity risk assessment. Fed Reg
51(185):34006-34012.
U.S. EPA. (1988) Recommendations for and documentation of biological values for use in risk assessment. EPA
600/6-87/008. Available from: National Technical Information Service, Springfield, VA; PB88-179874/AS.
U.S. EPA. (1991) Guidelines for developmental toxicity risk assessment. FedReg56(234):63798-63826.
U.S. EPA. (1994) Methods for derivation of inhalation reference concentrations and application of inhalation
dosimetry. EPA/600/8-90/066F. Available from: National Technical Information Service, Springfield, VA;
PB2000-500023, and http//:www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (1995) Use of the benchmark dose approach in health risk assessment. U.S. Environmental Protection
Agency. EPA/630/R-94/007. Available from: National Technical Information Service (NTIS), Springfield, VA;
PB95-213765, and http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (1996) Guidelines for reproductive toxicity risk assessment. Fed Reg 61(212):56274-56322.
U.S. EPA. (1998a) Guidelines for neurotoxicity risk assessment. Fed Reg 63(93):26926-26954.
U.S. EPA. (1998b) Science policy council handbook: peer review. Prepared by the Office of Science Policy, Office
of Research and Development, Washington, DC. EPA 100-B-98-001. Available from: National Technical
Information Service, Springfield, VA; PB98-140726, and http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (1998c) Assessment of Thyroid Follicular Cell Tumors. Risk Assessment Forum. Office of Research
and Development, Washington, DC. EPA/630/R-97/002.
U.S. EPA. (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-OOl. Available from:
http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (2000b) Science policy council handbook: risk characterization. Prepared by the Office of Science
Policy, Office of Research and Development, Washington, DC. EPA 100-B-00-002. Available from:
http://www.epa.gov/iris/backgr-d.htm
247
-------
U.S. EPA (2000c) Benchmark dose technical guidance document [external review draft]. EPA/630/R-00/001.
Available from http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (2000d) Supplementary guidance for conducting for health risk assessment of chemical mixtures. Risk
Assessment Forum, Washington, DC; EPA/630/R-00/002. Available from: .
U.S. EPA. (2002) A review of the reference dose and reference concentration processes. Risk Assessment Forum,
Washington, DC; EPA/630/P-02/0002F. Available from http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (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 (2005a) Guidelines for carcinogen risk assessment. Risk Assessment Forum, Washington, DC;
EPA/630/P-03/001B. Available from: http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid= 116283,
U.S. EPA (2005b) Supplemental Guidance for Assessing Susceptibility from Early Life Exposure to Carcinogens.
Risk Assessment Forum, Washington, DC; EPA/630/R-03/003F.. Available from:
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm? deid=l 16283.
U.S. FDA (2006a) The US FDA 2006 Exposure Assessment for Acrylamide (slide presentation by Dr. Michael
DiNovi). Available from http://www.cfsan.fda.gov/~dms/acryexpo.html.
U.S. FDA (2006b) The US FDA / Center for Food Safety and Applied Nutrition (CFSAN) site for information on
acrylamide and FDA research. Available from http://www.cfsan.fda.gov/~lrd/pestadd.html#acrylamide.
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):20-4.
Verschueren, K. (2001) Handbook of Environmental Data on Organic Chemicals, 4th edition, Vol. 1-2. John Wiley
& Sons, Inc.: New York, NY. p. 124.
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 58:363-9.
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 from:
http://www.inchem.org/documents/ehc/ehc/ehc49.htm
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.
248
-------
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 2:215-20.
Xiao, Y; Tates, 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):388-99.
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.
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 Toxicity: A Risk Assessment
Approach. Chapter 27 in Principles and Methods of Toxicology, 3rd edition, edited by A. Wallace Hayes. Raven
Press, New York. pp947.
249
-------
APPENDIX A. Summary of External Peer Review and Public Comments and Disposition
A-l 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
E. coli WP2 uwA~
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
0.5-5000 ug/plate
± S9 activation
2-10 mg/mL
100
10,000
100
50
5
1
5000
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
Jungetal., 1992;
Mtilleretal., 1993
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
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
0.25-5 mM
larvae feeding
50
1.0
1
1
1.0
Negative
Positive
Weakly positive
Positive
Positive
Knaapetal., 1988
Tripathy et al., 1991
Knaapetal., 1988
Batiste-Alentorn et al.,
1991
Tripathy et al., 1991
B-l
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
Dose/Concentration
HID or
LEDb
Result
Reference
Mammalian gene mutation assays in vitro
Mouse lymphoma
L5178YTK+/-,tk locus
Mouse lymphoma
L5178YTK+/-,tk locus
Mouse lymphoma
L5 178Y TK+A, tk and HPRT
loci
Mouse lymphoma
L5178Y TK+/~, HPRT locus
Chinese hamster V79H3 cells,
HPRT locus
10 mM
0-0.85 mg/mL
without 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
10
0.5
0.3
7
Positive (more pronounced without
activation)
Positive
Equivocal, increases only at
cytotoxic concentrations
Positive
Negative
Barfknechtetal.,
1988
Moore etal., 1987
Knaapetal., 1988
Knaapetal., 1988
Tsudaetal., 1993
Mammalian gene mutation assays in vivo
Transgenic mouse
liver ell,
lymphocyte
HPRT
Transgenic mouse
lacZ
Mouse spot test
Morphological
specific locus
Big Blue Mouse
(M,F)
Muta® Mouse
Muta® Mouse
Mouse embryos
(T x HT)P!
Mouse (C3H/R1 x lOl/R^Fj
(M)
Mouse (102/E1 x CSH/E^Fj
(M)
100, 500 mg/L
AA or GA
Drinking water for
3-4 weeks
5 x 50 mg/kg-day
i.p. injection
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
100
(est. 19-25
mg/kg-
day)
5 x 50
100
50
3x50
50
100
Positive
Weakly positive, no statistical
analysis
Negative
Positive
Positive
Positive (postspermatogonia)
Positive (postspermatogonia;
spermatogonia)
Manjanatha et al.,
2006
Hoornetal., 1993
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
Chinese hamster cells
Chinese hamster cell line (V79)
0.5-5 mM
no activation used
0.1-3 mg/mL
± S9 activation
2
1
Positive
Positive, with or without metabolic
activation
Tsudaetal., 1993
Knaapetal., 1988
B-2
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Cell division
aberration
Chromosome
enumeration
Polyploidy
Spindle
disturbances
Micronucleus
Test system"
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)
Dose/Concentration
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
HID or
LEDb
0.75
0.2
0.01
0.5
1
0.01
50
Result
Positive
Positive
Positive
Positive
Positive
Positive
Negative
Reference
Moore etal, 1987
Warretal., 1990
Warretal., 1990
Warretal., 1990
Tsudaetal., 1993
Adleretal., 1993
Lahdetie et al., 1994
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)
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
50
100
200
78
100
125
100
150
125
5 x50
Positive
Positive
Negative
Negative
Negative
Negative
Negative
Negative
Negative
Negative
Adleretal., 1988
Cihak and
Vontorkova, 1988
Shiraishi, 1978
Shiraishi, 1978
Krishna and Theiss,
1995
Backer etal., 1989
Kligerman et al., 1991
Adleretal., 1988
Backer etal., 1989
Adler, 1990
B-3
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Polyploidy or
aneuploid
Spindle
disturbances
Micronucleus
Test system"
Mouse (102/E1 x CSH/E^Fj
(spermatocytes)
Mouse (CFO
(first cleavage embryos)
Mouse (B6C3F1) (M)
(first cleavage one-cell 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
Dose/Concentration
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
100-200 mg/kg
i.p. injection
500 ppm in the diet
for 7 to 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
HID or
LEDb
100
150
75
100
78
120
50
100
25
136
M: 42.5
F:55
100
100
50
25
Result
Positive
Positive in embryos from which the
males had mated 6-8 days
following treatment (early
spermatozoa stage)
Positive
Positive
Positive
Negative
Positive
Positive
Positive
Positive
Positive
Negative
Negative
Positive
Positive, but results were not
analyzed statistically
Reference
Adler, 1990
Valdiviaetal., 1989
Pacchierotti et al.,
1994
Shiraishi 1978
Shiraishi 1978
Adler etal., 1993
Adler etal., 1988
Cihak and
Vontorkova, 1988
Cihak and
Vontorkova, 1988
Knaapetal., 1988
Cihak and
Vontorkova, 1990
Paulsson etal., 2002
Krishna and Theiss,
1995
Russoetal., 1994
Paulsson etal., 2002
B-4
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"
Mouse (CBA)
reticulocytes
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
Rat (Sprague-Dawley)
spermatids
Mouse (C57BL/J6) (M)
germ cells
Mouse (C57BL/J6) (M)
germ cells
Mouse (C3H x lO^Fj (M)
Mouse (C3H/E1) (M)
Mouse (C3H/E1) (M)
Mouse (C3H/E1) (M)
Dose/Concentration
0.18,0.35,0.70
mmol/kg; i.p.
injection
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
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
0.35
0.7
50
100
50
50
100
4 x50
150
50
40
50
50
50
Result
Positive, but results were not
analyzed statistically
Positive, but nonmonotonic,
probably due to toxiciry at high
dose
Positive
Positive
Positive
Positive
Positive
Positive
Negative
Weakly positive, asynapsis in
meiotic prophase
Positive
Positive
Positive
Positive
Reference
Paulssonetal.,2003
Paulssonetal.,2003
Backer etal., 1989
Kligermanetal., 1991
Collins et al., 1992
Russoetal., 1994
Xiao and Tates, 1994
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
B-5
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Spore rec assay
DNA breakage
In vitro UDS
In vivo/m vitro
UDS
In vivo UDS
DNA adducts
Test system"
Bacillus subtilis
H17(rec+)andM45(rec-)
Mouse (C3H x CSVBL/IO^
(M)
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
Mouse (C3H x BLIO^
testis
Mouse (C3H x BLIO^ (M)
liver
Rat (Sprague-Dawley)
liver, lung, kidney, brain, testis
Mouse (BALB/c)
liver, kidney, brain
Dose/Concentration
1-50 mg/disk
25-125 mg/kg
i.p. injection
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
46 mg/kg
i.p. injection
46 mg/kg
i.p. injection
46 mg/kg
i.p. injection
53 mg/kg
i.p. injection
HID or
LEDb
10
25
17.5
1
1
1 x 100
5x30
5x30
7.8
46
46
46
53
Result
Positive
Positive
Weakly positive
Negative
Positive
Negative
Positive
Positive
Positive
Positive
Positive
Positive
Reference
Tsudaetal., 1993
Sega and Generoso,
1990
Barfknechtetal.,
1988
Butterworthetal.,
1992
Butterworth et al.,
1992
Butterworth et al.,
1992
Butterworth et al.,
1992
Segaetal., 1990
Segaetal., 1990
Segaetal., 1990
Segerbacketal., 1995
Segerbacketal., 1995
Sister chromatic! exchange
In vitro
In vivo
Chinese hamster V79 cells
Chinese hamster V79 cells
Mouse (C57BL/6J) (M)
spleen lymphocytes
Mouse (C57BL/6) (M)
splenocytes
0.1-lmg/mL
± S9 activation
0.5-2.5 mM
no activation used
50-125 mg/kg
i.p. injection
100 mg/kg
i.p. injection
0.3
1
50
100
Positive at 0.3 mg/mL without S9
and 1.0mg/mLwithS9
Positive
Positive
Positive
Knaapetal., 1988
Tsudaetal., 1993
Backer etal., 1989
Kligermanetal., 1991
DRAFT-DO NOT CITE OR QUOTE
-------
Table B-l. Results of acrylamide mutagenicity testing
Assay
Test system"
Mouse (BALB/c)
differentiating spermatogonia
Dose/Concentration
50-100 mg/kg
HID or
LEDb
50
Result
Positive
Reference
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., 1993
Germ cell effects
Sperm head DNA
alkylation
Sperm head
protamine
alkylation
Sperm head
abnormalities
Mouse (C3H x 10 1^
Mouse (C3H x 10 1^
Mouse (ddY)
125 mg/kg
i.p. injection
125 mg/kg
i.p. injection
0.3-1.2 mM in
drinking water for 4
weeks
125
125
1.2
Weakly positive
Positive
Positive
Segaetal., 1989
Segaetal., 1989
Sakamoto and
Hashimoto, 1986
aM = male, F = female.
bHID, highest ineffective dose/concentration for negative tests; LED, lowest effective dose/concentration for positive tests.
B-7
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
lower than a 10% extra risk was selected for the following reasons (1) the BMDLs 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 BMDLs is consistent with the
technical guidance for benchmark dose analysis, which states that "while it is important to
always report EDi0s and LEDi0s for comparison purposes, the actual 'benchmark dose' used as a
POD may correspond to response levels below (or sometimes above) 10%, although for
convenience standard levels of 1%, 5%, or 10% have typically been used."
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
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
16/61C
26/49c
-
3.0
-
-
-
38/86c
aTwo 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 chi-square 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.
C- 1 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
%2 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
"Slope restricted to >1.
bRestrict power >1.
'Restrict betas >0, degree of polynomial = 4.
Data source: Johnson etal. (1986).
Log-Logistic IVbdel with 0.95 Confidence Level
0.4
0.35
5 °-3
| 0.25
5 0.2
=8
20.15
0.1
0.05
Log-Logistic
BMD
0
0.5
15:0906/062003
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.
Data source: Johnson et al. (1986).
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/>-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
"Slope restricted to >1.
bRestrict power >1.
'Restrict betas >0, degree of polynomial = 3.
Data source: Johnson etal. (1986).
Probit Model with 0.95 Confidence Level
0.4
0.35
2 0.25
1 0.15
^ 0.1
0.05
0
r i uui i
; , - ^// ^["
" li J^-^
' \-^~~^~\
- i} J-
BMDL
^—"^ ~L
BMD
0 0.5 1 1.5 2
dose
19:0506/102003
Figure C-2. Observed and predicted incidences for nerve changes in female
rats exposed to acrylamide in drinking water for 2 years.
Data source: Johnson et al. (1986).
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
(LED,)
0.05
0.11
Data 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 chi-square 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.
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.
Data source: Friedman etal. (1995).
Quantal Quadratic Model with 0.95 Confidence Level
0.7 : Quantal Quadratic
0.6
0.5
0.4
0.3
0.2
BMDL
BMD
0.5
15:3006/062003
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.
Data source: Friedman et al. (1995).
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
X2/>-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.
Data source: Friedman etal. (1995).
0.6
0.5
S 0.4
a
^ 0.3
o
'"S 0.2
Ll_
o
LJ.
Garrma Multi-hit Model wth 0.95 Confidence Level
Gamma Multi-hit
T
1
1.5
dose
2.5
0 0.5
18:5406/102003
Figure C-4. Observed and predicted incidences for nerve changes in female rats
exposed to acrylamide in drinking water for 2 years.
Data source: Friedman et al. (1995).
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
(LED,)
0.25
0.09
aRestrictpower>l.
Data source: Friedman etal. (1995).
C- 7 DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX D. DOSE-RESPONSE MODELING FOR CANCER
METHODS
Data: Tumor data from the latest 2-year bioassay with F344 rats (Friedman et al., 1995)
were modeled to obtain potential points of departure for deriving an oral slope factor and
inhalation unit risk (Table D-l). An earlier study (Johnson et al., 1986) had abnormally high
CNS and oral cavity tumors in control males, and possible confounding effects from a viral
infection. The new cancer bioassay was subsequently conducted with a larger number of rats
(increased statistical power) and dose spacing that improved the characterization of the dose-
response relationships for thyroid tumors, mammary gland tumors, and TVM. Tumor data from
Friedman et al. (1995) were expected to provide a more reliable characterization of dose-
response relationships and were selected for dose-response modeling analysis.
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
Reference/tumor type
Friedman et al., 1995/males"
Follicular cell adenoma/carcinoma
Tunica vaginalis mesotheliomab
Friedman et al., 1995/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
27/948J
2.0
17/751
13/751
-
3.0
-
23/1001
30/95J
48/95^
aTwo 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.
8Three 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).
D-l
DRAFT-DO NOT CITE OR QUOTE
-------
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, 2005).
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.
The multistage model is given by:
P(d) = 1 exp[-(q0 + qid + q2d2 + . . . + qtf?)],
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 (Benchmark Dose
Software, version 1.3.2; U.S. EPA, 2001) was used for all multistage model fits.
The multistage-Weibull model is given by:
P(d,t) = 1- exp[-(q0 - qid- q2d. . . - 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 to, 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
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
D-2 DRAFT-DO NOT CITE OR QUOTE
-------
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-day)"1 (see Table D-2).
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 0.1/(0.94 mg/kg-day), or 0.11 (mg/kg-day)"1 (see Table D-2).
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
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
Slope factor"
(mg/kg-day)'1
0.13
0.11
0.23
aSlope 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.
Data 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 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, 2005) 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
D-3
DRAFT-DO NOT CITE OR QUOTE
-------
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 approach involves adding
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 0.27(0.88 mg/kg-day),
or 0.23 (mg/kg-day)"1, approximately two-fold 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, 2005; 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:
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.
D-4 DRAFT-DO NOT CITE OR QUOTE
-------
2. The models previously fitted to estimate the BMCs and BMCLs were used to extrapolate
to a low level of risk (R), as low as 1CT6 risk, in order to reach the region of each
estimated dose-response function where the slope was reasonably constant and upper
bound estimation was still numerically stable.10 The unit risk for each site was then
estimated by R/BMDLR, as for the estimates for each tumor site above.
3. 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 two sites for female F344
rats.
4. 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.
Table D-3 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)"1 compared with 0.13
(mg/kg-day)"1 using the POD approach (Table D-2), and both methods lead to the same slope
factor for thyroid tumors, 0.11 (mg/kg-day)"1.
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
10 Although this step appears to differ from the explicit recommendation of the cancer guidelines (U.S.
EPA, 2005) to estimate cancer risk from a point of departure near the lower end of the observed range, without
significant extrapolation to lower doses, this method is recommended in the guidelines as a method for combining
multiple extrapolations. For this purpose, a quantitative combination of individual risks within the range of
observation is not generally practicable. More significantly, numerical combination of risks in the range of
observation does not generally lead to a numerically unique result, due to different dose-response relationships.
When risk is expected to be low-dose linear, the approach followed here leads to the most reliable estimate of the
summed risk.
D-5 DRAFT-DO NOT CITE OR QUOTE
-------
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 0.23 (mg/kg-day)"1, and the summed central
tendency was 0.17 (mg/kg-day)"1, about a 1.4-fold difference (Table D-3). 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 two-fold higher than either of the individual risks. For these data, the two approaches
yield very similar results.
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
Tumor site
Mammary
(benign and malignant)
Thyroid
(adenomas and carcinomas)
BMDR
(mg/kg-day)
1.1 x 1(T4
1.2 x 1(T4
BMDLR
(mg/kg-day)
7.4 x 1(T5
8.9 x 1(T5
Risk of either mammary or thyroid tumors
Central tendency
oral potency"
(mg/kg-day)"1
8.9 x 1(T2
8.1 x 1(T2
0.17
Upper bound on
lifetime extra risk
(mg/kg-day)"1
0.14
0.11
0.23b
aCentral tendency oral potency = R/BMDR, where R = 1 x 10 5. The combined central tendency risk is the sum of
the individual oral potencies.
bThe 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 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 etal. (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.
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
D-6
DRAFT-DO NOT CITE OR QUOTE
-------
slope factor associated with this site, using the POD approach, is 0.17(0.75 mg/kg-day), or 0.13
(mg/kg-day)"1 (see Table D-4).
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-day)"1 (see Table D-4).
Table D-4. 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 tumorsb
BMDRa
(mg/kg-day)
1.2
0.71
0.70
BMDLRa
(mg/kg-day)
0.75
0.45
0.30
Slope factor
(risk level/BMDL)
(mg/kg-day)"1
1.3 x 1Q-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 the NRC (1994) approach, the slope factor was 0.34 (see discussion below).
Data source: Friedman etal. (1995).
The first recommended method in the EPA cancer guidelines for assessing total risk from
multiple tumor sites (U.S. EPA, 2005; 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,7 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
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-5 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-4), 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-2). 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
D-7 DRAFT-DO NOT CITE OR QUOTE
-------
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-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
Tumor site
TVM
Thyroid
(adenomas and carcinomas)
BMDR
(mg/kg-day)
1.1 x 1(T5
6.7 x 1Q-6
BMDLR
(mg/kg-day)
7.1 x 1(T6
4.3 x 1(T6
Risk of either TVM or thyroid tumors
Central tendency
oral potency"
(mg/kg-dayjr1
8.8 x 1(T2
0.15
0.24
Upper bound on
lifetime extra risk
(mg/kg-day)"1
0.14
0.23
0.34b
aCentral tendency oral potency = R/BMDR, where R = 1 x 10 6. The combined central tendency risk is the sum of
the individual oral potencies.
bThe 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 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.
Data 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 0.34 (mg/kg-day)"1, and the summed central
tendency was 0.24 (mg/kg-day)"1, about a 1.4-fold difference (Table D-5). The estimated risk
for thyroid tumors was the more variable, contributing about 70% 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
human extra cancer risk from continuous, lifetime oral exposure to acrylamide is
0.3 (mg/kg-day)"1, rounding the summed risk for male rats above to one significant digit.11 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.
11 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, andy = 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 the NRC (1994) approach.
D-8 DRAFT-DO NOT CITE OR QUOTE
-------
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, 2005). 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 species 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.
D-9 DRAFT-DO NOT CITE OR QUOTE
-------
DATA PRINTOUTS FOR BMD MODELING
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:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l-beta2*dose^2)]
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
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.131573
Beta(l) = 0.0827018
Beta(2) = 0
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
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 -0.71
Beta(l) -0.71 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.124597 0.0835445 -0.0391471 0.288342
Beta(l) 0.0887157 0.0565531 -0.0221264 0.199558
Beta(2) 0 NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
D-10 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 -143.354 3
Fitted model
Reduced model
AIC:
-143.609
-149.278
291.219
0.51136 1
11.8483 2
0.4746
0.002674
Dose
Goodness of Fit
Scaled
Est._Prob. Expected Observed Size
Residual
0.0000
1.0000
3 .0000
0.1246
0.1989
0.3292
11.961
18.698
31.270
11
21
30
96 -0.297
94 0.595
95 -0.277
= 0.52
d.f. = 1
P-value = 0.4713
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD =
BMDL =
1.18762
0.776448
Specified effect = 0.0001
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.00112725
BMDL = 0.000736981
Specified effect = le-005
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.00011272
BMDL = 7.36948e-005
Specified effect = le-006
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.12726-005
BMDL = 1.127176-005
D-11 DRAFT-DO NOT CITE OR QUOTE
-------
0.45
0.4
0.35
£ 0.3
si
< 0.25
o
0.2
03
0.15
0.1
0.05
Multistage Model with 0.95 Confidence Level
Multistage
0
BMDL
BMD
0.5
1.5
dose
2.5
11:3206/052006
Figure D-l. Observed and predicted incidences for mammary gland tumors
in female rats exposed to acrylamide in drinking water for 2 years.
Source: Friedman et al. (1995).
D-12
DRAFT-DO NOT CITE OR QUOTE
-------
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
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l-beta2*dose^2)]
The parameter betas are restricted to be positive
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
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0220015
Beta(l) = 0.0800466
Beta(2) = 0
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
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 -0.71
Beta(l) -0.71 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0204321 0.0851609 -0.14648 0.187344
Beta(l) 0.0813062 0.0507808 -0.0182223 0.180835
Beta(2) 0 NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
D-13 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
AIC:
196.495
Dose
Goodness of Fit
Scaled
Est._Prob. Expected Observed Size
Residual
0.0000
1.0000
3 .0000
0.0204
0.0969
0.2325
2 .043
9.693
23 .246
2
10
23
100
100
100
= 0.02
d.f. = 1
P-value = 0.9021
-0.031
0.104
-0.058
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
BMD = 0.000122993
BMDL = 8.93171e-005
Specified effect = 0.0001
Risk Type = Extra risk
Confidence level = 0.95
BMD = 0.00122998
BMDL = 0.000893211
Specified effect = le-006
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1.229926-005
BMDL = 1.216636-005
D-14 DRAFT-DO NOT CITE OR QUOTE
-------
0.35
0.3
0.25
T3
CD
I °2
o 0.15
03
0.1
0.05
0
Multistage Model with 0.95 Confidence Level
Multistage
BMDL
BMR
0.5
1.5
dose
2.5
11:3806/052006
Figure D-2: Observed and predicted incidences for thyroid tumors in female
rats exposed to acrylamide in drinking water for 2 years.
Source: Friedman et al. (1995).
D-15
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
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l-beta2*dose^2)]
The parameter betas are restricted to be positive
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
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.130609
Beta(l) = 0.188994
Beta(2) = 0
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
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 -0.67
Beta(l) -0.67 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.127194 0.0836998 -0.0368541 0.291243
Beta(l) 0.192236 0.0612613 0.0721662 0.312306
Beta(2) 0 NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
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 -158.381 3
Fitted model -158.4 2 0.0370709 1 0.8473
Reduced model -175.349 1 33.9343 2 <.0001
AIC:
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.1272 12.211 12 96 -0.065
1.0000 0.2798 26.305 27 94 0.160
3.0000 0.5097 48.422 48 95 -0.087
i^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
D-17 DRAFT-DO NOT CITE OR QUOTE
-------
0.6
0.5
T3
0>
I °'4
0.3
o
03
0.2
0.1
Multistage Model with 0.95 Confidence Level
Multistage
BMD
BMD
0.5
1.5
dose
2.5
12:51 06/142006
Figure D-3: Observed and predicted incidences for mammary or thyroid
tumors in female rats exposed to acrylamide in drinking water for 2 years.
Source: Friedman et al. (1995).
D-18
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
D-19 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
D-20 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
CONFIDENCE
LOWER BOUND UPPER BOUND LIMIT
RISK MLEDOSE ON DOSE ON RISK INTERVAL TIME
.100000 1.20015 .747920 .155548 95.0% 108.000
1.000000E-03 1.139660E-02 7.102226E-03 1.604166E-03 95.0% 108.000
1.000000E-06 1.139090E-05 7.116445E-06 1.600645E-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 4.294526E-02 6.801233E-02 95.0% 108.000
2.00000 .161029 .252245 95.0% 108.000
NORMAL COMPLETION!
D-21 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
D-22
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 11 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 11 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 3 1 101.0 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 11 61.0 1 1
67.0 2 1 72.0 1 1
D-23 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!
D-24 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
D-25 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
D-26 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
D-27 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!
D-28 DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX E. KIRMAN ET AL. (2003) PBTK MODEL SUPPORTING
DOCUMENTATION
This appendix contains supporting documentation for the orginal Kirman et al. (2003)
PBTK model for acrylamide and metabolites, and for the EPA recalibrated model using more
recent data for rats and humans. The model schematic and original model parameter values
presented below were extracted directly from the original article — 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):253-74
The reader is referred to the orginal text for further details about the original Kirman et
al. (2003) model development and simulation results including the mass-balance equations. The
original Kirman et al (2003) ACSL code and command files, and the updated code in acslXtreme
are available from—[Note: website address or email address will be included in the final draft].
The diagram below of the Kirman et al. (2003) model illustrate 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. Different routes of exposure to AA are
represented in the Kirman et al. model including intravenous (i.v.), intraperitoneal (i.p.), oral
gavage, oral drinking water, and inhalation. Metabolism of AA and GA are represented only in
the liver.
Diagram of the Kirman et al. (2003) PBTK Model for Acrylamide and Glycidamide
Acrylamide Glycidamide
* I Ainu
-ung
T3
O
' 5
"I
"3
* LUIIE,
Tissues
r
^ Liver —
»L. - L
Liver
f
E-l
DRAFT-DO NOT CITE OR QUOTE
-------
The model parameter values and sources in the original Kirman model are presented in
Table E-l. The 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). The urinary metabolite data of Summer et al. (1992) was largely used
to calibrate the balance between different routes of metabolism, while the overall rate of
metabolism was calibrated to AA concentrations in blood and nervous tissue (Raymer et al.,
1993) and total radioactivity levels (Ramsey et al., 1984; Miller et al., 1982). Kirmal et al.
developed a single set of model parameter values that best fit these kinetic data. Optimized
values of Vmax =1.6 mg/hour-kg and Km =10 mg/L were determined for the metabolism of AA
via cytochrome P-450. The second-order rate constant for metabolism of AA via reaction with
GSH was best fit for 0.55 L/hour-mmol GSH. Similarly, the metabolism parameters for GA via
epoxide hydrolase were a Vmax of 1.9 mg/hour-kg, a Km of 100 mg/L, and a second-order rate
constant for GA-GSH conjugation of 0.8 L/hour-mmol GSH.
E-2 DRAFT-DO NOT CITE OR QUOTE
-------
Table E-l: Original Model Parameter Values for Rats in the Kirman et al. (2003)
PBTK Model. Source: Kirman et al. (2003)
Pjtjm' 1' i HI "up P:r,uni'i.'i
R.I phy-.]"]'^ Both w i^ht
( .idi; ipnf
^Ki-nki vuinLii'in
LIVI hlt» -dlU\
Fi. i lit m n lfii:l
Fii. linn u -nuns
LlV* 1 VH|U|MI
lY-Mii1 \olnnv
TIXXII,. hi,* .d ||< i».
V'.'lllllV li|n..w|
Mnil,,|
HU .kg
f » t • L
i J>I '< ' L
OLL ill
F^PC i|
FVBt u
VLt ML
VTX Hi,
QIC Hi
VP( ih,
iunit>i Value Rfi''irni-. swiiri "
i SiuJv HLI.]'. ^p"* iln
sp. > Ill>
h-kg 14 kol'lrrisi'i ,;l tir"'6)
h-k^i 14
* ih-nt," 1 1 »1-!ri
L,I IP >n \ B,i ii (iH
. U. .nBWi OH4 Bi..v,ri.'t;-l il(i«~.
itiunldVSi »»nr f.J.'uLi.--,| null -VLC .
.'. linn i.H ( l <<~r, i J. uLt', .1 > 1 -OLi.i
.' IP -n P\\ i • ' IN- l!i..\\n "l _,l i |H'»~I
FI..I limi fill »iil i t-llb FH( ili.ilii.ii VHi 0,44 Iv .Mnis rl J il'"«,i
Fk'.'tifin him'd y-mm FtiN ilurth >n VP" 0.56 k> 'Idriib .-t j| il'J'i'ii
^hv.ipimn Alwirpii«.n uti> k-\ i/h'i 5 M'di'l-simnl u-dT-lit t"
limn jv.iit. -ml' slmJ R.^niei "t, I 1,1'VHi.
tk . I I. ik-l d' tsv't ill i\1lll> I i.'l ."I ll^oji,
inn, piTiinn. J tjvin Rt HIM-\ i't 11 .|'«r,4.
np.K.M-i
Iniiibiun limi >\\ -|iwi FINF'h. u no,1
I'.ititii-n Bl"u- Uit. \\U> PMl'uniil"SM '.I mm ('"in Estiirui.-.l
• "i-HKK nts Li\' i I'l'Ti'k \\\[ > I'LI iimitlr'v-.! I'-'J! .I'nuhn « knshniii,
Ii->bU" hl'i'itl T.MIJ I'll iLinill-.-ssi 'l"ri I'*1'" P^'Wij, p'lihln
Pl'ind ^It, C.Vl |'MJ Hinilli. ss.l (IK 0. It M 'ill'
Liv i l.h..'|,i'L^ PLJ
Fwi* .hi", d i.L't I'TJ
Metabolism Cvt»i hi<.mi' l'-4rin V'M\\i i tmg'lvk^i 1,8 M"il< I sirnul:.tinn>
O *"' *
li\ilm.s<- \ \<,\\i J iin^'li-k^i 1.9
kMfJ unu Li 100
n -Aith l\< AK 1 0 "-5
ih n jnt, AMP i|_ mm»k"AH-hi i
n with M .Mi 1 ( lr,sH-hi
• In mt^M.m. KHi.HI 0.5 M^d. I simuLii"iT-.
. I ,^Hf,H-h) Mill- i •-I .'.I il"H
kH<,NJ 0.25 fcmi.i'V.'ttJ .1'
imlingt" hvfi KFKEL1 iln 0.2
m,.i iiininli'i nit s.
E-3 DRAFT-DO NOT CITE OR QUOTE
-------
Table E-l (continued):
Parameter group
l',r,m,,i,i
Rmdmi; tn livi
m. n»mn|il 11 until
KFKEL: • hi
KFKKFI Nil
KFEETJ i hi
Value KVt"i(.m •* v 'into
0.1
0,08
un4
m>i iunn ili-, Lik-s. CLY
I'mdingiM hl'HiH KFKEH1 i.IT
nil' iMiimlfriil'"!. A.MR
l!iniliiiL\iii M' 'ml kFEEf^ i hi
ITL.i l'M|lln|i:i uk'i. C.L^l
f"i>ii«in finriMvt i kl'F (-hi
Clutathione (,Sf 1 pi. di1.11. -n rir k< *HI' minn'I In i • i ij". D'Souza etal. (198-8)
I'SHL-bsLir KC.sHL. ill. i,» \-
IniiL.I GbiH GSHb1 unmnl Li ~ '^
i nnc i-ntuti' -n in livi
References cited by Kirman et al (2003) in the above Table of model parameters
Brown, R. P., Delp, M. D., Lindstedt, S. L., Rhomberg, L. R., and Beliles, R. P. 1997. Physiological
parameter values for physiologically based pharmacokinetic models. Toxicol Ind. Health 13:407-484.
D'Souza, R. W., Francis, W. R., and Andersen, M. E. 1988. Physiological model for tissue glutathione
depletion and increased resynthesis after ethylene dichloride exposure. J. Pharmacol. Exp. Ther.
245:563-568.
Kedderis, G. L., Teo, S. K., Batra, R., Held, S. D., and Gargas, M. L. 1996. Refinement and verification
of the physiologically based dosimetry description for acrylonitrile in rats. Toxicol. Appl. Pharmacol.
140:422-435.
Miller, M., Carter, D., and Sipes, I. 1982. Pharmacokinetics of acrylamide in Fischer 344 rats. Toxicol.
Appl. Pharmacol. 63:36-44.
Poulin, P., and Krishnan, K. 1995. A biologically-based algorithm for predicting human tissue:blood
partition coefficients of organic chemicals. Hum. Exp. Toxicol. 14:273-280.
Poulin, P., and Krishnan, K. 1996a. A mechanistic algorithm for predicting blood:air partition coefficients
of organic chemicals with the consideration of reversible binding in hemoglobin. Toxicol. Appl.
Pharmacol. 136:131-137.
Poulin, P., and Krishnan, K. 1996b. A tissue composition-based algorithm for predicting tissue:air
partition coefficients of organic chemicals. Toxicol. Appl. Pharmacol. 136:126-130.
Ramsey, J., Young, J., and Gorzinski, S. 1984. Acrylamide: Toxicodynamics in rats. Midland, MI: Dow
Chemical USA.
Raymer, J. H., Sparacino, C. M., Velez, G. R., Padilla, S., MacPhail, R. C, and Crofton, K. M. 1993.
Determination of acrylamide in rat serum and sciatic nerve by gas chromatography-electron-capture
detection. J. Chromatogr. 619:223-234.
Sumner, S., MacNeela, J., and Fennell, T. 1992. Characterization and quantitation of urinary metabolites
of [l,2,3-13C]acrylamide in rats and mice using 13C nuclear magnetic resonance spectroscopy. Chem.
Res. Toxicol. 5:81-89.
E-4 DRAFT-DO NOT CITE OR QUOTE
-------
Recalibrated Rat PBTK Model
The data used to calibrate the original Kirman et al. (2003) model were limited to urinary
metabolite data, tissue levels of AA radiolabel, and some information on binding to hemoglobin
and other tissue macromolecules. Additional kinetic and hemoglobin binding data in rats and
mice have subsequently been published (Sumner et al., 2003, Fennell et al. 2005; Doerge et al.,
2005a,b,c). The US EPA contracted Dale Hattis and Katherine Walker (George Perkins Marsh
Institute, Clark University, Worcester, MA) to assist in the recalibration and testing of the
Kirman et al. (2003) rat model.
Table E-2 summarizes the rat hemoglobin adduct data used to estimate area under the
time-concentration curves (AUC) for acrylamide and glycidamide. The basic adduct
measurements were translated into units of AUC in (imoles/liter-hour in blood with the aid of
second-order rate constants derived from in vitro measurement of the reaction of hemoglobin
with acrylamide and glycidamide.
Hemoglobin adducts on the terminal valine, as measured by Fennell et al. (2005):
Reaction rate constants L/g-hr
AAVal GAVal
Rat 3.82E-06 4.96E-06
Human 4.27E-06 6.72E-06
Adducts on the free cysteine of rat hemoglobin as measured by Bergmark et al. (1991):
Reaction rate constants L/g-hr
AACys GACys
Rat 1.8E-03 0.92E-03
For short-term measurements post exposure (e.g., at 24 hours following a single exposure),
adduct concentrations expressed in femptomoles/mg Hb are simply converted into AUC units:
(adducts in fmoles/mg Hb)(le- 9 umoles/fmole) * 1000 mg/g , , n
— E-B = (jmole - hr/L
rate constant in L/g - hr
When measuring adduct levels after repeated exposures over many days, one must take
into account the normal loss of red cells (e.g., lifespan of about 61 days in rats; Derelanko
[1987]12) and effective dilution of the adducted red cells via growth of the animal (and
12 Derelanko MJ. 1987. Determination of erythrocyte life span in F-344, wistar, and Sprague-Dawley rats
using a modification of the [3H]diisopropylfluorophosphate ([3HJDFP) method. Fundam. Appl Toxicol. 9: 271-276.
E-5 DRAFT-DO NOT CITE OR QUOTE
-------
consequent growth of the blood). Calleman et al. (1996)13 utilize the following equation for
repeated treatments over time:
Accumulated adducts = ^ a( 1 -1/61) * bt
t = l to 61
where "a" is the accumulation on each day based on 1-day experiments; 61 is the lifetime of
erythrocytes; and bt is a coefficient that corrects for the growth in blood volume from the start of
exposure to day t.
It can be seen in Table E-2 that the adduct-derived AUC observations fall into two
distinct groups: 1) recent measurements of Hb-terminal valine adducts in Fischer-344 rats and
humans (Fennell et al., 2003; Fennell et al, 2005; Sumner et al. 2003); and 2) older
measurements of Hb-cysteine adducts by Bergmark et al. (1991) in Sprague-Dawley rats. The
Fennell and Sumner measurements indicate several fold higher internal glycidamide AUC
exposures per unit of acrylamide external dose than derived from Bergmark et al. (1991) data.
The AUCs based on the Fennel and Sumner data, however, were given more weight for
calibrating the acrylamide PBTK model parameters because they used the same strain of rats
(F344 rats) as was used in the two co-principal rats bioassays (Johnson et al., 1986; Friedman et
al., 1995), the oral dose in Fennell et al. (2005) of 3 mg/kg bw corresponds roughly to the
highest daily dose in the chronic bioassays, and the measurement of the terminal valine adduct
are preferred to the cysteine adducts measured by Bergmark et al. (1991) because valine adducts
were also measured in the more recent human studies (Fennell et al., 2005) and are used in the
calibration of the human version of the PBTK model.
A caveat for the choice of the Fennell and Sumner data and AUCs as the primary target
for the model parameter calibration is that the measurements are for single exposures. These
data therefore will not reflect possible effects of enzyme induction that may arise during
achronic exposure, phenomena that may be reflected in the Bergmark et al. (1991) data from 10-
and 33-day repeated exposures.
13 Calleman CJ. 1996. The metabolism and pharmacokinetics of acrylamide: Implications for mechanisms
of toxicity and human risk estimation. Drug Metabolism Reviews 28:527-590.
E-6 DRAFT-DO NOT CITE OR QUOTE
-------
Table E-2: Data used to recalibrate the Kirman et al. (2003) model parameters
Reference
Source, Route
and Chemical
Administered
Fennell (2005a)
oral
Sumner (2003)
Inhalation
Sumner (2003) IP
Fennell (2003)
Observations
(oral)
Bergmark(1991)
IP acrylamide 30 d
Bergmark(1991)
IP acrylamide 10 d
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP glycidamide
Bergmark(1991)
IP glycidamide
Bergmark(1991)
IP glycidamide
Rat Strain
Male
Fischer-344
Male
Fischer-344
Male
Fischer-344
Male
Fischer-344
Sprague-
Dawley
Sprague-
Dawley
Sprague-
Dawley
Sprague-
Dawley
Sprague-
Dawley
Sprague-
Dawley
Sprague-
Dawley
Sprague-
Dawley
Type of
Adducts
Measured
terminal
valine
terminal
valine
terminal
valine
terminal
valine
cysteine
cysteine
cysteine
cysteine
cysteine
cysteine
cysteine
cysteine
Acrylamide
or
Glycidamide
Dose mg/kg
o
J
6.5 (retained
at 6 hr)
46.5
59.5
3.3
10
10
50
100
10
50
100
Acrylamide
AUC uM-hr
237
363
3395
5457
109
366
Not
Available
2617
4789
Acrylamide
AUC Std
Error
Not
Available
17
85
219
7
32
Not
Available
167
164
Glycidamide
AUC uM-hr
156
322
1861
1588
45
124
137
400
407
267
1185
2543
Glycidamide
AUC Std
Error
Not Available
21
47
76
2
11
17
35
28
Not Available
73
54
Observed
Acrylamide
AUC/dose
uM-
hr/(mg/kg
AA)
79
56
73
92
33
37
52
48
Observed
Glycidamide
AUC/dose uM-
hr/(mg/kg AA)
52
50
40
27
14
12
8
4
E-7
DRAFT-DO NOT CITE OR QUOTE
-------
The original Kirman et al. (2003) model parameter values were used to evalute the model
predictions against the AUCs derived from hemoglobin adduct data. The results of the observed
versus the predicted values are presented in Table E-3, and reflect a much closer fit to the AUCs
based on the Bergmark et al (1991) data, with two fold or higher model predicted levels than
estimated AUCs based on the Fennell and Sumner data. Since the AUCs based on Fennell and
Sumner adduct data were considered more relevant to the derivation of toxicity values for this
assessment, a recalibration of the model parameters was conducted to improve the fit to these
data. The following model parameters were chosen as good candidates for the recalibration
because their values were less certain and the more recent data was informative:
1. the tissue/blood partition coefficient multiplier of 3.2 for glycidamide;
2. the balance between P450 vs. GSH and other non-P450 metabolic routes for acrylamide,
on the basis that the urinary metabolite profile observed at 24 hours (considered to not
fully represent the complete metabolic fate of acrylamide, i.e., the fraction that does not
appear in urine but is irreversibly bound to tissues, or completely metabolized to building
blocks that are incorporated into tissue constituents or exhaled;
3. the tissue/blood partition coefficients for acrylamide, which were based on statistical
model projections (based on Poulin and Krishnan 1995; 1996a; 1996b) rather than direct
measurements.
A number of model runs were conducted against the hemoglobin adduct data and some of
the previous blood level data (Raymer et al, 1993) to evaluate model results for different
estimates of the tissue/blood partition coefficients including use of the original values, re-
estimation using a different algorithm based on octanol/water partition coefficients; and
ultimately calibration based on the volume of distributions derived directly from measurement of
serum AA and GA levels in mice and rats following i.v. low dose exposures (Doerge et al., 2005
b,c).
Table E-4 lists the recalibrated parameter values, and Table E-5 presents the fits of the
final recalibrated rat PBTK model predictions versus the data based estimates of the AUCs, as
limited by the Fennel et al. (2005) urinary metabolite data. Urinary output data provided a lower
bound on the amount of a particular metabolite that the model needs to simulate, but not
necessarily a precise estimate of the amount actually produced. This is because there is no
guarantee that all of the metabolite of a particular pathway produced in animals (or people) will
actually show up in the urine—in general, an unknown fraction of the metabolite that is actually
produced can undergo further metabolism to undetected or unquantified metabolites (including,
for example, exhaled carbon dioxide) and an unknown fraction may remain in the body at the
end of the urine collection period. By contrast, the model parameters were adjusted to fit the
E-8 DRAFT-DO NOT CITE OR QUOTE
-------
acrylamide and glycidamide hemoglobin adduct-derived AUCs at the approximately 3 mg/kg bw
dose. Departures of the model predictions from the observations at higher acute doses were
tolerated because as dose increases it is increasingly likely that phenomena such as GST enzyme
induction or induction of increased glutathione production levels following an initial depletion
will change the results via mechanisms that are not in the model and would not be expected to
substantially influence the toxicokinetics at the lower doses in the two principal chronic
bioassays of interest in this assessment. In particular, reducing the Km for the CYP2E1
metabolism of acrylamide to glycidamide (thus introducing this saturating nonlinearity at lower
doses) tends to worsen the fit of the model to the acrylamide AUC data at high doses, holding the
fit at the lowest dose level constant. The nonlinearity that was indicated by the observed
acrylamide AUC derivations at the higher doses is therefore in the opposite direction than would
be produced by a lower Km for CYP2E1.
Departures of the model predictions from even the low dose observations of Bergmark et
al (1991) were also tolerated because the Bergmark group used Sprague-Dawley rats, i.p doses,
measured cysteine (rather than valine) hemoglobin adducts, and administered much higher doses
than used in the chronic bioassays.
One possible model adaptation that would bring the model predictions closer to the
observations at higher doses is to assume, contrary to the current model formulation, that the
glutathione depletion at higher doses does not appreciably slow down the rates of the GST-
mediated reactions. There is some indication in the prior literature that the Km for binding of
glutathione to the active site of some GSTs can be very low (Sun and Morgenstern, 199714) —
.018 mM compared to a basal liver concentration in this version of our model of 7 mM —
indicating that modest changes in glutathione levels would not be reflected in proportionately
decreased rates of the GSH-acrylamide and GSH-glycidamide reactions . This was evaluated in
a series of model runs where liver glutathione was held constant at 7 mM rather than appreciably
depleted at high doses (approximately to 4.3 mM at minimum for the 59.5 mg/kg dose). The
results of this modification did indicate closer agreement to the AUC data at high does, while
causing only modest changes to the model fit of doses in the region where the bioassay was
conducted.
The current model reflects a considerable reduction in the estimates of the rates of
reaction between acrylamide and miscellaneous proteins in the liver, tissue, and blood, but
14 Sun TH, Morgenstern R. 1997. Binding of glutathione and an inhibitor to microsomal glutathione
transferase. Biochem J. 326 (Pt 1):193-196.
E-9 DRAFT-DO NOT CITE OR QUOTE
-------
relatively modest changes in other parameters to accommodate the revised tissue/blood partition
coefficients and the revised adduct-based AUC estimates.
E-10 DRAFT-DO NOT CITE OR QUOTE
-------
Table E-3: AUC Predictions from the Original Kirman Model versus AUCs Derived from Hemoglobin Adduct Data.
Data Source and
Route of Admin
Fennell (2005a)
Oral
Sumner (2003)
Inhalation
Sumner (2003)
IP
Fennell (2003)
Observations (oral)
Bergmark(1991)
IP acrylamide 30 d
Bergmark(1991)
IP acrylamide 10 d
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP acrylamide
Acrylamide AUC Comparisons
mg/kg
Acrylamide dose
3
6.5
46.5
59.5
3.3
10
10
50
100
ObsAA
AUC uM-hr
237
363
3395
5457
109
366
2617
4789
Model pred.
AAATJC
uM-hr
84
192
1919
2631
93
309
2104
5212
Ratio—Model
Predicted/Obs
0.354
0.529
0.565
0.482
0.852
0.844
0.804
1.088
Glvcidamide AUC
Comparisons
ObsGA
AUC uM-
hr
156
322
1861
1588
45
124
267
1185
2543
Model pred.
GAAUC
uM-hr
14
28
168
210
15
42
42
180
335
Ratio — Model
Predicted/Obs
0.087
0.088
0.090
0.133
0.328
0.339
0.157
0.152
0.132
E-ll
DRAFT-DO NOT CITE OR QUOTE
-------
Table E-4: Recalibrated PBTK Model Parameter Values for the Rat
Parameter
group
Basic physiology
Compartment
Volumes
Absorption
Partition
coefficients
(equilibrium
concentration
ratios)
Parameter
Body weight
Alveolar ventilation
Cardiac output (total body blood flow)
Liver blood flow
Tissue blood flow
Volume blood
Fraction arterial/total blood
Fraction venous/total blood
Liver volume
Tissue volume
Fraction blood cells/blood
Fraction blood serum/blood
Absorption rate from gastrointestinal tract (oral
dose) or intraperitoneal cavity (ip dose)
Blood: air, AA
Liver: blood, AA
Tissue: blood, AA
Blood: air, GA
Liver: blood, GA
Tissue: blood, GA
Symbol (units)
BW (kg)
QCC (L/h-kg° /4)
QPC (L/h-kg° /4)
QLC (fraction QCC)
QTC (fraction QCC)
VBC (fraction BW)
FABC (fraction VB)
FVBC (fraction VB)
VLC (fraction BW)
VTC (fraction BW)
FBC (fraction VB)
FBS (fraction VB)
KA(/h)
FBI (unitless)
PL1 (unitless)
PT1 (unitless)
PB2 (unitless)
PL2 (unitless)
PT2 (unitless)
Value
0.25
14
14
0.25
0.75
0.06
0.35
0.65
0.04
0.87
0.44
0.56
5
3.1E7
0.797
0.619
9.8E7
0.923
0.716
E-12
DRAFT-DO NOT CITE OR QUOTE
-------
Table E-4 continued:
Parameter
group
Metabolism
Tissue binding
Glutathione
Parameter
Cytochrome P-450 oxidation rate, AA
Cytochrome P-450 Michaelis-Menten
constant, AA
Epoxide hydrolase hydrolysis rate, GA
Epoxide hydrolase Michaelis-Menten
constant, GA
Reaction with glutathione, AA
Reaction with glutathione, GA
Binding to hemoglobin, AA
Binding to hemoglobin, GA
Binding to liver macromolecules, AA
Binding to liver macromolecules, GA
Binding to tissue macromolecules, AA
Binding to tissue macromolecules, GA
Binding to blood macromolecules other
than hemoglobin, AA
Binding to blood macromolecules other
than hemoglobin, GA
Protein turnover
GSH production rate (Based on a 0.25 Kg
rat)
GSH loss rate
Initial GSH concentration in liver
Symbol (units)
VMAXC1 (mg/h-kg0'7)
KMC1 (mg/L)
VMAXC2 (mg/h-kg0'7)
KMC2 (mg/L)
KGSTC1 [L/(mmolGSH-kg°'J -h)]
KGSTC2 [L/(mmolGSH-kg°3 -h)]
KHGB1 (L/gHGB-h)
KHGB2 (1/gHGB-h)
KFEEL1 (/h)
KFEEL2 (/h)
KFEET1 (/h)
KFEET2 (/h)
KFEEB1 (/h)
KFEEB2 (/h)
KPT (/h)
KGSHP (mmol/h)
KGSHL (/h)
GHSLO (mmol/L)
Value
4.00
31
0.988
100
0.225
0.630
0.00135
0.1300
0.00054
0.0520
0.000216
0.0208
0.000027
0.0026)
0.008
0.025
0.35
7.0
E-13
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
Data Source and
Route of Admin
Fennell (2005) oral
Fennell (2005) oral
Sumner (2003)
Inhalation
Sumner (2003)
IP
Fennell (2003)
Observations (oral)
Bergmark(1991)
IP acrylamide 30d
Bergmark(1991)
IP acrylamide lOd
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP acrylamide
Bergmark(1991)
IP acrylamide
mg/kg
Acrylamide
dose
2.86
2.86
6.5
46.5
59.5
3.3
10
10
50
100
Acrylamide ATJC Comparisons
Obs AA AUC
(iM-hr
237
363
3395
5457
109
366
2617
4789
Model pred.
AA AUC nM-
hr
210
479
5224
7432
232
773
5803
16200
Ratio— Model
Predicted/Obs.
0.886
1.322
1.544
1.362
2.134
2.113
2.218
3.383
Glycidamide ATJC Comparisons
ObsGA
AUC nM-
hr
156
322
1861
1588
45
124
267
1185
2543
Model pred.
GA AUC nM-
hr
152
332
2655
3499
167
516
516
2879
6240
Ratio — Model
Predicted/Obs
0.973
1.032
1.427
2.204
3.685
4.164
1.931
2.430
2.454
Urinary Metabolite Comparisons
Obs AA-GSH
excreted
(pinoles)
2.6
Obs GA-GSH
excreted
(pinoles)
1.8
Pred AA-GSH
produced
(pinoles)
2.62
Pred GA-GSH
produced
(pinoles)
6.17
Model/
Minimum
Obs
1.01
Model/
Minimum
Obs
3.43
E-14
DRAFT-DO NOT CITE OR QUOTE
-------
Use of the PBTK Model to Estimate the Internal Dose Metric from the Chronic Bioassay Data
To model the drinking water exposures to rats in the chronic bioasssays, values for
hourly percent of total water consumed were modeled based on data reported by Johnson and
Johnson (1990)15 for diurnal drinking patterns in rats. Johnson and Johnson (1990) observed that
drinking behavior in rats follows a nocturnal pattern, with the greatest intensity in the hours just
after the lights go out (by convention, labeled 6:00 PM), and just before the lights are due to go
back on again at 6:00 AM. About 94% of the total water consumption appears to take place
during the 12 hours of darkness.
Table E-6 contains the recalibrated rat model-predictions of the relationship between
delivered AUC doses of acrylamide and glycidamide, and mg/kg daily doses administered
according to the diurnal pattern of drinking behavior described by Johnson and Johnson (1990).
To derive these AUCs, the model was run for a 36 hour period, 24 hours with rat pattern of
drinking water consumption, followed by 12 hours with no further dosing. Infinite-time AUCs
for acrylamide and glycidamide were projected using the following formulae:
AUCinfinite = AUCo-36 hr + Cse/k
where Cse = the acrylamide or acrylamide blood concentration;
k = the rate constant for exponential decline in acrylamide or glycidamide blood
concentration between 35 and 36 hours.
It can be seen from Tables E-6 that the effect of this projection beyond 36 hours is
minimal, showing up sometimes as a change in only the third decimal place of the AUC results,
and sometimes with no additional effect. Overall, the current model predicts very little
nonlinearity in AA and GA AUCs between the limits of low dosage and the highest dose rates
used in the Friedman et al. (1995) bioassays. Linear interpolation formulae were therefore
derived between the closely spaced doses in Tables E-6 with the AUCs as the dependent
variables and mg/kg-day in all cases the independent variable yielding intercepts (b's) and slopes
(m's) of the interpolation lines. The intercepts and slopes were used to derive lifetime average
daily internal AUCs for benchmark dose levels derived for the noncancer and cancer dose-
response data. These internal AUC levels for blood AA or GA were used as the basis for
deriving the human equivalent concentration in lieu of using the default uncertainty factor for
interspecies toxicokinetic differences.
15 Johnson RF, Johnson AK. 1990. Light-dark cycle modulates drinking to homeostatic challenges. Am J
Physiol. 1990 Nov;259(5 Pt2):R1035-42.
E-15 DRAFT-DO NOT CITE OR QUOTE
-------
Table E-6: Estimated Internal AUC Acrylamide and Glycidamide Doses Produced by Various Drinking Water Intakes
mg/kg DW
0.001
0.01
0.1
0.5
1
1.5
2
2.5
o
3
3.5
4
Cum 36 hr
(iM-hr AA
in blood
6.68E-02
6.68E-01
6.69E+00
3.35E+01
6.73E+01
1.01E+02
1.35E+02
1.70E+02
2.04E+02
2.39E+02
2.74E+02
AABlood
Cone at 35
hr(nM)
1.76E-05
1.76E-04
1.76E-03
8.81E-03
1.77E-02
2.66E-02
3.56E-02
4.47E-02
5.38E-02
6.30E-02
7.22E-02
AA Blood
Cone at 36
hr (nM)
1.28E-05
1.28E-04
1.28E-03
6.43E-03
1.29E-02
1.94E-02
2.60E-02
3.26E-02
3.93E-02
4.60E-02
5.27E-02
35-36 In-
Half Life
forAA
decline (hr)
2.205
2.205
2.205
2.205
2.205
2.205
2.205
2.205
2.205
2.205
2.206
Infinite AUC
for AA nM-hr
6.69E-02
6.69E-01
6.69E+00
3.35E+01
6.73E+01
1.01E+02
1.35E+02
1.70E+02
2.05E+02
2.39E+02
2.74E+02
AAAUC
uM-hr/mg/kg
AA Dose
66.88
66.89
66.92
67.09
67.31
67.52
67.74
67.96
68.17
68.39
68.61
Cum 36 h
HM-hr GA
in blood
5.00E-02
5.00E-01
5.00E+00
2.51E+01
5.02E+01
7.54E+01
1.01E+02
1.26E+02
1.52E+02
1.77E+02
2.03E+02
GA Blood
Cone at 35
hr— (iM
8.00E-05
8.00E-04
8.01E-03
4.02E-02
8.08E-02
1.22E-01
1.63E-01
2.05E-01
2.48E-01
2.91E-01
3.34E-01
GA Blood
Cone at 36
hr— (iM
6.32E-05
6.32E-04
6.33E-03
3.18E-02
6.39E-02
9.64E-02
1.29E-01
1.62E-01
1.96E-01
2.30E-01
2.64E-01
35-36 hr
Half Life
forGA
decline (hr)
2.948
2.948
2.948
2.947
2.947
2.946
2.945
2.945
2.944
2.944
2.943
Infinite AUC
for GA nM-hr
5.03E-02
5.03E-01
5.03E+00
2.52E+01
5.05E+01
7.59E+01
1.01E+02
1.27E+02
1.53E+02
1.78E+02
2.04E+02
GAAUC
(iM-hr/mg/kg
AA Dose
50.29
50.29
50.31
50.38
50.48
50.57
50.66
50.75
50.85
50.94
51.03
E-16
DRAFT-DO NOT CITE OR QUOTE
-------
Table E-7: Available Data for Calibration of the Human PBTK model
Nominal
dose
(mg/kg)
0.5
1
3
Measured
dose mg/kg
0.437
0.925
2.864
(imole/kg
administered
5.9
12.5
38.7
Average
Std Dev
Std Error
Lower 95% CL
Upper 95% CL
AAVal AUC/dose
(iMol-hr/(mg/kg
AA)
275.7
231.4
202.7
236.6
36.8
21.2
215.4
257.8
GAVal
AUC/dose uMol-
hr/(mg/kg AA)
63.4
55.3
55.9
58.2
4.5
2.6
55.6
60.8
umole excreted
as AA-GSH
metabolite 24 hr
823
umole
excreted as
GA 24 hr
25.4
(imole excreted as
GA-Epoxide
hydrolase product 24
hr
103.8
Ratio of GA-GSH to
AA-GSH metabolite
excretion at low doses
0.206
The data are all from Fennell et al. (2005), with the exception of those in the last column, which are derived from the observations of general population urinary
excretion of glutathione metabolites by Boettcher et al. (2005).
E-17
DRAFT-DO NOT CITE OR QUOTE
-------
Parameter Values for the Human PBTK Model
To estimate the human external concentration that would result in the same internal AUC
of AA or GA in the blood as that prodiced from the BMDL, the Kirman et al. PBTK model
parameters values were adjusted to simulate an adult human male. Table E-7 lists the data used
for calibrating the human model including hemoglobin and urinary metabolite data primarily
from Fennell et al. (2005), with some observations of general population urinary excretion of
glutathione metabolites reported by Boettcher et al. (2005).
The Fennell et al. (2005) urinary data have inherent uncertainties in reflecting AA and
GA kinetics because of the possible incomplete excretion of both acrylamide and glycidamide by
the end of the 24 hour observation time, i.e., the model only predicts the amount of each
metabolite made at a particular observation time, not the amount that is actually excreted in the
urine, in the absence of renal clearance data. The Boettcher (2005) data were generated from
long term exposures and are not as sensitive to the timing issue, but still have inherent
uncertainties for total recovery since at least two potential glycidamide metabolites other than
GAMA remained unmeasured: 1) N-acetyl-S-(l-carbamoyl-2-hydroxyethyl)cysteine, and 2) a
hydrolysis product that is quantified by Fennell et al. (2005). Therefore, the molar ratio of total
glycidamide to acrylamide metabolites may well be understated by the urinary GAMA/AAMA
ratios reported by Boettcher et al (2005). For model calibration purposes, the urinary data can
only be considered to establish a lower limit on the amount of a particular metabolite that must
be produced. Any model output above the lower limit is acceptable. By contrast, the AUC
estimates, derived from the hemoglobin adduct data, should be considered central estimates of
the AUC that should be relatively precisely targeted in the model calibration.
The human model was parameterized for males because the study subjects in the Fennell
et al. (2005) study were males with an average body weight of 81.65 Kg. Central values for
body weight, tissue volumes and blood flows for the human PBTK model were developed from
the National Health and Nutrition Examination Survey (NHANES III) database using the P3M16
software program (Price et al., 2003). Initial values of metabolism parameters were estimated by
scaling overall body metabolic capacities (e.g. Vmax values for the liver as a whole) in
proportion to overall metabolic rates, which approximately scale across species with the three
16 The P3M model was created to use the anthropometric measurements in the NHANES III database (age,
gender, race, height, weight, circumferences, etc.) to predict physiological parameters for each individual in the
database using empirical equations describing these relationships in the published literature. The objective behind
the development of this model was to assist efforts to incorporate interindividual variability into PBTK modeling.
The databases generated using this program provide a set of internally consistent estimates of the physiological
parameters needed for PBTK modeling for each individual in the database. Estimates of interindividual variability
based on sampling from independent distributions representing each physiological parameter have to specify the
correlation structure for all parameters. These correlations are inherently preserved in the individual records
generated by P3M.
E-18 DRAFT-DO NOT CITE OR QUOTE
-------
quarters power of body weight (Rees and Hattis, 199417). Because the liver makes up only about
half the proportion of body weight in people as it does in rats, metabolism rates per unit of liver
tissue were further adjusted upward by a factor of (0.04 / 0.195) related to this proportion.
Parameters that were expressed in terms of a rate per unit time were adjusted downward in
proportion to body weight0'25. The liver glutathione parameters were adapted to establish steady
state at the same liver glutathione concentration (7 mmol/liter) as was used by Kirman et al.
(2003) for rats. For example, both the liver volume and the relative rate adjustments contributed
to the estimated baseline loss rate for glutathione in the liver of 0.164/hour, compared to the
comparable rat value of 0.35/hour. In general, only relatively modest adjustments in parameters
were needed to accommodate the new data. An extra pathway for renal removal of glycidamide
was added to reflect the newly reported observation of glycidamide in the urine of the studied
subjects.
Various model parameters were then adjusted in an iterative process that evaluated
various physiologically feasible modeling options to achieve the best fit to the data. The final
model parameter values are shown in Table E-8, and the model fits to the data are presented in
Table E-9.
The final iteration of the human model had nearly half of the acrylamide being converted
to glycidamide, (compared with only a little over a quarter for earlier versions) and an apparent
half-life for glycidamide of about 6 hours (the actual half life might be less than this because the
glycidamide levels present at 23-24 h levels might be limited by the conversion from the
declining level of acrylamide). In comparing different versions of the model, it was also noted
that the model parameters were underdetermined, that is, there is just not enough basic
pharmacokinetic data to derive a unique set of optimal parameter values, given the number of
"adjustable" parameters in the current model. The last two iterations of the model calibrations
closely fit the hemoglobin-adduct-derived AUC values, and do not yield appreciably different
estimates of mean "human equivalent doses" even though they reflect different assumptions
about the fraction of acrylamide that is processed by CYP2E1 and glutathione-S-transferase
dependent pathways. If additonal data were available to further resolve the model parameters for
these pathways, the human PBTK model could also be used to predict population distributions
and interindividual variability in internal AUCs for acrylamide and glycidamide based on
interindividual differences in metabolic rates. Until those data are available, however, the model
is only suitable for use in deriving a human equivalent concentration based on the rat bioassay
data in lieu of the interspecies toxicokinetic uncertainty factor, and not suitable to replace the
uncertainty factor for variability within the human population (i.e., to identify the most senstivie
subpopulation ) resulting from intrahuman differences in AA's or GA's toxicokinetics.
17 Rees DC, Hattis D 1994. "Developing Quantitative Strategies for Animal to Human Extrapolation" Chapter 8 in
Principles and Methods of Toxicology, 3rd Edition, A. W. Hayes, ed., Raven Press, New York, 1994, pp. 275- 315.
E-19 DRAFT-DO NOT CITE OR QUOTE
-------
Use of the Human PBTK Model to Derive the Human Equivalent Concentration for a Drining
Water Exposure to the Rat Neurotoxicity or Carcinogenicity Benchmark Doses
To derive human equivalent mg/kg-day doses and drinking water levels for the
acrylamide AUCs estimated in rats at the BMDL, oral exposures via drinking water were
modeled with the following schedule:
8:00 AM (Start of simulation; Time 0 hours)—25% of daily dose
10:30 AM—12.5% of daily dose
1:00 PM—25% of daily dose
3:30 PM—12.5% of daily dose
6:00 PM—25% of daily dose
Simulations were then continued until 48 hours after the start of dosing and infinite-time
AUCs for both acrylamide and glycidamide were projected using the equations described earlier.
The results are presented in Table E- 10.
Based on these results, interpolations of the the mg/kg-day dose corresponding to the
animal acrylamide AUCs estimated above were conducted similar to as those derived from the
rat modeling results. The resulting human equivalent drinking water exposure levels are before
application of uncertainty factors for human interindividual variability and/or any other
circumstances that are deemed necessary (except interspecies projection).
Human Equivalent Concnetrations from an Inhalation Exposure to the Rat Neurotoxicity or
Carcinogenicity Benchmark Doses
The inhalation exposure was modeled as a 24 hour continuous inhalation exposure
without allowances for diurnal changes in inhalation rates and tissue blood flows. In contrast to
the drinking water models, acrylamide is supplied directly to the arterial circulation, rather than
to an "unabsorbed" compartment that feeds in to the liver. Table E-l 1 has the AUC predictions
as a function of dose from the human model simulatation of an inhalation exposure. Overall the
differences between the inhalation and drinking water versions of the models are modest.
Because of the absence of first pass metabolism in the liver from an inhalation exposure, the
cumulative AUCs for acrylamide following inhalation are slightly higher and cumulative AUCs
for glycidamide are slightly lower than the corresponding drinking water AUCs.
E-20 DRAFT-DO NOT CITE OR QUOTE
-------
Table E-8. Parameters for the Human (male) Acrylamide PBTK Model
Parameter group
Basic physiology
Compartment
volumes
Absorption
Partition
coefficients
(Equilibrium
concentration
ratios)
Metabolism
Tissue binding
Renal Excretion
(new pathway)
Glutathione
Parameter
Body weight
Total Alveolar ventilation
Cardiac output (total body blood flow)
Liver blood flow
Tissue blood flow
Volume blood
Fraction arterial/total blood
Fraction venous/total blood
Liver volume
Tissue volume
Absorption rate from gastrointestinal tract
(oral dose) or intraperitoneal cavity (i.p. dose)
Blood:air, AA
Liverblood, AA
Tissue :blood, AA
Blood:air, GA
Liverblood, GA
Tissue :blood, GA
Cytochrome P-450 oxidation rate, AA
Cytochrome P-450 Michaelis-Menten constant,
AA
Epoxide hydrolase hydrolysis rate, GA
Epoxide hydrolase Michaelis-Menten constant,
GA
Reaction with glutathione, AA
Reaction with glutathione, GA
Binding to hemoglobin, AA (previously
eliminated from human model because of the
lack of the free cysteine present in rat
hemoglobin)
Binding to hemoglobin, GA (previously
eliminated from human model because of the
lack of the free cysteine present in rat
hemoglobin)
Binding to liver macromolecules, AA
Binding to liver macromolecules, GA
Binding to tissue macromolecules, AA
Binding to tissue macromolecules, GA
Binding to blood macromolecules other than
hemoglobin, AA
Binding to blood macromolecules other than
hemoglobin, GA
Direct renal elimination from arterial blood
compartment
GSH production rate
GSH loss rate
Initial GSH concentration in liver
Symbol (units)
BW (kg)
QC (L/hour-kg)
QP (L/hour-kg)
QLC (fraction QC)
QTC (fraction QC)
VBC (fraction BW)
FABC (fraction VB)
FVBC (fraction VB)
VLC (fraction BW)
VTC (fraction BW)
KA (/hour)
FBI (unitless)
PL1 (unitless)
PT1 (unitless)
PB2 (unitless)
PL2 (unitless)
PT2 (unitless)
Value
81.65
8.33
5.28
0.183
0.8842
0.0675
0.35
0.65
0.183
0.8842
5
3.1H107
0.88
0.40
3.1H107
0.88
0.40
VmaxCl (mg/hour-kg07)
KmCl (mg/L)
VmaxC2 (mg/hour-kg0 7)
KmC2 (mg/L)
KGSTC1 [L/(mmolGSH-
kg°3-hour)l
KGSTC2 [L/(mmolGSH-
kg°3-h)]
KHGB1 (L/gHGB-h)
KHGB2 (1/gHGB-h)
KFEEL1 (/hour)
KFEEL2 (/hour)
KFEET1 (/hour)
KFEET2 (/hour)
KFEEB1 (/hour)
KFEEB2 (/hour)
L/hour
KGSHP (umol/h)/kg
KGSHL (/hour)
GHSLO (mmol/L)
1.12
7
3.27
100
0.1769
0.5029
0
0
0.055
0.215
0.022
0.086
0.0028
0.01075
0.082
21.6687
0.16953
7.0
E-21
DRAFT-DO NOT CITE OR QUOTE
-------
Table E-9: Human PBTK Model Predictions versus AUCs and Urinary Metabolites
(Bolded numbers are the parameters used for comparison of model outputs with calibrating data)
mg/kg
dose
0.437
0.925
2.864
%AA
converted to
GA
49.67
48.95
46.42
%AA
directly
reacted
with GSH
34.91
35.34
36.81
%GA
eliminated
via GSH
reaction
56.84
56.67
56.00
%GA
eliminated
via epoxide
hydrolase
7.76
7.78
7.88
%GA
eliminated
via direct
renal
excretion
1.88
1.89
1.92
Average
model
prediction
Observed
AA AUC/dose
jiM-hr/mg/kg
AA
230.3
234.2
248.1
237.5
236.6
GA AUC/dose
(iM-hr/nig/kg
AA
60.11
59.23
56.12
58.49
58.22
Absolute
(imoles AA
Metabolized to
GSH Conjugate
164
352
1126
>823
Cum (imoles GA
metabolized to
epoxide hydrolase
metabolite by 24
hours
17.1
35.7
104.3
> 103.8
Cum (imoles
GA excreted in
urine by 24
hours
4.15
8.67
25.4
>25.4
Ratio of GA-
GSH
metabolite to
AA-GSH
metabolite
production
0.733
0.710
0.633
> 0.206
Apparent acrylamide half life for decline, 1-2 hours after dosing with the lowest dose: 5.8 hours
Apparent glycidamide half life for decline, 23-24 hours after dosing with the lowest dose: 6.1 hours (this value may be limited by the
rate of decline of acrylamide concentrations; the true elimination half life could be less if glycidamide were administered separately; rather
than being made from acrylamide)
E-22
DRAFT-DO NOT CITE OR QUOTE
-------
Table E-10: Estimated AUCs in Humans for Acrylamide and Glycidamide from a
Drinking Water Exposure
mg/kg
dose
0.001
0.01
0.1
0.3
0.5
1
1.5
2
2.5
o
J
Cum 48 hr
uM-hr AA in
blood
2.372E-01
2.373E+00
2.378E+01
7.168E+01
1.200E+02
2.429E+02
3.686E+02
4.971E+02
6.284E+02
7.623E+02
47-48 hr
AA Tl/2
5.49
5.49
5.49
5.49
5.49
5.49
5.50
5.50
5.50
5.50
48hrAA
blood cone
(MM)
1.50E-04
1.50E-03
1.51E-02
4.58E-02
7.731E-02
1.592E-01
2.459E-01
3.375E-01
4.341E-01
5.359E-01
Infinite
AUCuM-
hrAA
2.384E-01
2.384E+00
2.390E+01
7.20E+01
1.21E+02
2.44E+02
3.71E+02
5.00E+02
6.32E+02
7.67E+02
Cum 48 h
uM-hr GA
in blood
6.713E-02
6.712E-01
6.706E+00
2.01E+01
3.339E+01
6.644E+01
9.916E+01
1.316E+02
1.636E+02
1.954E+02
47-48 hr
GA Tl/2
5.58
5.58
5.59
5.59
5.59
5.60
5.61
5.63
5.64
5.65
48hrGA
blood cone
(MM)
9.21E-05
9.21E-04
9.26E-03
2.81E-02
4.73E-02
9.73E-02
1.50E-01
2.06E-01
2.64E-01
3.25E-01
Infinite
AUCuM-
hrGA
6.787E-02
6.786E-01
6.781E+00
2.030E+01
3.377E+01
6.723E+01
1.004E+02
1.332E+02
1.658E+02
1.980E+02
Table E-ll: Estimated AUCs in Humans for Acrylamide and Glycidamide from An
Inhalation Exposure
mg/kg
dose
0.001
0.01
0.1
0.3
0.5
1
1.5
2
2.5
o
J
Cum 48 hr
uM-hr AA in
blood
2.453E-01
2.454E+00
2.457E+01
7.394E+01
1.236E+02
2.491E+02
3.765E+02
5.057E+02
6.368E+02
7.697E+02
47-48 hr
AA Tl/2
5.49
5.49
5.49
5.49
5.50
5.50
5.51
5.51
5.52
5.52
48hrAA
blood cone
(MM)
4.76E-04
4.76E-03
4.78E-02
1.45E-01
2.430E-01
4.962E-01
7.600E-01
1.035E+00
1.321E+00
1.618E+00
Infinite
AUCuM-
hrAA
2.491E-01
2.492E+00
2.495E+01
7.51E+01
1.26E+02
2.53E+02
3.82E+02
5.14E+02
6.47E+02
7.83E+02
Cum 48 h
uM-hr GA
in blood
6.503E-02
6.502E-01
6.498E+00
1.95E+01
3.239E+01
6.455E+01
9.646E+01
1.281E+02
1.595E+02
1.907E+02
47-48 hr
GA Tl/2
5.79
5.79
5.80
5.80
5.81
5.84
5.86
5.88
5.91
5.93
48hrGA
blood cone
(MM)
2.80E-04
2.80E-03
2.81E-02
8.48E-02
1.42E-01
2.90E-01
4.42E-01
5.99E-01
7.62E-01
9.30E-01
Infinite
AUCuM-
hrGA
6.737E-02
6.736E-01
6.733E+00
2.018E+01
3.359E+01
6.698E+01
1.002E+02
1.332E+02
1.660E+02
1.987E+02
E-23
DRAFT-DO NOT CITE OR QUOTE
-------
APPENDIX F. YOUNG ET AL (2007) PBTK/TD MODEL SUPPORTING
DOCUMENTATION
The following tables and figures provide supporting documentation for the Young et al.
(2007) PBTK/TD model for acrylamide and glycidamide. All of the information presented in
Appendix F has been extracted directly from the original article — 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):388-99.
The reader is referred to the orginal text for further details about the Young et al. (2007) model
development and simulation results.
Schematic of the Young et al. (2007) PBTK/TD Model For Acrylamide and Glycidamide
Acrykimkk
1 J Ml tX't
' AcVauds
k- ';
™ „,#
OKA
i-jtn ovc-r
f
PD
Ai:r
P11PKO
UlycEtki.mu1e
fUAj
1' ;.T..!fr"
tr»'L»Ki!i-iti
PBPK-4
Figui'e 1 Bled: diagram for the nietsbclhm of AA to C;A and fjitiier
oetabol:sc3 of both to their ghnsdticMie ccnjugates. AL phancaccki^etic
c.ad phancacod^iigimc f?D) processes are firs? order.
F-l
DRAFT-DO NOT CITE OR QUOTE
-------
Table F-l: Data Generated at NCTR on AA and GA in rats and mice
fender
M and F
MandF
MaudF
MPMF
MadF
MaiidF
MaodF
fit M and F
MaiidF
MatdF
MiEdF
MaitdF
M ?nd F
• ikg]
012
012
0.1
0.1
-01
50
~1
012
012
01
0.1
-0.1
—1
compel
dc-ied
C-A
GA
AA
AA
AA
AA
AA
GA
GA
AA
AA
AA
AA
rone
IV
K"
gavaee
diet
savage
drinl:in£ water
IV
p,vase
IV
ffsvaffe
diet
drinl;ing water
single o:
midtidoi.e
single
sniffle
single
single, 30 nun
sniffle
42 days
single
iUiffle
i ingle
single
smsfe 30 nun
42 day;
ser.iin E
A.A
Yes
yes
yes
yes
yes
yes
yes
yes
ssah'Sis
GA
yes
yes
yes
yes
jes
yes
yes
yes
VBS
yes
yes
yes
}"es
sddxict dats
GA Hb addccts at 6 h
GA liver and GA Hb
adductc at S h
AA sad C-A Hb adductc ?.t 6 h
GA liver and AA and GA HI?
adducfsatBli
AA end GA Hb addr.cti 3t 12 h
GA liver adduct i^lcec. o'.it to S h
GA liver and .AA and GA Hb
addict valv.e1': out to
42 day-, and then decay
GA Eb addicK- at S h
GA liver and GA Hb
adduce at 1 3*h
AA and C-A Hb
addicts at S h
GA liver and AA and GA Hb
addicts at lOh
AA K:d C-A Hb addccr, -it 12 h
GA liver and AA md GA H?
adduce valuer ou: "o 42 day*
and then decay
F-2
DRAFT-DO NOT CITE OR QUOTE
-------
Table F-2: Pharmacokinetic and Pharmacodynamic Parameters from AA and GA Administration to Rats
[Mean ± Standard Deviation (Range)]
prinnu,.kir* ii. nn ji
1 i-ij 1 ,llj ifilifi mi "ill i l*?>±" »i II ±i, , ii'lit.M' II-
lil )l> i rtwliji 1 ! 1 I
h JS ! .!. .| US '!'
In I <•! K I 1 if_ ^ Ml !d1J
I "
I -.'41 I -.' ii
I - I I" < ± 1 - a II-, > I ± I ; i I i ^ ± - ,
I - 1 , ± i"1 i I .±.> ' I,
I '-, Ii I i-l >.i
10.0 8.8 ±1.8 a '! Il4±'4i 44±'i"'h
«»-(!.) >-) i , -LI
i i i i - ± ! i ' i, ± i: i,
ll-l'.i r -41
10.0 11.3±0.84 111 I I i± i 'b " '±n
(10-13) ,P-|ii I'-'i
tl.S
373
IS .4
ma
,
12,3
37
25 J
113
ICi.,2
t i1 (( i.s> J_ ' <,. !
, -±'i "I,
,1- n
Is il i ± 1 "I. !'. - " ± i -•
. — [ * s , — k t.^
1 ' ' i ' ± 1 ! ll,
1 — 1 "l
4.S ± i 'K II 4'±>' i^
,4 i- ii )-' 'i
. M -: ± 1 44
™^ j^
• 1 ± "i_ . ' 'h±n '-
,1 —. 'i 'i *-i i Ii
*- -J~ _ t> »1
" >,± *v '!- h
, — K j s_
- i±i 1 ,
, " i — ^ " ^ j
1 II ± ' >'•
i — 1 «- |
Hi| U „
j. — * H 1
"'•* ± i ^u
i "™. i
! '±1^1
i "*— h
I flHdll.il il i-l- till' 114 • ± I I l I ' ± . I I. 1 I il "1 i l '» ± I I j I ± _ • Ii
i,' J t'ui- 4 .- ' I i I I-l i 4 - 4i I 4- —i
I in-ill ,, ,1 1 h TO 4* 1 4~4± l_ ' tr 4: * ' l ± ' . t- '•'-' ± 12 . h M I 44 4' - U ±'" i .' ' ' " ± - ' ~ ' ± I 11. > i 1
tiXn'ltli, --".i. ,--KI ,_'-',i .s-".! ,4'-.-i 4'- "i
( 11 ill .., il In. I IIP I" ±.i 1 I n ± I Ii i. I -\" I f ± I " ' ± I - i|, > I
til |,*>l*l It I-' ll 111-""11 ' — Ii ,1 i-l I'll
Jaj', M! I I) l uu i >i-4 iiL4 ' i,L4 i nj4 ij(i.4 i n;4 1111.4 i M.'4 i'i>.4
,i 11 1 '
.-.. i, .1 Ii ,1 M4 mi i ill I ''.Hit '.nil > il II lit! i til iilf«
i^ 'i' J'
(-.•i i i!>) ,ln« " ', |., i.li Is t I i± f ', i %) i ± I N']I is'i i'."1! i I .'±'11 H%± -Ii
i. 'nl il l.|> "-Hi '!*- 'i .1 '- 'i I-1 'i
1 J l ,|iWu- t i 4 |; 1 ,) L 14* L'.±l),u 1 >*±!4(i At ±I'M. js I ', I il. ±4V I 4V I ±1 J [I ±Mi
i >IM, iK.i.i," 'it-i,n ''-ii . ! ,-i."'i .14-.''', 'ii-i i t-ii.i
ll i (i i fc'.'i. t- ii I i c -' ± i' 1 " . ± i , Ii 11* :*\ '±11 s I ± I
,i'i loilii ,45-.l.] >.'-4i 4 i-4 4i ' -i I 4i
Swl
F-3 DRAFT-DO NOT CITE OR QUOTE
-------
Table F-3: Pharmacokinetic and Pharmacodynamic Parameters from AA and GA
Administration to Mice"
*et
(mgig) n'2 1_ ' L in- 0 .4. . 12 f H_ n >1 _S 2 j
rciite T gi J2f> i gi* £" i!_e ^i ?ze A IL L z r ?™T lc(s si aj«* bet ^* is- Jinlm^
" ?'« V it 1
n *it»m US' E liiAii _JC!L IP.UI -ifoi nit-m ir«-t i>e,i'i iip.ai uieai mt k nr i i eih i i itiOb -iJi ^ '^ f 24 mir-ion iC ''^ 2i Jit ,v
.; it iiv
tin nil n«>J 1 S.73 0.62 1 1 0:74 033 i
till ' -'inJj
in- IL l-l O-I.KL nn 275 32! 333 224
toa ti n il cr .el OJ7 038 0.26 0.67
irui lite UP
til t yi IT d*e m tan 3
Jin "" ' 2r * 'i 11 aea ,rn 12* ^-< 1 ' " f w ~ i
run ** ^ ^^2 ^2 ^ ^ 1*4 ^i^ 4rf ^ st ~^^
G" > ^T^\ L ^ ~ii2i 1* ^ i *! ][("*, ^| U^. ll> t'«2 2M "'- 4 ••* 44" 141
^A LT t miiif* ^n " ^ ^ In 12- ('5~ 1 1 t
tj^ O! t' linn*1 ^HB '^ I 2 f _211 w^iJjnr^ ! *^^ _
jltn «cwiiaua ic [4 - "-4 41 i
^•\ i" in
t-aji.-iA! e, jii "11 4,,4 . ~-v 14 H 4 tl 2 2
UA (t hi
If T ilpi-iA nn ' «!" f fi I -3 li«%, n ^ i,i>i , in ij f ' nj
jlv f
l^j "1-^.iT" nil n ^i lit" i iin 1*1 u i ' HI » >n% h-. i'1 4 ' iiu iiif^-, I 'n
mi " '1 u < ) HI; l a ,u u (ill i Hi
insiierefiddiicfs
ijPiMHiln -i- IL Lr 19.5 11.9 13.4 1400 23.1 12.0 18.5
d 1i X I Z »l 11
iMuxiiilirL- p-*lc 143 146 S3 136 124 19350 151 14S 75.3 131 135
li?C-ili > HI 6.7 4.6 1271 259 6.7 4.6 :i:;3
ii, H f V;
E J i» ivi-aA c'il' a e -t r. ht ->Th thp* ajicf ' t~t -pj nt<.~ 'IEJ.^T in^dllnsj dt c nJ, « w a e» ru_Hi *j3'i a i, c* cr
irm'^luvtr • -^ u ci« "nnuc j nuii-'c ' To; ]Uct rf lin »• cth baJ n? n r » «* ^-nl^ Ji,r ei;h iiue ur • t (a frtn Te."j4
F-4 DRAFT-DO NOT CITE OR QUOTE
-------
Table F-5:. Pharmacokinetic Parameters from AA Administration to Human Volunteers
UEliS
literature ref
dose '.ffis.x;:
route
seacer I.H'I
itoosrii f'j^nrprjozi min
fraction absorbed from
5 101113 dl
AA to GA min" •
AA to AA-GS sum" •
GA to GA-GS liar :
AA to unne mur •
GA to vrme min" -
AA-GS to unne min" •
GA-GS tc unne nun" :
AA
AA-G^
GA
GA-GS
formation of Hb inkr 2
AA RddvKtt
formation of Hb mior J
GA adducTt
tormadcai of hver am""' ^
GA £ddlKt",
dec jv of Ho AA nun" •
addict:
decay of Kb GA raiir!
decay of bver GA IIHU" J
adciiictG
Fiilu-
0 0.24
oet. iiLgle
aials;3i feoalefj^
pharaucokir-ed: rate p?iaoe:en
30 32
0.7e±-)03 0.64 ±U. 15
fu'-j. -order iate cpn.taa'^
235 ± CL13 2.^2 ± 0.26
in 2 1C: 2
16.3 !£.:•
1359 ±273 lfC'5±3fi
222?5 ± 46S-3 1DP3 ± !C"42
3140 ± S3? i?C4 ± lf"3S
13Soi2:i8 9"2±315
ehniinatoK patten- !cc of sere hoc.!
•:5±1.7 f-±lS
76.3 ± 1 5 "5.7 ± 2.2
10.? ± 0 3 &.P ± 0 P
69 ±1.0 7.8 ±1(5
liifiir-Kcoayiiajuic rite cou,?rant?
Boe-cha4-
0 iJi'C23
oral, daily
male 1 3 ,
:?.aoe as -each
individual
froci r "jiff
5S4I 2P
:512 fj^-
25 6 ± 4.5
0.003
(i 003
0.0113
(3)
same a? each
individual
ioai Fiur
861 ± 196
4541 ±2731
32.6 ±16.9
".003
3.003
r..oii3
•'Reft :9ajci;{3.'Ref JJ.
F-5
DRAFT-DO NOT CITE OR QUOTE
-------
References Cited in the above Young et al. (2007 tables
(11) Raymer, J. H., Sparacino, C. M., Velez, G. R., Padilla, S., MacPhail, R. C., and Crofton, K.
M. (1993) Determination of acrylamide in rat serum and sciatic nerve by gas chromatography-
electron-capture detection. J. Chromatogr. 619, 223-234.
(12) Barber, D. S., Hunt, J. R., Ehrich, M. F., Lehning, E. J., and LoPachin, R. M. (2001)
Metabolism, toxicokinetics and hemoglobin adduct formation in rats following subacute and
subchronic acrylamide dosing. NeuroToxicology 22, 341-353.
(13) Kirman, C. R., Gargas, M. L., Deskin, R., Tonner-Navarro, L., and Andersen, M. E. (2003)
A physiologically based pharmacokinetic model for acrylamide and its metabolite, glycidamide,
in the rat. J. Toxicol. EnViron. Health, Part A 66, 253-274.
(14) Sumner, S. C. J., MacNeela, J. P., and Fennell, T. R. (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.
(15) Doerge, D. R., Young, J. F., McDaniel, L. P., Twaddle, N. C., and Churchwell, M. I. (2005)
Toxicokinetics of acrylamide and glycidamide in B6C3F1 mice. Toxicol. Appl. Pharmacol. 202,
258-267.
(16) Doerge, D. R., Young, J. F., McDaniel, L. P., Twaddle, N. C., and Churchwell, M. I. (2005)
Toxicokinetics of acrylamide and glycidamide in Fischer 344 rats. Toxicol. Appl. Pharmacol.
208, 199-209.
(17) Doerge, D. R., Gamboa da Costa, G., McDaniel, L. P., Churchwell, M. I, Twaddle, N. C.,
and Beland, F. A. (2005) DNA adducts derived from administration of acrylamide and
glycidamide to mice and rats. Mutat. Res. 580, 131-142.
(18) Sumner, S. C. J., Williams, C. C., Snyder, R. W., Krol, W. L., Asgharian, B., and Fennell,
T. R. (2003) Acrylamide: A comparison of metabolism and hemoglobin adducts in rodents
following dermal, intraperitoneal, oral, or inhalation exposure. Toxicol. Sci. 75, 260- 270.
(19) Fuhr, U., Boettcher, M. L, Kinzig-Schippers, M., Weyer, A., letter, A., Lazar, A., Taubert,
D., Tomalik-Scharte, D., Pournara, P., Jakob, V., Harlfinger, S., Klaassen, T., Berkessel, A.,
Angerer, J., So"rgel, F., and Schomig, E. (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 Pre V. 15 (2), 266-271.
(20) Fennell, T. R., Sumner, S. C. J., Snyder, R. W., Burgess, J., Spicer, R., Bridson, W. E., and
Friedman, M. A. (2005) Metabolism and hemoglobin adduct formation of acrylamide in humans.
Toxicol. Sci. 85, 447-459.
(21) Boettcher, M. L, Schettgen, T., Kutting, B., Pischetsrieder, M., and Angerer, J. (2005)
Mercapturic acids of acrylamide and glycidamide as biomarkers of the internal exposure to
acrylamide in the general population. Mutat. Res. 580, 167-176.
------- |