DRAFT - DO NOT CITE OR QUOTE EPA/63 5/R-09/005
www.epa.gov/iris
oEPA
TOXICOLOGICAL REVIEW
OF
1,4-Dioxane
(CAS No. 123-91-1)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
May 2009
NOTICE
This document is an External Review draft. This information is distributed solely for the
purpose of pre-dissemination peer review under applicable information quality guidelines. It has
not been formally disseminated by EPA. It does not represent and should not be construed to
represent any Agency determination or policy. It is being circulated for review of its technical
accuracy and science policy implications.
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document is a preliminary draft for review purposes only. This information is
distributed solely for the purpose of pre-dissemination peer review under applicable information
quality guidelines. It has not been formally disseminated by EPA. It does not represent and
should not be construed to represent any Agency determination or policy. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
May 2009 ii DRAFT - DO NOT CITE OR QUOTE
-------
TABLE OF CONTENTS
LIST OF TABLES vii
LIST OF FIGURES x
LIST OF ABBREVIATIONS AND ACRONYMS xiv
FOREWORD xvi
AUTHORS, CONTRIBUTORS, AND REVIEWERS xvii
1. INTRODUCTION 1
2. CHEMICAL AND PHYSICAL INFORMATION 3
3. TOXICOKINETICS 6
3.1. ABSORPTION 6
3.2. DISTRIBUTION 7
3.3. METABOLISM 8
3.4. ELIMINATION 11
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS 12
3.5.1. Available Pharmacokinetic Data 13
3.5.2. Published PBPK Models for 1,4-Dioxane 15
3.5.2.1. Leung and Paustenbach (1990) 15
3.5.2.2. Reitzetal. (1990) 16
3.5.2.3. Fisher etal. (1997) 17
3.5.3. Implementation of Published PBPK Models for 1,4-Dioxane 17
3.6. Rat Nasal Exposure via Drinking Water 20
4. HAZARD IDENTIFICATION 22
4.1. STUDIES IN HUMANS - EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS 22
4.1.1.Thiessetal. (1976) 24
4.1.2. Buffler etal. (1978) 25
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIO AS SAYS IN
ANIMALS - ORAL AND INHALATION 26
4.2.1. OralToxicity 26
4.2.1.1. Subchronic Oral Toxicity 26
4.2.1.1.1. Stoner etal. (1986) 26
4.2.1.1.2. Stott etal. (1981) 27
4.2.1.1.3. Kano etal. (2008) 27
4.2.1.1.4. Yamamoto etal. (1998a, b) 31
4.2.1.2. Chronic Oral Toxicity and Carcinogenicity 32
4.2.1.2.1. Argus etal. (1965) 32
4.2.1.2.2. Argus et al. (1973); Hoch-Ligeti et al. (1970) 32
4.2.1.2.3. Hoch-Ligeti and Argus (1970) 34
4.2.1.2.4. Kociba etal. (1974) 35
May 2009 iii DRAFT - DO NOT CITE OR QUOTE
-------
4.2.1.2.5. National Cancer Institute (NCI) (1978) 37
4.2.1.2.6. Japan Bioassay Research Center (JBRC) (1998a); Yamazaki et al. (1994). 41
4.2.2. Inhalation Toxicity 50
4.2.2.1. Subchronic Inhalation Toxicity 50
4.2.2.1.l.Fairleyetal. (1934) 50
4.2.2.2. Chronic Inhalation Toxicity and Carcinogenicity 51
4.2.2.2.l.Torkelsonetal. (1974) 51
4.2.3. Initiation/Promotion Studies 52
4.2.3.1. Bull etal. (1986) 52
4.2.3.2. King etal. (1973) 53
4.2.3.3. Lundberg etal. (1987) 54
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION 54
4.3.l.Giavini etal. (1985) 54
4.4. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES 55
4.4.1. Acute and Short-term Toxicity 55
4.4.1.1. Oral Toxicity 55
4.4.1.2. Inhalation Toxicity 55
4.4.2. Neurotoxicity 58
4.4.2.1. Frantik etal. (1994) 58
4.4.2.2. Goldberg etal. (1964) 59
4.4.2.3. Kanada etal. (1994) 59
4.4.2.4. Knoefel (1935) 60
4.5. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION 60
4.5.1. Genotoxicity 60
4.5.2. Mechanistic Studies 69
4.5.2.1. Free Radical Generation 69
4.5.2.2. Induction of Metabolism 69
4.5.2.3. Mechanisms of Tumor Induction 69
4.6. SYNTHESIS OF MAJOR NONCANCER EFFECTS 71
4.6.1. Oral 72
4.6.2. Inhalation 75
4.6.3. Mode of Action Information 75
4.7. EVALUATION OF CARCINOGENICITY 77
4.7.1. Summary of Overall Weight of Evidence 77
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence 77
4.7.3. Mode of Action Information 79
4.7.3.1. Identification of Key Events for Carcinogenicity 79
4.7.3.1.1. Liver 79
4.7.3.1.2. Nasal cavity 80
4.7.3.2. Strength, Consistency, Specificity of Association 81
4.7.3.2.1. Liver 81
4.7.3.2.2. Nasal cavity 82
4.7.3.3. Dose-Response Relationship 82
4.7.3.3.1. Liver 82
4.7.3.3.2. Nasal cavity 84
4.7.3.4. Temporal Relationship 84
4.7.3.4.1. Liver 84
4.7.3.4.2. Nasal cavity 84
May 2009 iv DRAFT - DO NOT CITE OR QUOTE
-------
4.7.3.5. Biological Plausibility and Coherence 85
4.7.3.5.1. Liver 85
4.7.3.5.2. Nasal cavity 85
4.7.3.6. Other Possible Modes of Action 85
4.7.3.7. Conclusions About the Hypothesized Mode of Action 86
4.7.3.7.1. Liver 86
4.7.3.7.2. Nasal cavity 86
4.7.3.8. Relevance of the Mode of Action to Humans 86
4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES 87
5. DOSE-RESPONSE ASSESSMENTS 88
5.1. ORAL REFERENCE DOSE (RfD) 88
5.1.1. Choice of Principal Studies and Critical Effect with Rationale and Justification 88
5.1.2. Methods of Analysis—Including Models (PBPK, BMD, etc.) 89
5.1.3. RfD Derivation - Including Application of Uncertainty Factors (UFs) 91
5.1.4. RfD Comparison Information 92
5.1.5. Previous RfD Assessment 97
5.2. INHALATION REFERENCE CONCENTRATION (RfC) 97
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE (RfD) 98
5.4. CANCER ASSESSMENT 100
5.4.1. Choice of Study/Data - with Rationale and Justification 100
5.4.2. Dose-Response Data 101
5.4.3. Dose Adjustments and Extrapolation Method(s) 102
5.4.3.1. Dose Adjustments 102
5.4.3.2. Extrapolation Method(s) 104
5.4.4. Oral Slope Factor and Inhalation Unit Risk 104
5.4.5. Previous Cancer Assessment 106
5.5. UNCERTAINTIES IN CANCER RISK VALUES 106
5.5.1. Sources of Uncertainty 106
5.5.1.1. Choice of Low-Dose Extrapolation Approach 106
5.5.1.2. Dose Metric 108
5.5.1.3. Cross-Species Scaling 108
5.5.1.4. Statistical Uncertainty at the POD 108
5.5.1.5. Bioassay Selection 108
5.5.1.6. Choice of Species/Gender 108
5.5.1.7. Relevance to Humans 109
5.5.1.8. Human Population Variability 109
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE Ill
6.1. HUMAN HAZARD POTENTIAL Ill
6.2. DOSE RESPONSE 112
6.2.1.Noncancer/Oral 112
6.2.2. Noncancer/Inhalation 113
6.2.3. Cancer/Oral 113
6.2.3.1. Choice of Low-Dose Extrapolation Approach 113
6.2.3.2. Dose Metric 114
6.2.3.3. Cross-Species Scaling 115
6.2.3.4. Statistical Uncertainty at the POD 115
May 2009 v DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.5. Bioassay Selection 115
6.2.3.6. Choice of Species/Gender 115
6.2.3.7'. Relevance to Humans 115
6.2.3.8. Human Population Variability 116
6.2.4. Cancer/Inhalation 116
7. REFERENCES 117
APPENDIX A. EXTERNAL REVIEW COMMENTS AND DISPOSITION A-1
APPENDIX B. EVALUATION OF EXISTING PBPK MODELS FOR 1,4-dioxane B-l
APPENDIX C. DETAILS OF BMD ANALYSIS FOR ORAL RfD FOR 1,4-dioxane C-l
APPENDIX D. DETAILS OF BMD ANALYSIS FOR ORAL CSF FOR 1,4-dioxane D-l
May 2009 vi DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF TABLES
Table 2-1. Physical properties and chemical identity of 1,4-dioxane 3
Table 4-1. Incidence of histopathological lesions in F344/DuCrj rats exposed to 1,4-dioxane in
drinking water for 13 weeks 29
Table 4-2. Incidence of histopathological lesions in Crj:BDFi mice exposed to 1,4-dioxane in
drinking water for 13 weeks 31
Table 4-3. Number of incipient liver tumors and hepatomas in male Sprague- Dawley rats
exposed to 1,4-dioxane in drinking water for 13 months 34
Table 4-4. Incidence of liver and nasal tumors in male and female Sherman rats (combined)
treated with 1,4-dioxane in the drinking water for 2 years 37
Table 4-5. Incidence of nonneoplastic lesions in Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water 38
Table 4-6. Incidence of nasal cavity squamous cell carcinoma and liver hepatocellular adenoma
in Osborne-Mendel rats exposed to 1,4-dioxane in drinking water 39
Table 4-7. Incidence of hepatocellular adenoma or carcinoma in B6C3Fi mice exposed to
1,4-dioxane in drinking water 41
Table 4-8. Incidence of histopathological lesions in male F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years 44
Table 4-9. Incidence of histopathological lesions in female F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years 45
Table 4-10. Incidence of nasal cavity, peritoneum, and mammary gland tumors in F344/DuCrj
rats exposed to 1,4-dioxane in drinking water for 2 years 47
Table 4-11. Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane in drinking
water for 2 years 47
Table 4-12. Incidence of histopathological lesions in male Crj:BDFi mice exposed to
1,4-dioxane in drinking water for 2 years 49
Table 4-13. Incidence of histopathological lesions in female Crj :BDFi mice exposed to
1,4-dioxane in drinking water for 2 years 49
Table 4-14. Incidence of liver tumors in Crj :BDFi mice exposed to 1,4-dioxane in drinking
water for 2 years 50
Table 4-15. Acute and short-term toxicity studies of 1,4-dioxane 56
Table 4-16a. Genotoxicity studies of 1,4-dioxane 63
Table 4-16b. Genotoxicity studies of 1,4-dioxane; mammalian in vivo 66
Table 4-17. Oral toxicity studies (noncancer effects) for 1,4-dioxane 72
Table 4-18. Temporal sequence and dose-response relationship for possible key events and liver
tumors in rats and mice 82
Table 5-1. Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed
tol,4-dioxane in drinking water for 2 years 90
Table 5-2. BMD and BMDL values derived from BMD modeling of cortical tubule
degeneration in male and female Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water for 2 years 91
Table 5-3. Incidence of liver hyperplasia in F344/DuCrj rats exposed to 1,4-dioxane in drinking
water for 2 years 91
Table 5-4. BMD and BMDL values derived from BMD modeling of liver hyperplasia in male
and female F344/DuCrj rats exposed to 1,4-dioxane in drinking water for 2 years.. 91
May 2009
vn
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-5. Incidence of liver, nasal cavity, peritoneal, and mammary gland tumors in rats and
mice exposed to 1,4-dioxane in drinking water for 2 years (based on survival to
12 months) 100
Table 5-6. Incidence of hepatocellular adenoma or carcinoma in rats and mice exposed to
1,4-dioxane in drinking water for 2 years 102
Table 5-7. Calculated HEDs for the tumor incidence data used for dose-response modeling ..103
Table 5-8. BMDio HED and BMDLio HED values from models fit to tumor incidence data for rats
and mice exposed to 1,4-dioxane in drinking water for 2 years and corresponding
oralCSFs 105
Table 5-9. Summary of uncertainty in the 1,4-dioxane cancer risk assessment 109
Table B-l. Human PBPK model parameter values for 1,4-dioxane B-9
Table B-2. PBPK metabolic and elimination parameter values resulting from re-calibration of
the human model using alternative values for physiological flow ratesa and tissue:air
partition coefficients B-ll
Table B-3. PBPK metabolic and elimination parameter values resulting from recalibration of the
human model using biologically plausible values for physiological flow ratesa and
selected upper and lower boundary values for tissue:air partition coefficients B-18
Table C-l. Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water for 2 years C-l
Table C-2. Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for cortical tubule degeneration in male and female Osborne-Mendel
rats exposed to 1,4-dioxane in drinking water C-2
Table C-3. Incidence of liver hyperplasia in F344/DuCrj rats exposed to 1,4-dioxane in drinking
water C-5
Table C-4. Benchmark dose modeling results based on the incidence of liver hyperplasias in
F344 male rats exposed to 1,4-dioxane in drinking water for 2 years C-5
Table C-5. Benchmark dose modeling results based on the incidence of liver hyperplasias in
F344 female rats exposed to 1,4-dioxane in drinking water for 2 years C-22
Table D-l. Recommended models for rodents exposed to 1,4-dioxane in drinking water (JBRC,
1998a) D-6
Table D-2. Data for hepatic adenomas and carcinomas in female F344 rats (JBRC, 1998a)... D-7
Table D-3. Summary of BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats D-8
Table D-4. Data for hepatic adenomas and carcinomas in male F344 rats (JBRC, 1998a) D-l5
Table D-5. Summary of BMDS dose-response modeling results for the combined incidence of
adenomas and carcinomas in livers of male F344 rats D-15
Table D-6. Data for significant tumors at other sites in male and female F344 rats D-20
Table D-7. Summary of BMDS dose-response modeling results for the incidence of nasal cavity
tumors in female F344 rats e D-21
Table D-8. Summary of BMDS dose-response modeling results for the incidence of nasal cavity
tumors in maleF344 rats D-26
Table D-9. Summary of BMDS dose-response modeling results for the incidence of mammary
gland adenomas in female F344 rats D-27
Table D-10. Summary of BMDS dose-response modeling results for the incidence of peritoneal
mesotheliomas in maleF344 rats D-32
Table D-l 1. Data for hepatic adenomas and carcinomas in female BDFi mice D-35
Table D-12. Summary of BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in female BDFi mice D-36
Table D-13. Data for hepatic adenomas and carcinomas in male BDFi mice D-38
May 2009
Vlll
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-14. Summary of BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in male BDFi mice D-39
Table D-15. Statistically significant MS dose-response models for F344 rats D-42
Table D-16. MS-combo analysis of excess risks for liver adenomas/ carcinomas, mammary
gland adenomas, or nasal cavity tumors in female F344 rats using MS models... D-43
Table D-17. MS-combo analysis of excess risks for liver adenomas, liver carcinomas, nasal
cavity tumors, or peritoneal mesotheliomas in male F344 rats using MS models D-48
Table D-18. Calculation of FED values for additional studies reporting the incidence of liver
and nasal cavity tumors in rats and mice exposed to 1,4-dioxane in the drinking
water for 2 years D-53
Table D-19. Summary of BMD modeling estimates and CSF values associated with liver and
nasal tumor incidence data resulting from chronic oral exposure to 1,4-dioxane in
rats and mice D-53
Table D-20. Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in male
and female Sherman rats (combined) treated with 1,4-dioxane in the drinking water
for 2 years D-54
Table D-21. Goodness-of-fit statistics and BMDio HED and BMDLio HED from multistage models
fit to incidence data for hepatocellular carcinoma and nasal tumors in male and
female Sherman rats (combined) exposed to 1,4-dioxane in the drinking water for
2 years D-55
Table D-22. Incidence of nasal cavity squamous cell carcinoma and liver hepatocellular
adenoma in Osborne-Mendel rats exposed to 1,4-dioxane in the drinking water. D-57
Table D-23. Goodness-of-fit statistics and BMDio HED and BMDLio HED from multistage models
fit to incidence data for hepatocellular adenoma and nasal tumors in male and female
Osborne-Mendel rats exposed to 1,4-dioxane in the drinking water for 2 years... D-57
Table D-24. Incidence of hepatocellular adenoma or carcinoma in B6C3Fi mice exposed to
1,4-dioxane in drinking water D-61
Table D-25. Goodness-of-fit statistics and BMDio HED and BMDLio HED values from multistage
models fit to incidence data for hepatocellular adenoma or carcinoma in male and
female B6C3Fi mice exposed to 1,4-dioxane in the drinking water for 2 years... D-61
May 2009 ix DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
Figure 2-1. 1,4-Dioxane chemical structure 3
Figure 3-1. Suggested metabolic pathways of 1,4-dioxane in the rat 9
Figure 3-2. Plasma 1,4-dioxane levels in rats following i.v. doses of 3-5,600 mg/kg 10
Figure 3-3. General PBPK model structure consisting of blood-flow limited tissue compartments
connected via arterial and venous blood flows 13
Figure 4-1. A schematic representation of the possible key events in the delivery of 1,4-dioxane
to the liver and the hypothesized MOA(s) for liver carcinogenicity 80
Figure 4-2. A schematic representation of the possible key events in the delivery of 1,4-dioxane
to the nasal cavity and the hypothesized MOA(s) for nasal cavity carcinogenicity.. 81
Figure 5-1. Points of departure (POD) for liver toxicity endpoints with corresponding applied
uncertainty factors and derived RfDs following oral exposure to 1,4-dioxane 94
Figure 5-2. Points of departure (POD) for kidney toxicity endpoints with corresponding applied
uncertainty factors and derived RfDs following oral exposure to 1,4-dioxane 95
Figure 5-3. Points of departure (POD) for nasal inflammation with corresponding applied
uncertainty factors and derived RfDs following oral expsorue to 1,4-dioxane 96
Figure 5-4. Points of departure (POD) for organ specific toxicity endpoints with corresponding
applied uncertainty factors and derived RfDs following oral exposure to 1,4-dioxane.
97
Figure B-l. Schematic representation of empirical model for 1,4-dioxane in rats B-3
Figure B-2. Schematic representation of empirical model for 1,4-dioxane in humans B-3
Figure B-3. Output of 1,4-dioxane blood level data from the acslXtreme implementation (left)
and published (right) empirical rat model simulations of i.v. administration
experiments B-5
Figure B-4. Output of HEAA urine level data from acslXtreme implementation (left) and
published (right) empirical rat model simulations of i.v. administration experiments.
B-5
Figure B-5. acslXtreme predictions of blood 1,4-dioxane and urine HEAA levels from the
empirical rat model simulations of a 6-hour, 50-ppm inhalation exposure B-6
Figure B-6. Output of 1,4-dioxane blood level data from the acslXtreme implementation (left)
and published (right) empirical human model simulations of a 6-hour, 50-ppm
inhalation exposure B-7
Figure B-7. Observations and acslXtreme predictions of cumulative HEAA in human urine
following a 6-hour, 50-ppm inhalation exposure B-8
Figure B-8. Predicted and observed blood 1,4-dioxane concentrations (left) and urinary HEAA
levels (right) following re-calibration of the human PBPK model with tissue:air
partition coefficient values B-12
Figure B-9. Predicted and observed blood 1,4-dioxane concentrations (left) and urinary HEAA
levels (right) following re-calibration of the human PBPK model with tissue:air
partition coefficient values B-12
Figure B-10. Predicted and observed blood 1,4-dioxane concentrations (left) and urinary HEAA
levels (right) using EPA estimated biologically plausible parameters (see Table B-l).
B-13
Figure B-l 1. The highest seven sensitivity coefficients (and associated parameters) for blood
1,4-dioxane concentrations (CV) at 1 (left) and 4 (right) hours of a 50-ppm
inhalation exposure B-l5
May 2009 x DRAFT - DO NOT CITE OR QUOTE
-------
Figure B-12. Comparisons of the range of PBPK model predictions from upper and lower
boundaries on partition coefficients with empirical model predictions and
experimental observations for blood 1,4-dioxane concentrations (left) and urinary
FtEAA levels (right) from a 6-hour, 50-ppm inhalation exposure B-17
Figure B-13. Comparisons of the range of PBPK model predictions from upper and lower
boundaries on partition coefficients with empirical model predictions and
experimental observations for blood 1,4-dioxane concentrations (left) and urinary
FtEAA levels (right) from a 6-hour, 50-ppm inhalation exposure B-17
Figure B-14. Predictions of blood 1,4-dioxane concentration following calibration of a first-
order metabolism rate constant, kLC, to the experimental data B-19
Figure B-15. Predictions of blood 1,4-dioxane concentration following calibration of a first-
order metabolism rate constant, kLC, to only the exposure phase of the experimental
data B-20
Figure B-16. Predictions of blood 1,4-dioxane concentration following simultaneous calibration
of a first-order metabolism rate constant, kLc, and slowly perfused tissue:air partition
coefficient to the experimental data B-21
Figure C-l. BMD log probit model of cortical tubule degeneration incidence data for male rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-2 C-3
Figure C-2. BMD log-logistic model of cortical tubule degeneration incidence data for female
rats exposed to 1,4-dioxane in drinking water for 2 years to support results Table
C-2 C-3
Figure C-3. BMD Weibull model of cortical tubule degeneration incidence data for female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-2 C-4
Figure C-4. BMD gamma model of liver hyperplasia incidence data for F344 male rats exposed
to 1,4-dioxane in drinking water for 2 years to support results Table C-4 C-6
Figure C-5. BMD logistic model of liver hyperplasia incidence data for F344 male rats exposed
to 1,4-dioxane in drinking water for 2 years to support results Table C-4 C-8
Figure C-6. BMD log-logistic model of liver hyperplasia incidence data for F344 male rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-4 C-10
Figure C-7. BMD multistage model of liver hyperplasia incidence data for F344 male rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-4 C-12
Figure C-8. BMD probit model of liver hyperplasia incidence data for F344 male rats exposed to
1,4-dioxane in drinking water for 2 years to support results Table C-4 C-14
Figure C-9. BMD probit model of liver hyperplasia incidence data for F344 male rats exposed to
1,4-dioxane in drinking water for 2 years, accounting for background incidence. C-l6
Figure C-10. BMD quantal linear model of liver hyperplasia incidence data for F344 male rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-4 C-18
Figure C-l 1. BMD Weibull model of liver hyperplasia incidence data for F344 male rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-4 C-20
Figure C-12. BMD gamma model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-22
May 2009 xi DRAFT - DO NOT CITE OR QUOTE
-------
Figure C-13. BMD logistic model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-24
Figure C-14. BMD log-logistic model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-26
Figure C-15. BMD multistage model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-28
Figure C-16. BMD probit model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-30
Figure C-17. BMD log-probit model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-32
Figure C-18. BMD quantal linear model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-34
Figure C-19. BMD Weibull model of liver hyperplasia incidence data for F344 female rats
exposed to 1,4-dioxane in drinking water for 2 years to support the results in Table
C-5 C-36
Figure D-1. Multistage BMD model (2 degree) for the combined incidence of hepatic adenomas
and carcinomas in female F344 rats D-9
Figure D-2. Multistage BMD model (1 & 2 degree) for the combined incidence of hepatic
adenomas and carcinomas in female F344 rats D-ll
Figure D-3. Log-logistic BMD model for the combined incidence of hepatic adenomas and
carcinomas in female F344 rats D-13
Figure D-4. Multistage BMD model (1 & 2 degree) for the combined incidence of hepatic
adenomas and carcinomas in maleF344 rats D-16
Figure D-5. Multistage BMD model (1 & 8 degree) for the combined incidence of hepatic
adenomas and carcinomas in maleF344 rats D-18
Figure D-6. Multistage BMD model (2 degree) for the nasal cavity tumors in female F344 rats.
D-22
Figure D-7. Multistage BMD model (8 degree) for the nasal cavity tumors in female F344 rats.
D-24
Figure D-8. Multistage BMD model (1 degree) for mammary gland adenomas in female F344
rats D-28
Figure D-9. Multistage BMD model (2 degree) for mammary gland adenomas in female F344
rats D-30
Figure D-10. Multistage BMD model (2 degree) for peritoneal mesotheliomas in male F344 rats.
D-33
Figure D-ll. Log-logistic BMD model (Fixed power=l) for the combined incidence of hepatic
adenomas and carcinomas in female BDFi mice D-36
Figure D-12. Log-logistic BMD model (Fixed power=l) for the combined incidence of hepatic
adenomas and carcinomas in male BDFi mice D-40
Figure D-13. BMD multistage model (1-degree polynomial) of the incidence of hepatocellular
carcinoma in male and female Sherman rats exposed to 1,4-dioxane in drinking
water D-55
May 2009 xii DRAFT - DO NOT CITE OR QUOTE
-------
Figure D-14. BMD multistage model (1-degree polynomial) of the incidence of nasal squamous
cell carcinoma in male and female Sherman rats exposed to 1,4-dioxane in drinking
water D-56
Figure D-15. BMD multistage model (1-degree polynomial) of the incidence of nasal squamous
cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water D-58
Figure D-16. BMD multistage model (1-degree polynomial) of the incidence of nasal squamous
cell carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water D-59
Figure D-17. BMD multistage model (1-degree polynomial) of the incidence of hepatocellular
adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water.
D-60
Figure D-18. BMD multistage model (2-degree polynomial) of the incidence of hepatocellular
adenoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking water D-62
Figure D-19. BMD multistage model (2-degree polynomial) of the incidence of hepatocellular
adenoma in female B6C3Fi mice exposed to 1,4-dioxane in drinking water D-63
May 2009 xiii DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS
AIC
ALP
ALT
AST
ATSDR
BMD
BMDL
BMDLio
BMDS
BMR
BrdU
BUN
BW(s)
CASE
CASRN
CHO
CI
CNS
CPK
CREST
CSF
CV
CYP450
DEN
FISH
G-6-Pase
GC
GGT
HEAA
HED(s)
HPLC
HSDB
Hz
IARC
i.p.
i.v.
IRIS
JBRC
ke
kLc
Km
kme
LAP
LD50
Akaike's Information Criterion
alkaline phosphatase
alanine aminotransferase
aspartate aminotransferase
Agency for Toxic Substances and Disease Registry
benchmark dose
benchmark dose at 10% extra risk
benchmark dose, lower 95% confidence limit
benchmark dose, lower 95% confidence limit at 10% extra risk
Benchmark Dose Software
benchmark response
5-bromo-2'-deoxyuridine
blood urea nitrogen
body weight(s)
computer automated structure evaluator
Chemical Abstracts Service Registry Number
Chinese hamster ovary (cells)
confidence interval(s)
central nervous system
creatinine phosphokinase
antikinetochore
cancer slope factor
concentration in venous blood
cytochrome P450
diethylnitrosamine
fluorescence in situ hybridization
glucose-6-phosphatase
gas chromatography
y-glutamyl transpeptidase
p-hydroxyethoxy acetic acid
human equivalent dose(s)
high-performance liquid chromatography
Hazardous Substances Data Bank
Hertz
International Agency for Research on Cancer
intraperitoneal
intravenous
Integrated Risk Information System
Japan Bioassay Research Center
1st order elimination rate of 1,4-dioxane
1 st order 1 ,4-dioxane inhalation rate constant
1st order, non-saturable metabolism rate constant for 1,4-dioxane in the liver
Michaelis constant for metabolism of 1,4-dioxane in the liver
1 st order elimination rate of HEAA ( 1 ,4-dioxane metabolite)
leucine aminopeptidase
median lethal dose
May 2009
xiv
DRAFT - DO NOT CITE OR QUOTE
-------
LDH lactate dehydrogenase
LOAEL lowest-observed-adverse-effect-level
MCV mean corpuscular volume
MOA mode of action
MS mass spectrometry, multi-stage
MTD maximum tolerated dose
MVK Moolgavkar-Venzon-Knudsen (model)
NCE normochromatic erythrocyte
NCI National Cancer Institute
ND no data, not detected
NE not estimated
NOAEL no-observed-adverse-effect-level
NRC National Research Council
NTP National Toxicology Program
OCT ornithine carbamyl transferase
ODC ornithine decarboxylase
PB blood:air partition coefficient
PBPK physiologically based pharmacokinetic
PC partition coefficient
PCB polychlorinated biphenyl
PCE polychromatic erthyrocyte
PFA fat:air partition coefficient
PLA liverair partition coefficient
POD point of departure
ppm parts per million
PRA rapidly perfused tissue:air partition coefficient
PSA slowy perfused tissue:air partition coefficient
QCC normalized cardiac output
QPC normalized alveolar ventilation rate
RBC red blood cell
RfC inhalation reference concentration
RfD oral reference dose
SCE sister chromatid exchange
SDH sorbitol dehydrogenase
SMR standardized mortality ratio
SRC Syracuse Research Corporation
TPA 12-O-tetradecanoylphorbol-13-acetate
TWA time-weighted average
UF uncertainty factor
UNEP United Nations Environment Programme
U.S. EPA U.S. Environmental Protection Agency
V volts
VAS visual analogue scale
Va volume of distribution
Vmax maximal rate of metabolism
Vmaxc normalized maximal rate of metabolism of 1,4-dioxane in liver
VOC(s) volatile organic compound(s)
WBC white blood cell
X2 Chi-squared
May 2009
xv
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
1,4-dioxane. It is not intended to be a comprehensive treatise on the chemical or toxicological
nature of 1,4-dioxane.
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 the data and related uncertainties. The discussion is intended to convey the
limitations of the assessment and to aid and guide the risk assessor in the ensuing steps of the
risk assessment process.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
May 2009 xvi DRAFT - DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGERS
Eva D. McLanahan
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Reeder Sams, II
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
AUTHORS AND CONTRIBUTORS
Hi sham El-Masri
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Karen Hogan
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
Fernando Llados
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Michael Lumpkin
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Allan Marcus
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Eva D. McLanahan
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
May 2009 xvii DRAFT - DO NOT CITE OR QUOTE
-------
Marc Odin
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Susan Rieth
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
Andrew Rooney
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Reeder Sams, II
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Paul Schlosser
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Julie Stickney
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
John Vandenberg
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
REVIEWERS
This document has been peer reviewed by EPA scientists, interagency reviewers from
other federal agencies, and the public, and peer reviewed by 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. A summary and EPA's disposition of the
comments received from the independent external peer reviewers and from the public is included
in Appendix A of the Toxicological Review of 1,4-dioxane.
May 2009 xviii DRAFT - DO NOT CITE OR QUOTE
-------
INTERNAL EPA REVIEWERS
Anthony DeAngelo
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Hi sham El-Masri
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Nagu Keshava
National Center for Environmental Assessment
Office of Research and Development
Jason Lambert
National Center for Environmental Assessment
Office of Research and Development
Connie Meacham
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Paul Schlosser
National Center for Environmental Assessment
Office of Research and Development
Debra Walsh
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Douglas Wolf
National Health and Environmental Effects Research Laboratory
Office of Research and Development
May 2009 xix DRAFT - DO NOT CITE OR QUOTE
-------
1. INTRODUCTION
1 This document presents background information and justification for the Integrated Risk
2 Information System (IRIS) Summary of the hazard and dose-response assessment of
3 1,4-dioxane. IRIS Summaries may include oral reference dose (RfD) and inhalation reference
4 concentration (RfC) values for chronic and subchronic exposure durations, and a carcinogenicity
5 assessment.
6 The RfD and RfC, if derived, provide quantitative information for use in risk assessments
7 for health effects known or assumed to be produced through a nonlinear (presumed threshold)
8 mode of action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with
9 uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
10 population (including sensitive subgroups) that is likely to be without an appreciable risk of
11 deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m3) is
12 analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
13 inhalation RfC considers toxic effects for both the respiratory system (portal-of-entry) and for
14 effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
15 values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
16 acute (< 24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
17 lifetime) exposure durations, all of which are derived based on an assumption of continuous
18 exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
19 derived for chronic exposure durations.
20 The carcinogenicity assessment provides information on the carcinogenic hazard
21 potential of the substance in question and quantitative estimates of risk from oral and inhalation
22 exposure may be derived. The information includes a weight-of-evidence judgment of the
23 likelihood that the agent is a human carcinogen and the conditions under which the carcinogenic
24 effects may be expressed. Quantitative risk estimates may be derived from the application of a
25 low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
26 the estimate of risk per mg/kg-day of oral exposure. Similarly, an inhalation unit risk is a
27 plausible upper bound on the estimate of risk per ug/m3 air breathed.
28 Development of these hazard identification and dose-response assessments for
29 1,4-dioxane has followed the general guidelines for risk assessment as set forth by the National
30 Research Council (NRC, 1983). EPA guidelines and Risk Assessment Forum Technical Panel
31 Reports that may have been used in the development of this assessment include the following:
32 Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S. EPA, 1986a), Guidelines
3 3 for Mutagenicity Risk Assessment (U. S. EPA, 1986b), Recommendations for and Documentation
34 of'Biological Values for Use in Risk Assessment (U.S. EP'A, 1988), Guidelines for
3 5 Developmental Toxicity Risk Assessment (U.S. EPA, 1991), Interim Policy for Particle Size and
36 Limit Concentration Issues in Inhalation Toxicity (U.S. EPA, 1994a), Methods for Derivation of
May 2009 1 DRAFT - DO NOT CITE OR QUOTE
-------
1 Inhalation Reference Concentrations and Application of Inhalation Dosimetry (U. S. EPA,
2 1994b), Use of the Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995),
3 Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA, 1996), Guidelines for
4 Neurotoxicity Risk Assessment (U.S. EPA, 1998), Science Policy Council Handbook. Risk
5 Characterization (U.S. EPA, 2000a), Benchmark Dose Technical Guidance Document (U.S.
6 EPA, 2000b), Supplementary Guidance for Conducting Health Risk Assessment of Chemical
1 Mixtures (U.S. EPA, 2000c), A Review of the Reference Dose and Reference Concentration
8 Processes (U.S. EPA, 2002a), Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a),
9 Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
10 (U.S. EPA, 2005b), Science Policy Council Handbook: Peer Review (U.S. EPA, 2006a), and A
11 Framework for Assessing Health Risks of Environmental Exposures to Children (U. S. EPA,
12 2006b).
13 The literature search strategy employed for this compound was based on the Chemical
14 Abstracts Service Registry Number (CASRN) and at least one common name. Any pertinent
15 scientific information submitted by the public to the IRIS Submission Desk was also considered
16 in the development of this document. The relevant literature was reviewed through August 2008.
May 2009 2 DRAFT - DO NOT CITE OR QUOTE
-------
2. CHEMICAL AND PHYSICAL INFORMATION
1 1,4-Dioxane, a volatile organic compound (VOC), is a colorless liquid with a pleasant
2 odor (Lewis, 2001, 2000). Synonyms include diethylene ether, 1,4-diethylene dioxide,
3 diethylene oxide, dioxyethylene ether, and dioxane (Lewis, 2001). The chemical structure of
4 1,4-dioxane is shown in Figure 2-1. Selected chemical and physical properties of this substance
5 are listed below:
Figure 2-1. 1,4-Dioxane chemical structure.
Table 2-1. Physical properties and chemical identity of 1,4-dioxane
CASRN:
Molecular weight:
Chemical formula:
Boiling point:
Melting point:
Vapor pressure:
Density:
Vapor density:
Water solubility:
Other solubilities:
Log Kow:
Henry's Law constant:
OH reaction rate constant:
Koc:
Bioconcentration factor:
Conversion factors (in air):
123-91-1 (Lide, 2000)
88.10(O'Neil, 2001)
C4H8O2(O'Neil, 2001)
101.l°C(O'Neil, 2001)
11.8°C (Lide, 2000)
40 mmHg at 25°C (Lewis, 2000)
1.0337 g/mL at 20°C (Lide, 2000)
3.03 (air = 1) (Lewis, 2000)
Miscible with water (Lewis, 2001)
Miscible with ethanol, ether, and acetone (Lide, 2000)
-0.27 (Hansch et al., 1995)
4.80 x 10"6 atm-nrVmolecule at 25°C (Park et al., 1987)
1.09 x 10"11 cnrVmolecule sec at25°C (Atkinson, 1989)
17 (estimated using log Kow) (Lyman et al., 1990)
0.4 (estimated using log Kow) (Meylan et al., 1999)
1 ppm = 3.6 mg/m3; 1 mg/m3 = 0.278 ppm
(25°C and 1 atm) (HSDB, 2007)
May 2009
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-Dioxane is produced commercially through the dehydration and ring closure
2 of diethylene glycol (Surprenant, 2002). Concentrated sulfuric acid is used as a catalyst
3 (Surprenant, 2002). This is a continuous distillation process with operating temperatures and
4 pressures of 130-200°C and 188-825 mmHg, respectively (Surprenant, 2002). During the years
5 1986 and 1990, the U.S. production of 1,4-dioxane reported by manufacturers was within the
6 range of 10-50 million pounds (U.S. EPA, 2002b). The production volume reported during the
7 years 1994, 1998, and 2002 was within the range of 1-10 million pounds (U.S. EPA, 2002b).
8 Historically, 1,4-dioxane has been used as a stabilizer for the solvent 1,1,1-trichloro-
9 ethane (Suprenant, 2002). However, this use is no longer expected to be important due to the
10 1990 Amendments to the Clean Air Act and the Montreal Protocol, which mandate the eventual
11 phase-out of 1,1,1-trichloroethane production in the U.S. (ATSDR, 2007; 2006; UNEP, 2000;
12 U.S. EPA, 1990). 1,4-Dioxane is also used as a solvent for cellulosics, organic products,
13 lacquers, paints, varnishes, paint and varnish removers, resins, oils, waxes, dyes, cements,
14 cosmetics, deodorants, fumigants, emulsions, and polishing compositions (Lewis, 2001; O'Neil,
15 2001; IARC, 1999). 1,4-Dioxane has been used as a solvent in the formulation of inks, coatings,
16 and adhesives and in the extraction of animal and vegetable oil (Suprenant, 2002). Reaction
17 products of 1,4-dioxane are used in the manufacture of insecticides, herbicides, plasticizers, and
18 monomers (Suprenant, 2002).
19 When 1,4-dioxane enters the air, it will exist as a vapor, as indicated by its vapor pressure
20 (HSDB, 2007). It is expected to be degraded in the atmosphere through photooxidation with
21 hydroxyl radicals (HSDB, 2007; Suprenant, 2002). The estimated half-life for this reaction is
22 6.7 hours (HSDB, 2007). It may also be broken down by reaction with nitrate radicals, although
23 this removal process is not expected to compete with hydroxyl radical photooxidation (Grosjean,
24 1990). 1,4-Dioxane is not expected to undergo direct photolysis (Wolfe and Jeffers, 2000).
25 1,4-Dioxane is primarily photooxidized to 2-oxodioxane and through reactions with nitrogen
26 oxides (NOX) results in the formation of ethylene glycol diformate (Platz et al., 1997).
27 1,4-Dioxane is expected to be highly mobile in soil based on its estimated Koc and is expected to
28 leach to lower soil horizons and groundwater (ATSDR, 2007; Lyman et al., 1990). This
29 substance may volatilize from dry soil surfaces based on its vapor pressure (HSDB, 2007). The
30 estimated bioconcentration factor value indicates that 1,4-dioxane will not bioconcentrate in
31 aquatic or marine organisms (Meylan et al., 1999; Franke et al., 1994). 1,4-Dioxane is not
32 expected to undergo hydrolysis or to biodegrade readily in the environment (HSDB, 2007;
33 ATSDR, 2007). Therefore, volatilization is expected to be the dominant removal process for
34 moist soil and surface water. Based on a Henry's Law constant of 4.8x 10"6 atm-m3/mole, the
35 half-life for volatilization of 1,4-dioxane from a model river is 5 days and that from a model lake
36 is 56 days (HSDB, 2007; Lyman et al., 1990; Park et al., 1987). 1,4-Dioxane may be more
37 persistent in groundwater where volatilization is hindered.
May 2009 4 DRAFT - DO NOT CITE OR QUOTE
-------
1 Recent environmental monitoring data for 1,4-dioxane are lacking. Existing data indicate
2 that 1,4-dioxane may leach from hazardous waste sites into drinking water sources located
3 nearby (Yasuhara et al., 2003, 1997; Lesage et al., 1990). 1,4-Dioxane has been detected in
4 contaminated surface and groundwater samples collected near hazardous waste sites and
5 industrial facilities (DeRosa et al., 1996).
May 2009 5 DRAFT - DO NOT CITE OR QUOTE
-------
3. TOXICOKINETICS
1 Data for the toxicokinetics of 1,4-dioxane in humans are very limited. However,
2 absorption, distribution, metabolism, and elimination of 1,4-dioxane are well described in rats
3 exposed via the oral, inhalation, or intravenous (i.v.) routes. 1,4-Dioxane is extensively absorbed
4 and metabolized in humans and rats to p-hydroxyethoxy acetic acid (HEAA), which is
5 predominantly excreted in the urine. Saturation of 1,4-dioxane metabolism has been observed in
6 rats and would be expected in humans; however, human exposure levels associated with
7 nonlinear toxicokinetics are not known.
8 Important data elements that have contributed to our current understanding of the
9 toxicokinetics of 1,4-dioxane are summarized in the following sections.
3.1. ABSORPTION
10 Absorption of 1,4-dioxane following inhalation exposure has been qualitatively
11 demonstrated in workers and volunteers. Workers exposed to a time-weighted average (TWA)
12 of 1.6 parts per million (ppm) of 1,4-dioxane in air for 7.5 hours showed a HEAA/l,4-dioxane
13 ratio of 118:1 in urine (Young et al., 1976). The authors assumed lung absorption to be 100%
14 and calculated an average absorbed dose of 0.37 mg/kg, although no exhaled breath
15 measurements were taken. In a study with four healthy male volunteers, Young et al. (1977)
16 reported 6-hour inhalation exposures of adult volunteers to 50 ppm of 1,4-dioxane in a chamber,
17 followed by blood and urine analysis for 1,4-dioxane and HEAA. The study protocol was
18 approved by a seven-member Human Research Review Committee of the Dow Chemical
19 Company, and written informed consent of study participants was obtained. At a concentration
20 of 50 ppm, uptake of 1,4-dioxane into plasma was rapid and approached steady-state conditions
21 by 6 hours. The authors reported a calculated absorbed dose of 5.4 mg/kg. However, the
22 exposure chamber atmosphere was kept at a constant concentration of 50 ppm and exhaled
23 breath was not analyzed. Accordingly, gas uptake could not be measured. As a result, the
24 absorbed fraction of inhaled 1,4-dioxane could not be accurately determined in humans. Rats
25 inhaling 50 ppm for 6 hours exhibited 1,4-dioxane and HEAA in urine with an HEAA to
26 1,4-dioxane ratio of over 3,100:1 (Young et al., 1978a, b). Plasma concentrations at the end of
27 the 6-hour exposure period averaged 7.3 ug/mL. The authors calculated an absorbed 1,4-dioxane
28 dose of 71.9 mg/kg; however, the lack of exhaled breath data and dynamic exposure chamber
29 precluded the accurate determination of the inhaled fraction of 1,4-dioxane.
30 No human data are available to evaluate the oral absorption of 1,4-dioxane.
31 Gastrointestinal absorption was nearly complete in male Sprague Dawley rats orally dosed with
32 10-1,000 mg/kg of [14C]-l,4-dioxane given as a single dose or as 17 consecutive daily doses
May 2009 6 DRAFT - DO NOT CITE OR QUOTE
-------
1 (Young et al., 1978a, b). Cumulative recovery of radiolabel in the feces was
-------
1 would be unknown had saline perfusion been performed. Covalent binding reached peak
2 percentages at 6 hours after dosing in liver (18.5%), spleen (22.6%), and colon (19.5%). At
3 16 hours after dosing, peak covalent binding percentages were observed in whole blood (3.1%),
4 kidney (9.5%), lung (11.2%), and skeletal muscle (11.2%). Within hepatocytes, radiolabel
5 distribution at 6 hours after dosing was greatest in the cytosolic fraction (43.8%) followed by the
6 microsomal (27.9%), mitochondrial (16.6%), and nuclear (11.7%) fractions. While little
7 covalent binding of radiolabel was measured in the hepatic cytosol (4.6%), greater binding was
8 observed at 16 hours after dosing in the nuclear (64.8%), mitochondrial (45.7%), and
9 microsomal (33.4%) fractions. Pretreatment with inducers of mixed-function oxidase activity
10 did not significantly change the extent of covalent binding in subcellular fractions.
3.3. METABOLISM
11 The major product of 1,4-dioxane metabolism appears to be HEAA, although there is
12 one report that identified l,4-dioxane-2-one as a major metabolite (Woo et al., 1977b).
13 However, the presence of this compound in the sample was believed to result from the acidic
14 conditions (pH of 4.0-4.5) of the analytical procedures. The reversible conversion of HEAA and
15 p-l,4-dioxane-2-one is pH-dependent (Braun and Young, 1977). Braun and Young (1977)
16 identified HEAA (85%) as the major metabolite, with most of the remaining dose excreted as
17 unchanged 1,4-dioxane in the urine of Sprague Dawley rats dosed with 1,000 mg/kg of
18 uniformly labeled l,4-[14C]dioxane. In fact, toxicokinetic studies of 1,4-dioxane in humans and
19 rats (Young et al., 1978a, b, 1977) employed an analytical technique that converted HEAA to the
20 more volatile dioxanone prior to gas chromatography (GC).
21 A proposed metabolic scheme for 1,4-dioxane metabolism (Woo et al., 1977b) in
22 Sprague Dawley rats is shown in Figure 3-1. Oxidation of 1,4-dioxane to diethylene glycol
23 (pathway a), l,4-dioxane-2-ol (pathway c), or directly to l,4-dioxane-2-one (pathway b) could
24 result in the production of HEAA. 1,4-Dioxane oxidation appears to be cytochrome P450
25 (CYP450)-mediated, as CYP450 induction with phenobarbital or Aroclor 1254 (a commercial
26 PCB mixture) and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine or cobaltous
27 chloride were effective in significantly increasing and decreasing, respectively, the appearance of
28 HEAA in the urine of Sprague Dawley rats (Woo et al., 1978, 1977c). 1,4-Dioxane itself
29 induced CYP450-mediated metabolism of several barbiturates in Hindustan mice given i.p.
30 injections of 25 and 50 mg/kg 1,4-dioxane (Mungikar and Pawar, 1978). Of the three possible
31 pathways proposed in this scheme, oxidation to diethylene glycol and HEAA appears to be the
32 most likely, because diethylene glycol was found as a minor metabolite in Sprague Dawley rat
33 urine following a single 1,000 mg/kg gavage dose of 1,4-dioxane (Braun and Young, 1977).
34 Additionally, i.p. injection of 100-400 mg/kg diethylene glycol in Sprague Dawley rats resulted
35 in urinary elimination of HEAA (Woo et al., 1977a).
36
May 2009 8 DRAFT - DO NOT CITE OR QUOTE
-------
O
[V] [VI] \^
HOH2C CHaOH HOH2C o
(b)""""-
-H20
Source: Adapted from Woo et al. (1977b, c).
Figure 3-1. Suggested metabolic pathways of 1,4-dioxane in the rat.
I = 1,4-dioxane; II = diethylene glycol; III = P-hydroxyethoxy acetic acid (HEAA);
IV = l,4-dioxane-2-one; V = l,4-dioxane-2-ol; VI = P-hydroxyethoxy acetaldehyde.
Note: Metabolite [V] is a likely intermediate in pathway b as well as pathway c.
The proposed pathways are based on the metabolites identified; the enzymes
responsible for each reaction have not been determined. The proposed pathways do
not account for metabolite degradation to the labeled carbon dioxide (COi)
identified in expired air after labeled 1,4-dioxane exposure.
1 Metabolism of 1,4-dioxane in humans is extensive. In a survey of 1,4-dioxane plant
2 workers exposed to a TWA of 1.6 ppm of 1,4-dioxane for 7.5 hours, Young et al. (1976) found
3 HEAA and 1,4-dioxane in the worker's urine at a ratio of 118:1. Similarly, in adult male
4 volunteers exposed to 50 ppm for 6 hours (Young et al., 1977), over 99% of inhaled 1,4-dioxane
5 (assuming negligible exhaled excretion) appeared in the urine as HEAA. The linear elimination
6 of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism was a
7 nonsaturated, first-order process at this exposure level.
8 Like humans, rats extensively metabolize inhaled 1,4-dioxane, as HEAA content in urine
9 was over 3,000-fold higher than that of 1,4-dioxane following exposure to 50 ppm for 6 hours
10 (Young et al., 1978a, b). 1,4-Dioxane metabolism in rats was a saturable process, as exhibited
11 by oral and i.v. exposures to various doses of [14C]-1,4-dioxane (Young et al., 1978a, b). Plasma
May 2009 9 DRAFT - DO NOT CITE OR QUOTE
-------
1 data from Sprague Dawley rats given single i.v. doses of 3, 10, 30, 100, 300, or 1,000 mg
2 [14C]-l,4-dioxane/kg demonstrated a dose-related shift from linear, first-order to nonlinear,
3 saturable metabolism of 1,4-dioxane between plasma 1,4-dioxane levels of 30 and 100 ug/mL
4 (Figure 3-2). Similarly, in rats given, via gavage in distilled water, 10, 100, or 1,000 mg
5 [14C]-l,4-dioxane/kg singly or 10 or 1,000 mg [14C]-l,4-dioxane/kg in 17 daily doses, the
6 percent urinary excretion of the radiolabel decreased significantly with dose while radiolabel in
7 expired air increased. Specifically, with single [14C]-l,4-dioxane/kg doses, urinary radiolabel
8 decreased from 99 to 76% and expired 1,4-dioxane increased from <1 to 25% as dose increased
9 from 10 to 1,000 mg/kg. Likewise, with multiple daily doses 10 or 1,000 mg
10 [14C]-l,4-dioxane/kg, urinary radiolabel decreased from 99 to 82% and expired 1,4-dioxane
11 increased from 1 to 9% as dose increased. The differences between single and multiple doses in
12 urinary and expired radiolabel support the notion that 1,4-dioxane may induce its own
13 metabolism.
10,000
T.OOO -
' 5 10 IS 20 25 30 35 4O 45 5O b
Source: Young etal. (1978a).
Figure 3-2. Plasma 1,4-dioxane levels in rats following i.v. doses of 3-5,600 mg/kg.
May 2009
10
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-Dioxane has been shown to induce several isoforms of CYP450 in various tissues
2 following acute oral administration by gavage or drinking water (Nannelli et al., 2005). Male
3 Sprague Dawley rats were exposed to either 2,000 mg/kg 1,4-dioxane via gavage for
4 2 consecutive days or by ingestion of a 1.5% 1,4-dioxane drinking water solution for 10 days.
5 Both exposures resulted in significantly increased CYP2B1/2, CYP2C11, and CYP2E1 activities
6 in hepatic microsomes. The gavage exposure alone resulted in increased CYP3A activity. The
7 increase in 2C11 activity was unexpected, as that isoform has been observed to be under
8 hormonal control and was typically suppressed in the presence of 2B1/2 and 2E1 induction. In
9 the male rat, hepatic 2C11 induction is associated with masculine pulsatile plasma profiles of
10 growth hormone (compared to the constant plasma levels in the female), resulting in
11 masculinization of hepatocyte function (Waxman et al., 1991). The authors postulated that
12 1,4-dioxane may alter plasma growth hormone levels, resulting in the observed 2C11 induction.
13 However, growth hormone induction of 2C11 is primarily dependent on the duration between
14 growth hormone pulses and secondarily on growth hormone plasma levels (Agrawal and
15 Shapiro, 2000; Waxman et al., 1991). Thus, the induction of 2C11 by 1,4-dioxane may be
16 mediated by changes in the time interval between growth hormone pulses rather than changes
17 in growth hormone levels. This may be accomplished by 1,4-dioxane temporarily influencing
18 the presence of growth hormone cell surface binding sites (Agrawal and Shapiro, 2000).
19 However, no studies are available to confirm the influence of 1,4-dioxane on either growth
20 hormone levels or changes in growth hormone pulse interval.
21 In nasal and renal mucosal cell microsomes, CYP2E1 activity, but not CYP2B1/2
22 activity, was increased. Pulmonary mucosal CYP450 activity levels were not significantly
23 altered. Observed increases in 2E1 mRNA in rats exposed by gavage and i.p. injection suggest
24 that 2E1 induction in kidney and nasal mucosa is controlled by a transcriptional activation of
25 2E1 genes. The lack of increased mRNA in hepatocytes suggests that induction is regulated via
26 a post-transcriptional mechanism. Differences in 2E1 induction mechanisms in liver, kidney,
27 and nasal mucosa suggest that induction is controlled in a tissue-specific manner.
3.4. ELIMINATION
28 In workers exposed to a TWA of 1.6 ppm for 7.5 hours, 99% of 1,4-dioxane eliminated in
29 urine was in the form of HEAA (Young et al., 1976). The elimination half-life was 59 minutes
30 in adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours, with 90% of urinary
31 1,4-dioxane and 47% of urinary HEAA excreted within 6 hours of onset of exposure (Young
32 et al., 1977). There are no data for 1,4-dioxane elimination in humans from oral exposures.
33 Elimination of 1,4-dioxane in rats (Young et al., 1978a, b) was primarily via urine. Like
34 humans, the elimination half-life in rats exposed to 50 ppm 1,4-dioxane for 6 hours was
35 calculated to be 1.01 hours. In Sprague Dawley rats given single daily doses of 10, 100, or
36 1,000 mg [14C]-l,4-dioxane/kg or multiple doses of 10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary
May 2009 11 DRAFT - DO NOT CITE OR QUOTE
-------
1 radiolabel ranged from 99% down to 76% of total radiolabel. Fecal elimination was less than
2 2% for all doses. The effect of saturable metabolism on expired 1,4-dioxane was apparent, as
3 expired 1,4-dioxane in singly dosed rats increased with dose from 0.4 to 25% while expired
4 14CC>2 changed little (between 2 and 3%) across doses. The same relationship was seen in
5 Sprague Dawley rats dosed i.v. with 10 or 1,000 mg [14C]-l,4-dioxane/kg. Higher levels of
6 14CC>2 relative to 1,4-dioxane were measured in expired air of the 10 mg/kg group, while higher
7 levels of expired 1,4-dioxane relative to 14CC>2 were measured in the 1,000 mg/kg group.
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS
8 PBPK models have been developed for 1,4-dioxane in rats and humans (Leung and
9 Paustenbach, 1990; Reitz et al., 1990) and lactating humans (Fisher et al., 1997). Each of the
10 models simulates the body as a series of compartments representing tissues or tissue groups that
11 receive blood from the central vascular compartment (Figure 3-3). Modeling was conducting
12 under the premise that transfers of 1,4-dioxane between blood and tissues occur sufficiently fast
13 to be effectively blood flow-limited, which is consistent with the available data (Ramsey and
14 Andersen, 1984). Blood time course and metabolite production data in rats and humans suggest
15 that absorption and metabolism are accomplished through common mechanisms in both species
16 (Young et al., 1978a, b, 1977), allowing identical model structures to be used for both species
17 (and by extension, for mice as well). In all three models, physiologically relevant, species-
18 specific parameter values for tissue volume, blood flow, and metabolism and elimination are
19 used. The models and supporting data are reviewed below, from the perspective of assessing
20 their utility for predicting internal dosimetry and for cross-species extrapolation of exposure-
21 response relationships for critical neoplastic and non-neoplastic endpoints (also see Appendix B).
May 2009 12 DRAFT - DO NOT CITE OR QUOTE
-------
IV
•a
o
a
DQ
en
O
£Z
QJ
infusion inhalation
I 4 t
Gl
Lung blood
t
Rapidly
perfused
Slowly
perfused
'
Fat
V J
*•
Liver
V .,
T
•a
o
DD
QJ
-id
absorption metabolism
Figure 3-3. General PBPK model structure consisting of blood-flow limited
tissue compartments connected via arterial and venous blood flows. Note: Orally
administered chemicals are absorbed directly into the liver while inhaled and
intravenously infused chemicals enter directly into the arterial and venous blood
pools, respectively.
3.5.1. Available Pharmacokinetic Data
1 Animal and human data sets available for model calibration derive from Young et al.
2 (1978a, b, 1977), Mikheev et al. (1990), and Woo et al. (1977a, b). Young et al. (1978a, b)
3 studied the disposition of radiolabeled [14C]-l,4-dioxane in adult male Sprague Dawley rats
4 following i.v., inhalation, and single and multiple oral gavage exposures. Plasma concentration-
5 time profiles were reported for i.v. doses of 3, 10, 30, 100, and 1,000 mg/kg. In addition,
6 exhaled 14CC>2 and urinary 1,4-dioxane and HEAA profiles were reported following i.v. doses of
7 10 and 1,000 mg/kg. The plasma 1,4-dioxane concentration-time course, cumulative urinary
8 1,4-dioxane and cumulative urinary HEAA concentrations were reported following a 6-hour
9 inhalation exposure to 50 ppm. Following oral gavage doses of 10-1,000 mg/kg, percentages of
10 total orally administered radiolabel were measured in urine, feces, expired air, and the whole
11 body.
12 Oral absorption of 1,4-dioxane was extensive, as only approximately 1% of the
13 administered dose appeared in the feces within 72 hours of dosing (Young et al., 1978a, b).
14 Although it may be concluded that the rate of oral absorption was high enough to ensure nearly
May 2009
13
DRAFT - DO NOT CITE OR QUOTE
-------
1 complete absorption by 72 hours, a more quantitative estimate of the rate of oral absorption is
2 not possible due to the absence of plasma time course data by oral exposure.
3 Saturable metabolism of 1,4-dioxane was observed in rats exposed by either the i.v. or
4 oral routes (Young et al., 1978a, b). Elimination of 1,4-dioxane from plasma appeared to be
5 linear following i.v. doses of 3-30 mg/kg, but was nonlinear following doses of 100-
6 1,000 mg/kg. Accordingly, 10 mg/kg i.v. doses resulted in higher concentrations of 14CC>2 (from
7 metabolized 1,4-dioxane) in expired air relative to unchanged 1,4-dioxane, while 1,000 mg/kg
8 i.v. doses resulted in higher concentrations of expired 1,4-dioxane relative to 14CC>2. Thus, at
9 higher i.v. doses, a higher proportion of unmetabolized 1,4-dioxane is available for exhalation.
10 Taken together, the i.v. plasma and expired air data from Young et al. (1978a, b) corroborate
11 previous studies describing the saturable nature of 1,4-dioxane metabolism in rats (Woo et al.
12 1977a, b) and are useful for optimizing metabolic parameters (Vmax and Km) in a PBPK model.
13 Similarly, increasing single or multiple oral doses of 10-1,000 mg/kg resulted in
14 increasing percentage of 1,4-dioxane in exhaled air and decreasing percentage of radiolabel
15 (either as 1,4-dioxane or a metabolite) in the urine, with significant differences in both metrics
16 being observed between doses of 10 and 100 mg/kg (Young et al., 1978a, b). These data identify
17 the region (10-100 mg/kg) in which oral exposures will result in nonlinear metabolism of
18 1,4-dioxane and can be used to test whether metabolic parameter value estimates derived from
19 i.v. dosing data are adequate for modeling oral exposures.
20 Post-exposure plasma data from a single 6-hour, 50 ppm inhalation exposure in rats were
21 reported (Young et al., 1978a, b). The observed linear elimination of 1,4-dioxane after
22 inhalation exposure suggests that, via this route, metabolism is in the linear region at this
23 exposure level.
24 The only human data adequate for use in PBPK model development (Young et al., 1977)
25 come from adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours. Plasma
26 1,4-dioxane and HEAA concentrations were measured both during and after the exposure period,
27 and urine concentrations were measured following exposure. Plasma levels of 1,4-dioxane
28 approached steady-state at 6 hours. HEAA data were insufficient to describe the appearance or
29 elimination of HEAA in plasma. Data on elimination of 1,4-dioxane and HEAA in the urine up
30 to 24 hours from the beginning of exposure were reported. At 6 hours from onset of exposure,
31 approximately 90% and 47% of the cumulative (0-24 hours) urinary 1,4-dioxane and HEAA,
32 respectively, were measured in the urine. The ratio of HEAA to 1,4-dioxane in urine 24 hours
33 after onset of exposure was 192:1 (similar to the ratio of 118:1 observed by Young et al. [1976]
34 in workers exposed to 1.6 ppm for 7.5 hours), indicating extensive metabolism of 1,4-dioxane
35 As with Sprague Dawley rats, the elimination of 1,4-dioxane from plasma was linear across all
36 observations (6 hours following end of exposure), suggesting that human metabolism of
37 1,4-dioxane is linear for a 50 ppm inhalation exposure to steady-state. Thus, estimation of
May 2009 14 DRAFT - DO NOT CITE OR QUOTE
-------
1 human Vmax and Km from these data will introduce uncertainty into internal dosimetry performed
2 in the nonlinear region of metabolism.
3 Further data were reported for the tissue distribution of 1,4-dioxane in rats. Mikheev
4 et al. (1990) administered i.p. doses of [14C]-1,4-dioxane to rats (strain not reported) and reported
5 time-to-peak blood, liver, kidney, and testes concentrations. They also reported ratios of tissue
6 to blood concentrations at various time points after dosing. Woo et al. (1977a, b) administered
7 i.p. doses of [14C]-1,4-dioxane to Sprague Dawley rats and measured radioactivity levels in
8 urine. However, since i.p. dosing is not relevant to human exposures, these data are of limited
9 use for PBPK model development.
3.5.2. Published PBPK Models for 1,4-Dioxane
3.5.2.1. Leung andPaustenbach (1990)
10 Leung and Paustenbach (1990) developed a PBPK model for 1,4-dioxane and its primary
11 metabolite, HEAA, in rats and humans. The model, based on the structure of a PBPK model for
12 styrene (Ramsey and Andersen, 1984), consists of a central blood compartment and four tissue
13 compartments: liver, fat, slowly perfused tissues (mainly muscle and skin), and richly perfused
14 tissues (brain, kidney, and viscera other than the liver). Tissue volumes were calculated as
15 percentages of total BW, and blood flow rates to each compartment were calculated as
16 percentages of cardiac output. Equivalent cardiac output and alveolar ventilation rates were
17 allometrically scaled to a power (0.74) of BW for each species. The concentration of
18 1,4-dioxane in alveolar blood was assumed to be in equilibrium with alveolar air at a ratio equal
19 to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane between
20 blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
21 between blood and tissue, governed by tissue:blood equilibrium partition coefficients. The latter
22 were derived from the quotient of blood:air and tissue:air partition coefficients, which were
23 measured in vitro (Leung and Paustenbach, 1990) for blood, liver, fat, and skeletal muscle
24 (slowly perfused tissue). Blood:air partition coefficients were measured for both humans and
25 rats. Rat tissue:air partition coefficients were used as surrogate values for humans, with the
26 exception of slowly perfused tissue:blood, which was estimated by optimization to the plasma
27 time-course data. Portals of entry included i.v. infusion (over a period of 36 seconds) into the
28 venous blood, inhalation by diffusion from the alveolar air into the lung blood at the rate of
29 alveolar ventilation, and oral administration via zero-order absorption from the gastrointestinal
30 tract to the liver. Elimination of 1,4-dioxane was accomplished through pulmonary exhalation
31 and saturable hepatic metabolism. Urinary excretion of HEAA was assumed to be instantaneous
32 with the generation of HEAA from the hepatic metabolism of 1,4-dioxane.
33 The parameter values for hepatic metabolism of 1,4-dioxane, Vmax and Km, were
34 optimized and validated against plasma and/or urine time course data for 1,4-dioxane and HEAA
May 2009 15 DRAFT - DO NOT CITE OR QUOTE
-------
1 in rats following i.v. and inhalation exposures and humans following inhalation exposure (Young
2 et al., 1978a, b, 1977); the exact data (i.e., i.v., inhalation, or both) used for the optimization and
3 calibration were not reported. Although the liver and fat were represented by tissue-specific
4 compartments, no tissue-specific concentration data were available for model development,
5 raising uncertainty as the model's ability to adequately predict exposure to these tissues. The
6 human inhalation exposure of 50 ppm for 6 hours (Young et al., 1977) was reported to be in the
7 linear range for metabolism; thus, uncertainty exists in the ability of the allometrically-scaled
8 value for the human metabolic Vmax to accurately describe 1,4-dioxane metabolism from
9 exposures resulting in metabolic saturation. Nevertheless, these values resulted in the model
10 producing good fits to the data. For rats, the values for Vmax had to be adjusted upwards by a
11 factor of 1.8 to reasonably simulate exposures greater than 300 mg/kg. The model authors
12 attributed this to metabolic enzyme induction by high doses of 1,4-dioxane.
3.5.2.2. Reitzetal. (1990)
13 Reitz et al. (1990) developed a model for 1,4-dioxane and HEAA in the mouse, rat, and
14 human. This model, also based on the styrene model of Ramsey and Andersen (1984), included
15 a central blood compartment and compartments for liver, fat, and rapidly and slowly perfused
16 tissues. Tissue volumes and blood flow rates were defined as percentages of total BW and
17 cardiac output, respectively. Physiological parameter values were similar to those used by
18 Andersen et al. (1987), except that flow rates for cardiac output and alveolar ventilation were
19 doubled in order to produce a better fit of the model to human blood level data (Young et al.,
20 1977). Portals of entry included i.v. injection into the venous blood, inhalation, oral bolus
21 dosing, and oral dosing via drinking water. Oral absorption of 1,4-dioxane was simulated, in all
22 three species, as a first-order transfer to liver (halftime approximately 8 minutes).
23 Alveolar blood levels of 1,4-dioxane were assumed to be in equilibrium with alveolar air
24 at a ratio equal to the experimentally measured blood:air partition coefficient. Transfers of
25 1,4-dioxane between blood and tissues were assumed to be blood flow-limited and to achieve
26 rapid equilibrium between blood and tissue, governed by tissue:blood equilibrium partition
27 coefficients. These coefficients were derived by dividing experimentally measured (Leung and
28 Paustenbach, 1990) in vitro blood:air and tissue:air partition coefficients for blood, liver, fat.
29 Blood:air partition coefficients were measured for both humans and rats. The mouse blood:air
30 partition coefficient was different from rat or human values; the source of the partition
31 coefficient for blood in mice was not reported. Rat tissue:air partition coefficients were used as
32 surrogate values for humans. Rat tissue partition coefficient values were the same values as used
33 in the Leung and Paustenbach (1990) model (with the exception of slowly perfused tissues) and
34 were used in the models for all three species. The liver value was used for the rapidly perfused
35 tissues, as well as slowly perfused tissues. Although slowly perfused tissue:air partition
36 coefficients for rats were measured, the authors suggested that 1,4-dioxane in the muscle and air
May 2009 16 DRAFT - DO NOT CITE OR QUOTE
-------
1 may not have reached equilibrium in the highly gelatinous tissue homogenate (Reitz et al., 1990).
2 Substitution of the liver value provided much closer agreement to the plasma data than when the
3 muscle value was used. Further, doubling of the measured human blood:air partition coefficient
4 improved the fit of the model to the human blood level data compared to the fit resulting from
5 the measured value (Reitz et al., 1990). The Reitz et al. (1990) model simulated three routes of
6 1,4-dioxane elimination: pulmonary exhalation, hepatic metabolism to HEAA, and urinary
7 excretion of HEAA. The elimination of HEAA was modeled as a first-order transfer of
8 1,4-dioxane metabolite to urine.
9 Values for the metabolic rate constants, Vmax and Km, were optimized to achieve
10 agreement with various observations. Reitz et al. (1990) optimized values for human Vmax and
11 Km against the experimental human 1,4-dioxane inhalation data (Young et al., 1977). As noted
12 previously, because the human exposures were below the level needed to exhibit nonlinear
13 kinetics, uncertainty exists in the ability of the optimized value of Vmax to simulate human
14 1,4-dioxane metabolism above the concentration that would result in saturation of metabolism.
15 Rat metabolic rate constants were obtained by optimization to simulated data from a
16 two-compartment empirical pharmacokinetic model, which was fitted to i.v. exposure data
17 (Young et al., 1978a, b). As with the Leung and Paustenbach (1990) model, the Reitz et al.
18 (1990) model included compartments for the liver and fat, although no tissue-specific
19 concentration data were available to validate dosimetry for these organs. The derivations of
20 human and rat HEAA elimination rate constants were not reported. Since no pharmacokinetics
21 data for 1,4-dioxane in mice were available, mouse metabolic rate constants were allometrically
22 scaled from rat and human values.
3.5.2.3. Fisher et al. (1997)
23 A PBPK model was developed by Fisher et al. (1997) to simulate a variety of volatile
24 organic compounds (VOCs, including 1,4-dioxane) in lactating humans. This model was similar
25 in structure to those of Leung and Paustenbach (1990) and Reitz et al. (1990) with the addition of
26 elimination of 1,4-dioxane to breast milk. Experimental measurements were made for blood:air
27 and milk:air partition coefficients. Other partition coefficient values were taken from Reitz et al.
28 (1990). The model was not optimized, nor was performance tested against experimental
29 exposure data. Thus, the ability of the model to simulate 1,4-dioxane exposure data is unknown.
3.5.3. Implementation of Published PBPK Models for 1,4-Dioxane
30 As previously described, several pharmacokinetic models have been developed to predict
31 the absorption, distribution, metabolism, and elimination of 1,4-dioxane in rats and humans.
32 Single compartment, empirical models for rats (Young et al., 1978a, b) and humans (Young
33 et al., 1977) were developed that predict blood levels of 1,4-dioxane and urine levels of the
34 primary metabolite, HEAA. PBPK models, which describe the kinetics of 1,4-dioxane using
May 2009 17 DRAFT - DO NOT CITE OR QUOTE
-------
1 biologically realistic flow rates, tissue volumes and affinities, metabolic processes, and
2 elimination behaviors, were also developed (Fisher et al., 1997; Leung and Paustenbach, 1990;
3 Reitzetal., 1990).
4 In developing updated toxicity values for 1,4-dioxane the available PBPK models were
5 evaluated for their ability to predict observations made in experimental studies of rat and human
6 exposures to 1,4-dioxane (Appendix B). The Reitz et al. (1990) and Leung and Paustenbach
7 (1990) PBPK models were both developed from a PBPK model of styrene (Ramsey and
8 Anderson, 1984), with the exception of minor differences in the use of partition coefficients and
9 biological parameters. The model code for Leung and Paustenbach (1990) was unavailable in
10 contrast to Reitz et al. (1990). The model of Reitz et al. (1990) was identified for further
11 consideration to assist in the derivation of toxicity values.
12 Issues related to the biological plausibility of parameter values in the human model were
13 identified. Specifically, the model is able to predict the only available human inhalation data set
14 (50 ppm 1,4-dioxane for 6 hours; Young et al., 1977) by increasing (i.e., doubling) the
15 parameter values for human alveolar ventilation, cardiac output, and the blood:air partition
16 coefficient above the measured values. Furthermore, the measured value for the slowly perfused
17 tissue:air partition coefficient (i.e., muscle) was replaced with the measured liver value to
18 improve the fit. Analysis of the Young et al. (1977) human data suggested that the apparent
19 volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold higher in rats than
20 humans, presumably due to species differences in tissue partitioning or other process not
21 represented in the model. Subsequent exercising of the model demonstrated that selecting a
22 human slowly perfused tissue:air partition coefficient much lower than the measured rat value
23 resulted in better agreement between model predictions of 1,4-dioxane in blood and experimental
24 observations. Based upon these observations the model (e.g., metabolism/elimination
25 parameters) was re-calibrated using biologically plausible values for flow rates and tissue:air
26 partition coefficients.
27 Appendix B describes all activities that have been conducted in the evaluation of the
28 empirical models and re-calibration and exercising of the Reitz et al. (1990) PBPK model to
29 determine the adequacy and preference for the potential use of the models for 1,4-dioxane
30 The evaluation consisted of implementation of the Young et al. (1978a, b, 1977)
31 empirical rat and human models using the acslXtreme simulation software and re-calibration of
32 the Reitz et al. (1990) human PBPK model. Using the model descriptions and equations given in
33 Young et al. (1978a, b, 1977), model code was developed for the empirical models and executed,
34 simulating the reported experimental conditions. The model output was then compared with the
35 model output reported in Young et al. (1978a, b, 1977).
36 The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for
37 cardiac and alveolar flow rates and tissue:air partition coefficients. The predictions of blood and
38 urine levels of 1,4-dioxane and HEAA, respectively, from the re-calibrated model were
May 2009 18 DRAFT - DO NOT CITE OR QUOTE
-------
1 compared with the empirical model predictions of the same dosimeters to determine whether the
2 re-calibrated PBPK model could perform similarly to the empirical model. As part of the PBPK
3 model evaluation, a sensitivity analysis was performed to identify the model parameters having
4 the greatest influence on the primary dosimeter of interest, the blood level of 1,4-dioxane.
5 Variability data for the experimental measurements of the tissue:air partition coefficients were
6 incorporated to determine a range of model outputs bounded by biologically plausible values for
7 these parameters.
8 The rat and human empirical models of Young et al. (1978a, b, 1977) were successfully
9 implemented in acslXtreme and perform identically to the models reported in the published
10 papers (Figures B-3 through B-6), with the exception of the lower predicted HEAA
11 concentrations and early appearance of the peak HEAA levels in rat urine. The early appearance
12 of peak HEAA levels cannot presently be explained, but may result from manipulations of kme or
13 other parameters by Young et al. (1978a, b) that were not reported. The lower predictions of
14 HEAA levels are likely due to reliance on a standard urine volume production rate in the absence
15 of measured (but unreported) urine volumes. While the human urinary HEAA predictions were
16 lower than observations, this is due to parameter fitting of Young et al. (1977). No model output
17 was published in Young et al. (1977) for comparison. The empirical models were modified to
18 allow for user-defined inhalation exposure levels. However, no modifications were made to
19 model oral exposures as adequate data to parameterize such modifications do not exist for rats or
20 humans.
21 Several procedures were applied to the human PBPK model to determine if an adequate
22 fit of the model to the empirical model output or experimental observations could be attained
23 using biologically plausible values for the model parameters. The re-calibrated model
24 predictions for blood 1,4-dioxane levels do not come within 10-fold of the experimental values
25 using measured tissue:air partition coefficients from Leung and Paustenbach (1990) or Sweeney
26 et al. (2008) (Figures B-8 and B-9). The utilization of a slowly perfused tissue:air partition
27 coefficient 10-fold lower than measured values produces exposure-phase predictions that are
28 much closer to observations, but does not replicate the elimination kinetics (Figure B-10).
29 Recalibration of the model with upper bounds on the tissue:air partition coefficients results in
30 predictions that are still six- to sevenfold lower than empirical model prediction or observations
31 (Figures B-12 and B-13). Exploration of the model space using an assumption of first-order
32 metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate fit to the
33 exposure and elimination data can be achieved only when unrealistically low values are assumed
34 for the slowly perfused tissue:air partition coefficient (Figure B-16). Artificially low values for
35 the other tissue:air partition coefficients are not expected to improve the model fit, as these
36 parameters are shown in the sensitivity analysis to exert less influence on blood 1,4-dioxane than
37 Vmaxc and Km. In the absence of actual measurements for the human slowly perfused tissue:air
38 partition coefficient, high uncertainty exists for this model parameter value. Differences in the
May 2009 19 DRAFT - DO NOT CITE OR QUOTE
-------
1 ability of rat and human blood to bind 1,4-dioxane may contribute to the difference in Vd.
2 However, this is expected to be evident in very different values for rat and human blood:air
3 partition coefficients, which is not the case (Table B-l). Therefore, some other, as yet unknown,
4 modification to model structure may be necessary.
5 Similarly, Sweeney et al. (2008) also evaluated the available PBPK models (Leung and
6 Paustenbach, 1990; Reitz et al., 1990) for 1,4-dioxane. To address uncertainties and deficiencies
7 in these models, the investigators conducted studies to fill data gaps and reduce uncertainties
8 pertaining to the pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. The
9 following studies were performed:
Partition coefficients, including measurements for mouse blood and tissues (liver, kidney, fat,
and muscle) and confirmatory measurements for human blood and rat blood and muscle.
Blood time course measurements in mice conducted for gavage administration of nominal
single doses (20, 200, or 2,000 mg/kg) of 1,4-dioxane administered in water.
Metabolic rate constants for rat, mouse, and human liver based on incubations of 1,4-dioxane
with rat, mouse, and human hepatocytes and measurement of HEAA.
10 Updated PBPK models were developed based on these new data and data from previous
11 kinetic studies in rats, workers, and human volunteers reported by Young et al. (1978a, b, 1977,
12 1976). The optimized rate of metabolism for the mouse was significantly higher than the value
13 previously estimated. The optimized rat kinetic parameters were similar to those in the 1990
14 models. Of the two available human studies (Young et al., 1977, 1976), model predictions were
15 consistent with one study, but did not fit the second as well.
3.6. RAT NASAL EXPOSURE VIA DRINKING WATER
16 Sweeney et al. (2008) conducted a rat nasal exposure study to explore the potential for
17 direct contact of nasal tissues with 1,4-dioxane-containing drinking water under bioassay
18 conditions. Two groups of male Sprague Dawley rats (5/group) received drinking water in
19 45-mL drinking water bottles containing a fluorescent dye mixture (Cell Tracker
20 Red/FluoSpheres). The drinking water for one of these two groups also contained 0.5%
21 1,4-dioxane, a concentration within the range used in chronic toxicity studies. A third group of
22 five rats received tap water alone (controls). Water was provided to the rats overnight. The next
23 morning, the water bottles were weighed to estimate the amounts of water consumed. Rats were
24 sacrificed and heads were split along the midline for evaluation by fluorescence microscopy.
25 One additional rat was dosed twice by gavage with 2 mL of drinking water containing
26 fluorescent dye (the second dose was 30 minutes after the first dose; total of 4 mL administered)
27 and sacrificed 5 hours later to evaluate the potential for systemic delivery of fluorescent dye to
28 the nasal tissues.
May 2009 20 DRAFT - DO NOT CITE OR QUOTE
-------
1 The presence of the fluorescent dye mixture had no measurable impact on water
2 consumption; however, 0.5% 1,4-dioxane reduced water consumption by an average of 62% of
3 controls following a single, overnight exposure. Fluorescent dye was detected in the oral cavity
4 and nasal airways of each animal exposed to the Cell Tracker Red/FluoSpheres mixture in their
5 drinking water, including numerous areas of the anterior third of the nose along the nasal
6 vestibule, maxillary turbinates, and dorsal nasoturbinates. Fluorescent dye was occasionally
7 detected in the ethmoid turbinate region and nasopharynx. 1,4-Dioxane had no effect on the
8 detection of the dye. Little or no fluorescence at the wavelength associated with the dye mixture
9 was detected in control animals or in the single animal that received the dye mixture by oral
10 gavage. The investigators concluded that the findings indicate rat nasal tissues are exposed by
11 direct contact with drinking water under bioassay conditions.
May 2009 21 DRAFT - DO NOT CITE OR QUOTE
-------
4. HAZARD IDENTIFICATION
4.1. STUDIES IN HUMANS - EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS
1 Case reports of acute occupational poisoning with 1,4-dioxane indicated that exposure to
2 high concentrations resulted in liver, kidney, and central nervous system (CNS) toxicity
3 (Johnstone, 1959; Barber, 1934). Barber (1934) described four fatal cases of hemorrhagic
4 nephritis and centrilobular necrosis of the liver attributed to acute inhalation exposure to high
5 (unspecified) concentrations of 1,4-dioxane. Death occurred within 5-8 days of the onset of
6 illness. Autopsy findings suggested that the kidney toxicity may have been responsible for
7 lethality, while the liver effects may have been compatible with recovery. Jaundice was not
8 observed in subjects and fatty change was not apparent in the liver. Johnstone (1959) presented
9 the fatal case of one worker exposed to high concentrations of 1,4-dioxane through both
10 inhalation and dermal exposure for a 1 week exposure duration. Measured air concentrations in
11 the work environment of this subject were 208-650 ppm, with a mean value of 470 ppm.
12 Clinical signs that were observed following hospital admission included severe epigastric pain,
13 renal failure, headache, elevation in blood pressure, agitation and restlessness, and coma.
14 Autopsy findings revealed significant changes in the liver, kidney, and brain. These included
15 centrilobular necrosis of the liver and hemorrhagic necrosis of the kidney cortex. Perivascular
16 widening was observed in the brain with small foci of demyelination in several regions (e.g.,
17 cortex, basal nuclei). It was suggested that these neurological changes may have been secondary
18 to anoxia and cerebral edema.
19 Several studies examined the effects of acute inhalation exposure in volunteers. In a
20 study performed at the Pittsburgh Experimental Station of the U.S. Bureau of Mines, eye
21 irritation and a burning sensation in the nose and throat were reported in five men exposed to
22 5,500 ppm of 1,4-dioxane vapor for 1 minute (Yant et al., 1930). Slight vertigo was also
23 reported by three of these men. Exposure to 1,600 ppm of 1,4-dioxane vapor for 10 minutes
24 resulted in similar symptoms with a reduced intensity of effect. In a study conducted by the
25 Government Experimental Establishment at Proton, England (Fairley et al., 1934), four men
26 were exposed to 1,000 ppm of 1,4-dioxane for 5 minutes. Odor was detected immediately and
27 one volunteer noted a constriction in the throat. Exposure of six volunteers to 2,000 ppm for 3
28 minutes resulted in no symptoms of discomfort. Wirth and Klimmer (1936), of the Institute of
29 Pharmacology, University of Wurzburg, reported slight mucous membrane irritation in the nose
30 and throat of several human subjects exposed to concentrations greater than 280 ppm for several
31 minutes. Exposure to approximately 1,400 ppm for several minutes caused a prickling sensation
32 in the nose and a dry and scratchy throat. Silverman et al. (1946) exposed 12 male and 12
33 female subjects to varying air concentrations of 1,4-dioxane for 15 minutes. A 200 ppm
May 2009 22 DRAFT - DO NOT CITE OR QUOTE
-------
1 concentration was reported to be tolerable, while a concentration of 300 ppm caused irritation to
2 the eyes, nose, and throat. The study conducted by Silverman et al. (1946) was conducted by the
3 Department of Industrial Hygiene, Harvard School of Public Health, and was sponsored and
4 supported by a grant from the Shell Development Company. These volunteer studies published
5 in the 1930s and 1940s (Silverman et al., 1946; Wirth and Klimmer, 1936; Fairley et al., 1934;
6 Yant et al., 1930) do not provide information on the human subjects research ethics procedures
7 undertaken in these study; however, there is no evidence that the conduct of the research was
8 fundamentally unethical or significantly deficient relative to the ethical standards prevailing at
9 the time the research was conducted.
10 Young et al. (1977) exposed four healthy adult male volunteers to a 50-ppm
11 concentration of 1,4-dioxane for 6 hours. The investigators reported that the protocol of this
12 study was approved by a seven-member Human Research Review Committee of the Dow
13 Chemical Company and was followed rigorously. Perception of the odor of 1,4-dioxane
14 appeared to diminish over time, with two of the four subjects reporting inability to detect the
15 odor at the end of the exposure period. Eye irritation was the only clinical sign reported in this
16 study. The pharmacokinetics and metabolism of 1,4-dioxane in humans were also evaluated in
17 this study (see Section 3.3). Clinical findings were not reported in four workers exposed in the
18 workplace to a TWA concentration of 1.6 ppm for 7.5 hours (Young et al., 1976).
19 Ernstagard et al. (2006) examined the acute effects of 1,4-dioxane vapor in male and
20 female volunteers. The study protocol was approved by the Regional Ethics Review Board in
21 Stockholm, and performed following informed consent and according to the Helsinki
22 declaration. In a screening study by these investigators, no self-reported symptoms (based on a
23 visual analogue scale (VAS) that included ratings for discomfort, breathing difficulty, headache,
24 fatigue, nausea, dizziness, or feeling of intoxication) were observed at concentrations up to 20
25 ppm; this concentration was selected as a tentative no-observed-adverse-effect-level (NOAEL) in
26 the main study. In the main study, six male and six female healthy volunteers were exposed to 0
27 or 20 ppm 1,4-dioxane, at rest, for 2 hours. This exposure did not significantly affect symptom
28 VAS ratings, blink frequency, pulmonary function or nasal swelling (measured before and at 0
29 and 3 hours after exposure), or inflammatory markers in the plasma (C-reactive protein and
30 interleukin-6) of the volunteers. Only ratings for "solvent smell" were significantly increased
31 during exposure.
32 Only two well documented epidemiology studies were available for occupational workers
33 exposed to 1,4-dioxane (Buffler et al., 1978; Thiess et al., 1976). These studies did not provide
34 evidence of effects in humans; however, the cohort size and number of reported cases were
35 small.
May 2009 23 DRAFT - DO NOT CITE OR QUOTE
-------
4.1.1. Thiess et al. (1976)
1 A cross-sectional survey was conducted in German workers exposed to 1,4-dioxane. The
2 study evaluated health effects in 74 workers, including 24 who were still actively employed in
3 1,4-dioxane production at the time of the investigation, 23 previously exposed workers who were
4 still employed by the manufacturer, and 27 retired or deceased workers. The actively employed
5 workers were between 32 and 62 years of age and had been employed in 1,4-dioxane production
6 for 5-41 years. Former workers (age range not given) had been exposed to 1,4-dioxane for 3-
7 38 years and retirees (age range not given) had been exposed for 12-41 years. Air
8 concentrations in the plant at the time of the study were 0.06-0.69 ppm. A simulation of
9 previous exposure conditions (prior to 1969) resulted in air measurements between 0.06 and
10 7.2 ppm.
11 Active and previously employed workers underwent a thorough clinical examination and
12 X-ray, and hematological and serum biochemistry parameters were evaluated. The examination
13 did not indicate pathological findings for any of the workers and no indication of malignant
14 disease was noted. Hematology results were generally normal. Serum transaminase levels were
15 elevated in 16 of the 47 workers studied; however, this finding was consistent with chronic
16 consumption of more than 80 g of alcohol per day, as reported for these workers. No liver
17 enlargement or jaundice was found. Renal function tests and urinalysis were normal in exposed
18 workers. Medical records of the 27 retired workers (15 living at the time of the study) were
19 reviewed. No symptoms of liver or kidney disease were reported and no cancer was detected.
20 Medical reasons for retirement did not appear related to 1,4-dioxane exposure (e.g., emphysema,
21 arthritis).
22 Chromosome analysis was performed on six actively employed workers and six control
23 persons (not characterized). Lymphocyte cultures were prepared and chromosomal aberrations
24 were evaluated. No differences were noted in the percent of cells with gaps or other
25 chromosome aberrations. Mortality statistics were calculated for 74 workers of different ages
26 and varying exposure periods. The proportional contribution of each of the exposed workers to
27 the total time of observation was calculated as the sum of man-years per 10-year age group.
28 Each person contributed one man-year per calendar year to the specific age group in which he
29 was included at the time. The expected number of deaths for this population was calculated from
30 the age-specific mortality statistics for the German Federal Republic for the years 1970-1973.
31 From the total of 1,840.5 person-years, 14.5 deaths were expected; however, only 12 deaths were
32 observed in exposed workers between 1964 and 1974. Two cases of cancer were reported,
33 including one case of lamellar epithelial carcinoma and one case of myelofibrotic leukemia.
34 These cancers were not considered to be the cause of death in these cases and other severe
35 illnesses were present. Standardized mortality ratios (SMRs) for cancer did not significantly
May 2009 24 DRAFT - DO NOT CITE OR QUOTE
-------
1 differ from the control population (SMR for overall population = 0.83; SMR for 65-75-year-old
2 men = 1.61; confidence intervals (CIs) not provided).
4.1.2. Buffler et al. (1978)
3 Buffler et al. (1978) conducted a mortality study on workers exposed to 1,4-dioxane at a
4 chemical manufacturing facility in Texas. 1,4-Dioxane exposure was known to occur in a
5 manufacturing area and in a processing unit located 5 miles from the manufacturing plant.
6 Employees who worked between April 1, 1954, and June 30, 1975, were separated into two
7 cohorts based on at least 1 month of exposure in either the manufacturing plant (100 workers) or
8 the processing area (65 workers). Company records and follow-up techniques were used to
9 compile information on name, date of birth, gender, ethnicity, job assignment and duration, and
10 employment status at the time of the study. Date and cause of death were obtained from copies
11 of death certificates and autopsy reports (if available). Exposure levels for each j ob category
12 were estimated using the 1974 Threshold Limit Value for 1,4-dioxane (i.e., 50 ppm) and
13 information from area and personal monitoring. Exposure levels were classified as low
14 (<25 ppm), intermediate (50-75 ppm), and high (>75 ppm). Monitoring was not conducted prior
15 to 1968 in the manufacturing areas or prior to 1974 in the processing area; however, the study
16 authors assumed that exposures would be comparable, considering that little change had been
17 made to the physical plant or the manufacturing process during that time. Exposure to
18 1,4-dioxane was estimated to be below 25 ppm for all individuals in both cohorts.
19 Manufacturing area workers were exposed to several other additional chemicals and processing
20 area workers were exposed to vinyl chloride.
21 Seven deaths were identified in the manufacturing cohort and five deaths were noted for
22 the processing cohort. The average exposure duration was not greater for those workers who
23 died, as compared to those still living at the time of the study. Cancer was the underlying cause
24 of death for two cases from the manufacturing area (carcinoma of the stomach, alveolar cell
25 carcinoma) and one case from the processing area (malignant mediastinal tumor). The workers
26 from the manufacturing area were exposed for 28 or 38 months and both had a positive smoking
27 history (>1 pack/day). Smoking history was not available for processing area workers. The
28 single case of cancer in this area occurred in a 21-year-old worker exposed to 1,4-dioxane for
29 1 year. The mortality data for both industrial cohorts were compared to age-race-sex specific
30 death rates for Texas (1960-1969). Person-years of observation contributed by workers were
31 determined over five age ranges with each worker contributing one person-year for each year of
32 observation in a specific age group. The expected number of deaths was determined by applying
33 the Texas 1960-1969 death rate statistics to the number of person years calculated for each
34 cohort. The observed and expected number of deaths for overall mortality (i.e., all causes) was
35 comparable for both the manufacturing area (7 observed versus 4.9 expected) and the processing
36 area (5 observed versus 4.9 expected). No significant excess in cancer-related deaths was
May 2009 25 DRAFT - DO NOT CITE OR QUOTE
-------
1 identified for both areas of the facility combined (3 observed versus 1.7 expected). A separate
2 analysis was performed to evaluate mortality in manufacturing area workers exposed to
3 1,4-dioxane for more than 2 years. Six deaths occurred in this group as compared to
4 4.1 expected deaths. The use of a conditional Poisson distribution indicated no apparent excess
5 in mortality or death due to malignant neoplasms in this study. It is important to note that the
6 cohorts evaluated were limited in size. In addition, the mean exposure duration was less than
7 5 years (<2 years for 43% of workers) and the latency period for evaluation was less than
8 10 years for 59% of workers. The study authors recommended a follow-up investigation to
9 allow for a longer latency period; however, no follow-up study of these workers has been
10 published.
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS - ORAL AND INHALATION
11 The majority of the subchronic (>30 days) and chronic (>1 year) studies conducted for
12 1,4-dioxane were oral drinking water studies. Longer-term inhalation studies consisted of only
13 one subchronic study (Fairley et al., 1934) and one chronic study (Torkelson et al., 1974). These
14 studies were not sufficient to characterize the inhalation risks of 1,4-dioxane (see Section 4.2.2.).
4.2.1. Oral Toxicity
4.2.1.1. Subchronic Oral Toxicity
15 Six rats and six mice (unspecified strains) were given drinking water containing 1.25%
16 1,4-dioxane for up to 67 days (Fairley et al. 1934). Using reference BWs and drinking water
17 ingestion rates for rats and mice (U.S. EPA, 1988), it can be estimated that these rats and mice
18 received doses of approximately 1,900 and 3,300 mg/kg-day, respectively. Gross pathology and
19 histopathology were evaluated in all animals. Five of the six rats in the study died or were
20 sacrificed in extremis prior to day 34 of the study. Mortality was lower in mice, with five of six
21 mice surviving up to 60 days. Kidney enlargement was noted in 5/6 rats and 2/5 mice. Renal
22 cortical degeneration was observed in all rats and 3/6 mice. Large areas of necrosis were
23 observed in the cortex, while cell degeneration in the medulla was slight or absent. Tubular casts
24 were observed and vascular congestion and hemorrhage were present throughout the kidney.
25 Hepatocellular degeneration with vascular congestion was also noted in five rats and three mice.
26 EPA identified the tested doses of 1,900 mg/kg-day in rats and 3,300 mg/kg-day in mice as the
27 lowest-observed-adverse-effect-levels (LOAELs) for liver and kidney degeneration in this study.
28 4.2.1.1.1. Stoner et al. (1986). 1,4-dioxane was evaluated for its ability to induce lung adenoma
29 formation in A/J mice. Six- to 8-week-old male and female A/J mice (16/sex/group) were given
30 1,4-dioxane by gavage or i.p. injection, 3 times/week for 8 weeks. Total cumulative dose levels
31 were given as 24,000 mg/kg (oral), and 4,800, 12,000, or 24,000 mg/kg (i.p.). Average daily
May 2009 26 DRAFT - DO NOT CITE OR QUOTE
-------
1 dose estimates were calculated to be 430 mg/kg-day (oral), and 86, 210, or 430 mg/kg-day (i.p.)
2 by assuming an exposure duration of 56 days. The authors indicated that i.p. doses represent the
3 maximum tolerated dose (MTD), 0.5 times the MTD, and 0.2 times the MTD. Mice were killed
4 24 weeks after initiation of the bioassay, and lungs, liver, kidney, spleen, intestines, stomach,
5 thymus, salivary, and endocrine glands were examined for gross lesions. Histopathology
6 examination was performed if gross lesions were detected. 1,4-Dioxane did not induce lung
7 tumors in male or female A/J mice in this study.
8 4.2.1.1.2. Stottetal. (1981). Male Sprague Dawley rats (4-6/group) were given average doses
9 of 0, 10, or 1,000 mg/kg-day 1,4-dioxane (>99% pure) in their drinking water, 7 days/week for
10 11 weeks. It should noted that the methods description in this report stated that the high dose
11 was 100 mg/kg-day, while the abstract, results, and discussion sections indicated that the high
12 dose was 1,000 mg/kg-day. Rats were implanted with a [6"3H]thymidine loaded osmotic pump
13 7 days prior to sacrifice. Animals were sacrificed by cervical dislocation and livers were
14 removed, weighed, and prepared for histopathology evaluation. [3H]-Thymidine incorporation
15 was measured by liquid scintillation spectroscopy.
16 An increase in the liver to BW ratio was observed in rats from the high dose group
17 (assumed to be 1,000 mg/kg-day). Histopathological alterations, characterized as minimal
18 centrilobular swelling, were also seen in rats from this dose group (incidence values were not
19 reported). Hepatic DNA synthesis, measured by [3H]-thymidine incorporation, was increased
20 1.5-fold in high-dose rats. No changes relative to control were observed for rats exposed to
21 10 mg/kg-day. EPA found a NOAEL value of 10 mg/kg-day and a LOAEL value of
22 1,000 mg/kg-day for this study based on histopathological changes in the liver.
23 Stott et al. (1981) also performed several acute experiments designed to evaluate
24 potential mechanisms for the carcinogenicity of 1,4-dioxane. These experiments are discussed
25 separately in Section 4.5.2 (Mechanistic Studies).
26 4.2.1.1.3. Kano et al (2008). Groups of 6-week-old F344/DuCrj rats (10/sex/group) and
27 Crj :BDFi mice (10/sex/group) were administered 1,4-dioxane (>99% pure) in the drinking water
28 for 13 weeks. The animals were observed daily for clinical signs of toxicity. Food consumption
29 and BWs were measured once per week and water consumption was measured twice weekly.
30 Food and water were available ad libitum. The concentrations of 1,4-dioxane in the water for
31 rats and mice were 0, 640, 1,600, 4,000, 10,000, or 25,000 ppm. The investigators used data
32 from water consumption and BW changes to calculate a daily intake of 1,4-dioxane by the male
33 and female animals. Thus, male rats received doses of approximately 0, 52, 126, 274, 657, and
34 1,554 mg 1,4-dioxane/kg-day and female rats received 0, 83, 185, 427, 756, and
35 1,614 mg/kg-day. Male mice received 0, 86, 231, 585, 882, or 1,570 mg/kg-day and female mice
36 received 0, 170, 387, 898, 1,620, or 2,669 mg/kg-day.
May 2009 27 DRAFT - DO NOT CITE OR QUOTE
-------
1 No information was provided as to when the blood and urine samples were collected.
2 Hematology analysis included red blood cell (RBC) count, hemoglobin, hematocrit, mean
3 corpuscular volume (MCV), platelet count, white blood cell (WBC) count, and differential
4 WBCs. Serum biochemistry included total protein, albumin, bilirubin, glucose, cholesterol,
5 triglyceride (rat only), alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate
6 dehydrogenase (LDH), leucine aminopeptidase (LAP), alkaline phosphatase (ALP), creatinine
7 phosphokinase (CPK) (rat only), urea nitrogen, creatinine (rat only), sodium, potassium,
8 chloride, calcium (rat only), and inorganic phosphorous (rat only). Urinalysis parameters were
9 pH, protein, glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ
10 weights (brain, lung, liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and
11 gross necropsy and histopathologic examination of tissues and organs were performed on all
12 animals (skin, nasal cavity, trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart,
13 tongue, salivary glands, esophagus, stomach, small and large intestine, liver, pancreas, kidney,
14 urinary bladder, pituitary thyroid adrenal, testes, epididymis, seminal vesicle, prostate, ovary,
15 uterus, vagina, mammary gland, brain, spinal cord, sciatic nerve, eye, Harderian gland, muscle,
16 bone, and parathyroid). Dunnett's test and £ test were used to assess the statistical significance
17 of changes in continuous and discrete variables, respectively.
18 Clinical signs of toxicity in rats were not discussed in the study report. One female rat in
19 the high dose group (1,614 mg/kg-day) group died, but cause and time of death were not
20 specified. Final BWs were reduced at the two highest dose levels in females (12 and 21%) and
21 males (7 and 21%), respectively. Food consumption was reduced 13% in females at
22 1,614 mg/kg-day and 8% in 1,554 mg/kg-day males. A dose-related decrease in water
23 consumption was observed in male rats starting at 52 mg/kg-day (15%) and in females starting at
24 185 mg/kg-day (12%). Increases in RBCs, hemoglobin, hematocrit, and neutrophils, and a
25 decrease in lymphocytes were observed in males at 1554 mg/kg-day. In females, MCV was
26 decreased at doses > 756 mg/kg and platelets were decreased at 1,614 mg/kg-day. With the
27 exception of the 30% increase in neutrophils in high-dose male rats, hematological changes were
28 within 2-15% of control values. Total serum protein and albumin were significantly decreased
29 in males at doses > 274 mg/kg-day and in females at doses > 427 mg/kg-day. Additional
30 changes in high-dose male and female rats included decreases in glucose, total cholesterol,
31 triglycerides, and sodium (and calcium in females), and increases in ALT (males only), AST,
32 ALP, and LAP. Serum biochemistry parameters in treated rats did not differ more than twofold
33 from control values. Urine pH was decreased in males at > 274 mg/kg-day and in females at
34 > 756 mg/kg-day.
35 Kidney weights were increased in females at >185 mg/kg-day with a maximum increase
36 of 15% and 44% at 1,614 mg/kg-day for absolute and relative kidney weight, respectively. No
37 organ weight changes were noted in male rats. Histopathology findings in rats that were related
38 to exposure included nuclear enlargement of the respiratory epithelium, nuclear enlargement of
May 2009 28 DRAFT - DO NOT CITE OR QUOTE
-------
1 the olfactory epithelium, nuclear enlargement of the tracheal epithelium, hepatocyte swelling of
2 the centrilobular area of the liver, vacuolar changes in the liver, granular changes in the liver,
3 single cell necrosis in the liver, nuclear enlargement of the proximal tubule of the kidneys,
4 hydropic changes in the proximal tubule of the kidneys, and vacuolar changes in the brain. The
5 incidence data for histopathological lesions in rats are presented in Table 4-1. The effects that
6 occurred at the lowest doses were nuclear enlargement of the respiratory epithelium in the nasal
7 cavity and hepatocyte swelling in the central area of the liver in male rats. Based on these
8 histopathological findings the study authors identified the LOAEL as 126 mg/kg-day and the
9 NOAEL as 52 mg/kg-day.
Table 4-1. Incidence of histopathological lesions in F344/DuCrj rats exposed
to 1,4-dioxane in drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; renal proximal tubule
Hydropic change; renal proximal tubule
Vacuolic change; brain
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; proximal tubule
Hydropic change; proximal tubule
Vacuolic change; brain
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
52
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
126
9/10b
0/10
0/10
9/10b
0/10
0/10
0/10
0/10
0/10
0/10
274
10/10b
10/10b
10/10b
10/10b
0/10
5/10c
5/10c
1/10
0/10
0/10
657
9/10b
9/10b
10/10b
10/10b
10/10b
2/10
2/10
5/10c
0/10
0/10
1,554
10/10b
10/10b
10/10b
10/10b
10/10b
10/10b
10/10b
9/10b
7/10b
10/10b
Female dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
2/10
2/10
0/10
0/10
0/10
83
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
185
5/10c
0/10
0/10
0/10
0/10
1/10
1/10
0/10
0/10
0/10
427
10/10b
9/10b
9/10b
0/10
0/10
5/10c
5/10
0/10
0/10
0/10
756
10/10b
10/10b
10/10b
9/10b
0/10
5/10c
5/10
8/10b
0/10
0/10
1,614
8/9b
8/9b
9/9b
9/9b
9/9b
8/9b
8/9b
9/9b
5/9c
9/9b
aData are presented for sacrificed animals.
bp < 0.01 by i2 test.
°p < 0.05.
Source: Kano etal. (2008).
May 2009
29
DRAFT - DO NOT CITE OR QUOTE
-------
1 Clinical signs of toxicity in mice were not discussed in the study report One male mouse
2 in the high-dose group (1,570 mg/kg-day) died, but no information was provided regarding cause
3 or time of death. Final BWs were decreased 29% in male mice at 1,570 mg/kg-day, but changed
4 less than 10% relative to controls in the other male dose groups and in female mice. Food
5 consumption was not significantly reduced in any exposure group. Water consumption was
6 reduced 14-18% in male mice exposed to 86, 231, or 585 mg/kg-day. Water consumption was
7 further decreased by 48 and 70% in male mice exposed to 882 and 1,570 mg/kg-day,
8 respectively. Water consumption was also decreased 31 and 57% in female mice treated with
9 1,620 and 2,669 mg/kg-day, respectively. An increase in MCV was observed in the two highest
10 dose groups in both male (882 and 1,570 mg/kg-day) and female mice (1,620 and
11 2,669 mg/kg-day). Increases in RBCs, hemoglobin, and hematocrit were also observed in high
12 dose males (1570 mg/kg-day). Hematological changes were within 2-15% of control values.
13 Serum biochemistry changes in exposed mice included decreased total protein (at 1,570
14 mg/kg-day in males, >1,620 mg/kg-day in females), decreased glucose (at 1,570 mg/kg-day in
15 males, >1,620 mg/kg-day in females), decreased albumin (at 1,570 mg/kg-day in males, 2,669
16 mg/ kg-day in females), decreased total cholesterol (> 585 mg/kg-day in males, >1,620
17 mg/kg-day in females), increased serum ALT (at 1,570 mg/kg-day in males, > 620 mg/kg-day in
18 females), increased AST (at 1,570 mg/kg-day in males, 2,669 mg/kg-day in females), increased
19 ALP (> 585 mg/kg-day in males, 2,669 mg/kg-day in females), and increased LDH (in females
20 only at doses > 1,620 mg/kg-day). With the exception of a threefold increase in ALT in male
21 and female mice, serum biochemistry parameters in treated rats did not differ more than twofold
22 from control values. Urinary pH was decreased in males at > 882 mg/kg-day and in females at
23 > 1,620 mg/kg-day.
24 Absolute and relative lung weights were increased in males at 1,570 mg/kg-day and in
25 females at 1,620 and 2,669 mg/kg-day. Absolute kidney weights were also increased in females
26 at 1,620 and 2,669 mg/kg-day and relative kidney weight was elevated at 2,669 mg/kg-day.
27 Histopathology findings in mice that were related to exposure included nuclear enlargement of
28 the respiratory epithelium, nuclear enlargement of the olfactory epithelium, eosinophilic change
29 in the olfactory epithelium, vacuolic change in the olfactory nerve, nuclear enlargement of the
30 tracheal epithelium, accumulation of foamy cells in the lung and bronchi, nuclear enlargement
31 and degeneration of the bronchial epithelium, hepatocyte swelling of the centrilobular area of the
32 liver, and single cell necrosis in the liver. The incidence data for histopathological lesions in
33 mice are presented in Table 4-2. Based on the changes in the bronchial epithelium in female
34 mice, the authors identified the dose level of 387 mg/kg-day as the LOAEL for mice; the
35 NOAEL was 170 mg/kg-day (Kano et al., 2008).
May 2009 30 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-2. Incidence of histopathological lesions in CrjrBDFi mice exposed
to 1,4-dioxane in drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
86
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
231
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
585
2/10
0/10
9/10c
0/10
0/10
7/10c
0/10
9/10c
0/10
10/10C
5/10b
882
5/10b
0/10
10/10C
0/10
0/10
9/10c
0/10
9/10c
0/10
10/10C
10/10C
1,570
0/9
5/9b
9/9c
6/9c
9/9c
9/9c
6/9c
9/9c
8/9c
9/9c
9/9c
Female dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
170
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
1/10
0/10
387
0/10
1/10
0/10
0/10
0/10
2/10
0/10
10/10C
0/10
1/10
0/10
898
3/10
1/10
6/10b
1/10C
0/10
9/10c
0/10
10/10C
0/10
10/10C
7/10c
1,620
3/10
5/10b
10/10C
6/10b
2/10
10/10C
10/10C
10/10C
7/10c
10/10C
10/10C
2,669
7/10c
9/10c
10/10C
6/10b
8/10c
10/10C
10/10C
10/10C
10/10C
9/10b
9/10c
"Data are presented for sacrificed animals.
V < 0.01 by x2 test.
cp < 0.05.
Source: Kano et al (2008).
1 4.2.1.1.4. Yamamoto et al. (1998a, b). Studies in rasH2 transgenic mice carrying the human
2 prototype c-Ha-ras gene have been investigated as a bioassay model for rapid carcinogenicity
3 testing. As part of validation studies of this model, 1,4-dioxane was one of many chemicals that
4 were evaluated. RasH2 transgenic mice were Fl offspring of transgenic male C57BLr6J and
5 normal female BALBrcByJ mice. CB6Fi mice were used as a nontransgenic control. Seven- to
6 nine-week-old mice (10-15/group) were exposed to 0, 0.5, or 1% 1,4-dioxane in drinking water
7 for 26 weeks. An increase in lung adenomas was observed in treated transgenic mice, as
May 2009 31 DRAFT - DO NOT CITE OR QUOTE
-------
1 compared to treated nontransgenic mice. The tumor incidence in transgenic animals, however,
2 was not greater than that observed in vehicle-treated transgenic mouse controls. Further study
3 details were not provided.
4.2.1.2. Chronic Oral Toxicity and Carcinogenicity
4 4.2.1.2.1. Argus et al. (1965), Twenty-six adult male Wistar rats weighing between 150 and
5 200 g were exposed to 1,4-dioxane (purity not reported) in the drinking water at a concentration
6 of 1% for 64.5 weeks. A group of nine untreated rats served as control. Food and water were
7 available ad libitum. The drinking water intake for treated animals was reported to be
8 30 mL/day, resulting in a dose/rat of 300 mg/day. Using a reference BW of 0.462 kg for chronic
9 exposure to male Wistar rats (U.S. EPA, 1988), it can be estimated that these rats received daily
10 doses of approximately 640 mg/kg-day. All animals that died or were killed during the study
11 underwent a complete necropsy. A list of specific tissues examined microscopically was not
12 provided; however, it is apparent that the liver, kidneys, lungs, lymphatic tissue, and spleen were
13 examined. No statistical analysis of the results was conducted.
14 Six of the 26 treated rats developed hepatocellular carcinomas, and these rats had been
15 treated for an average of 452 days (range, 448-455 days). No liver tumors were observed in
16 control rats. In two rats that died after 21.5 weeks of treatment, histological changes appeared to
17 involve the entire liver. Groups of cells were found that had enlarged hyperchromic nuclei. Rats
18 that died or were killed at longer intervals showed similar changes, in addition to large cells with
19 reduced cytoplasmic basophilia. Animals killed after 60 weeks of treatment showed small
20 neoplastic nodules or multifocal hepatocellular carcinomas. No cirrhosis was observed in this
21 study. Many rats had extensive changes in the kidneys often resembling glomerulonephritis,
22 however, incidence data was not reported for these findings. This effect progressed from
23 increased cellularity to thickening of the glomerular capsule followed by obliteration of the
24 glomeruli. One treated rat had an early transitional cell carcinoma in the kidney's pelvis; this rat
25 also had a large tumor in the liver. The lungs from many treated and control rats (incidence not
26 reported) showed severe bronchitis with epithelial hyperplasia and marked peribronchial
27 infiltration, as well as multiple abscesses. One rat treated with 1,4-dioxane developed leukemia
28 with infiltration of all organs, particularly the liver and spleen, with large, round, isolated
29 neoplastic cells. In the liver, the distribution of cells in the sinusoids was suggestive of myeloid
30 leukemia. The dose of 640 mg/kg-day tested in this study was a free-standing LOAEL,
31 identified by EPA, for glomerulonephritis in the kidney and histological changes in the liver
32 (hepatocytes with enlarged hyperchromic nuclei, large cells with reduced cytoplasmic
33 basophilia).
34 4.2.1.2.2. Argus et al. (1973); Hoch-Ligeti et al. (1970). Groups of 2-3-month-old male
35 Sprague Dawley rats (28-32/dose group) weighing 110-230 g at the beginning of the experiment
May 2009 32 DRAFT - DO NOT CITE OR QUOTE
-------
1 were administered 1,4-dioxane (purity not reported) in the drinking water for up to 13 months at
2 concentrations of 0, 0.75, 1.0, 1.4, or 1.8%. The drinking water intake was determined for each
3 group over a 3-day measurement period conducted at the beginning of the study and twice during
4 the study (weeks were not specified). The rats were killed with ether at 16 months or earlier if
5 nasal tumors were clearly observable. Complete autopsies were apparently performed on all
6 animals, but only data from the nasal cavity and liver were presented and discussed. The nasal
7 cavity was studied histologically only from rats in which gross tumors in these locations were
8 present; therefore, early tumors may have been missed and pre-neoplastic changes were not
9 studied. No statistical analysis of the results was conducted. Assuming a BW of 0.523 kg for an
10 adult male Sprague Dawley rat (U.S. EPA, 1988) and a drinking water intake of 30 mL/day as
11 reported by the study authors, dose estimates were 0, 430, 574, 803, and 1,032 mg/kg-day. The
12 progression of liver tumorigenesis was evaluated by an additional group of 10 male rats
13 administered 1% 1,4-dioxane in the drinking water (574 mg/kg-day), 5 of which were sacrificed
14 after 8 months of treatment and 5 were killed after 13 months of treatment. Liver tissue from
15 these rats and control rats was processed for electron microscopy examination.
16 Nasal cavity tumors were observed upon gross examination in six rats (one rat in the
17 0.75% group, one in the 1.0% group, two in the 1.4% group, and two in the 1.8% group). Gross
18 observation showed the tumors visible either at the tip of the nose, bulging out of the nasal
19 cavity, or on the back of the nose covered by intact or later ulcerated skin. As the tumors
20 obstructed the nasal passages, the rats had difficulty breathing and lost weight rapidly. No
21 neurological signs or compression of the brain were observed. In all cases, the tumors were
22 squamous cell carcinomas with marked keratinization and formation of keratin pearls. Bony
23 structure was extensively destroyed in some animals with tumors, but there was no invasion into
24 the brain. In addition to the squamous carcinoma, two adenocarcinomatous areas were present.
25 One control rat had a small, firm, well-circumscribed tumor on the back of the nose, which
26 proved to be subcutaneous fibroma. The latency period for tumor onset was 329-487 days.
27 Evaluation of the latent periods and doses received did not suggest an inverse relationship
28 between these two parameters.
29 Argus et al. (1973) studied the progression of liver tumorigenesis by electron microscopy
30 of liver tissues obtained following interim sacrifice at 8 and 13 months of exposure (5 rats/group,
31 574 mg/kg-day). The first change observed in the liver was an increase in the size of the nucleus
32 of the hepatocytes, mostly in the periportal area. Precancerous changes were characterized by
33 disorganization of the rough endoplasmic reticulum, an increase in smooth endoplasmic
34 reticulum, and a decrease in glycogen and increase in lipid droplets in hepatocytes. These
35 changes increased in severity in the hepatocellular carcinomas in rats exposed to 1,4-dioxane for
36 13 months.
37 Three types of liver nodules were observed in exposed rats at 13-16 months. The first
38 consisted of groups of cells with reduced cytoplasmic basophilia and a slightly nodular
May 2009 33 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
appearance as viewed by light microscopy. The second type of circumscribed nodule was
described consisting of large cells, apparently filled and distended with fat. The third type of
nodule was described as finger-like strands, 2-3 cells thick, of smaller hepatocytes with large
hyperchromic nuclei and dense cytoplasm. This third type of nodule was designated as an
incipient hepatoma, since it showed all the histological characteristics of a fully developed
hepatoma. All three types of nodules were generally present in the same liver. Cirrhosis of the
liver was not observed. The numbers of incipient liver tumors and hepatomas in rats from this
study (treated for 13 months and observed at 13-16 months) are presented in Table 4-3.
Table 4-3. Number of incipient liver tumors and hepatomas in male
Sprague- Dawley rats exposed to 1,4-dioxane in drinking water for
13 months
Dose (mg/kg-day)a
430
574
803
1,032
Incipient tumors
4
9
13
11
Hepatomas
0
0
o
J
12
Total
4
9
16
23
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
"Precise incidences cannot be calculated since the number of rats per group was reported as 28-32; incidence in
control rats was not reported; no statistical analysis of the results was conducted in the study.
Source: Argus etal. (1973).
Treatment with all dose levels of 1,4-dioxane induced marked kidney alterations, but
quantitative incidence data were not provided. Qualitatively, the changes indicated
glomerulonephritis and pyelonephritis, with characteristic epithelial proliferation of Bowman's
capsule, periglomerular fibrosis, and distension of tubules. No kidney tumors were found. No
tumors were found in the lungs. One rat at the 1.4% treatment level showed early peripheral
adenomatous change of the alveolar epithelium and another rat in the same group showed
papillary hyperplasia of the bronchial epithelium. The lowest dose tested (430 mg/kg-day) was
considered a LOAEL by EPA for hepatic and renal effects in this study.
4.2.1.2.3. Hoch-Ligeti and Argus (1970). Hoch-Ligeti and Argus (1970) provided a brief
account of the results of exposure of guinea pigs to 1,4-dioxane. A group of 22 male guinea pigs
(neither strain nor age provided) was administered 1,4-dioxane (purity not provided) in the
drinking water for at least 23 months and possibly up to 28 months. The authors stated that the
concentration of 1,4-dioxane was regulated so that normal growth of the guinea pigs was
maintained, and varied 0.5-2% (no further information provided). The investigators further
stated that the amount of 1,4-dioxane received by the guinea pigs over a 23-month period was
588-635 g. Using a reference BW of 0.89 kg for male guinea pigs in a chronic study (U.S. EPA,
1988) and assuming an exposure period of 700 days (23 months), the guinea pigs received doses
May 2009
34
DRAFT - DO NOT CITE OR QUOTE
-------
1 between 944 and 1,019 mg 1,4-dioxane/kg-day. A group often untreated guinea pigs served as
2 controls. All animals were sacrificed within 28 months, but the scope of the postmortem
3 examination was not provided.
4 Nine treated guinea pigs showed peri- or intrabronchial epithelial hyperplasia and nodular
5 mononuclear infiltration in the lungs. Also, two guinea pigs had carcinoma of the gallbladder,
6 three had early hepatomas, and one had an adenoma of the kidney. Among the controls, four
7 guinea pigs had peripheral mononuclear cell accumulation in the lungs, and only one had
8 hyperplasia of the bronchial epithelium. One control had formation of bone in the bronchus. No
9 further information was presented in the brief narrative of this study. Given the limited reporting
10 of the results, a NOAEL or LOAEL value was not provided for this study.
11 4.2.1.2.4. Kociba et al. (1974). Groups of 6-8-week-old Sherman rats (60/sex/dose level) were
12 administered 1,4-dioxane (purity not reported) in the drinking water at levels of 0 (controls),
13 0.01, 0.1, or 1.0% for up to 716 days. The drinking water was prepared twice weekly during the
14 first year of the study and weekly during the second year of the study. Water samples were
15 collected periodically and analyzed for 1,4-dioxane content by routine gas liquid
16 chromatography. Food and water were available ad libitum. Rats were observed daily for
17 clinical signs of toxicity, and BWs were measured twice weekly during the first month, weekly
18 during months 2-7, and biweekly thereafter. Water consumption was recorded at three different
19 time periods during the study: days 1-113, 114-198, and 446-460. Blood samples were
20 collected from a minimum of five male and five female control and high-dose rats during the 4th,
21 6th, 12th, and 18th months of the study and at termination. Each sample was analyzed for
22 packed cell volume, total erythrocyte count, hemoglobin, and total and differential WBC counts.
23 Additional endpoints evaluated included organ weights (brain, liver, kidney, testes, spleen, and
24 heart) and gross and microscopic examination of major tissues and organs (brain, bone and bone
25 marrow, ovaries, pituitary, uterus, mesenteric lymph nodes, heart, liver, pancreas, spleen,
26 stomach, prostate, colon, trachea, duodenum, kidneys, esophagus, jejunum, testes, lungs, spinal
27 cord, adrenals, thyroid, parathyroid, nasal turbinates, and urinary bladder). The number of rats
28 with tumors, hepatic tumors, hepatocellular carcinomas, and nasal carcinomas were analyzed for
29 statistical significance with Fisher's Exact test (one-tailed), comparing each treatment group
30 against the respective control group. Survival rates were compared using £ Contingency Tables
31 and Fisher's Exact test. Student's t test was used to compare hematological parameters, body
32 and organ weights, and water consumption of each treatment group with the respective control
33 group.
34 Male and female rats in the high-dose group (1% in drinking water) consumed slightly
35 less water than controls. BW gain was depressed in the high-dose groups relative to the other
36 groups almost from the beginning of the study (food consumption data were not provided).
37 Based on water consumption and BW data for specific exposure groups, Kociba et al. (1974)
May 2009 3 5 DRAFT - DO NOT CITE OR QUOTE
-------
1 calculated mean daily doses of 9.6, 94, and 1,015 mg/kg-day for male rats and 19, 148, and
2 1,599 mg/kg-day for female rats during days 114-198 for the 0.01, 0.1, and 1.0% concentration
3 levels, respectively. Treatment with 1,4-dioxane significantly increased mortality among high-
4 dose males and females beginning at about 2-4 months of treatment. These rats showed
5 degenerative changes in both the liver and kidneys. From the 5th month on, mortality rates of
6 control and treated groups were essentially the same. There were no treatment-related alterations
7 in hematological parameters. At termination, the only alteration in organ weights noted by the
8 authors was a significant increase in absolute and relative liver weights in male and female high-
9 dose rats (data not shown). Histopathological lesions were restricted to the liver and kidney from
10 the mid- and high-dose groups and consisted of variable degrees of renal tubular epithelial and
11 hepatocellular degeneration and necrosis (no quantitative incidence data were provided). Rats
12 from these groups also showed evidence of hepatic regeneration, as indicated by hepatocellular
13 hyperplastic nodule formation and evidence of renal tubular epithelial regenerative activity
14 (observed after 2 years of exposure). These changes were not seen in controls or in low-dose
15 rats. The authors determined a LOAEL of 94 mg/kg-day based on the liver and kidney effects in
16 male rats. The corresponding NOAEL value was 9.6 mg/kg-day.
17 Histopathological examination of all the rats in the study revealed a total of 132 tumors in
18 114 rats. Treatment with 1% 1,4-dioxane in the drinking water resulted in a significant increase
19 in the incidence of hepatic tumors (hepatocellular carcinomas in six males and four females). In
20 addition, nasal carcinomas (squamous cell carcinoma of the nasal turbinates) occurred in one
21 high-dose male and two high-dose females. Since 128 out of 132 tumors occurred in rats from
22 the 12th to the 24th month, Kociba et al. (1974) assumed that the effective number of rats was
23 the number surviving at 12 months, which was also when the first hepatic tumor was noticed.
24 The incidences of liver and nasal tumors from Kociba et al. (1974) are presented in Table 4-4.
25 Tumors in other organs were not elevated when compared to control incidence and did not
26 appear to be related to 1,4-dioxane administration.
May 2009 36 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-4. Incidence of liver and nasal tumors in male and female Sherman
rats (combined) treated with 1,4-dioxane in the drinking water for 2 years
Dose in mg/kg-day
(average of male
and female dose)
0
14
121
1307
Effective
number of
animals3
106
110
106
66
Number of tumor-
bearing animals
31
34
28
21
Number of animals
Hepatic tumors
(all types)
2
0
1
12b
Hepatocellular
carcinomas
1
0
1
10C
Nasal
carcinomas
0
0
0
3d
aRats surviving until 12 months on study.
bp = 0.00022 by one-tailed Fisher's Exact test.
cp = 0.00033 by one-tailed Fisher's Exact test.
dp = 0.05491 by one-tailed Fisher's Exact test.
Source: Kocibaetal. (1974).
1 The only dose level that increased the formation of liver tumors over control (average
2 dose for male and female rats, 1,307 mg/kg-day) was also demonstrated to cause significant liver
3 and kidney toxicity in these animals. The mid-dose group (average dose for male and female
4 rats, 121 mg/kg-day) experienced hepatic and renal degeneration and necrosis, as well as
5 regenerative hyperplasia in hepatocytes and renal tubule epithelial cells. No increase in tumor
6 formation was seen in the mid-dose group. No toxicity or tumor formation was observed in the
7 low-dose group of rats (average dose for male and female rats, 14 mg/kg-day).
8 4.2.1.2.5. National Cancer Institute (NCI) (1978). Groups of Osborne-Mendel rats
9 (35/sex/dose) and B6C3Fi mice (50/sex/dose) were administered 1,4-dioxane (> 99.95% pure) in
10 the drinking water for 110 or 90 weeks, respectively, at levels of 0 (matched controls), 0.5, or
11 1%. Solutions of 1,4-dioxane were prepared with tap water. The report indicated that at
12 105 weeks from the earliest starting date, a new necropsy protocol was instituted. This affected
13 the male controls and high-dose rats, which were started a year later than the original groups of
14 rats and mice. Food and water were available ad libitum. Endpoints monitored in this bioassay
15 included clinical signs (twice daily), BWs (once every 2 weeks for the first 12 weeks and every
16 month during the rest of the study), food and water consumption (once per month in 20% of the
17 animals in each group during the second year of the study), and gross and microscopic
18 appearance of all major organs and tissues (mammary gland, trachea, lungs and bronchi, heart,
19 bone marrow, liver, bile duct, spleen, thymus, lymph nodes, salivary gland, pancreas, kidney,
20 esophagus, thyroid, parathyroid, adrenal, gonads, brain, spinal cord, sciatic nerve, skeletal
21 muscle, stomach, duodenum, colon, urinary bladder, nasal septum, and skin). Based on the
22 measurements of water consumption and BWs, the investigators calculated average daily intakes
23 of 1,4-dioxane of 0, 240, and 530 mg/kg-day in male rats, 0, 350, and 640 mg/kg-day in female
24 rats, 0, 720, and 830 mg/kg-day in male mice, and 0, 380, and 860 mg/kg-day in female mice.
May 2009 37 DRAFT - DO NOT CITE OR QUOTE
-------
1 According to the report, the doses of 1,4-dioxane in high-dose male mice were only slightly
2 higher than those of the low-dose group due to decreased fluid consumption in high-dose male
3 mice.
4 During the second year of the study, the BWs of high-dose rats were lower than controls,
5 those of low-dose males were higher than controls, and those of low-dose females were
6 comparable to controls. The fluctuations in the growth curves were attributed to mortality by the
7 investigators; quantitative analysis of BW changes was not done. Mortality was significantly
8 increased in treated rats, beginning at approximately 1 year on study. Analysis of Kaplan-Meier
9 curves (plots of the statistical estimates of the survival probability function) revealed significant
10 positive dose-related trends (p < 0.001, Tarone test). In male rats, 33/35 (94%) in the control
11 group, 26/35 (74%) in the mid-dose group, and 33/35 (94%) in the high-dose group were alive
12 on week 52 of the study. The corresponding numbers for females were 35/35 (100%), 30/35
13 (86%), and 29/35 (83%). Nonneoplastic lesions associated with treatment with 1,4-dioxane were
14 seen in the kidneys (males and females), liver (females only), and stomach (males only). Kidney
15 lesions consisted of vacuolar degeneration and/or focal tubular epithelial regeneration in the
16 proximal cortical tubules and occasional hyaline casts. Elevated incidence of hepatocytomegaly
17 also occurred in treated female rats. Gastric ulcers occurred in treated males, but none were seen
18 in controls. The incidence of pneumonia was increased above controls in high-dose female rats.
19 The incidence of nonneoplastic lesions in rats following drinking water exposure to 1,4-dioxane
20 is presented in Table 4-5. EPA identified the LOAEL in rats from this study as 240 mg/kg-day
21 for increased incidence of gastric ulcer and cortical tubular degeneration in the kidney in males;
22 a NOAEL was not established.
Table 4-5. Incidence of nonneoplastic lesions in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water
Cortical tubule degeneration
Hepatocytomegaly
Gastric ulcer
Pneumonia
Males (mg/kg-day)
0
0/3 la
5/31
(16%)
0/30a
8/30
(27%)
240
20/3 lb
(65%)
3/32
(9%)
5/28b
(18%)
15/31
(48%)
530
27/3 3b
(82%)
11/33
(33%)
5/30b
(17%)
14/33
(42%)
Females (mg/kg-day)
0
0/3 la
7/3 r
(23%)
0/31
6/30a
(20%)
350
0/34
11/33
(33%)
1/33
(3%)
5/34
(15%)
640
10/32b
(31%)
17/32b
(53%)
1/30
(3%)
25/32b
(78%)
""Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this
review.
b-
Incidence significantly elevated compared to control by Fisher's Exact test (p < 0.05) performed for this review.
Source: NCI (1978).
May 2009
38
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Neoplasms associated with 1,4-dioxane treatment were limited to the nasal cavity
(squamous cell carcinomas, adenocarcinomas, and one rhabdomyoma) in both sexes, liver
(hepatocellular adenomas) in females, and testis/epididymis (mesotheliomas) in males. The first
tumors were seen at week 52 in males and week 66 in females. The incidence of squamous cell
carcinomas in the nasal turbinates in male and female rats is presented in Table 4-6. Squamous
cell carcinomas were first seen on week 66 of the study. Morphologically, these tumors varied
from minimal foci of locally invasive squamous cell proliferation to advanced growths consisting
of extensive columns of epithelial cells projecting either into free spaces of the nasal cavity
and/or infiltrating into the submucosa. Adenocarcinomas of the nasal cavity were observed in
3 of 34 high-dose male rats, 1 of 35 low-dose female rats, and 1 of 35 high-dose female rats.
The single rhabdomyoma (benign skeletal muscle tumor) was observed in the nasal cavity of a
male rat from the low-dose group. A subsequent re-examination of the nasal tissue sections by
Goldsworthy et al. (1991) concluded that the location of the tumors in the nasal apparatus was
consistent with the possibility that the nasal tumors resulted from inhalation of water droplets by
the rats (see Section 4.5.2 for more discussion of Goldsworthy et al., 1991).
Table 4-6. Incidence of nasal cavity squamous cell carcinoma and liver
hepatocellular adenoma in Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water
Males (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/33 (0%)
2/31(6%)
240b
12/33 (36%)
2/32 (6%)
530
16/34 (47%)c
1/33 (3%)
Females (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/34 (0%)d
0/31 (0%)f
350
10/35 (29%)e
10/33 (30%)e
640
8/35 (23%)c
ll/32(34%)e
16
17
18
19
20
aTumor incidence values were not adjusted for mortality.
bGroup not included in statistical analysis by NCI because the dose group was started a year earlier without
appropriate controls.
°p < 0.003 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.008 by Cochran-Armitage test.
ep < 0.001 by Fisher's Exact test pair-wise comparison with controls.
{p = 0.001 by Cochran-Armitage test.
Source: NCI (1978).
The incidence of hepatocellular adenomas in male and female rats is presented in
Table 4-7. Hepatocellular adenomas were first observed in high-dose females in week 70 of the
study. These tumors consisted of proliferating hepatic cells oriented as concentric cords.
Hepatic cell size was variable; mitoses and necrosis were rare. Mesothelioma of the vaginal
tunics of the testis/epididymis was seen in male rats (2/33, 4/33, and 5/34 in controls, low-, and
May 2009
39
DRAFT - DO NOT CITE OR QUOTE
-------
1 high-dose animals, respectively). The difference between the treated groups and controls was
2 not statistically significant. These tumors were characterized as rounded and papillary
3 projections of mesothelial cells, each supported by a core of fibrous tissue. Other reported
4 neoplasms were considered spontaneous lesions not related to treatment with 1,4-dioxane.
5 In mice, mean BWs of high-dose female mice were lower than controls during the second
6 year of the study, while those of low-dose females were higher than controls. In males, mean
7 BWs of high-dose animals were higher than controls during the second year of the study.
8 According to the investigators, these fluctuations could have been due to mortality; no
9 quantitative analysis of BWs was done. No other clinical signs were reported. Mortality was
10 significantly increased in female mice (p < 0.001, Tarone test), beginning at approximately
11 80 weeks on study. The numbers of female mice that survived to 91 weeks were 45/50 (90%) in
12 the control group, 39/50 (78%) in the low-dose group, and 28/50 (56%) in the high-dose group.
13 In males, at least 90% of the mice in each group were still alive at week 91. Nonneoplastic
14 lesions that increased significantly due to treatment with 1,4-dioxane were pneumonia in males
15 and females and rhinitis in females. The incidences of pneumonia were 1/49 (2%), 9/50 (18%),
16 and 17/47 (36%) in control, low-dose, and high-dose males, respectively; the corresponding
17 incidences in females were 2/50 (4%), 33/47 (70%), and 32/36 (89%). The incidences of rhinitis
18 in female mice were 0/50, 7/48 (14%), and 8/39 (21%) in control, low-dose, and high-dose
19 groups, respectively. Pair-wise comparisons of low-dose and high-dose incidences with controls
20 for incidences of pneumonia and rhinitis in females using Fisher's Exact test (done for this
21 review) yielded ^-values < 0.001 in all cases. Incidences of other lesions were considered to be
22 similar to those seen in aging mice. The authors stated that hepatocytomegaly was commonly
23 found in dosed mice, but the incidences were not significantly different from controls and
24 showed no dose-response trend. EPA concluded the LOAEL for 1,4-dioxane in mice was
25 380 mg/kg-day based on the increased incidence of pneumonia and rhinitis in female mice; a
26 NOAEL was not established in this study.
27 As shown in Table 4-7, treatment with 1,4-dioxane significantly increased the incidence
28 of hepatocellular carcinomas or adenomas in male and female mice in a dose-related manner.
29 Tumors were first observed on week 81 in high-dose females and in week 58 in high-dose males.
30 Tumors were characterized by parenchymal cells of irregular size and arrangement, and were
31 often hypertrophic with hyperchromatic nuclei. Mitoses were seldom seen. Neoplasms were
32 locally invasive within the liver, but metastasis to the lungs was rarely observed.
May 2009 40 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-7. Incidence of hepatocellular adenoma or carcinoma in B6C3Fi
mice exposed to 1,4-dioxane in drinking water
Males (mg/kg-day)a
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
0
2/49 (4%)b
8/49 (16%)b
720
18/50 (36%)c
19/50 (38%)d
830
24/47 (51%)c
28/47 (60%)c
Females (mg/kg-day)a
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
0
0/50 (0%)b
0/50 (0%)b
380
12/48 (25%)c
21/48 (44%)c
860
29/37 (78%)c
35/37 (95%)c
"Tumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
°p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (1978).
1 In addition to liver tumors, a variety of other benign and malignant neoplasms occurred.
2 However, the report (NCI, 1978) indicated that each type had been encountered previously as a
3 spontaneous lesion in the B6C3Fi mouse. The report further stated that the incidences of these
4 neoplasms were unrelated by type, site, group, or sex of the animal, and hence, not attributable to
5 exposure to 1,4-dioxane. There were a few nasal adenocarcinomas (1/48 in low-dose females
6 and 1/49 in high-dose males) that arose from proliferating respiratory epithelium lining of the
7 nasal turbinates. These growths extended into the nasal cavity, but there was minimal local
8 tissue infiltration. Nasal mucosal polyps were rarely observed. The polyps were derived from
9 mucus-secreting epithelium and were otherwise unremarkable. There was a significant negative
10 trend for alveolar/bronchiolar adenomas or carcinomas of the lung in male mice, such that the
11 incidence in the matched controls was higher than in the dosed groups. The report (NCI, 1978)
12 indicated that the probable reason for this occurrence was that the dosed animals did not live as
13 long as the controls, thus diminishing the possibility of the development of tumors in the dosed
14 groups.
15 4.2.1.2.6. Japan Bioassay Research Center (JBRC) (1998a); Yamazaki et al. (1994).
16 Groups of F344/DuCrj rats (50/sex/dose level) were exposed to 1,4-dioxane (>99% pure) in the
17 drinking water at levels of 0, 200, 1,000, or 5,000 ppm for 2 years. Groups of Crj :BDFi mice
18 (50/sex/dose level) were similarly exposed to 0, 500, 2,000, or 8,000 ppm of 1,4-dioxane in the
19 drinking water. Both rats and mice were 6 weeks old at the beginning of the study. Food and
20 water were available ad libitum. The animals were observed daily for clinical signs of toxicity,
21 and BWs were measured once per week for 14 weeks and once every 2 weeks until the end of
22 the study. Food consumption was measured once a week for 14 weeks and once every 4 weeks
23 for the remainder of the study. The investigators used data from water consumption and BW
May 2009
41
DRAFT - DO NOT CITE OR QUOTE
-------
1 changes to calculate the daily intake of 1,4-dioxane by the male and female animals and
2 presented these estimates as ranges. In order to simplify the summary of the results, the doses
3 presented here represent the midpoint of the ranges calculated by JBRC (1998a). Thus, male rats
4 received doses of approximately 0, 16, 81, or 398 mg/kg-day and female rats received 0, 21, 103,
5 or 514 mg/kg-day. Male mice received 0, 66, 251, or 768 mg/kg-day and female mice received
6 0, 77, 323, or 1,066 mg/kg-day.
7 No information was provided as to when urine samples were collected. Blood samples
8 were collected only at the end of the 2-year study (email from Dr. Kazunori Yamazaki, JBRC, to
9 Dr. Julie Stickney, SRC, dated 12/18/06). Hematology analysis included RBCs, hemoglobin,
10 hematocrit, MCV, platelets, WBCs and differential WBCs. Serum biochemistry included total
11 protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only), phospholipid, ALT, AST,
12 LDH, LAP, ALP, y-glutamyl transpeptidase (GGT), CPK, urea nitrogen, creatinine (rat only),
13 sodium, potassium, chloride, calcium, and inorganic phosphorous. Urinalysis parameters were
14 pH, protein, glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ
15 weights (brain, lung, liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and
16 gross necropsy and histopathologic examination of tissues and organs were performed on all
17 animals (skin, nasal cavity, trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart,
18 tongue, salivary glands, esophagus, stomach, small and large intestine, liver, pancreas, kidney,
19 urinary bladder, pituitary, thyroid, adrenal, testes, epididymis, seminal vesicle, prostate, ovary,
20 uterus, vagina, mammary gland, brain, spinal cord, sciatic nerve, eye, Harderian gland, muscle,
21 bone, and parathyroid). Dunnett's test and ^ test were used to assess the statistical significance
22 of changes in continuous and discrete variables, respectively.
23 Survival was significantly decreased in the rat high-dose groups (80% in control males
24 versus 44% in high-dose males; 76% in control females versus 48% in high-dose females). The
25 effect on survival in high-dose rats occurred in the second year of the study, as all control and
26 exposed rats lived at least 12 months following study initiation (email from Dr. Kazunori
27 Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). The extra mortality in the high-
28 dose groups was primarily related to tumors in these groups (peritoneal mesothelioma, liver and
29 nasal tumors) (email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated
30 12/18/06). Neither food nor water consumption were significantly affected by treatment in males
31 or females. Terminal BWs were reduced 10% in high-dose males and 19% in high-dose females.
32 RBC (males only), hemoglobin, hematocrit, and MCV were decreased, and platelets were
33 increased in high-dose groups. These changes (except for MCV) also occurred in mid-dose
34 males. With the exception of a 23% decrease in hemoglobin in high-dose male rats and a 27%
35 increase in platelets in high-dose female rats, hematological changes were within 15% of control
36 values. Significant changes in serum chemistry parameters occurred only in high-dose rats
37 (males: increased phospholipids, AST, ALT, LDH, ALP, GGT, CPK, potassium, and inorganic
38 phosphorus and decreased total protein, albumin, and glucose; females: increased total bilirubin,
May 2009 42 DRAFT - DO NOT CITE OR QUOTE
-------
1 cholesterol, phospholipids, AST, ALT, LDH, GGT, ALP, CPK, and potassium, and decreased
2 blood glucose). Increases in serum enzyme activities ranged from <2- to 17-fold above control
3 values, with the largest increases seen for ALT, AST, and GGT. Urine pH was significantly
4 decreased at 398 mg/kg-day in males (not tested at other dose levels) and at 103 and
5 514 mg/kg-day in females. Also, blood in the urine was seen in females at 103 and
6 514 mg/kg-day. In males, relative liver weights were increased at 81 and 398 mg/kg-day and
7 absolute liver weights were increased at 398 mg/kg-day. In females, relative and absolute lung
8 and liver weights and relative kidney weights were increased at 514 mg/kg-day.
9 Microscopic examination of the tissues showed nonneoplastic alterations in the nasal
10 cavity, liver, and kidneys mainly in high-dose rats and, in a few cases, in mid-dose rats (Tables
11 4-8 and 4-9). Alterations in high-dose males consisted of nuclear enlargement and metaplasia of
12 the olfactory and respiratory epithelia, atrophy of the olfactory epithelium, hydropic changes and
13 sclerosis of the lamina propria, adhesion, and inflammation. In females, nuclear enlargement of
14 the olfactory epithelium occurred at doses >103 mg/kg-day, and nuclear enlargement and
15 metaplasia of the respiratory epithelium, squamous cell hyperplasia, respiratory metaplasia of the
16 olfactory epithelium, hydropic changes and sclerosis of the lamina propria, adhesion,
17 inflammation, and proliferation of the nasal gland occurred at 514 mg/kg-day. Alterations were
18 seen in the liver at >81 mg/kg-day in males (spongiosis hepatis, hyperplasia, and clear and mixed
19 cell foci) and at 514 mg/kg-day in females (hyperplasia, spongiosis hepatis, cyst formation, and
20 mixed cell foci). Nuclear enlargement of the renal proximal tubule occurred in males at
21 398 mg/kg-day and in females at > 103 mg/kg-day.
May 2009 43 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8. Incidence of histopathological lesions in male F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years
Nuclear enlargement; nasal respiratory epithelium
Squamous cell metaplasia; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Respiratory metaplasia; nasal olfactory epithelium
Atrophy; nasal olfactory epithelium
Hydropic change; lamina propria
Sclerosis; lamina propria
Adhesion; nasal cavity
Inflammation; nasal cavity
Hyperplasia; liver
Spongiosis hepatis; liver
Clear cell foci; liver
Basophilic cell foci; liver
Mixed-cell foci; liver
Nuclear enlargement; kidney proximal tubule
Dose (mg/kg-day)a
0
0/40
0/40
0/40
10/40
0/40
0/40
0/40
0/40
0/40
3/40
12/40
3/40
7/40
2/40
0/40
16
0/45
0/45
0/45
11/45
0/45
0/45
0/45
0/45
0/45
2/45
20/45
3/45
11/45
8/45
0/45
81
0/35
0/35
4/35
17/35
0/35
0/35
1/35
0/35
0/35
9/3 5C
21/35C
9/3 5C
6/35
14/35b
0/35
398
12/22b
15/22b
20/22b
22/22b
17/22b
20/22b
20/22b
21/22b
7/22b
12/22b
21/22b
7/22c
8/22c
22/22b
22/22b
aData presented for sacrificed animals.
bp< 0.01 by x2 test.
cp< 0.05.
Source: JBRC(1998a).
May 2009
44
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-9. Incidence of histopathological lesions in female F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years
Nuclear enlargement; nasal respiratory epithelium
Squamous cell metaplasia; nasal respiratory epithelium
Squamous cell hyperplasia; nasal cavity
Nuclear enlargement; nasal olfactory epithelium
Respiratory metaplasia; nasal olfactory epithelium
Atrophy; nasal olfactory epithelium
Hydropic change; lamina propria
Sclerosis; lamina propria
Adhesion; nasal cavity
Inflammation; nasal cavity
Proliferation; nasal gland
Hyperplasia; liver
Spongiosis hepatis; liver
Cyst formation; liver
Mixed-cell foci; liver
Nuclear enlargement; kidney proximal tubule
Dose (mg/kg-day)a
0
0/38
0/38
0/38
0/38
1/38
0/38
0/38
0/38
0/38
0/38
0/38
2/38
0/38
0/38
1/38
0/38
21
0/37
0/37
0/37
0/37
0/37
0/37
0/37
0/37
0/37
0/37
0/37
2/37
0/37
1/37
1/37
0/37
103
0/38
0/38
0/38
24/3 8b
1/38
1/38
0/38
0/38
0/38
1/38
0/38
9/38
1/38
0/38
3/38
6/38b
514
7/24b
18/24b
4/24c
22/24b
24/24b
22/24b
23/24b
23/24b
24/24b
7/24b
8/24b
24/24b
14/24b
5/24c
7/24a
22/24a
"Data presented for sacrificed animals.
bp< 0.01 by x2 test.
cp < 0.05.
Source: JBRC(1998a).
1 NOAEL and LOAEL values for rats in this study were identified by EPA as 81 and
2 398 mg/kg-day, respectively, based on toxicity observed in nasal tissue of male rats (i.e., atrophy
3 of olfactory epithelium, adhesion, and inflammation). Metaplasia and hyperplasia of the nasal
4 epithelium were also observed in high-dose male and female rats. These effects are likely to be
5 associated with the formation of nasal cavity tumors in these dose groups. Nuclear enlargement
6 was observed in the nasal olfactory epithelium and the kidney proximal tubule at a dose of
7 103 mg/kg-day in female rats; however, it is unclear whether these alterations represent adverse
8 toxicological effects. Hematological effects noted in male rats given 81 and 398 mg/kg-day
9 (decreased RBCs, hemoglobin, hematocrit, increased platelets) were within 20% of control
10 values. In female rats decreases in hematological effects were observed in the high dose group
11 (514 mg/kg-day). A reference range database for hematological effects in laboratory animals
12 (Wolford et al., 1986) indicates that a 20% change in these parameters may fall within a normal
13 range (10th-90th percentile values) and may not represent a treatment-related effect of concern.
14 Liver lesions were also seen at a dose of 81 mg/kg-day in male rats; these changes are likely to
15 be associated with liver tumorigenesis. Clear and mixed-cell foci are commonly considered
16 preneoplastic changes and would not be considered evidence of noncancer toxicity. The nature
May 2009
45
DRAFT - DO NOT CITE OR QUOTE
-------
1 of spongiosis hepatis as a preneoplastic change is less well understood (Bannash, 2003; Karbe
2 and Kerlin, 2002; Stroebel et al., 1995). Spongiosis hepatis is a cyst-like lesion that arises from
3 the perisinusoidal Ito cells of the liver. It is commonly seen in aging rats, but has been shown to
4 increase in incidence following exposure to hepatocarcinogens. Spongiosis hepatis can be seen
5 in combination with preneoplastic foci in the liver or with hepatocellular adenoma or carcinoma
6 and has been considered a preneoplastic lesion (Bannash et al., 2003; Stroebel et al., 1995). This
7 change can also be associated with hepatocellular hypertrophy and liver toxicity and has been
8 regarded as a secondary effect of some liver carcinogens (Karbe and Kerlin, 2002). In the case
9 of the JBRC (1998a) study, spongiosis hepatis was associated with other preneoplastic changes
10 in the liver (clear and mixed-cell foci). No other lesions indicative of liver toxicity were seen in
11 this study; therefore, spongiosis hepatis was not considered indicative of noncancer effects.
12 Serum chemistry changes (increases in total protein, albumin, and glucose; decreases in AST,
13 ALT, LDH, and ALP, potassium, and inorganic phosphorous) were observed in both male and
14 female rats (JBRC, 1998a) in the high dose groups, 398 and 514 mg/kg-day, respectively. These
15 serum chemistry changes seen in terminal blood samples from high-dose male and female rats
16 are likely related to tumor formation in these dose groups.
17 Significantly increased incidences of liver tumors (adenomas and carcinomas) and tumors
18 of the nasal cavity occurred in high-dose male and female rats (Tables 4-11 and 4-12) treated
19 with 1,4-dioxane for 2 years. The first liver tumor was seen at 85 weeks in high-dose male rats
20 and 73 weeks in high-dose female rats (vs. 101-104 weeks in lower dose groups and controls)
21 (email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). In
22 addition, a significant increase (p < 0.01, Fisher's Exact test) in mesotheliomas of the
23 peritoneum was seen in high-dose males (28/50 versus 2/50 in controls). Mesotheliomas were
24 the single largest cause of death among high-dose male rats, accounting for 12 of 28
25 pretermination deaths (email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC,
26 dated 12/18/06). Also, in males, there were increasing trends in mammary gland fibroadenoma
27 and fibroma of the subcutis, both statistically significant (p < 0.01) by the Peto test of dose-
28 response trend. Females showed a significant increasing trend in mammary gland adenomas
29 (p = 0.006 Cochran-Armitage trend test). The tumor incidence values presented in Tables 4-10
30 and 4-11 were not adjusted for survival because all rats lived longer than 12 months on study.
May 2009 46 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-10. Incidence of nasal cavity, peritoneum, and mammary gland
tumors in F344/DuCrj rats exposed to 1,4-dioxane in drinking water for
2 years
Dose (mg/kg-day)
Males
0
16
81
398
Females
0
21
103
514
Nasal Cavity
Squamous cell carcinoma
Sarcoma
Rhabdomyosarcoma
Esthesioneuro-epithelioma
All Nasal Cavity Tumors
0/50
0/50
0/50
0/50
0/50a
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
3/50
2/50
1/50
1/50
7/50b
0/50a
0/50
0/50
0/50
0/50a
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
7/50b
0/50
0/50
1/50
8/50c
Peritoneum
Mesothelioma
2
2
5
28
1
0
0
0
Mammary Gland
Fibroadenoma
Adenoma
All Mammary Gland Tumors
1
0
1
1
0
1
0
0
0
4a
0
4
3
6
9
2
7
9
1
10
11
o
6
16d
19
ap < 0.01 by Peto test for trend.
V < 0.05.
°p < 0.01 by Fisher's Exact test.
dp = 0.006 by Cochran-Armitage trend test.
Source: JBRC(1998a).
Table 4-11. Incidence of liver tumors in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years
Dose (mg/kg-day)
Hepatocellular adenoma
Hepatocellular carcinoma
Adenoma or carcinoma
Males
0
0/50a
0/50a
0/50a
16
2/50
0/50
2/50
81
4/49
0/49
4/49
398
24/50a
14/50a
33/50a
Females
0
l/50b
l/50b
l/50b
21
0/50
0/50
0/50
103
5/50
0/50
5/50
514
38/50a
10/50a
40/50a
ap < 0.01 by Fisher's Exact test.
bp < 0.01 by Peto test for trend.
Source: JBRC(1998a).
1 In the study in mice, survival was low in all male groups (31/50, 33/50, 25/50, and 26/50
2 in control, low-, mid-, and high-dose groups, respectively) and particularly low in high-dose
3 females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-, and high-dose groups,
4 respectively). Deaths occurred primarily during the second year of the study. Survival at
5 12 months in male mice was 50/50, 48/50, 50/50, and 48/50 in control, low-, mid-, and high-dose
6 groups, respectively. Female mouse survival at 12 months was 50/50, 50/50, 48/50, and 48/50 in
7 control, low-, mid-, and high-dose groups, respectively (email from Dr. Kazunori Yamazaki,
May 2009
47
DRAFT - DO NOT CITE OR QUOTE
-------
1 JBRC, to Dr. Julie Stickney, Syracuse Research Corporation (SRC), dated 12/18/06). The deaths
2 were primarily tumor-related (e.g., liver tumors were listed as the cause of death for 31 of the 45
3 pretermination deaths in high-dose female rats) (email from Dr. Kazunori Yamazaki, JBRC, to
4 Dr. Julie Stickney, SRC, dated 12/18/06). Food consumption was not significantly affected, but
5 water consumption was reduced 26% in high-dose males and 28% in high-dose females. Final
6 BWs were reduced 43% in high-dose males and 15 and 45% in mid- and high-dose females,
7 respectively. Males showed increases in RBC counts, hemoglobin, and hematocrit, whereas in
8 females, there was a decrease in platelets in mid- and high-dose rats. With the exception of a
9 60% decrease in platelets in high-dose female mice, hematological changes were within 15% of
10 control values. Serum AST, ALT, LDH, and ALP activities were significantly increased in mid-
11 and high-dose males, whereas LAP and CPK were increased only in high-dose males. AST,
12 ALT, LDH, and ALP activities were increased in mid- and high-dose females, but CPK activity
13 was increased only in high-dose females. Increases in serum enzyme activities ranged from less
14 than two- to sevenfold above control values. Glucose and triglycerides were decreased in high-
15 dose males and in mid- and high-dose females. High-dose females also showed decreases in
16 serum phospholipid and albumin concentrations (not reported in males). Blood calcium was
17 lower in high-dose females and was not reported in males. Urinary pH was decreased in high-
18 dose males, whereas urinary protein, glucose, and occult blood were increased in mid- and high-
19 dose females. Relative and absolute lung weights were increased in high-dose males and in mid-
20 and high-dose females. Microscopic examination of the tissues for nonneoplastic lesions showed
21 significant alterations in the epithelium of the respiratory tract, mainly in high-dose animals,
22 although some changes occurred in mid-dose mice (Tables 4-12 and 4-13). Commonly seen
23 alterations included nuclear enlargement, atrophy, and inflammation of the epithelium. Other
24 notable changes observed included nuclear enlargement of the proximal tubule of the kidney and
25 angiectasis in the liver in high-dose males.
May 2009 48 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-12. Incidence of histopathological lesions in male CrjrBDFi mice
exposed to 1,4-dioxane in drinking water for 2 years
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Atrophy; nasal olfactory epithelium
Inflammation; nasal cavity
Atrophy; tracheal epithelium
Nuclear enlargement; tracheal epithelium
Nuclear enlargement; bronchial epithelium
Atrophy; lung/bronchial epithelium
Accumulation of foamy cells; lung
Angiectasis; liver
Nuclear enlargement; kidney proximal tubule
Dose (mg/kg-day)a
0
0/31
0/31
0/31
1/31
0/31
0/31
0/31
0/31
1/31
2/31
0/31
66
0/33
0/33
0/33
1/33
0/33
0/33
0/33
0/33
0/33
2/33
0/33
251
0/25
7/25b
0/25
1/25
0/25
0/25
0/25
0/25
0/25
3/25
0/25
768
19/26b
26/26b
26/26b
15/26b
24/26b
12/26b
24/26b
26/26b
22/26b
8/26c
22/26b
aData presented for sacrificed animals.
bp< 0.01 by x2 test.
°p< 0.05.
Source: JBRC(1998a).
Table 4-13. Incidence of histopathological lesions in female CrjrBDFi mice
exposed to 1,4-dioxane in drinking water for 2 years
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Atrophy; nasal olfactory epithelium
Inflammation; nasal cavity
Atrophy; tracheal epithelium
Nuclear enlargement; bronchial epithelium
Atrophy; lung/bronchial epithelium
Accumulation of foamy cells; lung
Dose (mg/kg-day)a
0
0/29
0/29
0/29
0/29
0/29
0/29
0/29
0/29
77
0/29
0/29
0/29
0/29
0/29
1/29
0/29
1/29
323
0/17
17/17b
0/17
5/17b
1/17
13/17b
3/17
3/17
1,066
5/5b
1/5
5/5b
5/5b
5/5b
5/5b
5/5b
5/5b
2
3
4
5
"Data presented for sacrificed animals.
bp< 0.01 by x2 test.
Source: JBRC(1998a).
NOAEL and LOAEL values for mice in this study were identified by EPA as 77 and
323 mg/kg-day, respectively, based on nasal inflammation observed in female mice. Nuclear
enlargement of the nasal olfactory epithelium and bronchial epithelium was also observed at a
dose of 323 mg/kg-day in female rats; however, it is unclear whether these alterations represent
adverse toxicological effects. The serum chemistry changes seen in terminal blood samples from
May 2009
49
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
male and female mice (mid- and high-dose groups) are likely related to tumor formation in these
animals. Liver angiectasis, an abnormal dilatation and/or lengthening of a blood or lymphatic
vessel, was seen in male mice given 1,4-dioxane at a dose of 768 mg/kg-day.
Treatment with 1,4-dioxane resulted in an increase in the formation of liver tumors
(adenomas and carcinomas) in male and female mice. The incidence of hepatocellular adenoma
was increased in male mice in the mid-dose group only. The incidence of male mice with
hepatocellular carcinoma or either tumor type (adenoma or carcinoma) was increased in the low,
mid, and high-dose groups. The appearance of the first liver tumor occurred in male mice at 64,
74, 63, and 59 weeks in the control, low- mid-, and high-dose groups, respectively (email from
Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). In female mice,
increased incidence was observed for each liver tumor type in all treatment groups (Table 4-14).
The appearance of the first liver tumor in female mice occurred at 95, 79, 71, and 56 weeks in
the control, low- mid-, and high-dose groups, respectively (email from Dr. Kazunori Yamazaki,
JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). The tumor incidence data presented for male
and female mice were based on mice that survived at least 12 months on study. Mice that died
prior to 12 months were not included in Table 4-14.
Table 4-14. Incidence of liver tumors in CrjrBDFi mice exposed to
1,4-dioxane in drinking water for 2 years
Dose (mg/kg-day)
Hepatocellular adenoma
Hepatocellular carcinoma
Adenoma or carcinoma
Males
0
7/50
15/503
21/50
66
16/48
20/48
31/48C
251
22/50b
23/50
37/50b
768
8/48
36/48b
39/48b
Females
0
4/50
0/50a
4/50a
77
30/50C
6/50b
34/50c
323
20/48C
30/48C
41/48C
1,066
2/48
45/48c
46/48c
17
18
19
20
21
22
zp < 0.05; positive dose-related trend (Cochran-Armitage test or Peto test)
bp < 0.05 by Fisher's Exact test.
°p < 0.01 by Fisher's Exact test.
Sources: JBRC (1998a); email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC (12/18/06).
A weight of evidence evaluation of the carcinogenicity studies presented in Section
4.2.1.2 is located in Section 4.7 and Table 4-18.
4.2.2. Inhalation Toxicity
4.2.2.1. Subchronic Inhalation Toxicity
4.2.2.1.1. Fairley et al. (1934). Rabbits, guinea pigs, rats, and mice (3-6/species/group) were
exposed to 1,000, 2,000, 5,000, or 10,000 ppm of 1,4-dioxane vapor for 3 hours/day,
5 days/week and 1.5 hours on the 6th day (16.5 hours/week). Animals were exposed until death
occurred or were sacrificed at varying time periods. At the 10,000 ppm concentration, only one
May 2009
50
DRAFT - DO NOT CITE OR QUOTE
-------
1 animal (rat) survived a 7-day exposure. The rest of the animals (six guinea pigs, three mice, and
2 two rats) died within the first five exposures. Severe liver and kidney damage and acute vascular
3 congestion of the lungs were observed in these animals. Kidney damage was described as patchy
4 degeneration of cortical tubules with vascular congestion and hemorrhage. Liver lesions varied
5 from cloudy hepatocyte swelling to large areas of necrosis. At 5,000 ppm, mortality was
6 observed in two mice and one guinea pig following 15-34 exposures. The remaining animals
7 were sacrificed following 49.5 hours (3 weeks) of exposure (three rabbits) or 94.5 hours
8 (5 weeks) of exposure (three guinea pigs). Liver and kidney damage in both dead and surviving
9 animals was similar to that described for the 10,000 ppm concentration. Animals (four rabbits,
10 four guinea pigs, six rats, and five mice) were exposed to 2,000 ppm for 45-102 total exposure
11 hours (approximately 2-6 weeks). Kidney and liver damage was still apparent in animals
12 exposed to this concentration. Animals exposed to 1,000 ppm were killed at intervals with the
13 total exposure duration ranging between 78 and 202.5 hours (approximately 4-12 weeks).
14 Cortical kidney degeneration and hepatocyte degeneration and liver necrosis were observed in
15 these animals (two rabbits, three guinea pigs, three rats, and four mice). The low concentration
16 of 1,000 ppm was identified by EPA as a LOAEL for liver and kidney degeneration in rats, mice,
17 rabbits, and guinea pigs in this study.
4.2.2.2. Chronic Inhalation Toxicity and Carcinogenicity
18 4.2.2.2.1. Torkelson et al. (1974), Whole body exposures of male and female Wistar rats
19 (288/sex) to 1,4-dioxane vapors (99.9% pure) at a concentration of 0.4 mg/L (111 ppm), were
20 carried out 7 hours/day, 5 days/week for 2 years. The age of the animals at the beginning of the
21 study was not provided. The concentration of 1,4-dioxane vapor during exposures was
22 determined with infrared analyzers. Food and water were available ad libitum except during
23 exposures. Endpoints examined included clinical signs, eye and nasal irritation, skin condition,
24 respiratory distress, and tumor formation. BWs were determined weekly. Standard
25 hematological parameters were determined on all surviving animals after 16 and 23 months of
26 exposure. Blood collected at termination was used also for determination of clinical chemistry
27 parameters (serum AST and ALP activities, blood urea nitrogen [BUN], and total protein).
28 Liver, kidneys, and spleen were weighed and the major tissues and organs were processed for
29 microscopic examination (lungs, trachea, thoracic lymph nodes, heart, liver, pancreas, stomach,
30 intestine, spleen, thyroid, mesenteric lymph nodes, kidneys, urinary bladder, pituitary, adrenals,
31 testes, ovaries, oviduct, uterus, mammary gland, lacrimal gland, lymph nodes, brain, vagina, and
32 bone marrow, and any abnormal growths). Nasal tissues were not obtained for histopathological
33 evaluation. Control and experimental groups were compared statistically using Student's t test,
34 Yates corrected $ test, or Fisher's Exact test.
35 Exposure to 1,4-dioxane vapors had no significant effect on mortality or BW gain and
36 induced no signs of eye or nasal irritation or respiratory distress. Slight, but statistically
May 2009 51 DRAFT - DO NOT CITE OR QUOTE
-------
1 significant, changes in hematological and clinical chemistry parameters were within the normal
2 physiological limits and were considered to be of no toxicological importance by the
3 investigators. Altered hematological parameters included decreases in packed cell volume, RBC
4 count, and hemoglobin, and an increase in WBC count in male rats. Clinical chemistry changes
5 consisted of a decrease in BUN (control—23 ± 9.9; 111-ppm dioxane—19.8 ± 8.8) and an
6 increase in ALP activity (control—34.4 ± 12.1; 111-ppm dioxane—29.9 ± 9.2) and total protein
7 (control—7.5 ± 0.37; 111-ppm dioxane—7.9 ± 0.53) in male rats (values are mean ± standard
8 deviation). Organ weights were not significantly affected. Microscopic examination of organs
9 and tissues did not reveal any treatment-related effects. Based of the lack of significant effects
10 on several endpoints, EPA identified the exposure concentration of 0.4 mg/L (111 ppm) as a free
11 standing NOAEL. The true NOAEL was likely to be higher.
12 Tumors, observed in all groups including controls, were characteristic of the rat strain
13 used and were considered unrelated to 1,4-dioxane inhalation. The most common tumors were
14 reticulum cell sarcomas and mammary tumors. Using Fisher's Exact test and a significance level
15 ofp < 0.05, no one type of tumor occurred more frequently in treated rats than in controls. No
16 hepatic or nasal cavity tumors were seen in any rat.
4.2.3. Initiation/Promotion Studies
4.2.3.1. Bulletal (1986)
17 Bull et al. (1986) tested 1,4-dioxane as a cancer initiator in mice using oral,
18 subcutaneous, and topical routes of exposure. A group of 40 female SENCAR mice (6-8 weeks
19 old) was administered a single dose of 1,000 mg/kg dioxane (purity >99%) by gavage,
20 subcutaneous injection, or topical administration (vehicle was not specified). A group of rats
21 was used as a vehicle control (number of animals not specified). Food and water were provided
22 ad libitum. Two weeks after administration of 1,4-dioxane, 12-O-tetradecanoylphorbol-13-
23 acetate (TPA) (1.0 jig in 0.2 mL of acetone) was applied to the shaved back of mice
24 3 times/week for a period of 20 weeks. The yield of papillomas at 24 weeks was selected as a
25 potential predictor of carcinoma yields at 52 weeks following the start of the promotion
26 schedule. Acetone was used instead of TPA in an additional group of 20 mice in order to
27 determine whether a single dose of 1,4-dioxane could induce tumors in the absence of TPA
28 promotion.
29 1,4-Dioxane did not increase the formation of papillomas compared to mice initiated with
30 vehicle and promoted with TPA, indicating lack of initiating activity under the conditions of the
31 study. Negative results were obtained for all three exposure routes. A single dose of
32 1,4-dioxane did not induce tumors in the absence of TPA promotion.
May 2009 52 DRAFT - DO NOT CITE OR QUOTE
-------
4.2.3.2. Kingetal. (1973)
1 1,4-Dioxane was evaluated for complete carcinogenicity and tumor promotion activity in
2 mouse skin. In the complete carcinogenicity study, 0.2 mL of a solution of 1,4-dioxane (purity
3 not specified) in acetone was applied to the shaved skin of the back of Swiss Webster mice
4 (30/sex) 3 times/week for 78 weeks. Acetone was applied to the backs of control mice (30/sex)
5 for the same time period. In the promotion study, each animal was treated with 50 ug of
6 dimethylbenzanthracene 1 week prior to the topical application of the 1,4-dioxane solution
7 described above (0.2 mL, 3 times/week, 78 weeks) (30 mice/sex). Acetone vehicle was used in
8 negative control mice (30/sex). Croton oil was used as a positive control in the promotion study
9 (30/sex). Weekly counts of papillomas and suspect carcinomas were made by gross
10 examination. 1,4-Dioxane was also administered in the drinking water (0.5 and 1%) to groups of
11 Osborne-Mendel rats (35/sex/group) and B6C3Fi mice for 42 weeks (control findings were only
12 reported for 34 weeks).
13 1,4-Dioxane was negative in the complete skin carcinogenicity test using dermal
14 exposure. One treated female mouse had malignant lymphoma; however, no papillomas were
15 observed in male or female mice by 60 weeks. Neoplastic lesions of the skin, lungs, and kidney
16 were observed in mice given the promotional treatment with 1,4-dioxane. In addition, the
17 percentage of mice with skin tumors increased sharply after approximately 10 weeks of
18 promotion treatment. Significant mortality was observed when 1,4-dioxane was administered as
19 a promoter (only 4 male and 5 female mice survived for 60 weeks), but not as a complete
20 carcinogen (22 male and 25 female mice survived until 60 weeks). The survival of acetone-
21 treated control mice in the promotion study was not affected (29 male and 26 female mice
22 survived until 60 weeks); however, the mice treated with croton oil as a positive control
23 experienced significant mortality (0 male and 1 female mouse survived for 60 weeks). The
24 incidence of mice with papillomas was similar for croton oil and 1,4-dioxane; however, the
25 tumor multiplicity (i.e., number of tumors/mouse) was higher for the croton oil treatment.
26 Oral administration of 1,4-dioxane in drinking water caused appreciable mortality in rats,
27 but not mice, and increased weight gain in surviving rats and male mice. Histopathological
28 lesions (i.e., unspecified liver and kidney effects) were also reported in exposed male and female
29 rats; however, no histopathological changes were indicated for mice.
30 1,4-Dioxane was demonstrated to be a tumor promoter, but not a complete carcinogen in
31 mouse skin, in this study. Topical administration for 78 weeks following initiation with
32 dimethylbenzanthracene caused an increase in the incidence and multiplicity of skin tumors in
33 mice. Tumors were also observed at remote sites (i.e., kidney and lung), and survival was
34 affected. Topical application of 1,4-dioxane for 60 weeks in the absence of the initiating
35 treatment produced no effects on skin tumor formation or mortality in mice.
May 2009 53 DRAFT - DO NOT CITE OR QUOTE
-------
4.2.3.3. Lundberg et al. (1987)
1 Lundberg et al. (1987) evaluated the tumor promoting activity of 1,4-dioxane in rat liver.
2 Male Sprague Dawley rats (8/dose group, 19 for control group) weighing 200 g underwent a
3 partial hepatectomy followed 24 hours later by an i.p. injection of 30 mg/kg diethylnitrosamine
4 (DEN) (initiation treatment). 1,4-Dioxane (99.5% pure with 25 ppm butylated hydroxytoluene
5 as a stabilizer) was then administered daily by gavage (in saline vehicle) at doses of 0, 100, or
6 1,000 mg/kg-day, 5 days/week for 7 weeks. Control rats were administered saline daily by
7 gavage, following DEN initiation. 1,4-Dioxane was also administered to groups of rats that were
8 not given the DEN initiating treatment (saline used instead of DEN). Ten days after the last
9 dose, animals were sacrificed and liver sections were stained for GGT. The number and total
10 volume of GGT-positive foci were determined.
11 1,4-Dioxane did not increase the number or volume of GGT-foci in rats that were not
12 given the DEN initiation treatment. The high dose of 1,4-dioxane (1,000 mg/kg-day) given as a
13 promoting treatment (i.e., following DEN injection) produced an increase in the number of
14 GGT-positive foci and the total foci volume. Histopathological changes were noted in the livers
15 of high-dose rats. Enlarged, foamy hepatocytes were observed in the midzonal region of the
16 liver, with the foamy appearance due to the presence of numerous fat-containing cytoplasmic
17 vacuoles. These results suggest that cytotoxic doses of 1,4-dioxane may be associated with
18 tumor promotion of 1,4-dioxane in rat liver.
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION
4.3.1. Giavini et al. (1985)
19 Pregnant female Sprague Dawley rats (18-20 per dose group) were given 1,4-dioxane
20 (99% pure, 0.7% acetal) by gavage in water at concentrations of 0, 0.25, 0.5, or 1 mL/kg-day,
21 corresponding to dose estimates of 0, 250, 500, or 1,000 mg/kg-day (density of 1,4-dioxane is
22 approximately 1.03 g/mL). The chemical was administered at a constant volume of 3 mL/kg on
23 days 6-15 of gestation. Food consumption was determined daily and BWs were measured every
24 3 days. The dams were sacrificed with chloroform on gestation day 21 and the numbers of
25 corpora lutea, implantations, resorptions, and live fetuses were recorded. Fetuses were weighed
26 and examined for external malformations prior to the evaluation of visceral and skeletal
27 malformations (Wilson's free-hand section method and staining with Alizarin red) and a
28 determination of the degree of ossification.
29 Maternal weight gain was reduced by 10% in the high-dose group (1,000 mg/kg-day).
30 Food consumption for this group was 5% lower during the dosing period, but exceeded control
31 levels for the remainder of the study. No change from control was observed in the number of
32 implantations, live fetuses, or resorptions; however, fetal birth weight was 5% lower in the
33 highest dose group (p < 0.01). 1,4-Dioxane exposure did not increase the frequency of major
May 2009 54 DRAFT - DO NOT CITE OR QUOTE
-------
1 malformations or minor anomalies and variants. Ossification of the sternebrae was reduced in
2 the 1,000 mg/kg-day dose group (p < 0.05). The study authors suggested that the observed delay
3 in sternebrae ossification combined with the decrease in fetal birth weight indicated a
4 developmental delay related to 1,4-dioxane treatment. NOAEL and LOAEL values of 500 and
5 1,000 mg/kg-day were identified from this study by EPA and based on delayed ossification of
6 the sternebrae and reduced fetal BWs.
4.4. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES
4.4.1. Acute and Short-term Toxicity
7 The acute (< 24 hours) and short-term toxicity studies (<30 days) of 1,4-dioxane in
8 laboratory animals are summarized in Table 4-15. Several exposure routes were employed in
9 these studies, including dermal application, drinking water exposure, gavage, vapor inhalation,
10 and i.v. or i.p. injection.
4.4.1.1. Oral Toxicity
11 Mortality was observed in many acute high-dose studies, and LD50 values for
12 1,4-dioxane were calculated for rats, mice, and guinea pigs (see Table 4-15; Pozzani et al., 1959;
13 Smyth et al., 1941; Laug et al., 1939). Clinical signs of CNS depression were observed,
14 including staggered gait, narcosis, paralysis, coma, and death (Nelson, 1951; Laug et al., 1939;
15 Schrenk and Yant, 1936; de Navasquez, 1935). Severe liver and kidney degeneration and
16 necrosis were often seen in acute studies (JBRC, 1998b; David, 1964; Kesten et al., 1939; Laug
17 et al., 1939; Schrenk and Yant, 1936; de Navasquez, 1935). JBRC (1998b) additionally reported
18 histopathological lesions in the nasal cavity and the brain of rats following 2 weeks of exposure
19 to 1,4-dioxane in the drinking water.
4.4.1.2. Inhalation Toxicity
20 Acute and short-term toxicity studies (all routes) are summarized in Table 4-15.
21 Mortality occurred in many high-concentration studies (Pozzani et al., 1959; Nelson, 1951;
22 Wirth and Klimmer, 1936). Inhalation of 1,4-dioxane caused eye and nasal irritation, altered
23 respiration, and pulmonary edema and congestion (Yant et al., 1930). Clinical signs of CNS
24 depression were observed, including staggered gait, narcosis, paralysis, coma, and death (Nelson,
25 1951; Wirth and Klimmer, 1936). Liver and kidney degeneration and necrosis were also seen in
26 acute and short-term inhalation studies (Drew et al., 1978; Fairley et al., 1934).
May 2009 55 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-15. Acute and short-term toxicity studies of 1,4-dioxane
Animal
Exposure route
Test conditions
Results
Dose3
Reference
Oral studies
Rat (inbred strain
and gender
unspecified)
Rat (strain and
gender unspecified)
F344/DuCrj rat
Female
Sprague Dawley rat
Female Carworth
Farms-Nelson rat
Male Wistar rat,
guinea pig
Rat, mouse, guinea
Pig
Rabbit
Rat, rabbit
Oral via
drinking water
Oral via
drinking water
Oral via
drinking water
Gavage
Gavage
Gavage
Gavage
Gavage
Gavage
1-10 Days of
exposure
5-12 Days of
exposure
14-Day exposure
0, 168, 840, 2550,
or 4,200 mg/kg by
gavage, 21 and
4 hours prior to
sacrifice
Determination of a
single dose LD50
Single dose,
LD50 determination
Single dose;
several dose
groups
Single gavage dose
of 0,207, 1,034, or
2,068 mg/kg-day
Single dose;
mortality after
2 weeks
Ultrastructural
changes in the
kidney, degenerative
nephrosis, hyaline
droplet accumulation,
crystal formation in
mitochondria
Extensive
degeneration of the
kidney, liver damage,
mortality in
8/10 animals by
12 days
Mortality, decreased
BWs,
histopathological
lesions in the nasal
cavity, liver, kidney,
and brain
Increased ODC
activity, hepatic
CYP450 content, and
DNA single-strand
breaks
Lethality
Lethality
Clinical signs of CNS
depression, stomach
hemorrhage, kidney
enlargement, and
liver and kidney
degeneration
Clinical signs of CNS
depression, mortality
at 2068 mg/kg, renal
toxicity (polyuria
followed by anuria),
histopathological
changes in liver and
kidneys
Mortality and
narcosis
11, 000 mg/kg-day
(5%)
11, 000 mg/kg-day
(5%)
2,500 mg/kg-day
(nuclear
enlargement of
olfactory epithelial
cells),
>7,500 mg/kg-day
for all other effects
840 mg/kg (ODC
activity only)
LD50= 6,400 mg/kg
(14,200 ppm)
LD50 (mg/kg):
rat = 7, 120
guinea pig = 3,150
LD50 (mg/kg):
mouse = 5,900
rat = 5,400
guinea pig = 4,030
1,034 mg/kg-day
3, 160 mg/kg
David, 1964
Kestenetal.,
1939
JBRC, 1998b
Kitchin and
Brown, 1990
Pozzani et al.,
1959
Smyth etal.,
1941
Laug et al.,
1939
de
Navasquez,
1935
Nelson, 1951
May 2009
56
DRAFT - DO NOT CITE OR QUOTE
-------
Animal
CrjiBDFj mouse
Dog
Exposure route
Oral via
drinking water
Drinking water
ingestion
Test conditions
14-Day exposure
3-10 Days of
exposure
Results
Mortality, decreased
BWs,
histopathological
lesions in the nasal
cavity, liver, kidney,
and brain
Clinical signs of CNS
depression, and liver
and kidney
degeneration
Dose"
10,800 mg/kg-day;
hepatocellular
swelling
11, 000 mg/kg-day
(5%)
Reference
JBRC, 1998b
Schrenk and
Yant, 1936
Inhalation studies
Male CD1 rat
Rat
Female Carworth
Farms-Nelson rat
Mouse, cat
Guinea pig
Rabbit, guinea pig,
rat, mouse
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Serum enzymes
measured before
and after a single
4 hour exposure
5 Hours of
exposure
Determination of a
4-hour inhalation
LC50
8 Hours/day for
17 days
8-Hour exposure to
0.1-3% by volume
3 Hours exposure,
for 5 days;
1.5 hour exposure
for 1 day
Increase in ALT,
AST, and OCT; no
change in G-6-Pase
Mortality and
narcosis
Lethality
Paralysis and death
Eye and nasal
irritation, retching
movements, altered
respiration, narcosis,
pulmonary edema
and congestion,
hyperemiaof the
brain
Degeneration and
necrosis in the kidney
and liver, vascular
congestion in the
lungs
1,000 ppm
6,000 ppm
LC50=51.3mg/L
8,400 ppm
0.5% by volume
10,000 ppm
Drew et al.,
1978
Nelson, 1951
Pozzani et al.,
1959
Wirth and
Klimmer,
1936
Yantetal.,
1930
Fairley etal.,
1934
Other routes
Male COBS/Wistar
rat
Rabbit, cat
Dermal
i.v. injection
Nonoccluded
technique using
shaved areas of the
back and flank;
single application,
14-day observation
Single injection of
0, 207, 1,034,
1,600 mg/kg-day
Negative; no effects
noted
Clinical signs of CNS
depression, narcosis
at 1,034 mg/kg,
mortality at 1,600
mg/kg
8,300 mg/kg
1,034 mg/kg-day
Clark et al.,
1984
de
Navasquez,
1935
May 2009
57
DRAFT - DO NOT CITE OR QUOTE
-------
Animal
Female
Sprague Dawley rat
CBA/J mouse
Exposure route
i.p. injection
i.p. injection
Test conditions
Single dose;
LD50 values
determined
24 hours and
14 days after
injection
Daily injection for
7 days, 0,0.1, 1,5,
and 10%
Results
Increased serum SDH
activity at l/16th of
the LD50 dose; no
change at higher or
lower doses
Slightly lower
lymphocyte response
to mitogens
Dose"
LD50 (mg/kg):
24 hours = 4,848
14 days = 799
2,000 mg/kg-day
(10%)
Reference
Lundberg
etal., 1986
Thurman
etal., 1978
aLowest effective dose for positive results/ highest dose tested for negative results.
ND = no data; OCT = ornithine carbamyl transferase; ODC = ornithine decarboxylase; SDH = sorbitol
dehydrogenase
4.4.2. Neurotoxicity
1 Clinical signs of CNS depression have been reported in humans and laboratory animals
2 following high dose exposure to 1,4-dioxane (see Sections 4.1 and 4.2.1.1). Neurological
3 symptoms were reported in the fatal case of a worker exposed to high concentrations of
4 1,4-dioxane through both inhalation and dermal exposure (Johnstone, 1959). These symptoms
5 included headache, elevation in blood pressure, agitation and restlessness, and coma. Autopsy
6 findings demonstrated perivascular widening in the brain, with small foci of demyelination in
7 several regions (e.g., cortex, basal nuclei). It was suggested that these neurological changes may
8 have been secondary to anoxia and cerebral edema. In laboratory animals, the neurological
9 effects of acute high-dose exposure included staggered gait, narcosis, paralysis, coma, and death
10 (Nelson, 1951; Laug et al., 1939; Schrenk and Yant, 1936; de Navasquez, 1935; Yant et al.,
11 1930). The neurotoxicity of 1,4-dioxane was further investigated in several studies described
12 below (Frantik et al., 1994; Kanada et al., 1994; Goldberg et al., 1964; Knoefel, 1935).
4.4.2.1. Frantik etal. (1994)
13 The acute neurotoxicity of 1,4-dioxane was evaluated following a 4-hour inhalation
14 exposure to male Wistar rats (four per dose group) and a 2-hour inhalation exposure to female
15 H-strain mice (eight per dose group). Three exposure groups and a control group were used in
16 this study. Exposure concentrations were not specified, but apparently were chosen from the
17 linear portion of the concentration-effect curve. The neurotoxicity endpoint measured in this
18 study was the inhibition of the propagation and maintenance of an electrically-evoked seizure
19 discharge. This endpoint has been correlated with the behavioral effects and narcosis that occur
20 following acute exposure to higher concentrations of organic solvents. Immediately following
21 1,4-dioxane exposure, a short electrical impulse was applied through ear electrodes (0.2 seconds,
22 50 hertz (Hz), 180 volts (V) in rats, 90 V in mice). Several time characteristics of the response
23 were recorded; the most sensitive and reproducible measures of chemically-induced effects were
May 2009
58
DRAFT - DO NOT CITE OR QUOTE
-------
1 determined to be the duration of tonic hind limb extension in rats and the velocity of tonic
2 extension in mice.
3 Linear regression analysis of the concentration-effect data was used to calculate an
4 isoeffective air concentration that corresponds to the concentration producing a 30% decrease in
5 the maximal response to an electrically-evoked seizure. The isoeffective air concentrations for
6 1,4-dioxane were 1,860 ± 200 ppm in rats and 2,400 ± 420 ppm in mice. A NOAEL value was
7 not identified from this study.
4.4.2.2. Goldberg et al. (1964)
8 Goldberg et al. (1964) evaluated the effect of solvent inhalation on pole climb
9 performance in rats. Female rats (Carworth Farms Elias strain) (eight per dose group) were
10 exposed to 0, 1,500, 3,000, or 6,000 ppm of 1,4-dioxane in air for 4 hours/day, 5 days/weeks, for
11 10 exposure days. Conditioned avoidance and escape behaviors were evaluated using a pole
12 climb methodology. Prior to exposure, rats were trained to respond to a buzzer or shock stimulus
13 by using avoidance/escape behavior within 2 seconds. Behavioral criteria were the abolishment
14 or significant deferment (>6 seconds) of the avoidance response (conditioned or buzzer response)
15 or the escape response (buzzer plus shock response). Behavioral tests were administered on day
16 1, 2, 3, 4, 5, and 10 of the exposure period. Rat BWs were also measured on test days.
17 1,4-Dioxane exposure produced a dose-related effect on conditioned avoidance behavior
18 in female rats, while escape behavior was generally not affected. In the 1,500 ppm group, only
19 one of eight rats had a decreased avoidance response, and this only occurred on days 2 and 5 of
20 exposure. A larger number of rats exposed to 3,000 ppm (two or three of eight) experienced a
21 decrease in the avoidance response, and this response was observed on each day of the exposure
22 period. The maximal decrease in the avoidance response was observed in the 6,000 ppm group
23 during the first 2 days of exposure (75-100% of the animals were inhibited in this response). For
24 exposure days 3-10, the percent of rats in the 6,000 ppm group with significant inhibition of the
25 avoidance response ranged from 37-62%. At the end of the exposure period (day 10), the BWs
26 for rats in the high exposure group were lower than controls.
4.4.2.3. Kanada et al. (1994)
27 Kanada et al. (1994) evaluated the effect of oral exposure to 1,4-dioxane on the regional
28 neurochemistry of the rat brain. 1,4-Dioxane was administered by gavage to male
29 Sprague Dawley rats (5/group) at a dose of 1,050 mg/kg, approximately equal to one-fourth the
30 oral LD50. Rats were sacrificed by microwave irradiation to the head 2 hours after dosing, and
31 brains were dissected into small brain areas. Each brain region was analyzed for the content of
32 biogenic amine neurotransmitters and their metabolites using high-performance liquid
33 chromatography (HPLC) or GC methods. 1,4-Dioxane exposure was shown to reduce the
34 dopamine and serotonin content of the hypothalamus. The neurochemical profile of all other
35 brain regions in exposed rats was similar to control rats.
May 2009 59 DRAFT - DO NOT CITE OR QUOTE
-------
4.4.2.4. Knoefel (1935)
1 The narcotic potency of 1,4-dioxane was evaluated following i.p. injection in rats and
2 gavage administration in rabbits. Rats were given i.p. doses of 20, 30, or 50 mmol/kg. No
3 narcotic effect was seen at the lowest dose; however, rats given 30 mmol/kg were observed to
4 sleep approximately 8-10 minutes. Rats given the high dose of 50 mmol/kg died during the
5 study. Rabbits were given 1,4-dioxane at oral doses of 10, 20, 50, 75, or 100 mmol/kg. No
6 effect on the normal erect animal posture was observed in rabbits treated with less than
7 50 mmol/kg. At 50 and 75 mmol/kg, a semi-erect or staggering posture was observed; lethality
8 occurred at both the 75 and 100 mmol/kg doses.
4.5. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION
4.5.1. Genotoxicity
9 The genotoxicity data for 1,4-dioxane are presented in Table 4-16. 1,4-Dioxane has been
10 tested for genotoxic potential using in vitro assay systems with prokaryotic organisms, non-
11 mammalian eukaryotic organisms, and mammalian cells, and in vivo assay systems using several
12 strains of rats and mice. In the large majority of in vitro systems, 1,4-dioxane was not genotoxic.
13 Where a positive genotoxic response was observed, it was generally observed in the presence of
14 toxicity. Similarly, in in vivo systems, 1,4-dioxane was not genotoxic in the majority of
15 available studies. 1,4-Dioxane did not bind covalently to DNA in a single study with calf
16 thymus DNA. Several investigators have reported that 1,4-dioxane caused increased DNA
17 synthesis indicative of cell proliferation. Overall, the available literature indicates that
18 1,4-dioxane is nongenotoxic or weakly genotoxic.
19 Negative findings were reported for mutagenicity in in vitro assays with the prokaryotic
20 organisms Salmonella typhimurium, Escherichia coli, and Photobacterium phosphoreum
21 (Mutatox assay) (Morita and Hayashi, 1998; Hellmer and Bolcsfoldi, 1992; Kwan et al., 1990;
22 Khudoley et al., 1987; Nestmann et al., 1984; Haworth et al., 1983; Stott et al., 1981). In in vitro
23 assays with nonmammalian eukaryotic organisms, negative results were obtained for the
24 induction of aneuploidy in yeast (Saccharomyces cerevisiae) and in the sex-linked recessive
25 lethal test in Drosophila melanogaster (Yoon et al., 1985; Zimmerman et al., 1985). In the
26 presence of toxicity, positive results were reported for meiotic nondisjunction in Drosophila
27 (Munoz and Barnett, 2002).
28 The ability of 1,4-dioxane to induce genotoxic effects in mammalian cells in vitro has
29 been examined in model test systems with and without exogenous metabolic activation and in
30 hepatocytes that retain their xenobiotic-metabolizing capabilities. 1,4-Dioxane was reported as
31 negative in the mouse lymphoma cell forward mutation assay (Morita and Hayashi, 1998;
32 McGregor et al., 1991). 1,4-Dioxane did not produce chromosomal aberrations or micronucleus
May 2009 60 DRAFT - DO NOT CITE OR QUOTE
-------
1 formation in Chinese hamster ovary (CHO) cells (Morita and Hayashi, 1998; Galloway et al.,
2 1987). Results were negative in one assay for sister chromatid exchange (SCE) in CHO (Morita
3 and Hayashi, 1998) and were weakly positive in the absence of metabolic activation in another
4 (Galloway et al., 1987). In rat hepatocytes, 1,4-dioxane exposure in vitro caused single-strand
5 breaks in DNA at concentrations also toxic to the hepatocytes (Sina et al., 1983) and produced a
6 positive genotoxic response in a cell transformation assay with BALB/3T3 cells also in the
7 presence of toxicity (Sheu et al., 1988).
8 1,4-Dioxane was not genotoxic in the majority of available in vivo mammalian assays.
9 Studies of micronucleus formation following in vivo exposure to 1,4-dioxane produced mostly
10 negative results, including studies of bone marrow micronucleus formation in B6C3Fi, BALB/c,
11 CBA, and C57BL6 mice (McFee et al., 1994; Mirkova, 1994; Tinwell and Ashby, 1994) and
12 micronucleus formation in peripheral blood of CD1 mice (Morita and Hayashi, 1998; Morita,
13 1994). Mirkova (1994) reported a dose-related increase in the incidence of bone marrow
14 micronuclei in male and female C57BL6 mice 24 or 48 hours after administration of
15 1,4-dioxane. At a sampling time of 24 hours, a dose of 450 mg/kg produced no change relative
16 to control, while doses of 900, 1,800, and 3,600 mg/kg increased the incidence of bone marrow
17 micronuclei by approximately two-, three-, and fourfold, respectively. A dose of 5,000 mg/kg
18 also increased the incidence of micronuclei by approximately fourfold at 48 hours. This
19 compares with the negative results for BALB/c male mice tested in the same study at a dose of
20 5,000 mg/kg and sampling time of 24 hours. Tinwell and Ashby (1994) could not explain the
21 difference in response in the mouse bone marrow micronucleus assay with C57BL6 mice
22 obtained in their laboratory (i.e., nonsignificant 1.6-fold increase over control) with the dose-
23 related positive findings reported by Mirkova (1994) using the same mouse strain, 1,4-dioxane
24 dose (3,600 mg/kg) and sampling time (24 hours). Morita and Hayashi (1998) demonstrated an
25 increase in micronucleus formation in hepatocytes following 1,4-dioxane dosing and partial
26 hepatectomy to induce cellular mitosis. DNA single-strand breaks were demonstrated in
27 hepatocytes following gavage exposure to female rats (Kitchin and Brown, 1990).
28 Roy et al. (2005) examined micronucleus formation in male CD1 mice exposed to
29 1,4-dioxane to confirm the mixed findings from earlier mouse micronucleus studies and to
30 identify the origin of the induced micronuclei. Mice were administered 1,4-dioxane by gavage at
31 doses of 0, 1,500, 2,500, and 3,500 mg/kg-day for 5 days. The mice were also implanted with
32 5-bromo-2-deoxyuridine (BrdU)-releasing osmotic pumps to measure cell proliferation in the
33 liver and to increase the sensitivity of the hepatocyte assay. The frequency of micronuclei in the
34 bone marrow erythrocytes and in the proliferating BrdU-labeled hepatocytes was determined
35 24 hours after the final dose. Significant dose-related increases in micronuclei were seen in the
36 bone-marrow at all the tested doses (> 1,500 mg/kg-day). In the high-dose (3,500-mg/kg) mice,
37 the frequency of bone marrow erythrocyte micronuclei was about 10-fold greater than the control
38 frequency. Significant dose-related increases in micronuclei were also observed at the two
May 2009 61 DRAFT - DO NOT CITE OR QUOTE
-------
1 highest doses (> 2,500 mg/kg-day) in the liver. Antikinetochore (CREST) staining or
2 pancentromeric fluorescence in situ hybridization (FISH) was used to determine the origin of the
3 induced micronuclei. The investigators determined that 80-90% of the micronuclei in both
4 tissues originated from chromosomal breakage; small increase in micronuclei originating from
5 chromosome loss was seen in hepatocytes. Dose-related statistically significant decreases in the
6 ratio of bone marrow polychromatic erythrocytes (PCE):normochromatic erthyrocytes (NCE), an
7 indirect measure of bone marrow toxicity, were observed. Decreases in hepatocyte proliferation
8 were also observed. Based on these results, the authors concluded that at high doses 1,4-dioxane
9 exerts genotoxic effects in both the mouse bone marrow and liver; the induced micronuclei are
10 formed primarily from chromosomal breakage; and 1,4-dioxane can interfere with cell
11 proliferation in both the liver and bone marrow. The authors noted that reasons for the
12 discrepant micronucleus assay results among various investigators was unclear, but could be
13 related to the inherent variability present when detecting moderate to weak responses using small
14 numbers of animals, as well as differences in strain, dosing regimen, or scoring criteria.
15 1,4-Dioxane did not affect in vitro or in vivo DNA repair in hepatocytes or in vivo DNA
16 repair in the nasal cavity (Goldsworthy et al., 1991; Stott et al., 1981), but increased hepatocyte
17 DNA synthesis indicative of cell proliferation in several in vivo studies (Miyagawa et al., 1999;
18 Uno et al., 1994; Goldsworthy et al., 1991; Stott et al., 1981). 1,4-Dioxane caused a transient
19 inhibition of RNA polymerase A and B in the rat liver (Kurl et al., 1981), indicating a negative
20 impact on the synthesis of ribosomal and messenger RNA (DNA transcription). Intravenous
21 administration of 1,4-dioxane at doses of 10 or 100 mg/rat produced inhibition of both
22 polymerase enzymes, with a quicker and more complete recovery of activity for RNA
23 polymerase A, the polymerase for ribosomal RNA synthesis.
24 1,4-Dioxane did not covalently bind to DNA under in vitro study conditions (Woo et al.,
25 1977a). DNA alkylation was also not detected in the liver 4 hours following a single gavage
26 exposure (1,000 mg/kg) in male Sprague Dawley rats (Stott et al., 1981).
27 Rosenkranz and Klopman (1992) analyzed 1,4-dioxane using the computer automated
28 structure evaluator (CASE) structure activity method to predict its potential genotoxicity and
29 carcinogenicity. The CASE analysis is based on information contained in the structures of
30 approximately 3,000 chemicals tested for endpoints related to mutagenic/genotoxic and
31 carcinogenic potential. CASE selects descriptors (activating [biophore] or inactivating
32 [biophobe] structural fragments) from a learning set of active and inactive molecules. Using the
33 CASE methodology, Rosenkranz and Klopman (1992) predicted that 1,4-dioxane would be
34 inactive for mutagenicity in several in vitro systems, including Salmonella, induction of
35 chromosomal aberrations in CHO cells, and unscheduled DNA synthesis in rat hepatocytes.
36 1,4-Dioxane was predicted to induce SCE in cultured CHO cells, micronuclei formation in rat
37 bone marrow, and carcinogenicity in rodents.
May 2009 62 DRAFT - DO NOT CITE OR QUOTE
-------
1 Gene expression profiling in cultured human hepatoma HepG2 cells was performed using
2 DNA microarrays to discriminate between genotoxic and other carcinogens (van Delft et al.,
3 2004). Van Delft et al. (2004) examined this method using a training set of 16 treatments (nine
4 genotoxins and seven nongenotoxins) and a validation set (three and three), with discrimination
5 models based on Pearson correlation analyses for the 20 most discriminating genes. 1,4-Dioxane
6 was included in the training set as a "nongenotoxic" carcinogen. The gene expression profile for
7 1,4-dioxane indicated a classification of this chemical as a "nongenotoxic" carcinogen. The
8 correctness for carcinogen classification using this method ranged from 33 to 100%, depending
9 on which chemical data sets and gene expression signals were included in the analysis.
Table 4-16a. Genotoxicity studies of 1,4-dioxane
Test system
Endpoint
Test conditions
Results3
Without
activation
With
activation
Doseb
Source
Prokaryotic organisms in vitro
S. typhimurium
strains TA98, TA100,
TA1535, TA1537
S. typhimurium
strains TA98, TA100,
TA1530, TA1535,
TA1537
S. typhimurium
strains TA98, TA100,
TA1535, TA1537
S. typhimurium
strains TA100,
TA1535
S. typhimurium
strains TA98, TA100,
TA1535, TA1537,
TA1538
E. coli K-12
uvrB/recA
E. coli
WP2/WP2uvrA
P. phosphoreum
M169
Reverse
mutation
Reverse
mutation
Reverse
mutation
Reverse
mutation
Reverse
mutation
DNA repair
Reverse
mutation
Mutagenicity,
DNA damage
Plate incorporation
assay
Plate incorporation
assay
Plate incorporation
and preincubation
assays
Preincubation
assay
Plate incorporation
assay
Host mediated
assay
Plate incorporation
and preincubation
assays
Mutatox assay
—
—
—
—
—
-
—
—
—
—
—
ND
10,000 ug/plate
ND
5,000 ug/plate
103 mg
103 mg
l,150mmol/L
5,000 ug/plate
ND
Haworth
etal., 1983
Khudoley
etal., 1987
Morita and
Hayashi,
1998
Nestmann
etal., 1984
Stott et al.,
1981
Hellmer and
Bolcsfoldi,
1992
Morita and
Hayashi,
1998
Kwan et al.,
1990
May 2009
63
DRAFT - DO NOT CITE OR QUOTE
-------
Test system
Endpoint
Test conditions
Results3
Without
activation
With
activation
Doseb
Source
Nonmammalian eukaryotic organisms in vitro
S. cerevisiae D6 1 .M
D. melanogaster
D. melanogaster
Aneuploidy
Meiotic
nondisjunction
Sex-linked
recessive lethal
test
Standard 16-hour
incubation or cold-
interruption
regimen
Oocytes were
obtained for
evaluation 24 and
48 hours after
mating
Exposure by
feeding and
injection
-T
+r
ND
NDd
NDd
4.75%
2% in sucrose
media
35,000ppmin
feed, 7 days or
50,000 ppm
(5% in water)
by injection
Zimmerman
etal., 1985
Munoz and
Barnett, 2002
Yoon et al.,
1985
Mammalian cells in vitro
Rat hepatocytes
Primary hepatocyte
culture from male
F344 rats
L5178Y mouse
lymphoma cells
L5178Y mouse
lymphoma cells
BALB/3T3 cells
DNA damage;
single-strand
breaks measured
by alkaline
elution
DNA repair
Forward
mutation assay
Forward
mutation assay
Cell
transformation
3 -Hour exposure
to isolated primary
hepatocytes
Autoradiography
Thymidine kinase
mutagenicity assay
(trifluorothymidin
e resistance)
Thymidine kinase
mutagenicity assay
(trifluorothymidin
e resistance)
48-Hour exposure
followed by
4 weeks
incubation; 13 day
exposure followed
by 2.5 weeks
incubation
+Te
—
+Tf
NDd
NDd
-T
NDd
0.3 mM
ImM
5,000 ug/mL
5,000 ug/mL
0.5 mg/mL
Sinaetal.,
1983
Goldsworthy
etal., 1991
McGregor
etal., 1991
Morita and
Hayashi,
1998
Sheu et al.,
1988
May 2009
64
DRAFT - DO NOT CITE OR QUOTE
-------
Test system
CHO cells
CHO cells
CHO cells
CHO cells
Endpoint
SCE
Chromosomal
aberration
SCE
Chromosomal
aberration
Test conditions
BrdU was added
2 hours after
1,4-dioxane
addition; chemical
treatment was
2 hours with S9
and 25 hours
without S9
Cells were
harvested 8-
12 hours or 18-
26 hours after
treatment (time of
first mitosis)
3 Hour pulse
treatment;
followed by
continuous
treatment of BrdU
for 23 or 26 hours
5 Hour pulse
treatment, 20 hour
pulse and
continuous
treatments, or 44
hour continuous
treatment; cells
were harvested 20
or 44 hours
following
exposure
Results3
Without
activation
±g
With
activation
_
Doseb
10,520 ug/mL
10,520 ug/mL
5,000 ug/mL
5,000 ug/mL
Source
Galloway
etal., 1987
Galloway
etal., 1987
Morita and
Hayashi,
1998
Morita and
Hayashi,
1998
May 2009
65
DRAFT - DO NOT CITE OR QUOTE
-------
Test system
CHO cells
CalfthymusDNA
Endpoint
Micronucleus
formation
Covalent
binding to DNA
Test conditions
5 Hour pulse
treatment or 44
hour continuous
treatment; cells
were harvested 42
hours following
exposure
Incubation with
microsomes from
3 -methy Icholanthr
ene treated rats
Results3
Without
activation
With
activation
Doseb
5,000 ug/mL
0.04 pmol/mg
DNA (bound)
Source
Morita and
Hayashi,
1998
Woo et al.,
1977a
a + = positive, ± = equivocal or weak positive, - = negative, T = toxicity. Endogenous metabolic
activation is not applicable for in vivo studies.
b Lowest effective dose for positive results/highest dose tested for negative results; ND = no data.
0 A dose-related decrease in viability was observed with 0, 2.4, 8.1, 51.7, and 82.8% mortality at
concentrations of 1, 1.5, 2, 3, and 3.5%, respectively. In mature oocytes, meiotic nondisjunction was
decreased at 2, 3, and 3.5%; however, a dose-response trend was not evident.
d Exogenous metabolic activation not used for most tests of fungi and many mammalian cell types in
vitro, or in vivo studies in mammals, due to endogenous metabolic ability in many of these systems.
e Cell viability was 98, 57, 54, 31, and 34% of control at concentrations 0, 0.03, 0.3, 10, and 30 mM. DNA
damage was observed at 0.3, 3, 10, and 30 mM; however, no dose-response trend was observed for the
extent of DNA damage (severity score related to the elution rate).
f For the 13-day exposure, relative survival was 92, 85, 92, and 61% of control for concentrations of 0.25,
0.5, 1, and 2 mg/mL, respectively. A significant increase in transformation frequency was observed at
the highest dose level (2 mg/mL). Similar results were observed for the 48-hour exposure, with
increased transformation frequency seen at concentrations of 2, 3, and 4 mg/mL. Concentrations >2
mg/mL also caused a significant decrease in cell survival (relative survival ranged between 6 and 52% of
control).
8 The highest concentration tested (10,520 ug/L) produced a 27% increase in the number of SCE/cell in the
absence of S9 mix. No effect was seen at lower doses (1,050 and 3,500 ug/L) in the absence of S9 mix
or at any concentration level (1,050, 3,500, 10,500 ug/L) tested in the presence of S9.
Table 4-16b. Genotoxicity studies of 1,4-dioxane; mammalian in vivo
Test system
Female
Sprague Dawley
Rat
Male
Sprague Dawley
Rat
Endpoint
DNA damage;
single-strand breaks
measured by alkaline
elution
DNA alkylation in
hepatocytes
Test Conditions
Two gavage doses given 21
and 4 hours prior to
sacrifice
Gavage; DNA isolation and
HPLC analysis 4 hours after
dosing
Results
+h
—
Dose
2,550 mg/kg
1,000 mg/kg
Source
Kitchin and
Brown, 1990
Stott et al.,
1981
May 2009
66
DRAFT - DO NOT CITE OR QUOTE
-------
Test system
Male
B6C3FJ
Mouse
Male and female
C57BL6
Mouse;
male BALB/c
Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
CBAand
C57BL6 Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
Sprague Dawley
Rat
Male
F344
Rat
Male
F344
Rat
Male
F344
Rat
Endpoint
Micronucleus
formation in bone
marrow
Micronucleus
formation in bone
marrow
Micronucleus
formation in
peripheral blood
Micronucleus
formation in
hepatocytes
Micronucleus
formation in
peripheral blood
Micronucleus
formation in bone
marrow
Micronuclei
formation in bone
marrow
Micronuclei
formation in
hepatocytes
DNA repair in
hepatocytes
DNA repair in
hepatocytes
(autoradiography)
DNA repair in nasal
epithelial cells from
the nasoturbinate or
maxilloturbinate
Replicative DNA
synthesis (i.e., cell
proliferation) in
hepatocytes
Test Conditions
i.p. injection; analysis of
polychromatic erythrocytes
24 or 48 hours after dosing
Gavage; analysis of
polychromatic erythrocytes
24 or 48 hours after dosing
Two i.p. injections (I/day);
micronucleated
reticulocytes measured 24,
48, and 72 hours after the
2nd dose
Gavage, partial
hepatectomy 24 hours after
dosing, hepatocytes
analyzed 5 days after
hepatectomy
Gavage, partial
hepatectomy 24 hours after
dosing, peripheral blood
obtained from tail vein 24
hours after hepatectomy
Gavage; analysis of
polychromatic erythrocytes
from specimens prepared
24 hours after dosing
Gavage; analysis for
micronucleated erythrocytes
24 hours after dosing
Gavage; analysis for
micronuclei 24 hours after
dosing
Drinking water; thymidine
incorporation with
hydroxyurea to repress
normal DNA synthesis
Gavage and drinking water
exposure; thymidine
incorporation
Gavage and drinking water
exposure; thymidine
incorporation
Gavage and drinking water
exposure; thymidine
incorporation
Results
+ (C57BL6)1
- (BALB/c)
+J
+k
+1
+m
(1-2 -week
exposure)
Dose
Single dose of
4,000 mg/kg;
3 daily doses of
2,000
900 mg/kg
(C57BL6);
5,000 mg/kg
(BALB/c)
3,200 mg/kg
2,000 mg/kg
3,000 mg/kg
3,600 mg/kg
1,500 mg/kg-day
for 5 days
2,500 mg/kg-day
for 5 days
1,000 mg/kg-day
for 1 1 weeks
1,000 mg/kg for
2 or 12 hours;
1,500 mg/kg-day
for 2 weeks or
3,000 mg/kg-day
for 1 week
1,500 mg/kg-day
for 8 days +
1,000 mg/kg
gavage dose
12 hours prior to
sacrifice
1,000 mg/kg for
24 or 48 hours;
1,500 mg/kg-day
for 1 or 2 weeks
Source
McFee et al.,
1994
Mirkova,
1994
Morita, 1994
Morita and
Hayashi,
1998
Morita and
Hayashi,
1998
Tinwell and
Ashby, 1994
Roy etal.,
2005
Roy etal.,
2005
Stott et al.,
1981
Goldsworthy
etal., 1991
Goldsworthy
etal., 1991
Goldsworthy
etal., 1991
May 2009
67
DRAFT - DO NOT CITE OR QUOTE
-------
Test system
Male
F344
Rat
Male
Sprague Dawley
Rat
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Endpoint
Replicative DNA
synthesis (i.e., cell
proliferation) in nasal
epithelial cells
RNA synthesis;
inhibition of RNA
polymerase A and B
DNA synthesis in
hepatocytes
DNA synthesis in
hepatocytes
DNA synthesis in
hepatocytes
Test Conditions
Drinking water exposure;
thymidine incorporation
i.v. injection; activity
measured in isolated
hepatocytes
Gavage; thymidine and
BrdU incorporation
Thymidine incorporation
Drinking water; thymidine
incorporation
Results
+n
+°
±P
+q
Dose
1,500 mg/kg-day
for 2 weeks
10 mg/rat
1,000 mg/kg
2,000 mg/kg
1,000 mg/kg-day
for 1 1 weeks
Source
Goldsworthy
etal., 1991
Kurletal.,
1981
Miyagawa
etal., 1999
Uno et al.,
1994
Stott et al.,
1981
a + = positive, ± = equivocal or weak positive, - = negative, T = toxicity. Endogenous metabolic
activation is not applicable for in vivo studies.
b Lowest effective dose for positive results/highest dose tested for negative results; ND = no data.
hRats were given doses of 0, 168, 840, 2,550, or 4,200 mg/kg at 4 and 21 hours prior to sacrifice. A 43
and 50% increase in the fraction of DNA eluted was observed for doses of 2,550 and 4,200 mg/kg,
respectively. Alkaline elution of DNA was not significantly different from control in the two lowest
dose groups (168 and 840 mg/kg).
1A dose-related increase in the incidence of bone marrow micronuclei was observed in male and female
C57BL6 mice 24 or 48 hours after administration of 1,4-dioxane. A dose of 450 mg/kg produced no
change relative to control, while doses of 900, 1,800, 3,600, and 5,000 mg/kg increased the incidence of
bone marrow micronuclei by approximately two-,three-, four- and fourfold, respectively.
J A dose-related increase in the incidence of hepatocyte micronuclei was observed in partially
hepatectomized mice 6 days after administration of 1,4-dioxane. A dose of 1,000 mg/kg produced no
change relative to control, while doses of 2,000 and 3,000 mg/kg increased the incidence of hepatocyte
micronuclei by 2.4- and 3.4-fold, respectively.
k Significant increases in the frequency of micronucleated erythrocytes were observed at each test dose of
1,4-dioxane (1,500, 2,500 and 3,500 mg/kg-day, 5 days/week).
1A dose-related increase in the frequency of micronuclei was observed in proliferating cells with micronuclei at
2,500 and 3,500 mg/kg-day, 5 days/week. No increase in the frequency of micronuclei was seen in the non-
proliferating cells.
m No increase in the hepatocyte labeling index was observed 24 or 48 hours following a single gavage
exposure of 1,000 mg/kg. Continuous administration of 1% 1,4-dioxane in the drinking water for up to 2
weeks produced a twofold increase in the hepatocyte labeling index.
n A similar pattern of RNA polymerase inhibition was observed at doses of 10 and 100 mg/rat. Inhibition
was more pronounced at the higher dose.
"Hepatocyte viability was 86, 89, 87, 88, 78, and 86% 24 hours following exposure to 0, 1,000, 1,500,
2,000, or 4,000 mg/kg. The incidence (%) of replicative DNA synthesis was increased by 2.5-fold (1,000
mg/kg) or 4.5-fold (1,500 and 2,000 mg/kg). No increase in replicative DNA synthesis was observed at
the highest dose (4,000 mg/kg).
p Replicative DNA synthesis was measured 24, 39, and 48 hours following a single dose of 0, 1,000, or
2,000 mg/kg. Hepatocyte viability ranged from 71 to 82%. The only increase in replicative DNA
synthesis was observed 24 hours after administration of 2,000 mg/kg (threefold increase). Cell viability
for this group was 79%.
qReplicative DNA synthesis was increased 1.5-fold in rats given 1,000 mg/kg of 1,4-dioxane for 11
weeks. No change from control was observed in rats exposed to 10 mg/kg for 11 weeks or rats acutely
exposed to 10, 100, or 1,000 mg/kg.
May 2009
68
DRAFT - DO NOT CITE OR QUOTE
-------
4.5.2. Mechanistic Studies
4.5.2.1. Free Radical Generation
1 Burmistrov et al. (2001) investigated the effect of 1,4-dioxane inhalation on free radical
2 processes in the rat ovary and brain. Female rats (6-9/group, unspecified strain) were exposed to
3 0, 10, or 100 mg/m3 of 1,4-dioxane vapor for 4 hours/day, 5 days/week, for 1 month. Rats were
4 sacrificed during the morning or evening following exposure and the ovaries and brain cortex
5 were removed and frozen. Tissue preparations were analyzed for catalase activity, glutathione
6 peroxidase activity, and protein peroxidation. 1,4-Dioxane inhalation was shown to increase
7 glutathione peroxidase activity at the 100 mg/m3 exposure level only in both rat ovary and rat
8 brain. No change in catalase activity or protein peroxidation was observed at either
9 concentration. A circadian rhythm for glutathione peroxidase activity was absent in control rats,
10 but occurred in rat brain and ovary following 1,4-dioxane exposure.
4.5.2.2. Induction of Metabolism
11 The metabolism of 1,4-dioxane is discussed in detail in Section 3.3. 1,4-Dioxane has
12 been shown to induce its own metabolism (Young et al., 1978a, b). Nannelli et al. (2005)
13 characterized the CYP450 isozymes that were induced by 1,4-dioxane in the liver, kidney, and
14 nasal mucosa of the rat. In the liver, the activities of several CYP450 isozymes were increased
15 (i.e., CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was inducible in the kidney and
16 nasal mucosa. CYP2E1 mRNA was increased approximately two- to threefold in the kidney and
17 nasal mucosa, but mRNA levels were not increased in the liver, suggesting that regulation of
18 CYP2E1 is organ-specific. Induction of hepatic CYPB1/2 and CYP2E1 levels by phenobarbital
19 or fasting did not increase the liver toxicity of 1,4-dioxane, as measured by hepatic glutathione
20 content or serum ALT activity. This result suggested that highly reactive and toxic intermediates
21 did not play a large role in the liver toxicity of 1,4-dioxane, even under conditions where
22 metabolism was enhanced. This finding was supported by a previous comparison of the
23 pharmacokinetic profile of 1,4-dioxane with the toxicology data from a chronic drinking water
24 study (Kociba et al., 1975). This analysis indicated that liver toxicity and eventual tumor
25 formation occurred only at doses where clearance pathways were saturated and elimination of
26 1,4-dioxane from the blood was reduced. Nannelli et al. (2005) further suggested that a
27 sustained induction of CYP2E1 may lead to generation of reactive oxygen species contributing
28 to target organ toxicity and regenerative cell proliferation; however, no data were provided to
29 support this hypothesis.
4.5.2.3. Mechanisms of Tumor Induction
30 Several studies have been performed to evaluate potential mechanisms for the
31 carcinogenicity of 1,4-dioxane (Goldsworthy et al., 1991; Kitchin and Brown, 1990; Stott et al.,
32 1981). Stott et al. (1981) evaluated 1,4-dioxane in several test systems, including salmonella
May 2009 69 DRAFT - DO NOT CITE OR QUOTE
-------
1 mutagenicity in vitro, rat hepatocyte DNA repair activity in vitro, DNA synthesis determination
2 in male Sprague Dawley rats following acute gavage dosing or an 11-week drinking water
3 exposure (described in Section 4.2.1), and hepatocyte DNA alkylation and DNA repair following
4 a single gavage dose. This study used doses of 0, 10, 100, or 1,000 mg/kg-day, with the highest
5 dose considered to be a tumorigenic dose level. Liver histopathology and liver to BW ratios
6 were also evaluated in rats from acute gavage or repeated dose drinking water experiments.
7 The histopathology evaluation indicated that liver cytotoxicity (i.e., centrilobular
8 hepatocyte swelling) was present in rats from the 1,000 mg/kg-day dose group that received
9 1,4-dioxane in the drinking water for 11 weeks (Stott et al., 1981). An increase in the liver to
10 BW ratio accompanied by an increase in hepatic DNA synthesis was also seen in this group of
11 animals. No effect on histopathology, liver weight, or DNA synthesis was observed in acutely
12 exposed rats or rats that were exposed to a lower dose of 10 mg/kg-day for 11 weeks.
13 1,4-Dioxane produced negative findings in the remaining genotoxicity assays conducted as part
14 of this study (i.e., Salmonella mutagenicity, in vitro and in vivo rat hepatocyte DNA repair, and
15 DNA alkylation in rat liver). The study authors suggested that the observed lack of genotoxicity
16 at tumorigenic and cytotoxic dose levels indicates an epigenetic mechanism for 1,4-dioxane
17 hepatocellular carcinoma in rats.
18 Goldsworthy et al. (1991) evaluated potential mechanisms for the nasal and liver
19 carcinogenicity of 1,4-dioxane in the rat. DNA repair activity was evaluated as a measure of
20 DNA reactivity and DNA synthesis was measured as an indicator of cell proliferation or
21 promotional activity. In vitro DNA repair was evaluated in primary hepatocyte cultures from
22 control and 1,4-dioxane-treated rats (1 or 2% in the drinking water for 1 week). DNA repair and
23 DNA synthesis were also measured in vivo following a single gavage dose of 1,000 mg/kg, a
24 drinking water exposure of 1% (1,500 mg/kg-day) for 1 week, or a drinking water exposure of
25 2% (3,000 mg/kg-day) for 2 weeks. Liver to BW ratios and palmitoyl CoA oxidase activity were
26 measured in the rat liver to determine whether peroxisome proliferation played a role in the liver
27 carcinogenesis of 1,4-dioxane. In vivo DNA repair was evaluated in rat nasal epithelial cells
28 derived from either the nasoturbinate or the maxilloturbinate of 1,4-dioxane-treated rats. These
29 rats received 1% 1,4-dioxane (1,500 mg/kg-day) in the drinking water for 8 days, followed by a
30 single gavage dose of 10, 100, or 1,000 mg/kg 12 hours prior to sacrifice. Archived tissues from
31 the NCI (1978) bioassay were reexamined to determine the primary sites for tumor formation in
32 the nasal cavity following chronic exposure in rats. Histopathology and cell proliferation were
33 determined for specific sites in the nasal cavity that were related to tumor formation. This
34 evaluation was performed in rats that were exposed to drinking water containing 1% 1,4-dioxane
35 (1,500 mg/kg-day) for 2 weeks.
36 1,4-Dioxane and its metabolite l,4-dioxane-2-one did not affect in vitro DNA repair in
37 primary hepatocyte cultures (Goldsworthy et al., 1991). In vivo DNA repair was also unaffected
38 by acute gavage exposure or ingestion of 1,4-dioxane in the drinking water for a 1- or 2-week
May 2009 70 DRAFT - DO NOT CITE OR QUOTE
-------
1 period. Hepatocyte cell proliferation was not affected by acute gavage exposure, but was
2 increased approximately twofold following a 1-2-week drinking water exposure. A 5-day
3 drinking water exposure to 1% 1,4-dioxane (1,500 mg/kg-day) did not increase the activity of
4 palmitoyl coenzyme A or the liver to BW ratio, suggesting that peroxisome proliferation did not
5 play a role in the hepatocarcinogenesis of 1,4-dioxane. Nannelli et al. (2005) also reported a lack
6 of hepatic palmitoyl CoA induction following 10 days of exposure to 1.5% 1,4-dioxane in the
7 drinking water (2,100 mg/kg-day).
8 Treatment of rats with 1% (1,500 mg/kg-day) 1,4-dioxane for 8 days did not alter DNA
9 repair in nasal epithelial cells (Goldsworthy et al., 1991). The addition of a single gavage dose
10 of up to 1,000 mg/kg 12 hours prior to sacrifice also did not induce DNA repair. Reexamination
11 of tissue sections from the NCI (1978) bioassay suggested that the majority of nasal tumors were
12 located in the dorsal nasal septum or the nasoturbinate of the anterior portion of the dorsal
13 meatus (Goldsworthy et al., 1991). No histopathological lesions were observed in nasal section
14 of rats exposed to drinking water containing 1% 1,4-dioxane (1,500 mg/kg-day) for 2 weeks and
15 no increase was observed in cell proliferation at the sites of highest tumor formation in the nasal
16 cavity.
17 Female Sprague Dawley rats (three to nine per group) were given 0, 168, 840, 2,550, or
18 4,200 mg/kg 1,4-dioxane (99% purity) by corn oil gavage in two doses at 21 and 4 hours prior to
19 sacrifice (Kitchin and Brown, 1990). DNA damage (single-strand breaks measured by alkaline
20 elution), ODC activity, reduced glutathione content, and CYP450 content were measured in the
21 liver. Serum ALT activity and liver histopathology were also evaluated. No changes were
22 observed in hepatic reduced glutathione content or ALT activity. Light microscopy revealed
23 minimal to mild vacuolar degeneration in the cytoplasm of hepatocytes from three of five rats
24 from the 2,550 mg/kg dose group. No histopathological lesions were seen in any other dose
25 group, including rats given a higher dose of 4,200 mg/kg. 1,4-Dioxane caused 43 and 50%
26 increases in DNA single-strand breaks at dose levels of 2,550 and 4,200 mg/kg, respectively.
27 CYP450 content was also increased at the two highest dose levels (25 and 66% respectively).
28 ODC activity was increased approximately two-, five-, and eightfold above control values at
29 doses of 840, 2,550, and 4,200 mg/kg, respectively. The results of this study demonstrated that
30 hepatic DNA damage can occur in the absence of significant cytotoxicity. Parameters associated
31 with tumor promotion (i.e., ODC activity, CYP450 content) were also elevated, suggesting that
32 promotion may play a role in the carcinogenesis of 1,4-dioxane.
4.6. SYNTHESIS OF MAJOR NONCANCER EFFECTS
33 Liver and kidney toxicity were the primary noncancer health effects associated with
34 exposure to 1,4-dioxane in humans and laboratory animals. Several fatal cases of hemorrhagic
35 nephritis and centrilobular necrosis of the liver were related to occupational exposure (i.e.,
36 inhalation and dermal contact) to 1,4-dioxane (Johnstone, 1959; Barber, 1934). Neurological
May 2009 71 DRAFT - DO NOT CITE OR QUOTE
-------
1 changes were also reported in one case; including, headache, elevation in blood pressure,
2 agitation and restlessness, and coma (Johnstone, 1959). Perivascular widening was observed in
3 the brain of this worker, with small foci of demyelination in several regions (e.g., cortex, basal
4 nuclei). Liver and kidney degeneration and necrosis were observed in acute oral and inhalation
5 studies (JBRC, 1998b; Drew et al., 1978; David, 1964; Kesten et al., 1939; Laug et al., 1939;
6 Schrenk and Yant, 1936; deNavasquez, 1935; Fairley et al., 1934). The results of subchronic
7 and chronic studies are discussed below.
4.6.1. Oral
8 Table 4-17 presents a summary of the noncancer results for the subchronic and chronic
9 oral studies of 1,4-dioxane toxicity in experimental animals. Liver and kidney toxicity were the
10 primary noncancer health effects of oral exposure to 1,4-dioxane in animals. Kidney damage at
11 high doses was characterized by degeneration of the cortical tubule cells, necrosis with
12 hemorrhage, and glomerulonephritis (NCI, 1978; Kociba et al., 1974; Argus et al., 1973, 1965;
13 Fairley et al., 1934). Renal cell degeneration generally began with cloudy swelling of cells in the
14 cortex (Fairley et al., 1934). Nuclear enlargement of proximal tubule cells was observed at doses
15 below those producing renal necrosis (Kano et al., 2008; JBRC, 1998a), but is of uncertain
16 toxicological significance. The lowest dose reported to produce kidney damage was 94 mg/kg-
17 day, which produced renal degeneration and necrosis of tubule epithelial cells in male rats in the
18 Kociba et al. (1974) study. Cortical tubule degeneration was seen at higher doses in the NCI
19 (1978) bioassay (240 mg/kg-day, male rats), and glomerulonephritis was reported for rats given
20 doses of > 430 mg/kg-day (Argus et al., 1965, 1973).
Table 4-17. Oral toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration
NOAEL
(mg/kg-day)
LOAEL
(mg/kg-day)
Effect
Reference
Subchronic studies
Rat and mouse
(6/species);
unknown strain
Male
Sprague Dawley
Rat
(4-6/group)
F344/DuCrj rat
(10/sex/group)
Rats 0 or 1,900 mg/kg-
day; mice 0 or
3,300 mg/kg-day for
67 days
0, 10, or 1,000 mg/kg-day
for 1 1 weeks
Males 0, 52, 126, 274,
657, or 1,554 mg/kg-day;
females 0, 83, 185, 427,
756, or 1,614 mg/kg-day
for 13 weeks
NA
10
52
1,900 rats
3,300 mice
1,000
126
Renal cortical degeneration
and necrosis, hemorrhage;
hepatocellular degeneration
Minimal centrilobular
hepatocyte swelling;
increased DNA synthesis
Nuclear enlargement of
nasal respiratory
epithelium; hepatocyte
swelling
Fairley etal.,
1934
Stott et al.,
1981
Kano etal.,
2008
May 2009
72
DRAFT - DO NOT CITE OR QUOTE
-------
Species
CrjiBDFj Mouse
(10/sex/group)
Dose/duration
Males 0, 86, 231,585,
882, or 1,570 mg/kg-day;
females 0, 170, 387, 898,
1,620, or 2,669 mg/kg-
day for 13 weeks
NOAEL
(mg/kg-day)
170
LOAEL
(mg/kg-day)
387
Effect
Nuclear enlargement of
bronchial epithelium
Reference
Kano etal.,
2008
Chronic studies
Male
Wistar
Rat (26 treated,
9 controls)
Male
Sprague Dawley
rats (30/group)
Sherman rat
(60/sex/dose
group)
Osborne-Mendel
rat (35/sex/dose
level)
B6C3F! mouse
(50/sex/dose
level)
F344/DuCrj rat
(50/sex/dose
level)
F344/DuCrj rat
(50/sex/dose
level)
F344/DuCrj rat
(50/sex/dose
level)
Crj:BDFi mouse
(50/sex/dose
level)
Crj:BDFi mouse
(50/sex/dose
level)
0 or 640 mg/kg-day for
63 weeks
0, 430, 574, 803, or
1,032 mg/kg-day for
13 months
Males 0, 9.6, 94, or
1,015 mg/kg-day; females
0, 19, 148, or
1,599 mg/kg-day for
2 years
Males 0, 240, or
530 mg/kg-day; females
0, 350, or 640 mg/kg-day
for 110 weeks
Males 0, 720, or
830 mg/kg-day; females
0,380, or 860 mg/kg-day
for 90 weeks
Males 0, 16, 81, or
398 mg/kg-day; females
0,21, 103, or 5 14 mg/kg-
day for 2 years
Males 0, 16, 81, or
398 mg/kg-day; females
0,21, 103, or 5 14 mg/kg-
day for 2 years
Males 0, 16, 81, or
398 mg/kg-day; females
0,21, 103, or 5 14 mg/kg-
day for 2 years
Males 0, 66, 251 or
768 mg/kg-day; females
0, 77, 323, or
1,066 mg/kg-day for
2 years
Males 0, 66, 251 or
768 mg/kg-day; females
0, 77, 323, or
1,066 mg/kg-day for
2 years
NA
NA
9.6
NA
NA
81
16
81
77
66
640
430
94
240
380
398
81
398
323
251
Hepatocytes with enlarged
hyperchromic nuclei;
glomerulonephritis
Hepatocy tomegaly ;
glomerulonephritis
Degeneration and necrosis
of renal tubular cells and
hepatocytes
Pneumonia, gastric ulcers,
and cortical tubular
degeneration in the kidney
Pneumonia and rhinitis
Atrophy of nasal olfactory
epithelium; nasal adhesion
and inflammation
Liver hyperplasia
Increases in serum liver
enzymes (GOT, GPT, LDH,
and ALP)
Nasal inflammation
Increases in serum liver
enzymes (GOT, GPT, LDH,
and ALP)
Argus et al.,
1965
Argus et al.,
1973
Kocibaetal.,
1974
NCI, 1978
NCI, 1978
JBRC, 1998a
JBRC, 1998a
JBRC, 1998a
JBRC, 1998a
JBRC, 1998a
Developmental studies
Sprague Dawley
rat
(18-20/group)
Pregnant dams 0, 250,
500, or 1,000 mg/kg-day
on gestation days 6-15
500
1,000
Delayed ossification of the
sternebrae and reduced fetal
BWs
Giavini et al.,
1985
May 2009
73
DRAFT - DO NOT CITE OR QUOTE
-------
1 Liver effects included degeneration and necrosis, hepatocyte swelling, cells with
2 hyperchromic nuclei, spongiosis hepatis, hyperplasia, and clear and mixed cell foci of the liver
3 (Kano et al., 2008; NCI, 1978; Kociba et al., 1974; Argus et al., 1965, 1973; Fairley et al., 1934).
4 Hepatocellular degeneration and necrosis were seen at high doses in a subchronic study
5 (1,900 mg/kg-day in rats) (Fairley et al., 1934) and at lower doses in a chronic study
6 (94 mg/kg-day, male rats) (Kociba et al., 1974). Argus et al. (1973) described a progression of
7 preneoplastic effects in the liver of rats exposed to a dose of 575 mg/kg-day. Early changes
8 (8 months exposure) were described as an increased nuclear size of hepatocytes, disorganization
9 of the rough endoplasmic reticulum, an increase in smooth endoplasmic reticulum, a decrease in
10 glycogen, an increase in lipid droplets in hepatocytes, and formation of liver nodules.
11 Spongiosis hepatis, hyperplasia, and clear and mixed-cell foci were also observed in the liver of
12 rats (doses > 81 mg/kg-day in male rats) (JBRC, 1998a). Clear and mixed-cell foci are
13 commonly considered preneoplastic changes and would not be considered evidence of noncancer
14 toxicity when observed in conjunction with tumor formation. If exposure to 1,4-dioxane had not
15 resulted in tumor formation, these lesions could represent potential noncancer toxicity. The
16 nature of spongiosis hepatis as a preneoplastic change is less well understood (Bannash, 2003;
17 Karbe and Kerlin, 2002; Stroebel et al., 1995). Spongiosis hepatis is a cyst-like lesion that arises
18 from the perisinusoidal Ito cells of the liver. This change is sometimes associated with
19 hepatocellular hypertrophy and liver toxicity (Karbe and Kerlin, 2002), but may also occur in
20 combination with preneoplastic foci, or hepatocellular adenoma or carcinoma (Bannash et al.,
21 2003; Stroebel et al., 1995). In the case of the JBRC (1998a) study, spongiosis hepatis was
22 associated with other preneoplastic changes in the liver (hyperplasia, clear and mixed-cell foci).
23 No other lesions indicative of liver toxicity were seen in this study; therefore, spongiosis hepatis
24 was not considered indicative of noncancer effects. The activity of serum enzymes (i.e., AST,
25 ALT, LDH, and ALP) was increased in rats and mice exposed to 1,4-dioxane, although only in
26 groups with high incidence of liver tumors. Blood samples were collected only at the end of the
27 2-year study, so altered serum chemistry may be associated with the tumorigenic changes in the
28 liver.
29 Hematological changes were reported in the JBRC (1998a) study only. Observed
30 increases in RBCs, hematocrit, hemoglobin in high-dose male mice (768 mg/kg-day) may be
31 related to lower drinking water consumption (74% of control drinking water intake).
32 Hematological effects noted in male rats given 81 mg/kg-day (decreased RBCs, hemoglobin,
33 hematocrit, increased platelets) were within 20% of control values. A reference range database
34 for hematological effects in laboratory animals (Wolford et al., 1986) indicates that a 20%
35 change in these parameters may fall within a normal range (10th-90th percentile values) and
36 may not represent a treatment-related effect of concern.
May 2009 74 DRAFT - DO NOT CITE OR QUOTE
-------
1 Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice
2 only, dose >380 mg/kg-day) and JBRC (1998a) studies (>98 mg/kg-day in rats, >323 mg/kg-day
3 in mice). The JBRC (1998a) study also demonstrates atrophy of the nasal epithelium and
4 adhesion in rats and mice. Nasal inflammation may be a response to direct contact of the nasal
5 mucosa with drinking water containing 1,4-dioxane (Sweeney et al., 2008; Goldsworthy et al.,
6 1991) or could result from systemic exposure. Regardless, inflammation may indicate toxicity
7 due to 1,4-dioxane exposure. A significant increase in the incidence of pneumonia was reported
8 in mice from the NCI (1978) study. The significance of this effect is unclear, as it was not
9 observed in other studies that evaluated lung histopathology (Kano et al., 2008; JBRC, 1998a;
10 Kociba et al., 1974). No studies were available regarding the potential for 1,4-dioxane to cause
11 immunological effects. Metaplasia and hyperplasia of the nasal epithelium were also observed in
12 high-dose male and female rats (JBRC, 1998a); however, these effects are likely to be associated
13 with the formation of nasal cavity tumors in these dose groups. Nuclear enlargement of the nasal
14 olfactory epithelium was observed at a dose of 103 mg/kg-day in female rats (JBRC, 1998a);
15 however, it is unclear whether this alteration represents an adverse toxicological effect. Nuclear
16 enlargement of the tracheal and bronchial epithelium and an accumulation of foamy cells in the
17 lung were also seen in male and female mice give 1,4-dioxane at doses of >323 mg/kg for
18 2 years.
4.6.2. Inhalation
19 Only one subchronic study (Fairley et al., 1934) and one chronic inhalation study
20 (Torkelson et al., 1974) were identified. In the subchronic study, rabbits, guinea pigs, rats, and
21 mice (3-6/species/group) were exposed to 1,000, 2,000, 5,000, or 10,000 ppm of 1,4-dioxane
22 vapor for 16.5 hours/week. Animals were exposed until death occurred or were sacrificed at
23 varying time periods. Severe liver and kidney damage and acute vascular congestion of the
24 lungs were observed. Kidney damage was described as patchy degeneration of cortical tubules
25 with vascular congestion and hemorrhage. Liver lesions varied from cloudy hepatocyte swelling
26 to large areas of necrosis. Torkelson et al. (1974) performed a chronic inhalation study in which
27 male and female Wistar rats (288/sex) were exposed to 111 ppm 1,4-dioxane vapor for
28 7 hours/day, 5 days/week for 2 years. Control rats (192/sex) were exposed to filtered air. No
29 significant effects were observed on BWs, survival, organ weights, hematology, clinical
30 chemistry, or histopathology. These studies were not sufficient to characterize the inhalation
31 risks of 1,4-dioxane, due to the nature of the available data (i.e., free-standing LOAEL and
32 NOAEL values).
4.6.3. Mode of Action Information
33 The metabolism of 1,4-dioxane in humans was extensive at low doses (<50 ppm). The
34 linear elimination of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism
May 2009 75 DRAFT - DO NOT CITE OR QUOTE
-------
1 was a nonsaturated, first-order process at this exposure level (Young et al., 1977, 1976). Like
2 humans, rats extensively metabolized inhaled 1,4-dioxane; however, plasma data from rats given
3 single i.v. doses of 3, 10, 30, 100, or 1,000 mg [14C]-l,4-dioxane/kg demonstrated a dose-related
4 shift from linear, first-order to nonlinear, saturable metabolism of 1,4-dioxane (Young et al.,
5 1978a, b).
6 1,4-Dioxane oxidation appeared to be CYP450-mediated, as CYP450 induction with
7 phenobarbital or Aroclor 1254 and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine
8 or cobaltous chloride was effective in significantly increasing and decreasing, respectively, the
9 appearance of HEAA in the urine of rats (Woo et al., 1978, 1977c). 1,4-Dioxane itself induced
10 CYP450-mediated metabolism of several barbiturates in Hindustan mice given i.p. injections of
11 25 and 50 mg/kg of 1,4-dioxane (Mungikar and Pawar, 1978). The differences between single
12 and multiple doses in urinary and expired radiolabel support the notion that 1,4-dioxane may
13 induce its own metabolism. 1,4-Dioxane has been shown to induce several isoforms of CYP450
14 in various tissues following acute oral administration by gavage or drinking water (Nannelli
15 et al., 2005). In the liver, the activity of several CYP450 isozymes was increased (i.e.,
16 CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was inducible in the kidney and nasal
17 mucosa. CYP2E1 mRNA was increased approximately two- to threefold in the kidney and nasal
18 mucosa, but mRNA levels were not increased in the liver, suggesting that regulation of CYP2E1
19 was organ-specific.
20 Nannelli et al. (2005) investigated the role of CYP450 isozymes in the liver toxicity of
21 1,4-dioxane. Hepatic CYPB1/2 and CYP2E1 levels were induced by phenobarbital or fasting
22 and liver toxicity was measured as hepatic glutathione content or serum ALT activity. No
23 increase in glutathione content or ALT activity was observed, suggesting that highly reactive and
24 toxic intermediates did not play a large role in the liver toxicity of 1,4-dioxane, even under
25 conditions where metabolism was enhanced. Pretreatment with inducers of mixed-function
26 oxidases also did not significantly change the extent of covalent binding in subcellular fractions
27 (Woo et al., 1977a). Covalent binding was measured in liver, kidney, spleen, lung, colon, and
28 skeletal muscle 1-12 hours after i.p. dosing with 1,4-dioxane. Covalent binding was highest in
29 liver, spleen, and colon. Within hepatocytes, 1,4-dioxane distribution was greatest in the
30 cytosolic fraction, followed by the microsomal, mitochondrial, and nuclear fractions.
31 The absence of an increase in toxicity following an increase in metabolism suggests that
32 accumulation of the parent compound may be related to 1,4-dioxane toxicity. This hypothesis is
33 supported by a comparison of the pharmacokinetic profile of 1,4-dioxane with the toxicology
34 data from a chronic drinking water study (Kociba et al., 1975). This analysis indicated that liver
35 toxicity did not occur unless clearance pathways were saturated and elimination of 1,4-dioxane
36 from the blood was reduced. Alternative metabolic pathways (i.e., not CYP450 mediated) may
37 be present at high doses of 1,4-dioxane; however, the available studies have not characterized
May 2009 76 DRAFT - DO NOT CITE OR QUOTE
-------
1 these pathways or identified any possible reactive intermediates. The mechanism by which
2 1,4-dioxane induces tissue damage is not known.
4.7. EVALUATION OF CARCINOGENICITY
4.7.1. Summary of Overall Weight of Evidence
3 Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), 1,4-dioxane
4 can be described as likely to be carcinogenic to humans, based on adequate evidence of liver
5 carcinogenicity in several 2-year bioassays conducted in three strains of rats, two strains of mice,
6 and in guinea pigs (JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Argus et al., 1973;
7 Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al., 1965). Additionally,
8 mesothiolomas of the peritoneam (JBRC, 1998a), mammary (JBRC, 1998a), and nasal tumors
9 (JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Argus et al., 1973; Hoch-Ligeti et al., 1970) have
10 been observed in rats due to exposure to 1,4-dioxane. Studies in humans are inconclusive
11 regarding evidence for a causal link between occupational exposure to 1,4-dioxane and increased
12 risk for cancer; however, only two studies were available and these were limited by small cohort
13 size and a small number of reported cancer cases (Buffler et al., 1978; Thiess et al., 1976).
14 The available evidence is inadequate to establish a mode of action (MOA) by which
15 1,4-dioxane induces liver tumors in rats and mice. A MOA hypothesis involving sustained
16 proliferation of spontaneously transformed liver cells has some support from data indicating that
17 1,4-dioxane acts as a tumor promoter in mouse skin and rat liver bioassays (Lundberg
18 et al.,1987; King et al., 1973). Dose-response and temporal data support the occurrence of cell
19 proliferation and hyperplasia prior to the development of liver tumors (JBRC, 1998a; Kociba
20 et al., 1974) in the rat model. However, the dose-response relationship for induction of hepatic
21 cell proliferation has not been characterized, and it is unknown if it would reflect the dose-
22 response relationship for liver tumors in the 2-year rat and mouse studies. Conflicting data from
23 rat and mouse bioassays (JBRC, 1998a; Kociba et al., 1974) suggest that cytotoxicity may not be
24 a required precursor event for 1,4-dioxane-induced cell proliferation. Data regarding a plausible
25 dose response and temporal progression (see Table 4-18) from cytotoxicity and cell proliferation
26 to eventual liver tumor formation are not available.
27 The MOA by which 1,4-dioxane produces liver, nasal, peritoneal (mesothiolomas), and
28 mammary gland tumors is unknown, and the available data do not support any hypothesized
29 carcinogenic MOA for 1,4-dioxane.
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence
30 Human studies of occupational exposure to 1,4-dioxane were inconclusive; in each case,
31 the cohort size and number of reported cases were of limited size (Buffler et al., 1978; Thiess
32 etal., 1976).
May 2009 77 DRAFT - DO NOT CITE OR QUOTE
-------
1 Several carcinogen!city bioassays have been conducted for 1,4-dioxane in mice, rats, and
2 guinea pigs (JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Torkelson et al., 1974; Argus et al.,
3 1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al., 1965). Liver tumors
4 have been observed following drinking water exposure in male Wistar rats (Argus et al., 1965),
5 male guinea pigs (Hoch-Ligeti and Argus, 1970), male Sprague Dawley rats (Argus et al., 1973;
6 Hoch-Ligeti et al., 1970), male and female Sherman rats (Kociba et al., 1974), female Osborne-
7 Mendel rats (NCI, 1978), male and female F344/DuCrj rats (JBRC, 1998a), male and female
8 B6C3Fi mice (NCI, 1978), and male and female Crj :BDFi mice (JBRC, 1998a). In the earliest
9 cancer bioassays, the liver tumors were described as hepatomas (Argus et al., 1973; Hoch-Ligeti
10 and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al., 1965); however, later studies made a
11 distinction between hepatocellular carcinoma and hepatocellular adenoma (JBRC, 1998a; NCI,
12 1978; Kociba et al., 1974). Both tumor types have been seen in rats and mice exposed to
13 1,4-dioxane. Kociba et al. (1974) noted evidence of liver toxicity at or below the dose levels that
14 produced liver tumors but did not report incidence data for these effects. Hepatocellular
15 degeneration and necrosis were observed in the mid- and high-dose groups of male and female
16 Sherman rats exposed to 1,4-dioxane, while tumors were only observed at the highest dose.
17 Hepatic regeneration was indicated in the mid- and high-dose groups by the formation of
18 hepatocellular hyperplastic nodules. Findings from JBRC (1998a) also provided evidence of
19 liver hyperplasia in male F344/DuCrj rats at a dose level below the dose that induced a
20 statistically significant increase in tumor formation.
21 Nasal cavity tumors were also observed in Sprague Dawley rats (Argus et al., 1973;
22 Hoch-Ligeti et al., 1970), Osborne-Mendel rats (NCI, 1978), Sherman rats (Kociba et al., 1974),
23 and F344/DuCrj rats (JBRC, 1998a). Most tumors were characterized as squamous cell
24 carcinomas. Nasal tumors were not elevated in B6C3Fi or Crj :BDFi mice. JBRC (1998a) was
25 the only study that evaluated nonneoplastic changes in nasal cavity tissue following prolonged
26 exposure to 1,4-dioxane in the drinking water. Histopathological lesions in female F344/DuCrj
27 rats were suggestive of toxicity and regeneration in this tissue (i.e., atrophy, adhesion,
28 inflammation, nuclear enlargement, and hyperplasia and metaplasia of respiratory and olfactory
29 epithelium). Some of these effects occurred at a lower dose (103 mg/kg-day) than that shown to
30 produce nasal cavity tumors (513 mg/kg-day). Reexamination of tissue sections from the NCI
31 (1978) bioassay suggested that the majority of nasal tumors were located in the dorsal nasal
32 septum or the nasoturbinate of the anterior portion of the dorsal meatus. Nasal tumors were not
33 observed in an inhalation study in Wistar rats exposed to 111 ppm for 5 days/week for 2 years
34 (Torkelson et al., 1974). It is unlikely that 1,4-dioxane in expired air following a drinking water
35 exposure could exceed this air concentration.
36 Tumor initiation and promotion studies in mouse skin and rat liver suggested that
37 1,4-dioxane does not initiate the carcinogenic process, but instead acts as a tumor promoter
38 (Lundberg et al., 1987; Bull et al., 1986; King et al., 1973) (see Section 4.2.3).
May 2009 78 DRAFT - DO NOT CITE OR QUOTE
-------
1 In addition to the liver and nasal tumors observed in several studies, a statistically
2 significant increase in mesotheliomas of the peritoneum was seen in male rats from the JBRC
3 (1998a) study. Female rats also showed a statistically significant increase in mammary gland
4 adenomas. A significant increase in the incidence of these tumors was not observed in other
5 chronic oral bioassays of 1,4-dioxane (NCI, 1978; Kociba et al., 1974).
4.7.3. Mode of Action Information
6 The MO A by which 1,4-dioxane produces liver, nasal, peritoneal (mesothiolomas), and
7 mammary gland tumors is unknown, and the available data do not support any hypothesized
8 mode of carcinogenic action for 1,4-dioxane. The hypothesized MO As for 1,4-dioxane
9 carcinogenicity are discussed below within the context of the modified Hill criteria of causality
10 as recommended in the most recent Agency guidelines (U.S. EPA, 2005a). The hypothesized
11 MOA(s) presented in the following sections were not explored for peritoneal or mammary gland
12 tumors due to the absence of any chemical specific information for these tumor types.
4.7.3.1. Identification of Key Events for Carcinogenicity
13 4.7.3.1.1. Liver. A key event in this MOA hypothesis is sustained proliferation of
14 spontaneously transformed liver cells, resulting in the eventual formation of liver tumors.
15 Precursor events in which 1,4-dioxane may promote proliferation of transformed liver cells are
16 uncertain. One study suggests that induced liver cytotoxicity may be a key precursor event to
17 cell proliferation leading to the formation of liver tumors (Kociba et al., 1974), however, they did
18 not report incidence data for these effects. Other studies suggest that cell proliferation can occur
19 in the absence of liver cytotoxicity (JBRC, 1998a). Figure 4-1 presents a schematic
20 representation of possible key events in the MOA for 1,4-dioxane liver carcinogenicity. These
21 include: (1) oxidation by CYP2E1 and CYP2B1/2 (i.e., detoxification pathway for 1,4-dioxane),
22 (2) saturation of metabolism/clearance leading to accumulation of the parent 1,4-dioxane,
23 (3) liver damage followed by regenerative cell proliferation, or (4) cell proliferation in the
24 absence of cytotoxicity (i.e., mitogenesis), (5) hyperplasia, and (6) tumor formation. It is
25 suggested that liver toxicity is related to the accumulation of the parent compound following
26 metabolic saturation at high doses (Kociba et al., 1975); however, no in vivo or in vitro assays
27 have examined the toxicity of metabolites resulting from 1,4-dioxane to support this hypothesis.
28 Nanelli et al. (2005) demonstrated that an increase in the oxidative metabolism of 1,4-dioxane
29 via CYP450 induction using phenobarbital or fasting does not result in an increase in liver
30 toxicity. This result suggested that highly reactive and toxic intermediates did not play a large
31 role in the liver toxicity of 1,4-dioxane, even under conditions where metabolism was enhanced.
32 Alternative metabolic pathways (e.g., not CYP450 mediated) may be present at high doses of
33 1,4-dioxane; although the available studies have not characterized these pathways nor identified
34 any possible reactive intermediates. Tumor promotion studies in mouse skin and rat liver
May 2009 79 DRAFT - DO NOT CITE OR QUOTE
-------
1 suggest that 1,4-dioxane may enhance the growth of previously initiated cells (Lundberg
2 et al.,1987; King et al., 1973). This is consistent with the increase in hepatocyte cell
3 proliferation observed in several studies (Miyagawa et al., 1999; Uno et al., 1994; Goldsworthy
4 et al., 1991; Stott et al., 1981). These mechanistic studies provide evidence of cell proliferation,
5 but do not indicate whether mitogenesis or cytotoxicity is responsible for increased cell turnover.
MOA for Liver Tumors
Oral absorption of
1,4-dioxane
Metabolism by
CYP2E1 and
CYP2B1/2
Metabolic
saturation and
accumulation of
1,4-dioxane in the
blood
HEAA elimination
in the urine
Hepatocellular
cytotoxicity
Cell proliferation in
absence of
cytotoxicity
Regenerative cell
proliferation
Hyperplasia
Hyperplasia
Tumor promotion
Tumor formation
Figure 4-1. A schematic representation of the possible key events in the delivery of
1,4-dioxane to the liver and the hypothesized MOA(s) for liver carcinogenicity.
6 4.7.3.1.2. Nasal cavity. A possible key event in the MOA hypothesis for nasal tumors is
7 sustained proliferation of spontaneously transformed nasal epithelial cells, resulting in the
8 eventual formation of nasal cavity tumors. Precursor events in which 1,4-dioxane may promote
9 proliferation of transformed nasal cells are highly uncertain. Figure 4-2 presents a schematic
10 representation of possible key events leading to the formation of nasal cavity tumors.
11 Histopathological lesions in female rats were suggestive of toxicity and regeneration in this
12 tissue (i.e., atrophy, adhesion, inflammation, nuclear enlargement, and hyperplasia and
13 metaplasia of respiratory and olfactory epithelium) (JBRC, 1998a).
May 2009
80
DRAFT - DO NOT CITE OR QUOTE
-------
HEAA
elimination in
the urine
Metabolic
saturation and
accumulation of
1,4-dioxane in the
blood, exhalation
of 1,4-dioxane in
breath
MOA for Nasal Cavity
Tumors
Chronic
irritation due to
direct contact
with nasal
epithelium
Cytotoxicity to
nasal cell
epithelium
Regenerative
cell
proliferation
Hyperplasia
Tumor formation
Figure 4-2. A schematic representation of the possible key events in the delivery of
1,4-dioxane to the nasal cavity and the hypothesized MOA(s) for nasal cavity
carcinogenicity.
4.7.3.2. Strength, Consistency, Specificity of Association
1 4.7.3.2.1. Liver. The plausibility of a MOA that would include liver cytotoxicity, with
2 subsequent reparative cell proliferation, as precursor events to liver tumor formation is
3 minimally supported by findings that nonneoplastic liver lesions occurred at exposure levels
4 lower than those resulting in significantly increased incidences of hepatocellular tumors (Kociba
5 et al., 1974) and the demonstration of nonneoplastic liver lesions in subchronic (Kano et al.,
6 2008) and acute and short-term oral studies (see Table 4-15). Because the incidence of
7 nonneoplastic lesions was not reported by Kociba et al. (1974), it is difficult to know whether the
8 incidence of liver lesions increased with increasing 1,4-dioxane concentration. Contradicting the
9 observations by Kociba et al. (1974), liver tumors were observed in female rats and female mice
10 in the absence of lesions indicative of cytotoxicity (Kano et al., 2008; JBRC, 1998a; NCI, 1978).
11 This suggests that cytotoxicity may not be a requisite step in the MOA for liver cancer.
12 Mechanistic and tumor promotion studies suggest that enhanced cell proliferation without
13 cytotoxicity may be a key event; however, data showing a plausible dose response and temporal
14 progression from cell proliferation to eventual liver tumor formation are not available (see
15 Sections 4.7.3.3 and 4.7.3.4). Mechanistic studies that demonstrated cell proliferation after
16 short-term exposure did not evaluate liver cytotoxicity (Miyagawa et al., 1999; Uno et al., 1994;
17 Goldsworthy et al., 1991). Studies have not investigated possible precursor events that may lead
18 to cell proliferation in the absence of cytotoxicity (i.e., genetic regulation of mitogenesis).
May 2009
81
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.2.2. Nasal cavity. Nasal cavity tumors have been demonstrated in several rat strains
2 (JBRC, 1998a; NCI, 1978; Kociba et al., 1974), but were not elevated in two strains of mice
3 (JBRC, 1998a; NCI, 1978). Chronic irritation was indicated by the observation of rhinitis and
4 inflammation of the nasal cavity in rats from the JBRC (1998a) study. This study also showed
5 atrophy of the nasal epithelium and adhesion in rats. Regeneration of the nasal epithelium is
6 demonstrated by metaplasia and hyperplasia observed in rats exposed to 1,4-dioxane (JBRC,
7 1998a).
4.7.3.3. Dose-Response Relationship
8 4.7.3.3.1. Liver. Table 4-18 presents the temporal sequence and dose-response relationship for
9 possible key events in the liver carcinogenesis of 1,4-dioxane. Dose-response information
10 provides some support for enhanced cell proliferation as a key event in the liver tumorigenesis of
11 1,4-dioxane; however, the role of cytotoxicity as a required precursor event is not supported by
12 data from more than one study. Kociba et al. (1974) demonstrated that liver toxicity and
13 hepatocellular regeneration occurred at a lower dose level than tumor formation. Hepatocellular
14 degeneration and necrosis were observed in the mid- and high-dose groups of Sherman rats
15 exposed to 1,4-dioxane, although it is not possible to discern whether this effect was observed in
16 both genders due to the lack of incidence data (Kociba et al., 1974). Hepatic tumors were only
17 observed at the highest dose (Kociba et al., 1974). Hepatic regeneration was indicated in the
18 mid- and high-dose group by the formation of hepatocellular hyperplastic nodules. Liver
19 hyperplasia was also seen in rats from the JBRC (1998a) study, at or below the dose level that
20 resulted in tumor formation; however, hepatocellular degeneration and necrosis were not
21 observed. These results suggest that hepatic cell proliferation and hyperplasia may occur in the
22 absence of significant cytotoxicity. Liver angiectasis (i.e., dilation of blood or lymphatic
23 vessels) was observed in male mice at the same dose that produced liver tumors; however, the
24 relationship between this vascular abnormality and tumor formation is unclear.
Table 4-18. Temporal sequence and dose-response relationship for possible
key events and liver tumors in rats and mice
Dose (mg/kg-day)
Key event (time — >)
Metabolism
1,4-dioxane
Liver damage
Cell proliferation
Hyperplasia
Adenomas
and/or
carcinomas
Kociba et al., 1974 — Sherman rats (male and female combined)
0
14
121
1,307
a
[b
[b
[b
a
r
r
a
a
a
a
r
r
a
a
a
r
May 2009
82
DRAFT - DO NOT CITE OR QUOTE
-------
Dose (mg/kg-day)
Key event (time — >)
Metabolism
1,4-dioxane
Liver damage
Cell proliferation
Hyperplasia
Adenomas
and/or
carcinomas
NCI, 1978— female Osborne-Mendel rats
0
350
640
a
[b
[b
a
a
a
a
a
a
a
a
a
a
r
r
NCI, 1978— male B6C3FJ mice
0
720
830
a
[b
[b
a
a
a
a
a
a
a
a
a
a
r
r
NCI, 1978— female B6C3FJ mice
0
380
860
a
[b
[b
a
a
a
a
a
a
a
a
a
a
r
r
JBRC, 1998a— male F344/DuCrj rats
0
16
81
398
a
[b
[b
[b
a
a
a
rd
a
a
a
a
a
a
r
r
a
a
a
r
JBRC, 1998a— female F344/DuCrj rats
0
21
103
514
a
[b
[b
[b
a
a
a
a
a
a
a
a
a
a
a
r
a
a
a
r
JBRC, 1998a— male Crj:BDFi mice
0
66
251
768
a
[b
[b
[b
a
a
a
rd
a
a
a
a
a
a
a
a
a
r
r
tc
JBRC, 1998a— female CrjiBDFi mice
0
77
323
1,066
a
[b
[b
[b
a
a
a
pd
a
a
a
a
a
a
a
a
a
[c
[c
[c
a— No evidence demonstrating key event.
b[ 1,4-dioxane metabolism was not evaluated as part of the chronic bioassays. Data from pharmacokinetic studies
suggest that metabolism of 1,4-dioxane by CYP2E1 and CYP2B2 occurs immediately and continues throughout the
duration of exposure at all exposure levels.
°[ Evidence demonstrating key event.
d[ Single cell necrosis was observed in a 13 week bioassay for male rats (274 mg/kg-day), male mice (585 mg/kg-
day), and female mice (898 mg/kg-day) exposed to 1,4-dioxane in drinking water.
May 2009
83
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.3.2. Nasal cavity. Jbxicity and regeneration in nasal epithelium (i.e., atrophy, adhesion,
2 inflammation, and hyperplasia and metaplasia of respiratory and olfactory epithelium) was
3 evident in one study at the same dose levels that produced nasal cavity tumors (JBRC, 1998a).
4.7.3.4. Temporal Relationship
4 4.7.3.4.1. Liver. Available information regarding temporal relationships between the key event
5 (sustained proliferation of spontaneously transformed liver cells) and the eventual formation of
6 liver tumors is limited. A comparison of 13-week and 2-year studies conducted in F344/DuCrj
7 rats and Crj :BDFi mice at the same laboratory revealed that tumorigenic doses of 1,4-dioxane
8 produced liver toxicity by 13 weeks of exposure (Kano et al., 2008; JBRC, 1998a). Hepatocyte
9 swelling of the centrilobular area of the liver, vacuolar changes in the liver, granular changes in
10 the liver, and single cell necrosis in the liver were observed in mice and rats given 1,4-dioxane in
11 the drinking water for 13 weeks. Sustained liver damage could presumably lead to regenerative
12 hyperplasia and tumor formation following chronic exposure. As discussed above,
13 histopathological evidence of regenerative hyperplasia has been seen following long-term
14 exposure to 1,4-dioxane (JBRC, 1998a; Kociba et al., 1974). Tumors occurred earlier at high
15 doses in both mice and rats from this study (email from Dr. Kazunori Yamazaki, JBRC, to Dr.
16 Julie Stickney, SRC, dated 12/18/06); however, temporal information regarding hyperplasia or
17 other possible key events was not available (i.e., interim blood samples not collected, interim
18 sacrifices were not performed). Argus et al. (1973) studied the progression of tumorigenesis by
19 electron microscopy of liver tissues obtained following interim sacrifices at 8 and 13 months of
20 exposure (five rats/group, 574 mg/kg-day). The first change observed was an increase in the size
21 of the nuclei of the hepatocytes, mostly in the periportal area. Precancerous changes were
22 characterized by disorganization of the rough endoplasmic reticulum, increase in smooth
23 endoplasmic reticulum, and decrease in glycogen and increase in lipid droplets in hepatocytes.
24 These changes increased in severity in the hepatocellular carcinomas in rats exposed to
25 1,4-dioxane for 13 months.
26 Three types of liver nodules were observed in exposed rats at 13-16 months. The first
27 consisted of groups of these cells with reduced cytoplasmic basophilia and a slightly nodular
28 appearance as viewed by light microscopy. The second type of nodule was described consisting
29 of large cells, apparently filled and distended with fat. The third type of nodule was described as
30 finger-like strands, 2-3 cells thick, of smaller hepatocytes with large hyperchromic nuclei and
31 dense cytoplasm. This third type of nodule was designated as an incipient hepatoma, since it
32 showed all the histological characteristics of a fully developed hepatoma. All three types of
33 nodules were generally present in the same liver.
34 4.7.3.4.2. Nasal cavity. No information was available regarding the temporal relationship
35 between toxicity in the nasal epithelium and the formation of nasal cavity tumors.
May 2009 84 DRAFT - DO NOT CITE OR QUOTE
-------
4.7.3.5. Biological Plausibility and Coherence
1 4.7.3.5.1. Liver, The hypothesis that sustained proliferation of spontaneously transformed liver
2 cells is a key event within a MOA is possible based on supporting evidence indicating that
3 1,4-dioxane is a tumor promoter of mouse skin and rat liver tumors (Lundberg et al., 1987; Bull
4 et al., 1986; King et al., 1973). Further support for this hypothesis is provided by studies
5 demonstrating that 1,4-dioxane increased hepatocyte DNA synthesis, indicative of cell
6 proliferation (Miyagawa et al., 1999; Uno et al., 1994; Goldsworthy et al., 1991; Stott et al.,
7 1981). In addition, the generally negative results for 1,4-dioxane in a number of genotoxicity
8 assays indicates the carcinogenicity of 1,4-dioxane may not be mediated by a mutagenic MOA.
9 The importance of cytotoxicity as necessary precursor to sustained cell proliferation is
10 biologically plausible, but is not supported by the dose-response in the majority of studies of
11 1,4-dioxane carcinogenicity.
12 4.7.3.5.2. Nasal cavity. Sustained cell proliferation in response to cell death from toxicity may
13 be related to the formation of nasal cavity tumors; however, this MOA is also not established .
14 Nasal carcinogens are generally characterized as potent genotoxins (Ashby, 1994); however,
15 other MO As have been proposed for nasal carcinogens that induce effects through other
16 mechanisms (Kasper et al. 2007; Green et al. 2000).
17 National Toxicological Program (NTP) database identified 12 chemicals from
18 approximately 500 bioassays as nasal carcinogens and 1,4-dioxane was the only identified nasal
19 carcinogen that showed little evidence of genotoxicity (Haseman and Hailey, 1997). Nasal
20 tumors were not observed in an inhalation study in Wistar rats exposed to 111 ppm for
21 5 days/week for 2 years (Torkelson et al., 1974). It is unlikely that 1,4-dioxane in expired air
22 following a drinking water exposure could exceed this air concentration.
4.7.3.6. Other Possible Modes of Action
23 An alternate MOA could be hypothesized that 1,4-dioxane alters DNA, either directly or
24 indirectly, which causes mutations in critical genes for tumor initiation, such as oncogenes or
25 tumor suppressor genes. Following these events, tumor growth may be promoted by a number of
26 molecular processes leading to enhanced cell proliferation or inhibition of programmed cell
27 death. The results from in vitro and in vivo assays do not provide overwhelming support for the
28 hypothesis of a genotoxic MOA for 1,4-dioxane carcinogenicity. The genotoxicity data for
29 1,4-dioxane were reviewed in Section 4.5.1 and were summarized in Table 4-16. Negative
30 findings were reported for mutagenicity in Salmonella typhimurium, Escherichia coli, and
31 Photobacterium phosphoreum (Mutatox assay) (Morita and Hayashi, 1998; Hellmer and
32 Bolcsfoldi, 1992; Kwan et al., 1990; Khudoley et al., 1987; Nestmann et al., 1984; Haworth
33 et al., 1983; Stott et al., 1981). Negative results were also indicated for the induction of
34 aneuploidy in yeast (Saccharomyces cerevisiae) and the sex-linked recessive lethal test in
May 2009 85 DRAFT - DO NOT CITE OR QUOTE
-------
1 Drosophila melanogaster (Zimmerman et al., 1985). In contrast, positive results were reported in
2 assays for sister chromatid exchange (Galloway et al., 1987), DNA damage (Kitchin and Brown,
3 1990), and in in vivo micronucleus formation in bone marrow (Roy et al., 2005; Mirkova, 1994),
4 and liver (Roy et al., 2005; Morita and Hayashi, 1998). Lastly, in the presence of toxicity,
5 positive results were reported for meiotic nondisjunction in drosophila (Munoz and Barnett,
6 2002), DNA damage (Sina et al., 1983), and cell transformation (Sheu et al., 1988).
7 Additionally, 1,4-dioxane metabolism did not produce reactive intermediates that
8 covalently bound to DNA (Stott et al., 1981; Woo et al., 1977a) and DNA repair assays were
9 generally negative (Goldsworthy et al., 1991; Stott et al., 1981). No studies were available to
10 assess the ability of 1,4-dioxane or its metabolites to induce oxidative damage to DNA.
4.7.3.7. Conclusions About the Hypothesized Mode of Action
11 4.7.3.7.1. Liver. The MOA by which 1,4-dioxane produces liver tumors is unknown, and
12 available evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
13 inconclusive. A MOA hypothesis involving 1,4-dioxane induced cell proliferation is possible
14 but data are not available to support this hypothesis. Pharmacokinetic data suggest that
15 clearance pathways were saturable and target organ toxicity occurs after metabolic saturation.
16 Liver toxicity preceded tumor formation in one study (Kociba et al., 1974) and a regenerative
17 response to tissue injury was demonstrated by histopathology. Liver hyperplasia and tumor
18 formation have also been observed in the absence of cytotoxicity (JBRC, 1998a). Cell
19 proliferation and tumor promotion have been shown to occur after prolonged exposure to
20 1,4-dioxane (Miyagawa et al., 1999; Uno et al., 1994; Goldsworthy et al., 1991; Lundberg et al.,
21 1987; Bull et al., 1986; Stott et al., 1981; King et al., 1973).
22 4.7.3.7.2. Nasal cavity. The MOA for the formation of nasal cavity tumors is unknown, and
23 evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
24 inconclusive.
4.7.3.8. Relevance of the Mode of Action to Humans
25 Several hypothesized MO As for 1,4-dioxane induced tumors in laboratory animals have
26 been discussed along with the supporting evidence for each. As was stated, the MOA by which
27 1,4-dioxane produces liver, nasal, peritoneal, and mammary gland tumors is unknown. Currently
28 there does exist some mechanistic information to present hypothesized MO As for liver and nasal
29 tumors but no information exists to present hypothesis for the observed peritoneal or mammary
30 gland tumors (JBRC, 1998a). At this time there is inadequate evidence to determine the human
31 relevance of any of the hypothesized MO As for 1,4-dioxane-induced tumors.
May 2009 86 DRAFT - DO NOT CITE OR QUOTE
-------
4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES
1 There is no direct evidence to establish that certain populations and lifestages may be
2 potentially susceptible to 1,4-dioxane. Changes in susceptibility with lifestage as a function of
3 the presence of microsomal enzymes that metabolize and detoxify this compound (i.e., CYP2E1
4 present in liver, kidney, and nasal mucosa can be hypothesized). Vieira et al. (1996) reported
5 that large increases in hepatic CYP2E1 protein occur postnatally between 1 and 3 months in
6 humans. Adult hepatic concentrations of CYP2E1 are achieved sometime between 1 and
7 10 years. To the extent that hepatic CYP2E1 levels are lower, children may be more susceptible
8 to liver toxicity from 1,4-dioxane than adults. CYP2E1 has been shown to be inducible in the rat
9 fetus. The level of CYP2E1 protein was increased by 1.4-fold in the maternal liver and 2.4-fold
10 in the fetal liver following ethanol treatment, as compared to the untreated or pair-fed groups
11 (Carpenter et al., 1996). Pre- and postnatal induction of microsomal enzymes resulting from
12 exposure to 1,4-dioxane or other drugs or chemicals may reduce overall toxicity following
13 sustained exposure to 1,4-dioxane.
14 Genetic polymorphisms have been identified for the human CYP2E1 gene (Watanabe
15 et al., 1994; Hayashi et al., 1991) and were considered to be possible factors in the abnormal
16 liver function seen in workers exposed to vinyl chloride (Huang et al., 1997). Individuals with a
17 CYP2E1 genetic polymorphism resulting in increased expression of this enzyme may be less
18 susceptible to toxicity following exposure to 1,4-dioxane.
19 Gender differences were noted in subchronic and chronic toxicity studies of 1,4-dioxane
20 in mice and rats (see Sections 4.6 and 4.7). No consistent pattern of gender sensitivity was
21 identified across studies.
May 2009 87 DRAFT - DO NOT CITE OR QUOTE
-------
5. DOSE-RESPONSE ASSESSMENTS
5.1. ORAL REFERENCE DOSE (RFD)
5.1.1. Choice of Principal Studies and Critical Effect with Rationale and Justification
1 Liver and kidney toxicity were the primary noncancer health effects associated with
2 exposure to 1,4-dioxane in humans and laboratory animals. Occupational exposure to
3 1,4-dioxane has resulted in hemorrhagic nephritis and centrilobular necrosis of the liver
4 (Johnstone, 1959; Barber, 1934). In animals, liver and kidney degeneration and necrosis were
5 observed frequently in acute oral and inhalation studies (JBRC, 1998b; Drew et al., 1978; David,
6 1964; Kesten et al., 1939; Laug et al., 1939; Schrenk and Yant, 1936; de Navasquez, 1935;
7 Fairley et al., 1934). Liver and kidney effects were also observed following chronic oral
8 exposure to 1,4-dioxane in animals (JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Argus et al.,
9 1973, 1965) (see Table 4-17).
10 Liver toxicity in the available chronic studies was characterized by necrosis, spongiosis
11 hepatic, hyperplasia, cyst formation, clear foci, and mixed cell foci. Kociba et al. (1974)
12 demonstrated hepatocellular degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in
13 male rats) or greater. The NOAEL for liver toxicity was 9.6 mg/kg-day and 19 mg/kg-day in
14 male and female rats, respectively. No quantitative incidence data were provided in this study.
15 Argus et al. (1973) described early preneoplastic changes in the liver and JBRC (1998)
16 demonstrated liver lesions that are primarily associated with the carcinogenic process. Clear and
17 mixed-cell foci in the liver are commonly considered preneoplastic changes and would not be
18 considered evidence of noncancer toxicity. In the JBRC (1998a) study, spongiosis hepatis was
19 associated with other preneoplastic changes in the liver (clear and mixed-cell foci) and no other
20 lesions indicative of liver toxicity were seen. Spongiosis hepatis was therefore not considered
21 indicative of noncancer effects in this study. The activity of serum enzymes (i.e., AST, ALT,
22 LDH, and ALP) was increased in mice and rats chronically exposed to 1,4-dioxane (JBRC,
23 1998a); however, these increases were seen only at tumorigenic dose levels. Blood samples
24 were collected at study termination and elevated serum enzymes may reflect changes associated
25 with tumor formation. Histopathological evidence of liver toxicity was not seen in rats from the
26 JBRC (1998a) study. The highest non-tumorigenic dose levels for this study approximated the
27 LOAEL derived from the Kociba et al. (1974) study (94 and 148 mg/kg-day for male and female
28 rats, respectively).
29 Kidney damage in chronic toxicity studies was characterized by degeneration of the
30 cortical tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCI, 1978; Kociba
31 et al., 1974; Argus et al., 1965, 1973; Fairley et al., 1934). Kociba et al. (1974) described renal
May 2009 88 DRAFT - DO NOT CITE OR QUOTE
-------
1 tubule epithelial cell degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats)
2 or greater, with a NOAEL of 9.6 mg/kg-day. No quantitative incidence data were provided in
3 this study. Doses of > 430 mg/kg-day 1,4-dioxane induced marked kidney alterations (Argus
4 et al., 1973). The observed changes included glomerulonephritis and pyelonephritis, with
5 characteristic epithelial proliferation of Bowman's capsule, periglomerular fibrosis, and
6 distension of tubules. Quantitative incidence data were not provided in this study. In the NCI
7 (1978) study, kidney lesions in rats consisted of vacuolar degeneration and/or focal tubular
8 epithelial regeneration in the proximal cortical tubules and occasional hyaline casts. Kidney
9 toxicity was not seen in rats from the JBRC (1998a) study at any dose level (highest dose was
10 398 mg/kg-day in male rats and 514 mg/kg-day in female rats).
11 Kociba et al. (1974) was chosen as the principal study for derivation of the RfD because
12 the liver and kidney effects in this study are adverse and represent the most sensitive effects
13 identified in the database (NOAEL 9.6 mg/kg-day, LOAEL 94 mg/kg-day in male rats). Kociba
14 et al. (1974) reported degenerative effects in the liver, while liver lesions reported in other
15 studies (JBRC, 1998a; Argus et al., 1973) appeared to be related to the carcinogenic process.
16 Kociba et al. (1974) also reported degenerative changes in the kidney. NCI (1978) and Argus
17 et al. (1973) provided supporting data for this endpoint; however, kidney toxicity was observed
18 in these studies at higher doses. JBRC (1998a) reported nasal inflammation in rats (NOAEL 81
19 mg/kg-day, LOAEL 398 mg/kg-day) and mice (NOAEL 77 mg/kg-day, LOAEL 323 mg/kg-
20 day).
5.1.2. Methods of Analysis—Including Models (PBPK, BMD, etc.)
21 Several procedures were applied to the human PBPK model to determine if an adequate
22 fit of the model to the empirical model output or experimental observations could be attained
23 using biologically plausible values for the model parameters. The re-calibrated model
24 predictions for blood 1,4-dioxane levels do not come within 10-fold of the experimental values
25 using measured tissue:air partition coefficients of Leung and Paustenbach (1990) or Sweeney
26 et al. (2008) (Figures B-8 and B-9). The utilization of a slowly perfused tissue:air partition
27 coefficient 10-fold lower than measured values produces exposure-phase predictions that are
28 much closer to observations, but does not replicate the elimination kinetics (Figure B-10). Re-
29 calibration of the model with upper bounds on the tissue:air partition coefficients results in
30 predictions that are still six- to sevenfold lower than empirical model prediction or observations
31 (Figures B-12 and B-13). Exploration of the model space using an assumption of first-order
32 metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate fit to the
33 exposure and elimination data can be achieved only when unrealistically low values are assumed
34 for the slowly perfused tissue:air partition coefficient (Figure B-16). Artificially low values for
35 the other tissue:air partition coefficients are not expected to improve the model fit, as these
36 parameters are shown in the sensitivity analysis to exert less influence on blood 1,4-dioxane than
May 2009 89 DRAFT - DO NOT CITE OR QUOTE
-------
1 Vmaxc and Km. This suggests that the model structure is insufficient to capture the apparent 10-
2 fold species difference in the blood 1,4-dioxane Vd between rats and humans. In the absence of
3 actual measurements for the human slowly perfused tissue:air partition coefficient, high
4 uncertainly exists for this model parameter value. Differences in the ability of rat and human
5 blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this is expected to
6 be evident in very different values for rat and human blood:air partition coefficients, which is not
7 the case (Table B-l). Therefore, some other, as yet unknown, modification to model structure
8 may be necessary.
9 Kociba et al. (1974) did not provide quantitative incidence or severity data for liver and
10 kidney degeneration and necrosis. Benchmark dose (BMD) modeling could not be performed
11 for this study and the NOAEL for liver and kidney degeneration (9.6 mg/kg-day in male rats)
12 was used as the point of departure (POD) in deriving the RfD for 1,4-dioxane.
13 Alternative PODs were calculated using incidence data reported for cortical tubule
14 degeneration in male and female rats (NCI, 1978) and liver hyperplasia (JBRC, 1998a). The
15 incidence data for cortical tubule cell degeneration in male and female rats exposed to
16 1,4-dioxane in the drinking water for 2 years are presented in Table 5-1. Details of the BMD
17 analysis of these data are presented in Appendix C. Male rats were more sensitive to the kidney
18 effects of 1,4-dioxane than females and the male rat data provided the lowest POD for cortical
19 tubule degeneration in the NCI (1978) study (BMDLio of 38.5 mg/kg-day) (see Table 5-2).
20 Incidence data for liver hyperplasia in male and female rats exposed to 1,4-dioxane in the
21 drinking water for 2 years are presented in Table 5-3. Details of the BMD analysis of these data
22 are presented in Appendix C. Male rats were more sensitive to developing liver hyperplasia due
23 to exposure to 1,4-dioxane than females and the male rat data provided the lowest POD for
24 hyperplasia in the JBRC (1998a) study (BMDLio of 34.7 mg/kg-day) (see Table 5-4). The
25 BMDLio values of 38.5 mg/kg-day and 34.7 mg/kg-day from the NCI (1978) and JBRC (1998a)
26 studies, respectively, supports the NOAEL of 9.6 mg/kg-day observed by Kociba et al. (1974).
Table 5-1. Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed tol,4-dioxane in drinking water for 2 years
Males (mg/kg-day)
0
0/3 r
240
20/3 lb
530
27/3 3b
Females (mg/kg-day)
0
0/3 la
350
0/34
640
10/32b
""Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this
review.
b-
Incidence significantly elevated compared to control by Fisher's Exact test (p < 0.001) performed for this review.
Source: NCI (1978).
May 2009
90
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-2. BMD and BMDL values derived from BMD modeling of cortical
tubule degeneration in male and female Osborne-Mendel rats exposed to
1,4-dioxane in drinking water for 2 years
Male rats
Female rats
BMD10 (mg/kg-day)
51.4
591.8
BMDL10 (mg/kg-day)
38.5
447.2
Source: NCI (1978).
Table 5-3. Incidence of liver hyperplasia in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years
Males (mg/kg-day)
0
3/40
16
2/45
81
9/3 5a
398
12/22b
Females (mg/kg-day)
0
0/3 8a
21
0/37
103
1/38
514
14/24b
"Statistically significant compared to controls by the Dunnett's test (p < 0.05).
blncidence significantly elevated compared to control by Chi2 test (p < 0.01).
Source: JBRC(1998a).
Table 5-4. BMD and BMDL values derived from BMD modeling of liver
hyperplasia in male and female F344/DuCrj rats exposed to 1,4-dioxane in
drinking water for 2 years
Male rats
Female rats
BMD10 (mg/kg-day)
52.3
69.0
BMDL10 (mg/kg-day)
34.7
38.7
Source: JBRC(1998a).
5.1.3. RfD Derivation - Including Application of Uncertainty Factors (UFs)
•v-2
1 The RfD of 3 x 10 mg/kg-day is based on liver and kidney toxicity in rats exposed to
2 1,4-dioxane in the drinking water for 2 years (Kociba et al., 1974). The Kociba et al. (1974)
3 study was chosen as the principal study because it provides the most sensitive measure of
4 adverse effects by 1,4-dioxane. The incidence of liver and kidney lesions was not reported for
5 each dose group. Therefore, BMD modeling could not be used to derive a POD. The RfD for
6 1,4-dioxane is derived by dividing the NOAEL of 9.6 mg/kg-day (Kociba et al., 1974) by a
7 composite UF of 300, as follows:
May 2009
91
DRAFT - DO NOT CITE OR QUOTE
-------
1 RfD = NOAEL/UF
2 =9.6 mg/kg-day / 300
3 = 0.03 or 3 x 10"2 mg/kg-day
4 The composite UF of 300 includes factors of 10 for animal-to-human extrapolation and
5 interindividual variability, and a partial UF of 3 for database deficiencies.
6 A default interspecies UF of 10 was used to account for pharmacokinetic and
7 pharmacodynamic differences across species. Existing PBPK models could not be used to derive
8 an oral RfD for 1,4-dioxane (see Appendix B).
9 A default interindividual variability UF of 10 is used to account for variation in
10 sensitivity within human populations because there is limited information on the degree to which
11 humans of varying gender, age, health status, or genetic makeup might vary in the disposition of,
12 or response to, 1,4-dioxane.
13 A default UF of 3 for database deficiencies is selected due to the lack of a
14 multigeneration reproductive toxicity study. A single oral prenatal developmental toxicity study
15 in rats was available for 1,4-dioxane (Giavini et al., 1985). This developmental study indicates
16 that the developing fetus may be a target of toxicity.
17 An UF to extrapolate from a subchronic to a chronic exposure duration was not necessary
18 because the RfD was derived from a study using a chronic exposure protocol.
19 An UF to extrapolate from a LOAEL to a NOAEL was not necessary because the RfD
20 was based on a NOAEL. Kociba et al. (1974) was a well-conducted, chronic drinking water
21 study with an adequate number of animals. Histopathological examination was performed for
22 many organs and tissues, but clinical chemistry analysis was not performed. NOAEL and
23 LOAEL values were derived from the study based on liver and kidney toxicity. Several
24 additional oral studies (acute/short-term, subchronic, and chronic durations) were available that
25 support liver and kidney toxicity as the critical effect (Kano et al., 2008; JBRC, 1998a; NCI,
26 1978; Argus et al., 1973, see Tables 4-15 and 4-17). Although degenerative liver and kidney
27 toxicity was not observed in rats from the JBRC (1998a) study at doses at or below the LOAEL
28 value in the Kociba et al. (1974) study, other endpoints such as metaplasia and hyperplasia of the
29 nasal epithelium, nuclear enlargement, and hematological effects, were noted.
5.1.4. RfD Comparison Information
30 PODs and sample oral RfDs based on selected studies included in Table 4-17 are arrayed
31 in Figures 5-1 to 5-3, and provide perspective on the RfD supported by Kociba et al. (1974).
32 These figures should be interpreted with caution because the PODs across studies are not
33 necessarily comparable, nor is the confidence in the data sets from which the PODs were derived
34 the same. PODs in these figures may be based on a NOAEL, LOAEL, or BMDL (as indicated),
35 and the nature, severity, and incidence of effects occurring at a LOAEL are likely to vary. To
36 some extent, the confidence associated with the resulting sample RfD is reflected in the
May 2009 92 DRAFT - DO NOT CITE OR QUOTE
-------
1 magnitude of the total UF applied to the POD (i.e., the size of the bar); however, the text of
2 Sections 5.1.1 and 5.1.2 should be consulted for a more complete understanding of the issues
3 associated with each data set and the rationale for the selection of the critical effect and principal
4 study used to derive the RfD.
5 The predominant noncancer effect of chronic oral exposure to 1,4-dioxane is
6 degenerative effects in the liver and kidney. Figure 5-1 provides a graphical display of effects
7 that were observed in the liver following chronic oral exposure to 1,4-dioxane. Information
8 presented includes the PODs and UFs that could be considered in deriving the oral RfD. As
9 discussed in Sections 5.1.1 and 5.1.2, among those studies that demonstrated liver toxicity, the
10 study by Kociba et al. (1974) provided the data set most appropriate for deriving the RfD. For
11 degenerative liver effects resulting from 1,4-dioxane exposure, the Kociba et al. (1974) study
12 represents the most sensitive effect and dataset observed in a chronic bioassay (Figure 5-1).
13 Kidney toxicity as evidenced by glomerulonephritis (Argus et al., 1973; 1965) and
14 degeneration of the cortical tubule (NCI, 1978; Kociba et al., 1974) has also been observed in
15 response to chronic exposure to 1,4-dioxane. As was discussed in Sections 5.1 and 5.2,
16 degenerative effects were observed in the kidney at the same dose level as effects in the liver
17 (Kociba et al., 1974). A comparison of the available datasets from which an RfD could
18 potentially be derived is presented in Figure 5-2.
19 Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice
20 only, dose > 380 mg/kg-day) and JBRC (1998a) studies (> 398 mg/kg-day in rats, >323 mg/kg-
21 day in mice). JBRC (1998a) reported nasal inflammation in rats (NOAEL 81 mg/kg-day,
22 LOAEL 398 mg/kg-day) and mice (NOAEL 77 mg/kg-day, LOAEL 323 mg/kg-day). A
23 comparison of the available datasets from which an RfD could potentially be derived is presented
24 in Figure 5-3.
25 Figure 5-4 displays PODs for the major targets of toxicity associated with oral exposure
26 to 1,4-dioxane. Studies in experimental animals have also found that relatively high doses of
27 1,4-dioxane (1,000 mg/kg-day) during gestation can produce delayed ossification of the
28 sternebrae and reduced fetal BWs (Giavini et al., 1985). This graphical display (Figure 5-4)
29 compares organ specific toxicity for 1,4-dioxane, including a single developmental study. The
30 most sensitive measures of degenerative liver are and kidney effects. The sample RfDs for
31 degenerative liver and kidney effects are identical since they were derived from the same study
32 and dataset (Kociba et al., 1974) and are presented for completeness.
May 2009 93 DRAFT - DO NOT CITE OR QUOTE
-------
100
Rat
Rat
Mouse
Rat
Rat
7
ra
• POD
UAnimal-to-human
QHuman variation
E3LOAELtoNOAEL
nSubchronic to Chronic
^Database deficiencies
ORfD
0.01
Liver hyperplasia; Hepatocellular Increase in serum liver Increase in serum liver Liver hyperplasia;
NOAEL;2yrrat degeneration and enzymes; NOAEL; 2 yr enzymes; NOAEL; 2 yr BMDL10; 2 yr rat
drinking water study necrosis; NOAEL; 2 yr mouse drinking water rat drinking water study drinking water study
(JBRC, 1998a) rat drinking water study study (JBRC, 1998a) (JBRC, 1998a) (JBRC, 1998a)
(Kocibaetal., 1974)
Figure 5-1. Points of departure (POD) for liver toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following oral exposure
to 1,4-dioxane.
May 2009
94
DRAFT - DO NOT CITE OR QUOTE
-------
Rat
Rat
Rat
1000
100
10
0.1 --
0.01
• POD
nTIAnimal-to-human
QHuman variation
E2LOAEL to NOAEL
dlSubchronic to Chronic
•Database deficiencies
ORfD
Glomerulonephritis; LOAEL; 13 month
rat drinking water study (Argus et al.,
1973)
Degeneration and necrosis of tubular
epithelium; NOAEL; 2 yr rat drinking
water study (Kociba et al., 1974)
Cortical tubule degeneration; BMDL10; 2
yr rat drinking water study (NCI, 1978)
Figure 5-2. Points of departure (POD) for kidney toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following oral exposure
to 1,4-dioxane.
May 2009
95
DRAFT - DO NOT CITE OR QUOTE
-------
100
10
•9
01
0.1
Mouse
nil
Rat
nil
I
I
o
• POD
[[jAnimal-to-human
QHuman variation
0LOAEL to NOAEL
EDsubchronic to Chronic
BDatabase deficiencies
ORfD
Nasal inflammation; NOAEL; 2 yr mouse drinking water Nasal inflammation; NOAEL; 2 yr rat drinking water study
study (JBRC, 1998a) (JBRC, 1998a)
Figure 5-3. Points of departure (POD) for nasal inflammation with corresponding
applied uncertainty factors and derived RfDs following oral expsorue to
1,4-dioxane.
May 2009
96
DRAFT - DO NOT CITE OR QUOTE
-------
1000
100
10 -
Rat
Rat
Rat
Mouse
0)
V)
8
0.1 -
0.01
• POD
Q]Animal-to-human
nHuman variation
0LOAEL to NOAEL
QSubchronic to Chronic
| Database deficiencies
ORfD
Degeneration and necrosis Hepatocellular Delayed ossification of Nasal inflammation;
of tubular epithelium; degeneration and necrosis; sternebrae and reduced NOAEL; 2 yr mouse
NOAEL; 2 yr rat drinking NOAEL; 2 yr rat drinking fetal body weight; NOAEL; drinking water study
water study (Kociba et al., water sduy (Kociba et al., rat study gestation days 6- (JBRC, 1998a)
1974) 1974) 15 (Giavini et al., 1985)
Figure 5-4. Points of departure (POD) for organ specific toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following oral exposure
to 1,4-dioxane.
5.1.5. Previous RfD Assessment
1 An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
2 oral RfD was not developed as part of the 1988 assessment.
5.2. INHALATION REFERENCE CONCENTRATION (RFC)
3 Inhalation studies for 1,4-dioxane evaluated in this assessment were not adequate for the
4 determination of an RfC value. Only one subchronic study (Fairley et al., 1934) and one chronic
5 inhalation study (Torkelson et al., 1974) were identified. In the subchronic study, rabbits, guinea
6 pigs, rats, and mice (3-6/species/group) were exposed to 1,000, 2,000, 5,000, or 10,000 ppm of
7 1,4-dioxane vapor for 16.5 hours/week. Animals were exposed until death occurred or were
8 sacrificed at varying time periods (up to 12 weeks). Severe liver and kidney damage and acute
9 vascular congestion of the lungs were observed at concentrations > 1,000 ppm. Kidney damage
10 was described as patchy degeneration of cortical tubules with vascular congestion and
11 hemorrhage. Liver lesions varied from cloudy hepatocyte swelling to large areas of necrosis.
12 Torkelson et al. (1974) performed a chronic inhalation study in which male and female
13 Wistar rats (288/sex) were exposed to 111 ppm 1,4-dioxane vapor for 7 hours/day, 5 days/week
14 for 2 years. Control rats (192/sex) were exposed to filtered air. No significant effects were
May 2009 97 DRAFT - DO NOT CITE OR QUOTE
-------
1 observed on BWs, survival, organ weights, hematology, clinical chemistry, or histopathology.
2 Because Fairley et al. (1934) identified a free-standing LOAEL only, and Torkelson et al. (1974)
3 identified a free-standing NOAEL only, neither study was sufficient to characterize the
4 inhalation risks of 1,4-dioxane. A route extrapolation from oral toxicity data was not performed
5 because 1,4-dioxane inhalation causes direct effects on the respiratory tract (i.e., respiratory
6 irritation in humans, pulmonary congestion in animals) (Wirth and Klimmer, 1936; Fairley et al.,
7 1934; Yant et al., 1930), which would not be accounted for in a cross-route extrapolation. In
8 addition, available kinetic models are not suitable for this purpose (see Appendix B).
9 An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
10 inhalation RfC was not developed as part of the 1988 assessment.
11 During the review of this assessment, new data regarding the toxicity of 1,4-dioxane
12 through the inhalation route of exposure have become available. The Agency will evaluate the
13 recently published 1,4-dioxane inhalation data for the potential to derive an RfC in a separate
14 document to follow this assessment.
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE (RfD)
15 Risk assessments need to portray associated uncertainty. The following discussion
16 identifies uncertainties associated with the RfD for 1,4-dioxane. As presented earlier in this
17 section (5.1.2 and 5.1.3), the uncertainty factor approach (U.S. EPA, 2002a, 1994b), was applied
18 to a POD. Factors accounting for uncertainties associated with a number of steps in the analyses
19 were adopted to account for extrapolating from an animal bioassay to human exposure, a diverse
20 population of varying susceptibilities, and to account for database deficiencies. These
21 extrapolations are carried out with current approaches given the paucity of experimental
22 1,4-dioxane data to inform individual steps.
23 An adequate range of animal toxicology data are available for the hazard assessment of
24 1,4-dioxane, as described throughout the previous section (Chapter 4). The database of oral
25 toxicity studies includes chronic drinking water studies in rats and mice, multiple subchronic
26 drinking water studies conducted in rats and mice, and a developmental study in rats. Toxicity
27 associated with oral exposure to 1,4-dioxane is observed predominately in the liver and kidney.
28 The database of inhalation toxicity studies in animals includes one subchronic bioassay in
29 rabbits, guinea pigs, and rats, and a chronic inhalation bioassay in rats. Although the subchronic
30 bioassay observed degenerative effects in the liver, kidney, and lungs of all species tested, the
31 information reported from the study was insufficient to determine an exposure level below which
32 these effects did not occur. The only available chronic inhalation bioassay did not indicate any
33 treatment related effects due to exposure to 1,4-dioxane. Thus, the inhalation database lacked
34 sufficient information to derive toxicity values relevant to this route of exposure for 1,4-dioxane.
35 In addition to oral and inhalation data, there are PBPK models and genotoxicity studies of
May 2009 98 DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-dioxane. Critical data gaps have been identified and uncertainties associated with data
2 deficiencies of 1,4-dioxane are more fully discussed below.
3 Consideration of the available dose-response data led to the selection of the two-year
4 drinking water bioassay in Sherman rats (Kociba et al., 1974) as the principal study and
5 increased liver and kidney degeneration as the critical effects for deriving the RfD for
6 1,4-dioxane. The dose-response relationship for oral exposure to 1,4-dioxane and cortical tubule
7 degeneration in Osborne-Mendel rats (NCI, 1978) was also suitable for deriving a RfD, but it is
8 associated with higher a POD and sample RfD compared to Kociba et al. (1974).
9 The RfD was derived by applying UFs to a NOAEL for degenerative liver and kidney
10 effects. The incidence data for the observed effects were not reported in the principal study
11 (Kociba et al., 1974), precluding modeling of the dose-response. However confidence in the
12 LOAEL can be derived from additional studies (JBRC, 1998a; NCI, 1978; Argus et al., 1973;
13 1965) that observed effects on the same organs at comparable dose levels and by the BMDL
14 generated by modeling of the kidney dose-response data from the chronic NCI (1978) study.
15 Extrapolating from animals to humans embodies further issues and uncertainties. The
16 effect and the magnitude associated with the dose at the POD in rodents are extrapolated to
17 human response. Pharmacokinetic models are useful to examine species differences in
18 pharmacokinetic processing; however, it was determined that dosimetric adjustment using
19 pharmacokinetic modeling was to reduce uncertainty following oral exposure to 1,4-dioxane was
20 not supported. Insufficient information was available to quantitatively assess toxicokinetic or
21 toxicodynamic differences between animals and humans, so a 10-fold UF was used to account
22 for uncertainty in extrapolating from laboratory animals to humans in the derivation of the RfD.
23 Heterogeneity among humans is another uncertainty associated with extrapolating doses
24 from animals to humans. Uncertainty related to human variation needs consideration. In the
25 absence of 1,4-dioxane-specific data on human variation, a factor of 10 was used to account for
26 uncertainty associated with human variation in the derivation of the RfD. Human variation may
27 be larger or smaller; however, 1,4-dioxane-specific data to examine the potential magnitude of
28 over- or under-estimation is unavailable.
29 Uncertainties in the assessment of the health hazards of ingested 1,4-dioxane are
30 associated with deficiencies in reproductive toxicity information. The oral database lacks a
31 multigeneration reproductive toxicity study. A single oral prenatal developmental toxicity study
32 in rats was available for 1,4-dioxane (Giavini et al., 1985). This developmental study indicates
33 that the developing fetus may be a target of toxicity. The database of inhalation studies is of
34 particular concern due to the lack of a basic toxicological studies, a multigenerational
35 reproductive study, and developmental toxicity studies.
May 2009 99 DRAFT - DO NOT CITE OR QUOTE
-------
5.4. CANCER ASSESSMENT
5.4.1. Choice of Study/Data - with Rationale and Justification
1 Three chronic drinking water bioassays provided incidence data for liver tumors in rats
2 and mice, and nasal cavity, peritoneal, and mammary gland tumors in rats only (JBRC, 1998a;
3 NCI, 1978; Kociba et al., 1974). The dose-response data from each of these studies are
4 summarized in Table 5-5. With the exception of the NCI (1978) study, the incidence of nasal
5 cavity tumors was generally lower than the incidence of liver tumors in exposed rats. The JBRC
6 (1998a) drinking water study was chosen as the principal study for derivation of an oral cancer
7 slope factor (CSF) for 1,4-dioxane. This study used three dose groups in addition to controls and
8 characterized the dose-response relationship at lower exposure levels, as compared to the high
9 doses employed in the NCI (1978) bioassay. The Kociba et al. (1974) study also used three dose
10 groups and low exposures; however, the study authors only reported the incidence of
11 hepatocellular carcinoma, which may underestimate the combined incidence of rats with
12 adenoma or carcinoma. In addition to increased incidence of liver tumors, chosen as the most
13 sensitive target organ for tumor formation, the JBRC (1998a) study also noted increased
14 incidence of peritoneal and mammary gland tumors. Nasal cavity tumors were also seen in high-
15 dose male and female rats; however, the incidence of nasal tumors was much lower than the
16 incidence of liver tumors in both rats and mice.
Table 5-5. Incidence of liver, nasal cavity, peritoneal, and mammary gland
tumors in rats and mice exposed to 1,4-dioxane in drinking water for 2 years
(based on survival to 12 months)
Study
Kociba etal., 1974
NCI, 1978
Species/strain/gender
Sherman rats, male and
female combined3 >b
Male Osborne-Mendel
ratsb
Female Osborne-
Mendel ratsb>c
Male B6C3FJ miced
Female B6C3FJ miced
Animal dose
(mg/kg-day)
0
14
121
1,307
0
240
530
0
350
640
0
720
830
0
380
Tumor Incidence
Liver
l/106h
0/110
1/106
10/661
NA
NA
NA
0/3 lh
10/301
11/291
8/49h
19/501
28/471
0/50h
21/481
Nasal
cavity
0/106h
0/110
0/106
3/66
0/3 3h
12/26
16/331
0/34h
10/301
8/291
NA
NA
NA
NA
NA
Peritoneal
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Mammary
gland
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
May 2009
100
DRAFT - DO NOT CITE OR QUOTE
-------
Study
JBRC, 1998a;
email from Dr.
Kazunori
Yamazaki, JBRC,
to Dr. Julie
Stickney, SRC,
dated 12/18/06.
Species/strain/gender
Male F344/DuCrj
ratsd,e,f,g
Female F344/DuCrj
ratsd,e,f,g
MaleCrj:BDF1miced
Female CrjiEDFj
miced
Animal dose
(mg/kg-day)
860
0
16
81
398
0
21
103
514
0
66
251
768
0
77
323
1,066
Tumor Incidence
Liver
35/371
0/50h
2/50
4/49
33/501
l/50h
0/50
5/50
40/501
21/50
31/48
37/50
39/481
4/50h
34/501
41/481
46/481
Nasal
cavity
NA
0/50h
0/50
0/50
7/501
0/50h
0/50
0/50
8/501
NA
NA
NA
NA
NA
NA
NA
NA
Peritoneal
NA
2/50
2/50
5/50
28/501
1/50
0/50
0/50
0/50
NA
NA
NA
NA
NA
NA
NA
NA
Mammary
gland
NA
1/50
1/50
0/50
4/50
9/50
9/50
11/50
19/50
NA
NA
NA
NA
NA
NA
NA
NA
"Incidence of hepatocellular carcinoma.
blncidence of nasal squamous cell carcinoma.
Incidence of hepatocellular adenoma.
Incidence of hepatocellular adenoma or carcinoma.
Incidence of all nasal tumors including squamous cell carcinoma, sarcoma, rhabdomyosarcoma, and esthesioneuro-
epithelioma.
Incidence of peritoneal tumors (mesothelioma).
Incidence of mammary gland tumors (fibroadenoma and adenoma combined)
hp < 0.05; positive dose-related trend (Cochran-Armitage orPeto test).
lp < 0.05; Fisher's Exact test.
NA = data were not available for modeling (no significant change from controls)
5.4.2. Dose-Response Data
1 Table 5-6 summarizes the incidence of hepatocellular adenoma or carcinoma in rats and
2 mice from the JBRC (1998a) 2-year drinking water study. There were statistically significant
3 increasing trends in tumorigenic response for males and females of both species. The dose-
4 response curve for female mice is steep, with 68% incidence of liver tumors occurring in the
5 low-dose group (77 mg/kg-day). Exposure to 1,4-dioxane increased the incidence of these
6 tumors in a dose-related manner.
7 A significant increase in the incidence of peritoneal mesothelioma was observed in high-
8 dose male rats only (28/50 rats, see Table 5-5). The incidence of peritoneal mesothelioma was
9 lower than the observed incidence of hepatocellular adenoma or carcinoma in male rats (see
10 Table 5-6); therefore, hepatocellular adenoma or carcinoma data were used to derive an oral CSF
11 for 1,4-dioxane.
May 2009
101
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-6. Incidence of hepatocellular adenoma or carcinoma in rats and
mice exposed to 1,4-dioxane in drinking water for 2 years
Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male CrjiBDFj mice
Female Crj:BDFi mice
Animal dose
(mg/kg-day)
0
16
81
398
0
21
103
514
0
66
251
768
0
77
323
1066
Incidence of liver tumors"
0/50b
2/50
4/49
33/50c
l/50b
0/50
5/50
40/50C
21/50
31/48
37/50
39/48c
4/50b
34/50c
41/48C
46/48c
""Incidence of hepatocellular adenoma or carcinoma.
bp < 0.05; positive dose-related trend (Cochran-Armitage orPeto test).
°p < 0.05; Fisher's Exact test.
Source: JBRC(1998a).
5.4.3. Dose Adjustments and Extrapolation Method(s)
5.4.3.1. Dose Adjustments
1 Human equivalent doses (HEDs) were calculated from the administered animal doses
2 using a BW scaling factor (BW0'75). This was accomplished using the following equation:
-|025
. TT__ • ij f n \ animal BW (kg)
3 HED = animal dose (mg/kg) x ±-^-
[_ human BW (kg)
4 HEDs for the principal study (JBRC, 1998a) are given in Table 5-7. HEDs were also calculated
5 for supporting studies (NCI, 1978; Kociba et al., 1974) and are also shown in Table 5-7.
May 2009
102
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-7. Calculated HEDs for the tumor incidence data used for dose-
response modeling
Study
JBRC, 1998a
Kocibaetal., 1974
NCI, 1978
Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male Crj:BDF! mice
Female Crj iBDFj mice
Male and female (combined)
Sherman rats
Male Osborne-Mendel rats
Female Osborne-Mendel rats
Male B6C3FJ mice
Female B6C3FJ mice
Animal BW (g)
TWA
380a
380a
380a
229a
229a
229a
37.3a
37.3a
37.3a
35.3a
35.3a
35.3a
325b
325b
285C
470b
470b
310b
310b
32b
32b
30b
30b
Animal dose
(mg/kg-day)
16
81
398
21
103
514
66
251
768
77
323
1066
14
121
1307
240
530
350
640
720
830
380
860
RED
(mg/kg-day)d
4.3
22
108
5.0
25
123
10
38
117
12
48
160
3.7
32
330
69
152
90
165
105
121
55
124
aJBRC (1998a) reported only terminal BWs. Default TWA BWs for F344 rats and B6C3FJ mice in a chronic study
were obtained from U.S. EPA (1988).
bTWA BWs were determined from BW curve provided for control animals.
°B Ws of high dose male and female rats were significantly lower than controls throughout the study. TWA
represents the mean of TWA for male and females (calculated separately from growth curves).
dHEDs are calculated as HED = (animal dose) x (animal BW / human BW)°25.
Sources: JBRC (1998a); Kociba et al. (1974); and NCI (1978).
May 2009
103
DRAFT - DO NOT CITE OR QUOTE
-------
5.4.3.2. Extrapolation Method(s)
1 The weight of evidence is inadequate to establish a MOA(s) by which 1,4-dioxane
2 induces peritoneal, mammary, or nasal tumors in rats and liver tumors in rats and mice (see
3 Section 4.7.3 for a more detailed discussion of 1,4-dioxane's hypothesized MOAs). Therefore,
4 based on the U.S. EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), a linear
5 low dose extrapolation was used as a default option. Accordingly, the CSF for 1,4-dioxane was
6 derived via a linear extrapolation from the POD calculated by curve fitting the experimental
7 dose-response data. The POD is the 95% lower confidence limit on the dose associated with a
8 benchmark response (BMR) near the lower end of the observed data. The BMD modeling
9 analysis used to estimate the POD is described in detail in Appendix D and is summarized below
10 in Section 5.4.4.
11 Model estimates were derived for all available bioassays and tumor endpoints (see
12 Appendix D); however, the POD used to derive the CSF is based on the most sensitive species
13 and target organ in the principal study (female mice; liver tumors; JBRC, 1998a).
14 The oral CSF was calculated using the following equation:
15 CSF= °J
BMDL10
5.4.4. Oral Slope Factor and Inhalation Unit Risk
16 The multistage model in the Benchmark Dose Software (BMDS, version 1.3.2) was fit to
17 the incidence data for hepatocellular carcinoma and/or adenoma in rats and mice and mammary
18 and peritoneal tumors in rats exposed to 1,4-dioxane in the drinking water (JBRC, 1998a) (Table
19 5-5). HEDs were used for BMD modeling (Table 5-7). Doses associated with a BMR of 10%
20 extra risk were calculated with the polynomial degree initially set at (n-1) and lower. BMDs and
21 BMDLs from the lowest degree polynomial models with an adequate fit (i p > 0.1) were
22 reported (see Appendix D). A summary of the BMDS model predictions for the JBRC (1998a),
23 NCI (1978), and Kociba et al. (1974) studies is shown in Table 5-8.
24 The multistage model did not provide an adequate fit (as determined by ^p > 0.1) to the
25 data for the incidence of hepatocellular adenoma or carcinoma in female mice (see Appendix D).
26 The high dose was dropped for the female mouse liver tumor dataset in an attempt to achieve an
27 adequate fit; however, an adequate fit was still not achieved. Because the female mice were
28 clearly the most sensitive group tested, other BMD models were applied to the female mouse
29 liver tumor dataset to achieve an adequate fit. The log-logistic model was the only model that
30 provided adequate fit for this data set due to the steep rise in the dose-response curve (68%
31 incidence at the low dose) followed by a plateau at near maximal tumor incidence in the mid-
32 and high-dose regions (85 and 96% incidence, respectively). The predicted BMDio and BMDLio
33 for the female mouse data are presented in Table 5-8. Similarly, the multistage model did not
May 2009 104 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
provide an adequate fit of mammary tumor incidence data for the female rat. The log logistic
model provided an adequate fit of this dataset. The predicted BMDio and BMDLio for female rat
mammary tumors and male peritoneal tumors are presented in Table 5-8.
A comparison of the model estimates derived for rats and mice from the JBRC (1998a),
NCI (1978), and Kociba et al. (1974) studies (Table 5-8) indicates that female mice are more
sensitive to liver carcinogenicity induced by 1,4-dioxane compared to other species or tumor
types. Additionally, the combined risk of multiple tumor sites in the rat model was considered
(See Appendix D, Tables D-16 through D-21) which also supports that liver carcinogenicity in
female mice is the most sensitive response. The BMDLio HED for the female mouse data was
chosen as the POD and the CSF of 0.19 (mg/kg-day)"1 was calculated as follows:
0.1
CSF =
= 0.19 (mg/kg-day)"
0.52 mg/kg - day (BMDL10HED for female mice)
Calculation of a CSF for 1,4-dioxane based on dose-response data for the most sensitive
species and gender represents a health-protective approach; however, no data currently exist to
determine which animal model (i.e., mouse or rat) is more representative of the potential cancer
risk in humans.
Table 5-8. BMDio HED and BMDLio HED values from models fit to tumor
incidence data for rats and mice exposed to 1,4-dioxane in drinking water for
2 years and corresponding oral CSFs
Study
JBRC, 1998a
Kociba et al.,
1974
NCI, 1978
Species/strain/gender
Male F344/DuCrj rats3
Female F344/DuCrj ratsb
MaleCrjiBDFj micec
Female Crj :BDP! mice0
Male F344/DuCrj ratsb
Female F344/DuCrj rats'
Male and female (combined)
Sherman rats3
Male Osborne Mendel rats3
Female Osborne Mendel rats3
Tumor type
Hepatocellular
adenoma or
carcinoma
Peritoneal
tumors
Mammary
tumors
Nasal
squamous cell
carcinomas
BMD10HED
(mg/kg-day)
21.9
31.1
4.74
0.79
39.3
40.9
880.8
18.8
36.9
BMDL10HED
(mg/kg-day)
11.9
27.3
2.41
0.52
33.6
20.9
387.8
13.9
25.6
Oral CSF
(mg/kg-day)1
8.4 x 10'3
3.7xlO'3
4.1xlO'2
0.19
3.0xlO'3
4.8 x 10'3
2.6 x 10'4
7.2 x 10'3
3.9 xlO'3
16
17
18
19
""Multistage model, degree of polynomial = 1.
bMultistage model, degree of polynomial = 2.
°Log logistic model, high dose dropped, degree of polynomial = 1.
Inhalation studies for 1,4-dioxane evaluated in this assessment were not adequate for the
determination of an inhalation unit risk value. No treatment-related tumors were noted in a
chronic inhalation study in rats; however, only a single exposure concentration was used
(111 ppm 1,4-dioxane vapor for 7 hours/day, 5 days/week for 2 years) (Torkelson et al., 1974).
May 2009 105 DRAFT - DO NOT CITE OR QUOTE
-------
1 A route extrapolation from oral bioassay data was not performed (see Section 5.2). In addition,
2 available kinetic models are not suitable for this purpose (see Appendix B).
3 During the review of this assessment, new data regarding the toxicity of 1,4-dioxane
4 through the inhalation route of exposure have become available. The Agency will evaluate the
5 recently published 1,4-dioxane inhalation data for the potential to derive an IUR in a separate
6 document to follow this assessment.
5.4.5. Previous Cancer Assessment
7 A previous cancer assessment was posted for 1,4-dioxane on IRIS in 1988. 1,4-Dioxane
8 was classified as a Group B2 Carcinogen (probable human carcinogen; sufficient evidence from
9 animal studies and inadequate eveident or no data from human epidemiology studies [U.S. EPA,
10 1986c]) based on the induction of nasal cavity and liver carcinomas in multiple strains of rats,
11 liver carcinomas in mice, and gall bladder carcinomas in guinea pigs. An oral CSF of 0.011
12 (mg/kg-day)"1 was derived from the tumor incidence data for nasal squamous cell carcinoma in
13 male rats exposed to 1,4-dioxane in drinking water for 2 years (NCI, 1978). The linearized
14 multistage extra risk procedure was used for linear low dose extrapolation.
5.5. UNCERTAINTIES IN CANCER RISK VALUES
15 As in most risk assessments, extrapolation of study data to estimate potential risks to
16 human populations from exposure to 1,4-dioxane has engendered some uncertainty in the results.
17 Several types of uncertainty may be considered quantitatively, but other important uncertainties
18 cannot be considered quantitatively. Thus an overall integrated quantitative uncertainty analysis
19 is not presented. Principal uncertainties are summarized below and in Table 5-9.
5.5.1. Sources of Uncertainty
5.5.1.1. Choice of Low-Dose Extrapolation Approach
20 The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
21 1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible
22 MOA(s) for the observed tumors. The MOA is a key consideration in clarifying how risks
23 should be estimated for low-dose exposure. Exposure to 1,4-dioxane has been observed in
24 animal models to induce multiple tumor types, including liver adenomas and carcinomas, nasal
25 carcinomas, mammary adenomas and flbroadenomas, and mesothiolomas of the peritoneal cavity
26 (JBRC, 1998a). MOA information that is available for the carcinogenicity of 1,4-dioxane has
27 largely focused on liver adenomas and carcinomas, with little or no MOA information available
28 for the remaining tumor types. In Section 4.7.3, hypothesized MO As, other than a mutagenic
29 MOA, were explored due to the lack of mutagenicity observed in genetic toxicology tests
30 performed for 1,4-dioxane. Information that would provide sufficient support for any MOA is
May 2009 106 DRAFT - DO NOT CITE OR QUOTE
-------
1 not available. In the absence of a MOA(s) for the observed tumor types, a linear low-dose
2 extrapolation approach was used to estimate human carcinogenic risk associated with
3 1,4-dioxane exposure.
4 It is not possible to predict how additional MOA information would impact the dose-
5 response assessment for 1,4-dioxane because of the variety of tumors observed and the lack of
6 data on how 1,4-dioxane or a metabolite thereof, interacts with cells starting the progression to
7 the observed tumors.
8 In general, the Agency has preferred to use the multistage model for analyses of tumor
9 incidence and related endpoints because they have a generic biological motivation based on
10 long-established mathematical models such as the Moolgavkar-Venzon-Knudsen (MVK) model.
11 The MVK model does not necessarily characterize all modes of tumor formation, but it is
12 a starting point for most investigations and, much more often than not, has provided at least an
13 adequate description of tumor incidence data.
14 In the studies evaluated (JBRC 1998a; NCI, 1978; Kociba et al., 1974), the multistage
15 model provided good descriptions of the incidence of many tumor types in male and female rats
16 and in male mice exposed to 1,4-dioxane (JBRC, 1998a). However, the multistage model did
17 not provide an adequate fit for the female mouse liver tumor dataset based upon the following
18 (U.S. EPA, 2000b):
Goodness-of-fit %2/7-value > 0.10.
Akaike's Information Criterion (AIC) less than any other competing models, even if the
alternative models did not have a biological motivation.
No data greatly deviating from the fitted model, as measured by their j^ residuals.
19 BMDS software typically implements the guidance in the BMD technical guidance
20 document (U.S.EPA, 2000b) by imposing constraints on the values of certain parameters of the
21 models. When these constraints were imposed, the multistage model and most other models did
22 not fit the incidence data for female mouse liver adenomas or carcinomas.
23 The log-logistic model provided an adequate fit for the female mouse data (JBRC,
24 1998a). Additionally, a log-logistic model with a slope of 1, as is the case for the female mouse
25 data (JBRC, 1998a), represents a low-dose linear extrapolation that is consistent with Agency
26 guidance (U.S. EPA, 2005a). Therefore, the log-logistic model was selected, with the
27 BMDLio HED derived by applying the constraints, as consistent with recommended use of BMDS
28 in the BMD technical guidance document (U.S. EPA, 2000b).
29 The human equivalent oral CSFs estimated from tumor datasets with statistically
30 significant increases ranged from 2.6 x 10"4 to 0.19 per mg/kg-day (Table 5-8), a range of about
31 three orders of magnitude, with the extremes coming from the combined male and female rat
32 data for nasal carcinomas (Kociba et al., 1974) and the female mouse liver adenoma and
33 carcinoma dataset (JBRC, 1998a).
May 2009 107 DRAFT - DO NOT CITE OR QUOTE
-------
5.5.1.2. Dose Metric
1 1,4-Dioxane is known to be metabolized in vivo. However, it is unknown whether a
2 metabolite or the parent compound, or some combination of parent compound and metabolites, is
3 responsible for the observed toxicity. If the actual carcinogenic moiety is proportional to
4 administered exposure, then use of administered exposure as the dose metric is the least biased
5 choice. On the other hand, if this is not the correct dose metric, then the impact on the CSF is
6 unknown.
5.5.1.3. Cross-Species Scaling
1 An adjustment for cross-species scaling (BW°75) was applied to address toxicological
8 equivalence of internal doses between each rodent species and humans, consistent with the 2005
9 Guidelines for Carcinogen Risk Assessment (US EPA, 2005a). It is assumed that equal risks
10 result from equivalent constant lifetime exposures.
5.5.1.4. Statistical Uncertainty at the POD
11 Parameter uncertainty can be assessed through confidence intervals. Each description of
12 parameter uncertainty assumes that the underlying model and associated assumptions are valid.
13 For the log-logistic model applied to the female mouse data, there is a reasonably small degree of
14 uncertainty at the 10% excess incidence level (the POD for linear low-dose extrapolation).
5.5.1.5. Bioassay Selection
15 The study by JBRC (1998a) was used for development of an oral CSF. This was a well-
16 designed study, conducted in both sexes in two species with a sufficient number of animals per
17 dose group. The number of test animals allocated among three dose levels and an untreated
18 control group was adequate, with examination of appropriate toxicological endpoints in both
19 sexes of rats and mice. Alternative bioassays (NCI, 1978; Kociba et al., 1974) are available and
20 were fully considered for the derivation of the oral CSF.
5.5.1.6. Choice of Species/Gender
21 The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the
22 female mouse, which was thought to be more sensitive than male mice or either sex of rats to the
23 carcinogenicity of 1,4-dioxane. While all data, both species and sexes reported from the JBRC
24 (1998a) study, were suitable for deriving an oral CSF, the female mouse data represented the
25 most sensitive indicator of carcinogenicity in the rodent model. The lowest exposure level
26 (77 mg/kg-day or 12 mg/kg-day [HED]) observed a considerable and significant increase in
27 combined liver adenomas and carcinomas. Additional testing of doses within the range of
28 control and the lowest dose (77 mg/kg-day or 12 mg/kg-day [HED]) could refine and reduce
29 uncertainty for the oral CSF.
May 2009 108 DRAFT - DO NOT CITE OR QUOTE
-------
5.5.1.7. Relevance to Humans
1 The derivation of the oral CSF is derived using the tumor incidence in the liver of female
2 mice. A thorough review of the available toxicological data available for 1,4-dioxane provides
3 no scientific justification to propose that the liver adenomas and carcinomas observed in animal
4 models due to exposure to 1,4-dioxane are not relevant to humans. As such, liver adenomas and
5 carcinomas were considered relevant to humans due to exposure to 1,4-dioxane.
5.5.1.8. Human Population Variability
6 The extent of inter-individual variability in 1,4-dioxane metabolism has not been
7 characterized. A separate issue is that the human variability in response to 1,4-dioxane is also
8 unknown. Although a mutagenic MOA would indicate increased early-life susceptibility, the
9 data exploring whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life
10 stages is unavailable. This lack of understanding about potential differences in metabolism and
11 susceptibility across exposed human populations thus represents a source of uncertainty. Also,
12 the lack of information linking a MOA for 1,4-dioxane to the observed carcinogenicity is a
13 source of uncertainty.
Table 5-9. Summary of uncertainty in the 1,4-dioxane cancer risk
assessment
Consideration/
approach
Low-dose
extrapolation
procedure
Dose metric
Cross-species
scaling
Bioassay
Impact on oral slope
factor
Departure from
EPA's Guidelines for
Carcinogen Risk
Assessment POD
paradigm, if justified,
could | or t unit risk
an unknown extent
Alternatives could t
or | CSF by an
unknown extent
Alternatives could J,
orf CSF [e.g., 3.5-
fold J, (scaling by
BW) or t twofold
(scaling by BW067)]
Alternatives could t
or | CSF by an
unknown extent
Decision
Log-logistic model
to determine POD,
linear low-dose
extrapolation from
POD
Used administered
exposure
BW075 (default
approach)
JBRC 1998a
Justification
A linear low-dose extrapolation approach was used to
estimate human carcinogenic risk associated with
1,4-dioxane exposure. Where data are insufficient to
ascertain the MOA, EPA's 2005 Guidelines for
Carcinogen Risk Assessment recommend application
of a linear low-dose extrapolation approach.
Experimental evidence supports a role for metabolism
in toxicity, but it is unclear if the parent compound,
metabolite or both contribute to 1,4-dioxane toxicity.
There are no data to support alternatives. BW0'75
scaling was used to calculate equivalent cumulative
exposures for estimating equivalent human risks.
PBPK modeling was conducted but not deemed
suitable for interspecies extrapolation.
Alternative bioassays were available and considered
for derivation of oral CSF.
May 2009
109
DRAFT - DO NOT CITE OR QUOTE
-------
Consideration/
approach
Species /gender
combination
Human
relevance of
mouse tumor
data
Human
population
variability in
metabolism and
response/
sensitive
subpopulations
Impact on oral slope
factor
Human risk could J,
or t, depending on
relative sensitivity
If rodent tumors
proved not to be
relevant to humans,
unit risk would not
apply i.e., could J,
CSF
Low-dose risk f or J,
to an unknown extent
Decision
Female mouse
Liver adenomas and
carcinomas are
relevant to humans
Considered
qualitatively
Justification
There are no MOA data to guide extrapolation
approach for any choice. It was assumed that humans
are as sensitive as the most sensitive rodent
gender/species tested; true correspondence is
unknown. Calculation of the CSF for 1,4-dioxane
was based on dose-response data from the most
sensitive species and gender. The carcinogenic
response occurs across species.
1,4-dioxane is a multi-site carcinogen in rodents and
the MOA(s) is unknown; carcinogenicity observed in
the rodent studies is considered relevant to human
exposure.
No data to support range of human
variability /sensitivity, including whether children are
more sensitive.
May 2009
110
DRAFT - DO NOT CITE OR QUOTE
-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE
6.1. HUMAN HAZARD POTENTIAL
1 1,4-Dioxane is absorbed rapidly following oral and inhalation exposure, with much less
2 absorption occurring from the dermal route. 1,4-Dioxane is primarily metabolized to HEAA,
3 which is excreted in the urine. Liver and kidney toxicity are the primary noncancer health
4 effects associated with exposure to 1,4-dioxane in humans and laboratory animals. Several fatal
5 cases of hemorrhagic nephritis and centrilobular necrosis of the liver were related to
6 occupational exposure (i.e., inhalation and dermal contact) to 1,4-dioxane (Johnstone, 1959;
7 Barber, 1934). Neurological changes were also reported in one case, including headache,
8 elevation in blood pressure, agitation and restlessness, and coma (Johnstone, 1959). Perivascular
9 widening was observed in the brain of this worker, with small foci of demyelination in several
10 regions (e.g., cortex, basal nuclei). Severe liver and kidney degeneration and necrosis were
11 observed frequently in acute oral and inhalation studies (> 1,000 mg/kg-day oral, > 1,000 ppm
12 inhalation) (JBRC, 1998b; Drew et al., 1978; David, 1964; Kesten et al., 1939; Laug et al., 1939;
13 Schrenk and Yant, 1936; de Navasquez, 1935; Fairley et al., 1934).
14 Liver and kidney toxicity were the primary noncancer health effects of subchronic and
15 chronic oral exposure to 1,4-dioxane in animals. Hepatocellular degeneration and necrosis were
16 observed (Kociba et al., 1974) and preneoplastic changes were noted in the liver following
17 chronic administration of 1,4-dioxane in drinking water (JBRC, 1998a, Argus et al., 1973).
18 Liver and kidney toxicity appear to be related to saturation of clearance pathways and an
19 increase in the 1,4-dioxane concentration in the blood (Kociba et al., 1975). Kidney damage was
20 characterized by degeneration of the cortical tubule cells, necrosis with hemorrhage, and
21 glomerulonephritis (NCI, 1978; Kociba et al., 1974; Argus et al., 1973, 1965; Fairley et al.,
22 1934).
23 Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and
24 guinea pigs (JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Torkelson et al., 1974; Argus et al.,
25 1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al., 1965). Liver tumors
26 (hepatocellular adenomas and carcinomas) have been observed following drinking water
27 exposure in several species and strains of rats, mice, and guinea pigs. Nasal (squamous cell
28 carcinomas), peritoneal, and mammary tumors were also observed in rats, but were not seen in
29 mice. With the exception of the NCI (1978) study, the incidence of nasal cavity tumors was
30 generally lower than that of liver tumors in the same study population.
31 Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), 1,4-dioxane
32 can be classified as likely to be carcinogenic to humans, based on adequate evidence of liver
May 2009 111 DRAFT - DO NOT CITE OR QUOTE
-------
1 carcinogenicity in several 2-year bioassays conducted in three strains of rats, two strains of mice,
2 and in guinea pigs (JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Argus et al., 1973; Hoch-
3 Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al., 1965). Studies in humans found
4 no conclusive evidence for a causal link between occupational exposure to 1,4-dioxane and
5 increased risk for cancer; however, only two studies were available and these were limited by
6 small cohort size and a small number of reported cancer cases (Buffler et al., 1978; Thiess et al.,
7 1976).
8 The available evidence is inadequate to establish a MOA by which 1,4-dioxane induces
9 liver tumors in rats and mice. The genotoxicity data for 1,4-dioxane is generally characterized as
10 negative, although several studies may suggest the possibility of genotoxic effects (Roy et al.,
11 2005; Morita and Hayashi, 1998; Mirkova, 1994; Kitchin and Brown, 1990; Galloway et al.,
12 1987). A MOA hypothesis involving sustained proliferation of spontaneously transformed liver
13 cells has some support by evidence that suggests 1,4-dioxane is a tumor promoter in mouse skin
14 and rat liver bioassays (Lundberg et al., 1987; King et al., 1973). Dose-response and temporal
15 evidence support the occurrence of cell proliferation and hyperplasia prior to the development of
16 liver tumors (JBRC, 1998a; Kociba et al., 1974). However, the dose-response relationship for
17 the induction of hepatic cell proliferation has not been characterized, and it is unknown if it
18 would reflect the dose-response relationship for liver tumors in the 2-year rat and mouse studies.
19 Conflicting data from rat and mouse bioassays (JBRC, 1998a; Kociba et al., 1974) suggest that
20 cytotoxicity is not a required precursor event for 1,4-dioxane-induced cell proliferation. Data
21 regarding a plausible dose response and temporal progression from cytotoxicity to cell
22 proliferation and eventual liver tumor formation are not available.
6.2. DOSE RESPONSE
6.2.1. Noncancer/Oral
23 The RfD of 3 x 10"2 mg/kg-day was derived based on liver and kidney toxicity in rats
24 exposed to 1,4-dioxane in the drinking water for 2 years (Kociba et al., 1974). This study was
25 chosen as the critical study because it provides the most sensitive measure of adverse effects by
26 1,4-dioxane. The incidence of liver and kidney lesions was not reported for each dose group.
27 Therefore, BMD modeling could not be used to derive a POD. Instead, the RfD is derived by
28 dividing the NOAEL of 9.6 mg/kg-day by a composite UF of 300 (factors of 10 for animal-to-
29 human extrapolation and interindividual variability, and a partial UF of 3 for database
30 deficiencies). Information was unavailable to quantitatively assess toxicokinetic or
31 toxicodynamic differences between animals and humans and the potential variability in human
32 susceptibility; thus, the interspecies and intraspecies uncertainty factors of 10 were applied. In
33 addition, a threefold database uncertainty factor was applied due to the lack of information
34 addressing the potential reproductive toxicity associated with 1,4-dioxane.
May 2009 112 DRAFT - DO NOT CITE OR QUOTE
-------
1 The overall confidence in this RfD assessment is medium. Confidence in the principal
2 study (Kociba et al., 1974) is medium. Confidence in the database is medium due to the lack of a
3 multigeneration reproductive toxicity study. Reflecting medium confidence in the principal
4 study and medium confidence in the database, confidence in the RfD is medium.
6.2.2. Noncancer/Inhalation
5 No inhalation RfC was derived for 1,4-dioxane. Inhalation data were inadequate and a
6 route extrapolation from oral toxicity data was not performed, due to direct effects of
7 1,4-dioxane on the respiratory tract (i.e., respiratory irritation in humans, pulmonary congestion
8 in animals) (Wirth and Klimmer, 1936; Fairley et al., 1934; Yant et al., 1930) and lack of a
9 suitable kinetic model (see Appendix B).
6.2.3. Cancer/Oral
10 An oral CSF for 1,4-dioxane of 0.19 (mg/kg-day)"1 was based on liver tumors in female
11 mice from a chronic study (JBRC, 1998a). Because the MOA for liver carcinogenicity of
12 1,4-dioxane is not known, the CSF was derived by linear low-dose extrapolation. The POD was
13 calculated by curve fitting the experimental dose-response data from the POD, the range of
14 observation (BMDLio HED of 0.52 mg/kg-day).
15 The uncertainties associated with the quantitation of the oral CSF are discussed below.
6.2.3.1. Choice of Low-Dose Extrapolation Approach
16 The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
17 1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible
18 MOA(s) for the observed tumors. The MOA is a key consideration in clarifying how risks
19 should be estimated for low-dose exposure. Exposure to 1,4-dioxane has been observed in
20 animal models to induce multiple tumor types, including liver adenomas and carcinomas, nasal
21 carcinomas, mammary adenomas and flbroadenomas, and mesothiolomas of the peritoneal cavity
22 (JBRC, 1998a). MOA information that is available for the carcinogenicity of 1,4-dioxane has
23 largely focused on liver adenomas and carcinomas, with little or no MOA information available
24 for the remaining tumor types. In Section 4.7.3, hypothesized MO As, other than a mutagenic
25 MOA, were explored due to the lack of mutagenicity observed in genetic toxicology tests
26 performed for 1,4-dioxane. Data are not available to support a carcinogenic MOA for
27 1,4-dioxane. In the absence of a MOA(s) for the observed tumor types due to exposure to
28 1,4-dioxane, a linear low-dose extrapolation approach was used to estimate human carcinogenic
29 risk associated with 1,4-dioxane exposure.
30 The extent to which the overall uncertainty in low-dose risk estimation could be reduced
31 if the MOA for 1,4-dioxane were known is of interest, but additional supporting data on the
32 MOA(s) of 1,4-dioxane is not available. Even if it were available, incorporation of MOA into
May 2009 113 DRAFT - DO NOT CITE OR QUOTE
-------
1 dose-response modeling might not be straightforward and might not significantly reduce the
2 uncertainty about low-dose extrapolation.
3 In general, the Agency has preferred to use the multistage model for analyses of tumor
4 incidence and related endpoints because they have a generic biological motivation based on
5 long-established mathematical models such as the MVK model. The MVK model does not
6 necessarily characterize all modes of tumor formation, but it is a starting point for most
7 investigations and, much more often than not, has provided at least an adequate description of
8 tumor incidence data.
9 In the studies evaluated (JBRC 1998a; NCI, 1978; Kociba et al., 1974) the multistage
10 model provided good descriptions of the incidence of many tumor types in male and female rats
1 1 and in male mice exposed to 1,4-dioxane (JBRC, 1998a). However, the multistage model did
12 not provide an adequate fit for female mouse liver tumor dataset based upon the following (U.S.
13 EPA, 2000b):
Goodness-of-fit -value > 0.10;
AIC less than any other competing models, even if the alternative models did not have a
biological motivation;
No data greatly deviating from the fitted model, as measured by their j^ residuals.
14 BMDS software typically implements the guidance in the BMD technical guidance
15 document (U.S. EPA, 2000b) by imposing constraints on the values of certain parameters of the
16 models. When these constraints were imposed, the multistage model and most other models did
17 not fit the incidence data for female mouse liver adenomas or carcinomas.
18 The log-logistic model provided an adequate fit for the female mouse data (JBRC,
19 1998a). Additionally, a log-logistic model with a slope of 1, as is the case for the female mouse
20 data (JBRC, 1998a), represents a low-dose linear extrapolation that is consistent with Agency
21 guidance (U.S. EPA, 2005a). Therefore, the log-logistic model was selected, with the BMDLio
22 derived by applying the constraints, as consistent with recommended use of BMDS in the BMD
23 technical guidance document (U.S. EPA, 2000b).
24 The human equivalent oral CSF estimated from liver tumor datasets with statistically
25 significant increases ranged from 2.58 x 10"4 to 0. 19 per mg/kg-day, a range of about three
26 orders of magnitude, with the extremes coming from the combined male and female data for
27 nasal carcinomas (Kociba et al., 1974) and the female mouse liver adenoma and carcinoma
28 dataset (JBRC, 1998a).
6.2.3.2. Dose Metric
29 1,4-Dioxane is known to be metabolized in vivo. However, it is unknown whether a
30 metabolite or the parent compound, or some combination of parent compound and metabolites, is
3 1 responsible for the observed toxicity. If the actual carcinogenic moiety is proportional to
May 2009 1 14 DRAFT - DO NOT CITE OR QUOTE
-------
1 administered exposure, then use of administered exposure as the dose metric is the least biased
2 choice. On the other hand, if this is not the correct dose metric, then the impact on the CSF is
3 unknown.
6.2.3.3. Cross-Species Scaling
4 An adjustment for cross-species scaling (BW°75) was applied to address toxicological
5 equivalence of internal doses between each rodent species and humans, consistent with the 2005
6 Guidelines for Carcinogen Risk Assessment (US EPA, 2005a). It is assumed that equal risks
7 result from equivalent constant lifetime exposures.
6.2.3.4. Statistical Uncertainty at the POD
8 Parameter uncertainty can be assessed through confidence intervals. Each description of
9 parameter uncertainty assumes that the underlying model and associated assumptions are valid.
10 For the log-logistic model applied to the female mouse data, there is a reasonably small degree of
11 uncertainty at the 10% excess incidence level (the POD for linear low-dose extrapolation).
6.2.3.5. Bioassay Selection
12 The study by JBRC (1998a) was used for development of an oral CSF. This was a well-
13 designed study, conducted in both sexes in two species with a sufficient number of animals per
14 dose group. The number of test animals allocated among three dose levels and an untreated
15 control group was adequate, with examination of appropriate toxicological endpoints in both
16 sexes of rats and mice. Alternative bioassays (NCI, 1978; Kociba et al., 1974) are available and
17 were fully considered for the derivation of the oral CSF.
6.2.3.6. Choice of Species/Gender
18 The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the
19 female mouse, which was thought to be more sensitive than male mice or either sex of rats to the
20 carcinogenicity of 1,4-dioxane. While all data, both species and sexes reported from the JBRC
21 (1998a) study, were suitable for deriving an oral CSF, the female mouse data represented the
22 most sensitive indicator of carcinogenicity in the rodent model. The lowest exposure level
23 (77 mg/kg-day or 12 mg/kg-day [HED]) observed a considerable and significant increase in
24 combined liver adenomas and carcinomas. Additional testing of doses within the range of
25 control and the lowest dose (77 mg/kg-day or 12 mg/kg-day [HED]) could refine and reduce
26 uncertainty for the oral CSF.
6.2.3.7. Relevance to Humans
27 The derivation of the oral CSF is derived using the tumor incidence in the liver of female
28 mice. A thorough review of the available toxicological data available for 1,4-dioxane provides
29 no scientific justification to propose the liver adenomas and carcinomas observed in animal
30 models due to exposure to 1,4-dioxane are not plausible in humans. Liver adenomas and
31 carcinomas were considered as a plausible outcome in humans due to exposure to 1,4-dioxane.
May 2009 115 DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.8. Human Population Variability
1 The extent of inter-individual variability in 1,4-dioxane metabolism has not been
2 characterized. A separate issue is that the human variability in response to 1,4-dioxane is also
3 unknown. Although a mutagenic MOA would indicate increased early-life susceptibility, the
4 data exploring whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life
5 stages is unavailable. This lack of understanding about potential differences in metabolism and
6 susceptibility across exposed human populations thus represents a source of uncertainty. Also,
7 the lack of information linking a MOA for 1,4-dioxane to the observed carcinogenicity is a
8 source of uncertainty.
6.2.4. Cancer/Inhalation
9 Inhalation studies for 1,4-dioxane were not adequate for the determination of an
10 inhalation unit risk value. No treatment-related tumors were noted in a chronic inhalation study
11 in rats; however only a single exposure concentration was used (111 ppm 1,4-dioxane vapor for
12 7 hours/day, 5 days/week for 2 years) (Torkelson et al., 1974). Route extrapolation from oral
13 bioassay data was not performed because available kinetic models were not considered suitable
14 for this purpose.
May 2009 116 DRAFT - DO NOT CITE OR QUOTE
-------
7. REFERENCES
ATSDR (Agency for Toxic Substances and Disease Registry). (2007) Toxicological profile for 1,4-dioxane. Draft
for public comment. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service. Available
online at http://www.atsdr.cdc.gov/toxpro2.html.
ATSDR. (2006) Toxicological profile for 1,1,1-trichloroethane. Atlanta, GA: U.S. Department of Health and
Human Services, Public Health Service. Available online at http://www.atsdr.cdc.gov/toxpro2.html.
Agrawal, AK; Shapiro, BH. (2000) Differential expression of gender-dependent hepatic isoforms of cytochrome P-
450 by pulse signals in the circulating masculine episodic growth hormone profile of the rat. J Pharmacol Exp
Therap 292:228-237.
Andersen, ME; Clewell, HJ,III; Gargas, ML; et al. (1987) Physiologically-based pharmacokinetics and the risk
assessment process for methylene chloride. Toxicol Appl Pharmacol 87:185-205.
Argus, MF; Arcos, JC; Hochligeti, C. (1965) Studies on the carcinogenic activity of protein-denaturing agents:
hepatocarcinogenicity of dioxane. J Natl Cancer Inst 35:949-958.
Argus, MF; Sohal, RS; Bryant, GM; et al. (1973) Dose-response and ultrastructural alterations in dioxane
carcinogenesis: Influence of methylcholanthrene on acute toxicity. Eur J Cancer 9:237-243.
Ashby, J. (1994) Series: current issues in mutagenesis and carcinogenesis, No. 45—The genotoxicity of 1,4-dioxane.
MutatRes 322(2): 141-150.
Atkinson, R. (1989) Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic
compounds. JPhys ChemRef Data. Monograph No. 1 p. 143.
Bannasch P. (2003) Comments on R. Karbe and R.L. Kerlin (2002) Cystic degeneration/spongiosis hepatic (Toxicol
Pathol 30 (2), 216-227). Toxicol Pathol 31:566-570.
Barber, H. (1934) Haemorrhagic nephritis and necrosis of the liver from dioxane poisoning. Guys Hosp Rep
84:267-280.
Braun, WH; Young, JD. (1977) Identification of B-hydroxyethoxyacetic acid as the major urinary metabolite of
1,4-dioxane in the rat. Toxicol Appl Pharmacol 39:33-38.
Bronaugh, RL. (1982) Percutaneous absorption of cosmetic ingredients. In: Frost, P; Horwitz, SN, eds. Principles
of cosmetics for the dermatologist. Minneapolis, MN: University of Minnesota Press, 277-284.
Brown, RP; Delp, MD; Lindstedt, SL; et al. (1997) Physiological parameter values for physiologically based
pharmacokinetic models. Toxicol Ind Health 13(4):407-484.
Buffler, PA; Wood, SM; Suarez, L; et al. (1978) Mortality follow-up of workers exposed to 1,4-dioxane. J Occup
Med 20(4):255-259.
Bull, PJ; Robinson, M; Laurie, RD. (1986) Association of carcinoma yield with early papilloma development in
SENCAR mice. Environ Health Perspect 68:11-17.
Burmistrov, SO; Arutyunyan, AV; Stepanov, MG; et al. (2001) Effect of chronic inhalation of toluene and dioxane
on activity of free radical processes in rat ovaries and brain. Bull Exp Biol Med 132(3):832-836.
Carpenter, SP; Lasker, JM; Raucy, JL. (1996) Expression, induction, and catalytic activity of the ethanol-inducible
cytochrome P450 (CYP2E1) in human fetal liver and hepatocytes. Mol Pharmacol 49:260-268.
Clark, B; Furlong, JW; Ladner, A; et al. (1984) Dermal toxicity of dimethyl acetylene dicarboxylate, n-methyl
pyrrolidone, triethylene glycol dimethyl ether, dioxane and tetraline in the rat. IRCS J Med Sci 12:296-297.
May 2009 117 DRAFT - DO NOT CITE OR QUOTE
-------
David, H. (1964) [Electron-microscopic findings in dioxane-dependent nephrosis in rat kidneys] Beitr Pathol Anat
130:187-212.
de Navasquez, S. (1935) Experimental tubular necrosis of the kidneys accompanied by liver changes due to dioxane
poisoning. JHyg 35:540-548.
Derosa, CT; Wilbur, S; Holler, J; et al. (1996) Health evaluation of 1,4-dioxane. Toxicol Ind Health 12(l):l-43.
Drew, RT; Patel, JM; Lin, F. (1978) Changes in serum enzymes in rats after inhalation of organic solvents singly
and in combination. Toxicol Appl Pharmacol 45:809-819.
Ernstard, L; Iregren, A; Sjogren, B; et al. (2006) Acute effects of exposure to vapours of dioxane in humans. Human
Experimental Toxicol 25:723-729.
Fairley, A; Linton, EC; Ford-Moore, AH. (1934) The toxicity to animals of 1,4-dioxane. J Hyg 34:486-501.
Fisher, J; Mahle, D; Bankston, L; et al. (1997) Lactational transfer of volatile chemicals in breast milk. Am Ind Hyg
AssocJ58(6):425-431.
Franke, C; Studinger, G; Berger, G; et al. (1994) The assessment of bioaccumulation. Chemosphere 29:1501-1514.
Frantik, R; Hornychova, M; Horvath, M. (1994) Relative acute neurotoxicity of solvents: isoeffective air
concentrations of 48 compounds evaluated in rats and mice. Environ Res 66:173-185.
Galloway, SM; Armstrong, MJ; Reuben, C; et al. (1987) Chromosome aberrations and sister chromatid exchanges in
Chinese hamster ovary cells: evaluations of 108 chemicals. Environ Mol Mutagen 10(supp 10): 1-175.
Giavini, E; Vismara, C; Broccia, ML. (1985) Teratogenesis study of dioxane in rats. Toxicol Lett 26:85-88.
Goldberg, ME; Johnson, HE; Pozzani, UC; et al. (1964) Effect of repeated inhalation of vapors of industrial solvents
on animal behavior. I. Evaluation of nine solvent vapors on pole-climb performance in rats. Am Ind Hyg Assoc J
25:369-375.
Goldsworthy, TL; Monticello, TM; Morgan, KT; et al. (1991) Examination of potential mechanisms of
carcinogenicity of 1,4-dioxane in rat nasal epithelial cells and hepatocytes. Arch Toxicol 65:1-9.
Green, T., Lee, R., Moore, R., et al. (2000) Acetochlor-induced rat nasal tumors: Further studies on the mode of
action and relevance to humans.
Grosjean, D. (1990) Atmospheric chemistry of toxic contaminants. 2. Saturated aliphatics:Acetaldehyde, dioxane,
ethylene glycol ethers, propylene oxide. J Air Waste Mgmt Assoc 40:1522-1531.
Hansch, C; Leo, A; Hoekman, D. (1995) Exploring QSAR. Hydrophobic, electronic, and steric constants. ACS
Professional Reference Book. Washington, DC: American Chemical Society., p. 9.
Haseman, JK; Hailey, JR, (1997) An update of the National Toxicology Program database on nasal carcinogens.
Mutat Res 380:3-11.
Haworth, S; Lawlor, T; Mortelmans, K; et al. (1983) Salmonella mutagenicity test results for 250 chemicals.
Environ Mutagen supp. 1:1, 10.
Hayashi, S; Watanabe, J; Kawajiri, K. (1991) Genetic polymorphisms in the 5'-flanking region change
transcriptional regulation of the human cytochrome P450IIE1 gene. J Biochem 110:559-565.
Hellmer, L; Bolcsfoldi, G. (1992) An evaluation of the E coli K-12 uvrB/recA DNA repair host- mediated assay I In
vitro sensitivity of the bacteria to 61 compounds. Mutat Res 272:145-160.
May 2009 118 DRAFT - DO NOT CITE OR QUOTE
-------
Hoch-Ligeti, C; Argus, MF. (1970) Effect of carcinogens on the lung of guinea pigs. In: Nettesheim, P; Hanna,
MG; Deatherage, JW, eds. Morphology of experimental respiratory carcinogenesis. AEC Symp Ser21National
Cancer Institute and U.S. Atomic Energy Commission, pp. 267-279. CONF700501.
Hoch-Ligeti, C; Argus. MF; Arcos, JC. (1970) Induction of carcinomas in the nasal cavity of rats by dioxane. Br J
Cancer 24(1): 164-167.
HSDB (Hazardous Substances Data Bank) (2007) 1,4-dioxane. Last update dated 4/14/2006. National Library of
Medicine, National Toxicology Program, Bethesda, Maryland.
Huang, CY; Huang, KL; Chong, TI; et al. (1997) The GSTT1 and CYP2E1 genotypes are possible factors causing
vinyl chloride induced abnormal liver function. Arch Toxicol 71:482-488.
IARC (International Agency for Research on Cancer). (1999) Some chemicals that cause tumours of the kidney or
urinary bladder in rodents and some other substances- 1,4-dioxane (group 2B). IARC Monogr Eval Carcinog Risks
Hum 71 Lyon, France: World Health Organization, p. 589. Available online at http://www-
cie.iarc.fr/htdocs/monographs/vol71/019-dioxane.html.
ICRP (International Commission of Radiological Protection). (1975) No. 23: Report of the Task Group on
Reference Man. Pergamon Press, New York, NY.
JBRC. (1998a) Two-year studies of 1,4-dioxane in F344 rats and BDFi mice (drinking water). Kanagawa, Japan:
Japan Bioassay Research Center.
JBRC. (1998b) Two-week studies of 1,4-dioxane inF344 rats and BDFi mice (drinking water studies). Kanagawa,
Japan: Japan Bioassay Research Center.
Johnstone, RT. (1959) Death due to dioxane? AMA Arch Ind Health 20:445^147.
Kanada, M; Miyagawa, M; Sato, M; et al. (1994) Neurochemical profile of effects of 28 neurotoxic chemicals on the
central nervous system in rats (1). Effects of oral administration on brain contents of biogenic amines and
metabolites. Ind Health 32:145-164.
Kano, J; Umeda, Y; Saito, M; et al. (2008) Thirteen-week oral toxicity of 1,4-dioxane in rats and mice. J Toxicol
Sci33(2):141-153.
Karbe, E; Kerlin, RL. (2002) Cystic degeneration/spongiosis hepatic in rats. Toxicol Pathol 30(2):216-227.
Kasper, P., Uno, Y., Mauthe, R. et al. (2007) Follow-up testing of rodent carcinogens not positive in the standard
genotoxicity testing battery: IWGT workgroup report. Mut Res 627:106-116.
Kesten, HD; Mulinos, MG; Pomerantz L. (1939) Pathologic effects of certain glycols and related compounds. Arch
Pathol 27:447-465.
Khudoley, W; Mizgireuv, I; Pliss, GB. (1987) The study of mutagenic activity of carcinogens and other chemical
agents with Salmonella typhimurium assays: testing of 126 compounds. Arch Geschwulstforsch 57:453^-62.
King, ME; Shefner, AM; Bates, RR. (1973) Carcinogenesis bioassay of chlorinated dibenzodioxins and related
chemicals. Environ Health Perspect 5:163-170.
Kitchin, KT; Brown, JL. (1990) Is 1,4-dioxane a genotoxic carcinogen? Cancer Lett 53(1):67-71.
Knoefel, PK. (1935) Narcotic potency of some cyclic acetals. J Pharmacol Exp Ther 53:440-444.
Kociba, RJ; McCollister, SB; Park, C; et al. (1974) 1,4-dioxane. I. Results of a 2-year ingestion study in rats.
Toxicol Appl Pharmacol 30:275-286.
May 2009 119 DRAFT - DO NOT CITE OR QUOTE
-------
Kociba, RJ; Torkelson, TR; Young, JD; etal. (1975) 1,4-dioxane: correlation of the results of chronic ingestionand
inhalation studies with its dose-dependent fate in rat. In: Proceedings of the 6th Annual Conference on
Environmental Toxicology. October 21-23, Aerospace Medical Research Laboratory, Wright-Patterson Air Force
Base, Ohio, pp. 345-354.
Kurl, RN; Poellinger, L; Lund, J;etal.(1981) Effects of dioxane on RNA synthesis in the rat liver. Arch Toxicol
49(l):29-33.
Kwan, KK; Dutka, BJ; Rao, SS; et al. (1990) Mutatox test: a new test for monitoring environmental genotoxic
agents. EnvironPollut65:323-332.
Laug, EP; Calvery, HO; Morris, HJ; et al. (1939) The toxicity of some glycols and derivatives. J Ind Hyg Toxicol
21:173-201.
Lesage S; Jackson RE; Priddle MW; et al. (1990) Occurrence and fate of organic solvent residues in
anoxic groundwater at the Gloucester Landfill, Canada. Environ Sci Technol 24:559-565.
Lewis, RJ, ed. (2000) Sax's dangerous properties of industrial materials. 10th edition. New York, NY: John Wiley
&Sons, Inc., p.1509.
Lewis, RJ, ed. (2001) Hawley's condensed chemical dictionary. 14th edition. New York, NY: John Wiley & Sons,
Inc., p. 406.
Lide, DR, ed. (2000) CRC handbook of chemistry and physics. 81st edition. Boca Raton, FL: CRC Press LLC, pp.
3-144.
Leung, H-W; Paustenbach, DJ. (1990) Cancer risk assessment for dioxane based upon a physiologically-based
pharmacokinetic approach. Toxicol Lett 51:147-162.
Lundberg, I; Ekdahl, M; Kronevi, T; et al. (1986) Relative hepatotoxicity of some industrial solvents after
intraperitoneal injection or inhalation exposure in rats. Environ Res 40:411-420.
Lundberg, I; Hogberg, J; Kronevi, T; et al. (1987) Three industrial solvents investigated for tumor promoting
activity in the rat liver. Cancer Lett 36:29-33.
Lyman, WJ; Reehl, WF; Rosenblatt; et al. (1990) Handbook of chemical property estimation methods. Washington,
DC: American Chemical Society, pp. 4-9.
Marzulli, F; Anjo, DM; Maibach, HI. (1981) In vivo skin penetration studies of 2,4-toluenediamine,
2,4-diaminoanisole, 2-nitro-p-phenylenediamine, p-dioxane and N-nitrosodiethanolamine in cosmetics. Food
Cosmet Toxicol 19:743-747.
McFee, AF; Abbott, MG; Gulati, DK; et al. (1994) Results of mouse bone marrow micronucleus studies on
1,4-dioxane. Mutat Res 322:141-150.
McGregor, DB; Brown, AG; Howgate, S; et al. (1991) Responses of the L5178Y mouse lymphoma cell forward
mutation assay. V. 27 coded chemicals. Environ Mol Mutagen 17(3): 196-219.
Meylan, WM; Howard, PH; Boethling, RS; et al. (1999) Improved method for estimating bioconcentration/
bioaccumulation factor from octanol/water partition coefficient. Environ Toxicol Chem 18:664-672.
Mikheev, MI; Gorlinskaya, YE P; Solovyova, TV. (1990) The body distribution and biological action of
xenobiotics. J Hyg Epidemiol Microbiol Immunol 34:329-336.
Mirkova, ET. (1994) Activity of the rodent carcinogen 1,4-dioxane in the mouse bone marrow micronucleus assay.
Mutat Res 322:142-144.
May 2009 120 DRAFT - DO NOT CITE OR QUOTE
-------
Miyagawa, M; Shirotori, T; Tsuchitani, M; et al. (1999) Repeat-assessment of 1,4-dioxane in a rat-hepatocyte
replicative DNA synthesis (RDS) test: evidence for stimulus of hepatocyte proliferation. Exp Toxicol Pathol
51:555-558.
Morita, T. (1994) No clastogenicity of 1,4-dioxane as examined in the mouse peripheral blood micronucleus test.
MMSCommun 2:7-8.
Morita, T; Hayashi, M. (1998) 1,4-dioxane is not mutagenic in five in vitro assays and mouse peripheral blood
micronucleus assay, but is in mouse liver micronucleus assay. Environ Mol Mutagen 32:269-280.
Mungikar, AM; Pawar, SS. (1978) Induction of the hepatic microsomal mixed function oxidase system in mice by
p-dioxane. Bull Environ Contam Toxicol 20:797-804.
Munoz, ER; Barnett, BM. (2002) The rodent carcinogens 1,4-dioxane and thiourea induce meiotic non-disjunction
inDrosophila melanogaster females. Mutat Res 517(l-2):231-238.
Nannelli, A; De Rubertis, A; Longo, V; et al. (2005) Effects of dioxane on cytochrome P450 enzymes in liver,
kidney, lung and nasal mucosa of rat. Arch Toxicol 79(2):74-82.
NCI (National Cancer Institute). (1978) Bioassay of 1,4-dioxane for possible carcinogenicity. Bethesda, MD:
National Cancer Institute. NIH Pub. No. 78-1330 NCICGTR-80.
Nelson, N. (1951) Solvent toxicity with particular reference to certain octyl alcohols and dioxanes. Med Bull
11:226-238.
Nestmann, ER; Otson, R; Kowbel, DJ; et al. (1984) Mutagenicity in a modified Salmonella assay of fabric-
protecting products containing 1,1,1-trichloroethane. Environ Mutagen 6:71-80.
NRC (National Research Council). (1983) Risk assessment in the Federal government: managing the process.
Committee on the Institutional Means for Assessment of Risks to Public Health, Commission on Life Sciences,
NRC. Washington, DC; National Academy Press.
O'Neil, MJ, ed. (2001) The Merck index - An encyclopedia of chemicals, drugs, and biologicals. 13th edition.
Whitehouse Station, NJ: Merck and Co., Inc., p. 581.
Park, JH; Hussam, A; Couasnon, P; et al. (1987) Experimental reexamination of selected partition coefficients from
Rohrschneider's data set. Anal Chem 59:1970-1976.
Platz, J., Megelberg, T., Nielson, O., et al. (1997) Atmospheric chemistry of 1,4-dioxane. J Chem Soc
Faraday Trans 93:2855-2863.
Pozzani, UC; Weil, CS; Carpenter, CP. (1959) The toxicological basis of threshold limit values: 5: The experimental
inhalation of vapor mixtures by rats, with notes upon the relationship between single dose inhalation and single dose
oral data. JIndHyg 20:364-369.
Ramsey, JC; Andersen, ME. (1984) A physiologically based description of the inhalation pharmacokinetics of
styrene in rats and humans. Toxicol Appl Pharmacol 73:159-175.
Reitz, RH; McCroskey, PS; Park, CN; et al. (1990) Development of a physiologically based pharmacokinetic model
for risk assessment with 1,4-dioxane. Toxicol Appl Pharmacol 105:37-54.
Rosenkranz, HS; Klopman, G. (1992) 1,4-dioxane: prediction of in vivo clastogenicity. Mutat Res 280:245-251.
Roy, SK; Thilagar, AK; Eastmond, DA. (2005) Chromosome breakage is primarily responsible for the micronuclei
induced by 1,4-dioxane in the bone marrow and liver of young GDI mice. Mutation Research 586:28-37.
Schrenk, HH; Yant, WP. (1936) Toxicity of dioxane. J Ind Hyg Toxicol 18:448-460.
May 2009 121 DRAFT - DO NOT CITE OR QUOTE
-------
Sheu, CW; Moreland, FM; Lee, JK; et al. (1988) In vitro BALB/3T3 cell transformation assay of nonoxynol-9 and
1,4-dioxane. Environ Mol Mutagen 11:41-48.
Silverman, L; Schulte, HF; First, MW. (1946) Further studies of sensory response to certain industrial solvent
vapors. J Ind Hyg Toxicol 28:262-266.
Sina, JF; Bean, CL; Dysart, GR; et al. (1983) Evaluation of the alkaline elution/rat hepatocyte assay as a predictor of
carcinogenic/mutagenicpotential. MutatRes 113:357-391.
Smyth, HFJ; Seaton, J; Fischer, L. (1941) The single dose toxicity of some glycols and derivatives. J Ind Hyg
Toxicol 23(6):259-268.
Stoner, GD, Conran, PB; Greisiger, EA; et al. (1986) Comparison of two routes of chemical administration on the
lung adenoma response in strain A/J mice. Toxicol Appl Pharmacol 82:19-31.
Stott, WT; Quast, JF; Watanabe, PG. (1981) Differentiation of the mechanisms of oncogenicity of 1,4-dioxane and
1,3-hexachlorobutadiene in the rat. Toxicol Appl Pharmacol 60:287-300.
Stroebel, P; Mayer, F; Zerban, H; et al. (1995) Spongiotic pericytoma: a benign neoplasm deriving from the
perisinusoidal (Ito) cells in rat liver. Am J Pathol 146(4):903-913.
Suprenant, KS. (2002) Dioxane. In: Ullmann's encyclopedia of industrial chemistry. Article Online Posting Date
June 15, 2000. Available from subscription to Wiley Interscience: Wiley-VCH Verlag GmbH and Co. KGaA, and
online at http://www.mrw.interscience.wiley.com/ueic/ueic_search_fs.html.
Sweeney, L., Thrall, K., Poet, T., et al. (2008) Physiologically based pharmacokinetic modeling of 1,4-dioxane in
rats, mice, and humans. Toxicol Sci 101(1):32-50.
Thiess, AM; Tress, E; Fleig, I. (1976) [Industrial-medical investigation results in the case of workers exposed to
dioxane.] Arbeitsmed Sozialmed Pravent 11:35-46.
Thurman, GB; Simms, BG; Goldstein, AL; et al. (1978) The effects of organic compounds used in the manufacture
of plastics on the responsivity of murine and human lymphocytes. Toxicol Appl Pharmacol 44(3):617-641.
Tinwell, H; Ashby, J. (1994) Activity of 1,4-dioxane in mouse bone marrow micronucleus assays. Mutat Res
322:148-150.
Torkelson, R; Leong, BKJ; Kociba, RJ; et al. (1974) 1,4-dioxane. II. Results of a 2-year inhalation study in rats.
Toxicol Appl Pharmacol 30:287-298.
UNEP (United Nations Environment Programme). (2000) The Montreal Protocol on substances that deplete the
ozone layer. United Nations Office at Nairobi (UNON), Nairobi, Kenya. Available online at
http://www.unep.org/ozone/Treaties_and_Ratification/2B_montreal_protocol.asp.
Uno, Y; Takaswas, H; Miyagawa, M; et al. (1994) An in vivo-in vitro replications DNA synthesis (RDS) test using
rat hepatocytes as an early prediction assay for nongenotoxic hepatocarcinogens: Screening of 22 known positives
and 25 noncarcinogens. Mutat Res 320(3): 189-205.
U.S. EPA (U.S. Environmental Protection Agency). (1986a) Guidelines for the health risk assessment of chemical
mixtures. Federal Register 51(185):34014-34025.
U.S. EPA. (1986b) Guidelines for mutagenicity risk assessment. Federal Register 51(185):34006-34012.
U.S. EPA (1986c). Guidelines for Carcinogen Risk Assessment. Report #EPA/630/R-00/004.
http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=54933 Federal Register 51(185):33992-34003
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.
May 2009 122 DRAFT - DO NOT CITE OR QUOTE
-------
U.S. EPA. (1990) Amendments to the Clean Air Act. Sec. 604. Phase-out of Production and Consumption of Class I
Substances. Available online at http://www.epa.gov/oar/caa/caa604.txt.
U.S. EPA. (1991) Guidelines for developmental toxicity risk assessment. Federal Register 56(234):63798-63826.
U.S. EPA. (1994a) Interim policy for particle size and limit concentration issues in inhalation toxicity studies.
Federal Register 59(206):53799.
U.S. EPA. (1994b) 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 online at 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 online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (1996) Guidelines for reproductive toxicity risk assessment. Federal Register 61(212):56274-56322.
U.S. EPA. (1998) Guidelines for neurotoxicity risk assessment. Federal Register 63(93):26926-26954.
U.S. EPA. (2000a) 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 online at
http://www.epa. gov/iris/backgr-d.htm.
U.S. EPA. (2000b) Benchmark dose technical guidance document [external review draft]. EPA/630/R-00/001.
Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (2000c) Supplemental guidance for conducting for health risk assessment of chemical mixtures.
EPA/630/R-00/002. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA. (2002a) A review of the reference dose and reference concentration processes. Risk Assessment Forum,
Washington, DC; EPA/630/P-02/0002F. Available online at http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (2002b) Toxic Substances Control Act (TSCA) Inventory Update Database. Available online at
http://www.epa.gov/oppt/iur/iur02/index.htm.
U.S. EPA. (2005a) Guidelines for carcinogen risk assessment. Risk Assessment Forum, Washington, DC;
EPA/630/P-03/001B. Available online at http://www.epa.gov/iris/backgr-d.htm
U.S. EPA. (2005b) Supplemental guidance for assessing susceptibility from early-life exposure to carcinogens. Risk
Assessment Forum, Washington, DC; EPA/630/R-03/003F. Available online at http://www.epa.gov/iris/backgr-
d.htm.
U.S. EPA. (2006a) Peer review handbook. 3rd edition. Science Policy Council, Washington, DC. Available online
at http://www.epa.gov/peerreview/pdfs/Peer%20Review%20HandbookMay06.pdf.
U.S. EPA. (2006b) A Framework for Assessing Health Risk of Environmental Exposures to Children. National
Center for Environmental Assessment, Washington, DC, EPA/600/R-05/093F. Available from:
.
van Delft, JHM; Van Agen, E; van Breda, SGJ; et al. (2004) Discrimination of genotoxic from non-genotoxic
carcinogens by gene expression profiling. Carcinogenesis 25(7): 1265-1276.
Vieira, I; Sonnier, M; Cresteil, T. (1996) Developmental expression of CYP2E1 in the human liver.
Hypermethylation control of gene expression during the neonatal period. Eur J Biochem 238:476-483.
Watanabe, J; Hayashi, S; Kawajiri, K. (1994) Different regulation and expression of the human CYP2E1 gene due to
the Rsal polymorphism in the 5'-flanking region. J Biochem 116:321-326.
May 2009 123 DRAFT - DO NOT CITE OR QUOTE
-------
Waxman, DJ; Pampori, NA; Prabha, AR. (1991) Interpulse interval in circulating growth hormone patterns regulates
sexually dimorphic expression of hepatic cytochrome P450. Proc Natl Acad Sci 88:6868-6872.
Wirth, W; Klimmer, O. (1936) [On the toxicology of organic solvents. 1,4-dioxane (diethylene dioxide).] Arch
Gewerbepathol Gewerbehyg 17:192-206.
Wolfe, NL; Jeffers PM. (2000) Hydrolysis. In: Boethling RS, Mackay D, eds. Handbook of property
estimation methods for chemicals: Environmental and health sciences. Boca Raton, FL: Lewis Publishers.
Wolford, ST; Schroer, RA; Gohs, FX; et al. (1986) Reference range data base for serum chemistry and hematology
values in laboratory animals. J Toxicol Environ Health 18:161-188.
Woo, Y-T; Argus, MF; Arcos, JC. (1977a) Tissue and subcellular distribution of 3H-dioxane in the rat and apparent
lack of microsome-catalyzed covalent binding in the target tissue. Life Sci 21(10): 1447-1456.
Woo, Y; Arcos, JC; Argus, MF; et al. (1977b) Structural identification of p-dioxane-2-one as the major urinary
metabolite of p-dioxane. Naunyn-Schmiedebergs Arch Pharmacol 299:283-287.
Woo, Y; Argus, MF; Arcos, JC. (1977c) Metabolism in vivo of dioxane: effect of inducers and inhibitors of hepatic
mixed-function oxidases. BiochemPharmacol 25:1539-1542.
Woo, YT; Argus, MF; Arcos, JC. (1978) Effect of mixed-function oxidase modifiers on metabolism and toxicity of
the oncogen dioxane. Cancer Res 38:1621-1625.
Woo, Y-T; Neuburger, BJ; Arcos, JC; et al. (1980) Enhancement of toxicity and enzyme-repressing activity of
p-dioxane by chlorination: stereoselective effects. Toxicol Lett 5(l):69-75.
Yamamoto, S; Urano, K; Nomura, T. (1998a) Validation of transgenic mice harboring the human prototype
c-Ha-ras gene as a bioassay model for rapid carcinogenicity testing. Toxicol Lett 102-103:473-478.
Yamamoto, S; Urano, K; Koizumi, H; et al. (1998b) Validation of transgenic mice carrying the human prototype
c-Ha-ras gene as a bioassay model for rapid carcinogenicity testing. Environ Health Perspect 106:57-69.
Yamazaki, K; Ohno, H; Asakura, M; et al. (1994) Two-year toxicological and carcingenesis studies of 1,4-dioxane
in F344 rats and BDFi mice. In: Sumino K, Sato S, eds. Second Asia-Pacific Symposium on Environmental and
Occupational Health, 22-24 July, 1993, Kobe: proceedings. Kobe: International Center for Medical Research Kobe,
University School of Medicine, 193-198.
Yant, WP; Schrenk, HH; Waite, CP; et al. (1930) Acute response of guinea pigs to vapors of some new commercial
organic compounds -VI Dioxane. Public Health Rep 45:2023-2032.
Yasuhara A; Shiraishi H; Nishikawa M; et al. (1997) Determination of organic components in leachates from
hazardous waste disposal sites in Japan by gas chromatography-mass spectrometry. J Chromatogr A 774:321-332.
Yasuhara A; Tanaka Y; Tanabe A; et al. (2003) Elution of 1,4-dioxane from waste landfill sites. Bull
Environ Contam Toxicol 71:641-647.
Yoon, JS; Mason, JM; Nalencia, R; et al. (1985) Chemical mutagenesis testing in drosophila. IV. Results of
45 coded compounds tested for the national toxicology program. Environ Mutagen 7:349-367.
Young, JD; Braun, WH; Gehring, PJ; et al. (1976) Short communication. 1,4-dioxane andbeta-
hydroxyethoxyacetic acid excretion in urine of humans exposed to dioxane vapors. Toxicol Appl Pharmacol
38:643-646.
Young, JD; Braun, WH; Rampy, LW. (1977) Pharmacokinetics of 1,4-dioxane in humans. J Toxicol Environ
Health 3:507-520.
Young, JD; Braun, WH; Gehring, PJ. (1978a) The dose-dependent fate of 1,4-dioxane in rats. J Environ Pathol
Toxicol 2:263-282.
May 2009 124 DRAFT - DO NOT CITE OR QUOTE
-------
Young, JD; Braun, WH; Gehring, PJ. (1978b) Dose-dependent fate of 1,4-dioxane in rats. J Toxicol Environ Health
4(5-6):709-726.
Zimmermann, FK; Mayer, VW; Scheel, I; et al. (1985) Acetone, methyl ethyl ketone, ethyl acetate, acetonitrile and
other polar aprotic solvents are strong inducers of aneuploidy in Saccharomyces cerevisiae. Mutat Res 149:339-
351.
May 2009 125 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX A. EXTERNAL REVIEW COMMENTS AND DISPOSITION
May 2009 A-1 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX B. EVALUATION OF EXISTING PBPK MODELS FOR 1,4-dioxane
B.I. BACKGROUND
1 Several pharmacokinetic models have been developed to predict the absorption,
2 distribution, metabolism, and elimination of 1,4-dioxane in rats and humans. Single
3 compartment, empirical models for rats (Young et al., 1978a, b) and humans (Young et al., 1977)
4 were developed that predict blood levels of 1,4-dioxane and urine levels of the primary
5 metabolite, p-hydroxyethoxy acetic acid (HEAA). Physiologically based pharmacokinetic
6 (PBPK) models, which describe the kinetics of 1,4-dioxane using biologically realistic flow
7 rates, tissue volumes and affinities, metabolic processes, and elimination behaviors, were also
8 developed (Fisher et al., 1997; Leung and Paustenbach, 1990; Reitz et al., 1990).
9 In developing updated toxicity values for 1,4-dioxane, the available PBPK models were
10 evaluated for their ability to predict observations made in experimental studies of rat and human
11 exposures to 1,4-dioxane. The model of Reitz et al. (1990) was identified for further
12 consideration to assist in the derivation of toxicity values. Issues related to the biological
13 plausibility of parameter values in the human model were identified. Specifically, the model's
14 ability to predict the only available human inhalation data set (50 ppm 1,4-dioxane for 6 hours;
15 Young et al., 1977) relies on increasing (i.e., doubling) of parameter values for human alveolar
16 ventilation, cardiac output, and the blood:air partition coefficient above the measured values.
17 Furthermore, the measured value for the slowly perfused tissue:air partition coefficient (i.e.,
18 muscle) was replaced with the measured liver value to improve the fit. Analysis of the Young
19 et al. (1977) human data suggested that the apparent volume of distribution (Vd) for 1,4-dioxane
20 was approximately 10-fold higher in rats than humans, presumably due to species differences in
21 tissue partitioning or other process not represented in the model. Subsequent exercising of the
22 model demonstrated that selecting a human slowly perfused tissue:air partition coefficient much
23 lower than the measured rat value resulted in better agreement between model predictions of
24 1,4-dioxane in blood and experimental observations. Based upon these observations, the model
25 (e.g., metabolism/elimination parameters) was re-calibrated using biologically plausible values
26 for flow rates and tissue:air partition coefficients.
27 This appendix describes activities conducted in the evaluation of the empirical models
28 (Young et al. 1978a, b, 1977) and re-calibration and exercising of the Reitz et al. (1990) PBPK
29 model to determine the potential utility of the PBPK models for 1,4-dioxane for interspecies and
30 route to route extrapolation.
May 2009 B-l DRAFT - DO NOT CITE OR QUOTE
-------
B.2. SCOPE
1 The scope of this effort consisted of implementation of the Young et al. (1978a, b, 1977)
2 empirical rat and human models using the acslXtreme simulation software and re-calibration of
3 the Reitz et al. (1990) human PBPK model. Using the model descriptions and equations given in
4 Young et al. (1978a, b, 1977), model code was developed for the empirical models and executed,
5 simulating the reported experimental conditions. The model output was then compared with the
6 model output reported in Young et al. (1978a, b, 1977).
7 The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for
8 cardiac and alveolar flow rates and tissue:air partition coefficients. The predictions of blood and
9 urine levels of 1,4-dioxane and HEAA, respectively, from the re-calibrated model were
10 compared with the empirical model predictions of the same dosimeters to determine whether the
11 re-calibrated PBPK model could perform similarly to the empirical model. As part of the PBPK
12 model evaluation, a sensitivity analysis was performed to identify the model parameters having
13 the greatest influence on the primary dosimeter of interest, the blood level of 1,4-dioxane.
14 Variability data for the experimental measurements of the tissue:air partition coefficients were
15 incorporated to determine a range of model outputs bounded by biologically plausible values for
16 these parameters.
B.3. IMPLEMENTATION OF THE EMPIRICAL MODELS IN ACSLXTREME
17 The empirical models of Young et al. (1978a, b, 1977) for 1,4-dioxane in rats and
18 humans were reproduced using acslXtreme, version 2.3 (Aegis Technologies, Huntsville, AL).
19 Model code files were developed using the equations described in the published papers.
20 Additional files containing experiment-specific information (i.e., BWs, exposure levels, and
21 duration) were also generated.
B.3.1. Model Descriptions
22 The empirical model of Young et al. (1978a, b) for 1,4-dioxane in rats is shown in Figure
23 B-l. This is a single-compartment model that describes the absorption and metabolism kinetics
24 of 1,4-dioxane in blood and urine. No information is reported describing pulmonary absorption
25 or intravenous (i.v.) injection/infusion of 1,4-dioxane. The metabolism of 1,4-dioxane and
26 subsequent appearance of HEAA is described by Michaelis-Menten kinetics governed by a
27 maximum rate (Vmax, ug/mL-hour) and affinity constant (Km, ug/mL) . Both 1,4-dioxane and
28 HEAA are eliminated via the first-order elimination rate constants, ke and kme, respectively
29 (hour"1) by which 35% of 1,4-dioxane and 100% of HEAA appear in the urine, while 65% of
30 1,4-dioxane is exhaled. Blood concentration of 1,4-dioxane is determined by dividing the
31 instantaneous amount of 1,4-dioxane in blood by a Vd of 301 mL/kg BW.
May 2009 B-2 DRAFT - DO NOT CITE OR QUOTE
-------
Inhalation (kINH)
i.v. admin
dt
V , n •
K +DioX
- - k „ x Dioxk,
body
dt
Exhaled
Dzax
z+
Urine
x/ffi/14
-> Urine
Source: Young et al. (1978a, b).
Figure B-l. Schematic representation of empirical model for 1,4-dioxane in rats.
1 Figure B-2 illustrates the empirical model for 1,4-dioxane in humans as described in
2 Young et al. (1977). Like the rat model, the human model predicts blood 1,4-dioxane and
3 urinary 1,4-dioxane and HEAA levels using a single-compartment structure. However, the
4 metabolism of 1,4-dioxane to HEAA in humans is modeled as a first-order process governed by
5 a rate constant, KM (hour"1). Urinary deposition of 1,4-dioxane and HEAA is described using the
6 first order rate constants, ke(diox) and kme(HEAA), respectively. Pulmonary absorption is described
7 by a fixed rate of 76.1 mg/hour (kiNn). Blood concentrations of 1,4-dioxane and HEAA are
8 calculated as instantaneous amount (mg) divided by Vd(diox) or Vd(HEAA), respectively (104 and
9 480 mL/kg BW, respectively).
Inhalation (k^
Dioxane
K
•M
HEAA
*d(HEAA) X
3 (diox)
: (HEAA)
Urine
Cumulative
Dioxane and
HEAA
Source: Young et al. (1977).
Figure B-2. Schematic representation of empirical model for 1,4-dioxane in
humans.
B.3.2. Modifications to the Empirical Models
10 Several modifications were made to the empirical models. The need for the
11 modifications arose in some cases from incomplete reporting of the Young et al. (1978a, b, 1977)
May 2009
B-3
DRAFT - DO NOT CITE OR QUOTE
-------
1 studies and in other cases from the desire to add capabilities to the models to assist in the
2 derivation of toxicity values.
3 For the rat model, no information was given by Young et al. (1978a, b) regarding the
4 parameterization of pulmonary absorption (or exhalation) or i.v. administration of 1,4-dioxane.
5 Therefore, additional parameters were added to simulate these processes in the simplest form.
6 To replicate 1,4-dioxane inhalation, a first-order rate constant, kiNH (hour"1), was introduced.
7 kiNH was multiplied by the inhalation concentration and the respiratory minute volume of
8 0.238 L/minute (Young et al., 1978a, b). The value for kiNn was estimated by optimization
9 against the blood time course data of Young et al. (1978a, b). Intravenous (i.v.) administration
10 was modeled as instantaneous appearance of the full dose at the start of the simulation. Rat
11 urinary HEAA data were reported by Young et al. (1978a, b) in units of concentration. To
12 simulate urinary HEAA concentration, an estimate of urine volume was required. Since
13 observed urinary volumes were not reported by Young et al. (1978a, b), a standard rat urine
14 production rate of 0.00145 L/hour was used.
15 For humans, Young et al. (1977) used a fixed 1,4-dioxane inhalation uptake rate of
16 76.1 mg/hour, which corresponded to observations during a 50 ppm exposure. In order to
17 facilitate user-specified inhalation concentrations, pulmonary absorption was modeled. The
18 modeling was performed identically to the rat model, but using a human minute volume of
19 7 L/minute. Urinary HEAA data were reported by Young et al. (1977) as a cumulative amount
20 (mg) of HEAA. Cumulative amount of HEAA in the urine is readily calculated from the rate of
21 transfer of HEAA from plasma to urine, so no modification was necessary to simulate this dose
22 metric for humans.
23 Neither empirical model of Young et al. (1978a, b;1977) described oral uptake of
24 1,4-dioxane. Adequate data to estimate oral absorption parameters are not available for either
25 rats or humans; therefore, neither empirical model was modified to include oral uptake.
B.3.3. Results
26 The acslXtreme implementation of the Young et al. (1978a, b) rat empirical model
27 simulates the 1,4-dioxane blood levels from the i.v. experiments identically to the model output
28 reported in the published paper (Figure B-3). However, the acslXtreme version predicts urinary
29 HEAA concentrations in rats that are approximately threefold lower and reach a maximum
30 sooner than the predicted levels reported in the paper (Figure B-4). These discrepancies may be
31 due, at least in part, to the reliance in the acslXtreme implementation on a constant, standard,
32 urine volume rather than experimental measurements, which may have been different from the
33 assumed value and may have varied over time. Unreported model parameters (e.g., lag times for
34 appearance of excreted HEAA in bladder urine) may also contribute to the discrepancy.
May 2009 B-4 DRAFT - DO NOT CITE OR QUOTE
-------
Observations and predictions of 1,4-dioxane in rat blood
following 3 to 1000 mg/kg IV injection
Source: Young et al. (1978a, b).
Figure B-3. Output of 1,4-dioxane blood level data from the acslXtreme
implementation (left) and published (right) empirical rat model simulations of i.v.
administration experiments.
Observations and predictions of HEAA in rat urine
following 10 or 1000 mg/kg IV injection
20 30
Time(hrs)
Source: Young et al. (1978a, b).
Figure B-4. Output of HEAA urine level data from acslXtreme implementation
(left) and published (right) empirical rat model simulations of i.v. administration
experiments.
May 2009
B-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 The Young et al. (1978a, b) report did not provide model predictions for the 50-ppm
2 inhalation experiment. However, the acslXtreme implementation produces blood 1,4-dioxane
3 predictions that are quite similar to the reported observations (Figure B-5). As with the urine
4 data from the i.v. experiment, the acslXtreme-predicted urinary HEAA concentrations are
5 approximately threefold lower than the observations, presumably for the same reasons discussed
6 above for the i.v. predictions.
Observations and predictions of 1,4-dioxane in rat blood
following a 6 hour 50 ppm inhalation exposure
£ m 1.0-
I E 1
D
acsl version - Young et al.
(1978a, b) empirical model
Young et al. (1978a, b)
observations
6 8 10
Time(hrs)
12
Observations and predictions of HEAA in rat urine
following a 6 hour 50 ppm inhalation exposure
25.0
20.0 -
D
acsl version - Young et
al. (1978a, b) empirical
model
Young et al. (1978a, b)
observations
9
10
11
12
Source: Young et al. (1978a, b).
Figure B-5. acslXtreme predictions of blood 1,4-dioxane and urine HEAA levels
from the empirical rat model simulations of a 6-hour, 50-ppm inhalation exposure.
Inhalation data for a single exposure level (50 ppm) are available for humans. The
acslXtreme predictions of the blood 1,4-dioxane observations are identical to the predictions
reported in Young et al. (1977) (Figure B-6). Limited blood HEAA data were reported, and the
specimen analysis was highly problematic (e.g., an analytical interference was sometimes present
from which HEAA could not be separated). For this reason, Young et al. (1977) did not compare
predictions of the blood HEAA data to observations in their manuscript.
May 2009
B-6
DRAFT - DO NOT CITE OR QUOTE
-------
Observations and predictions of 1,4-dioxane in human
blood following a 6 hour 50 ppm inhalation exposure
S E
D
acsl version - Young et
al (1977) empirical model
observed
Source: Young et al. (1978a, b).
Figure B-6. Output of 1,4-dioxane blood level data from the acslXtreme
implementation (left) and published (right) empirical human model simulations of a
6-hour, 50-ppm inhalation exposure.
1 Data for cumulative urinary HEAA amounts are provided in Young et al. (1977), and no
2 analytical problems for these data were reported. Nevertheless, model predictions for urinary
3 HEAA were not presented in the manuscript. The acslXtreme prediction of the HEAA kinetics
4 profile is similar to the observations, although predicted values are approximately 1.5- to 2-fold
5 lower than the observed values (Figure B-7). Unlike urinary HEAA observations in the rat,
6 human observations were reported as cumulative amount produced, negating the need for urine
7 volume data. Therefore, discrepancies between model predictions and experimental observations
8 for humans cannot be attributed to uncertainties in urine volumes in the subjects.
May 2009
B-7
DRAFT - DO NOT CITE OR QUOTE
-------
Observations and predictions of HEAA in human urine
following a 6 hour 50 ppm inhalation exposure
700.0
600.0 -
< "3 500.0 -
I 400.0 -
o
E
< 300.0 -
S
200.0 -
100.0 -
0.0
D
acsl version - Young et al
(1977) empirical model
observed
0
10 15
Time (hrs)
20
25
Source: Young et al. (1977).
Figure B-7. Observations and acslXtreme predictions of cumulative HEAA in
human urine following a 6-hour, 50-ppm inhalation exposure.
B.3.4. Conclusions for Empirical Model Implementation
1 The empirical models described by Young et al. (1978a, b, 1977) for rats and humans
2 were implemented using acslXtreme. The models were modified to allow for user-defined
3 inhalation levels by addition of a first-order rate constant for pulmonary uptake of 1,4-dioxane,
4 fitted to the inhalation data. No modifications were made for oral absorption as adequate data
5 are not available for parameter estimation. The acslXtreme predictions of 1,4-dioxane in the
6 blood are identical to the published predictions for simulations of 6-hour, 50-ppm inhalation
7 exposures in rats and humans and 3 to 1,000 mg/kg i.v. doses in rats (Figures B-3, B-5, and
8 B-6). However, the acslXtreme version predicts lower urinary HEAA concentrations in rats
9 appearing earlier than either the Young et al. (1978a, b) model predictions or the experimental
10 observations. The lower predicted urinary HEAA levels in the acslXtreme implementation for
11 rats is likely due to use of default values for urine volume in the absence of measured volumes.
12 The reason for differences in time-to-peak levels is unknown, but may be the result of an
13 unreported adjustment by Young et al. (1978a, b) in model parameter values. For humans,
14 Young et al. (1977) did not report model predictions of urinary HEAA levels. The urinary
15 HEAA levels predicted by acslXtreme were low relative to the observations. However, unlike
16 the situation in rats, these data are not dependent on unreported urine volumes (observations
17 were reported as cumulative HEAA amount rather than HEAA concentration), but reflect the
18 model parameter values reported by Young et al. (1977). Presently, there is no explanation for
19 the lack of fit of the reported urinary HEAA elimination rate constant to the observations.
May 2009
B-8
DRAFT - DO NOT CITE OR QUOTE
-------
B.4. INITIAL RE-CALIBRATION OF THE PBPK MODEL
1
2
3
4
5
6
7
10
11
12
13
14
15
16
17
18
19
20
21
Concern regarding adjustments made to some of the parameter values in Reitz et al.
(1990) prompted a re-calibration of the human PBPK model using more biologically plausible
values for all measured parameter values. Reitz et al. (1990) doubled the measured physiological
flows and blood:air partition coefficient and substituted the slowly-perfused tissue:air partition
coefficient with the liverair value in order to attain an adequate fit to the observations. This
approach increases uncertainty in these parameter values, and in the utilization of the model for
cross-species dose extrapolation. Therefore, the model was re-calibrated using parameter values
that are more biologically plausible to determine whether an adequate fit of the model to the
available data can be attained.
B.4.1. Sources of Values for Flow Rates
0.74
The cardiac output of 30 L/hour/kgu /4 (Table B-l) reported by Reitz et al. (1990) is
approximately double the mean resting value of 14 L/hour/kg0'74 reported in the widely accepted
compendium of Brown et al. (1997). Brown et al. (1997) cite the work of Astrand (1983) in
which resting cardiac output was measured to be 5.2 L/minute (or 14 L/hour/kg0'74), while
strenuous exercise resulted in a flow of 9.9 L/minute (or 26 L/hour/kg0'74). Brown et al. (1997)
also cite the ICRP (1975) as having a mean respiratory minute volume of 7.5 L/minute, which
results in an alveolar ventilation rate of 5 L/minute (assuming 33% lung dead space), or 13
L/minute/kg0'74. Again, this is roughly half the value of 30 L/hour/kg0'74 employed for this
parameter by Reitz et al. (1990). Young et al. (1977) reported that the human subjects exposed
to 50 ppm for 6 hours were resting inside a walk-in exposure chamber. Thus, use of cardiac
output and alveolar ventilation rates of 30 L/hour/kg074 is not consistent with the experimental
conditions being simulated.
Table B-l. Human PBPK model parameter values for 1,4-dioxane
Parameter
Reitz et al. (1990)
Leung and
Paustenbach (1990)
Sweeney et al.
(2008)
EPAC
Physiological Flows
Cardiac output (QCC)a
Alveolar ventilation (QPC)a
30
30
~
~
~
~
17.0
17.7
Partition Coefficients (PCs)
Blood:air (PB)
Fat:air (PFA)
Liverair (PLA)
Rapidly perfused tissue :air (PRA)
Slowly perfused tissue:air (PSA)
3,650
851
1,557
1,557
1,557
1,825 ± 94
851 ± 118
1,557 ±114
~
997 ± 254
1,666 ± 287
~
1,862 ± 739b
-
1,348 ± 290b
1,850
851
1,557
1,557
166
Metabolic Constants
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)d
6.35
~
~
5.49
May 2009
B-9
DRAFT - DO NOT CITE OR QUOTE
-------
Parameter
Metabolic dissociation constant
(Km)e
HEAA urinary elimination rate
constant (kme)f
Reitz et al. (1990)
3.00
0.56
Leung and
Paustenbach (1990)
~
—
Sweeney et al.
(2008)
~
—
EPAC
9.8
0.44
aL/hour/kgBW074
bMeasurement for rat tissue
'Biologically plausible values utilized by EPA in this assessment
dmg/hour/kg BW°75
emg/L
'hour1
1 Examination of the experimental data of Young et al. (1977) yields an estimated alveolar
2 ventilation to be 7 L/minute (or 16 L/hour/kg0'74) for volunteers having a mean BW of 84 kg.
3 This rate is based on the Young et al. (1977) estimate of 76.1 mg/hour for 1,4-dioxane uptake.
4 Based on these findings, the cardiac output and alveolar ventilation rates of 17.0 and 17.7
5 L/hour/kg0'74 were biologically plausible for the experimental subjects. These rate estimates are
6 based on calculations made using empirical data and are consistent with standard human values
7 and the experimental conditions (i.e., subject exertion level) reported by Young et al. (1977).
8 Therefore, these flow values were chosen for the model re-calibration.
B.4.2. Sources of Values for Partition Coefficients
9 Two data sources are available for the tissue:air equilibrium partition coefficients for
10 1,4-dioxane: Leung and Paustenbach (1990) and Sweeney et al. (2008). Both investigators
11 report mean values and standard deviations for human blood:air, rat liverair, and rat muscle:air
12 (e.g., slowly perfused tissue:air), while Leung and Paustenbach et al. (1990) also reported values
13 for rat fatair (Table B-l).
B.4.3. Calibration Method
14 The PBPK model was twice re-calibrated using the physiological flow values suggested
15 values (current EPA assessment, see Table B-l) and the partition coefficients of Leung and
16 Paustenbach (1990) and Sweeney et al. (2008) separately. For each calibration, the metabolic
17 parameters Vmaxc and Km, were simultaneously fit (using the parameter estimation tool provided
18 in the acslXtreme software) to the output of 1,4-dioxane blood concentrations generated by the
19 acslXtreme implementation of the Young et al. (1977) empirical human model for a 6 hour,
20 50 ppm inhalation exposure. Subsequently, the HEAA urinary elimination rate constant, kme,
21 was fitted to the urine HEAA predictions from the empirical model. The empirical model
22 predictions, rather than experimental observations, were used to provide a more robust data set
23 for model fitting, as the empirical model simulation provided 240 data points (one prediction
May 2009
B-10
DRAFT - DO NOT CITE OR QUOTE
-------
1 every 0.1 hour) compared with hourly experimental observations, and to avoid introducing error
2 by calibrating the model to data digitally captured from Young et al. (1977).
B.4.4. Results
3 Results of the model re-calibration are provided in Table B-2. The re-calibrated values
4 for Vmaxc and kme associated with the Leung and Paustenbach (1990) or Sweeney et al. (2008)
5 tissue:air partition coefficients are very similar. However, the fitted value for Km using the
6 Sweeney et al. (2008) partition coefficients is far lower (0.0001 mg/L) than that resulting from
7 use of the Leung and Paustenbach (1990) partition coefficients (2.5 mg/L). This appears to be
8 due to the higher slowly perfused tissue:air partition coefficient determined by Sweeney et al.
9 (2008) (1,348 vs. 997), resulting in a higher apparent Vd than if the Leung and Paustenbach
10 (1990) value is used. Thus, the optimization algorithm selects a low Km, artificially saturating
11 metabolism in an effort to drive predicted blood 1,4-dioxane levels closer to the empirical model
12 output. Saturation of metabolism during a 50 ppm inhalation exposure is inconsistent with the
13 observed kinetics.
Table B-2. PBPK metabolic and elimination parameter values resulting
from re-calibration of the human model using alternative values for
physiological flow rates" and tissuerair partition coefficients
Source of Partition Coefficients
Maximum rate for 1,4-dioxane metabolism (VmaxC)b
Metabolic dissociation constant (Km)°
HEAA urinary elimination rate constant (kme)d
Leung and Paustenbach (1990)
16.9
2.5
0.18
Sweeney et al. (2008)
20.36
0.0001
0.17
14
15
16
17
18
19
"Cardiac output = 17.0 L/hour/kg BW074, alveolar ventilation = 17.7 L/hour/kg BW074
bmg/hour/kg BW°75
cmg/L
dhouf1
Plots of predicted and experimentally observed blood 1,4-dioxane and urinary HEAA
levels are shown in Figures 4-1 and 4-2. Neither re-calibration resulted in an adequate fit to the
blood 1,4-dioxane data from the empirical model output or the experimental observations. Re-
calibration using either the Leung and Paustenbach (1990) or Sweeney et al. (2008) partition
coefficients resulted in blood 1,4-dioxane predictions that were at least 10-fold lower than
empirical model predictions or observations.
May 2009
B-ll
DRAFT - DO NOT CITE OR QUOTE
-------
Observations and predictions of 1,4-dioxane in human blood from
a 6-hour, 50 ppm exposure: VmaxC and Km fit while using PC
values from Leung and Paustenbach (1990)
100.0
1.0 -
024
10 12 14
Time (hrs)
Observations and predictions of HEAA in human urine from a
6-hour, 50 ppm exposure: kme fit while using PC values from
Leung and Paustenbach (1990)
700
600-
£-01500 -
5 = 400 -
" < 300 -
=> <
E <
5^200-1
100-
0
10 15
Time (hrs)
20
25
1
2
Source: Leung and Paustenbach (1990).
Figure B-8. Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following re-calibration of the human PBPK model
with tissue:air partition coefficient values.
The refitted values for kme resulted in HEAA levels in urine that were very similar to the
empirical model output (compare Figures B-7, B-8, and B-9), which was not surprising, given
the fitting of a single parameter to the data.
Observations and predictions of 1 ,4-dioxane in human blood from
a 6-hour, 50 ppm exposure: Vma
C and K fit while using PC
values from Sweeney et al. (2008)
100 o -p— — — — — — — — — — — — — — — — — — — ^
o
'•S
t
g 10.0-
c
o
0 ~
-o,
x £
o
9 1.0-
•o
o
o
m
0 1
,fH'H,
*m *
' I
'
•'T
i
pgPI/ predicted
a observed
. . . . empirical predricted
T
I
f
0 5
10 15
Time (hrs)
Observations and predictions of HEAA in human urine from a
6-hour, 50 ppm exposure: kme fit while using PC values from
Sweeney etal. (2008)
700.0
4
5
6
Source: Sweeney et al. (2008).
Figure B-9. Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following re-calibration of the human PBPK model
with tissue:air partition coefficient values.
Outputs of the blood 1,4-dioxane and urinary HEAA levels using the suggested (see
Table B-l) parameters are shown in Figure B-10. These outputs rely on a very low value for the
slowly perfused tissue:air partition coefficient (166) that is six- to eightfold lower than the
May 2009
B-12
DRAFT - DO NOT CITE OR QUOTE
-------
1 measured values reported in Leung and Paustenbach (1990) and Sweeney et al. (2008), and 10-
2 fold lower than the value used by Reitz et al. (1990). While the predicted maximum blood
3 1,4-dioxane levels are much closer to the observations, the elimination kinetics are markedly
4 different, producing higher predicted elimination rates compared to observations during the post-
5 exposure phase of the experiment.
Observations and predictions of 1,4-dioxane in
human blood from a 6-hour, 50 ppm exposure:
EPA parameter estimates used
100.0 -
Observations and predictions of HEAA in human
urine from a 6-hour, 50 ppm exposure:
EPA parameter estimates used
700.0
600.0 -
£-3500.0 -
'B § 400.0 -
.1 I
j| < 300.0 -
§ I 200.0 -
100.0 -
0.0
10 15
Time (hrs)
20
25
Figure B-10. Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) using EPA estimated biologically plausible parameters
(see Table B-l).
B.4.5. Conclusions for Empirical Model Implementation
6 Re-calibration of the human PBPK model was performed using experiment-specific
7 values for cardiac output and alveolar ventilation (values derived from Young et al., 1977) and
8 measured mean tissue:air 1,4-dioxane partition coefficients reported by Leung and Paustenbach
9 (1990) or Sweeney et al. (2008). The resulting predictions of 1,4-dioxane in blood following a
10 6-hour, 50-ppm inhalation exposure were 10-fold (or more) lower than either the observations or
11 the empirical model predictions, while the predictions of urinary HEAA by the PBPK and
12 empirical models were similar to each other, but lower than observed values (Figures B-8 and
13 B-9). Output from the model using biologically plausible parameter values (see Table B-l),
14 Figure B-10 shows that application of a value for the slowly perfused tissue:air partition
15 coefficient, which is 10-fold lower than the measured value reported by Leung and Paustenbach
16 (1990), results in closer agreement of the predictions to observations during the exposure phase,
17 but not during the elimination phase. Thus, model re-calibration using experiment-specific flow
18 rates and mean measured partition coefficients does not result in an adequate fit of the PBPK
19 model to the available data.
May 2009
B-13
DRAFT - DO NOT CITE OR QUOTE
-------
B.4.6. SENSITIVITY ANALYSIS
1 A sensitivity analysis of the Reitz et al. (1990) model was performed to determine which
2 PBPK model parameters exert the greatest influence on the outcome of dosimeters of interest—
3 in this case, the concentration of 1,4-dioxane in blood. Knowledge of model sensitivity is useful
4 for guiding the choice of parameter values to minimize model uncertainty.
B.4.7. Method
5 A univariate sensitivity analysis was performed on all of the model parameters for two
6 endpoints: blood 1,4-dioxane concentrations after 1 and 4 hours of exposure. These time points
7 were chosen to assess sensitivity during periods of rapid uptake (1 hour) and as the model
8 approached steady state (4 hours) for blood 1,4-dioxane. Model parameters were perturbated 1%
9 above and below nominal values and sensitivity coefficients were calculated as follows:
<> x
Ax f(x)
10 where x is the model parameter, f(x) is the output variable, Ax is the perturbation of the
11 parameter from the nominal value, and f (x) is the sensitivity coefficient. The sensitivity
12 coefficients were scaled to the nominal value of x and f(x) to eliminate the potential effect of
13 units of expression. As a result, the sensitivity coefficient is a measure of the proportional
14 change in the blood 1,4-dioxane concentration produced by a proportional change in the
15 parameter value, with a maximum value of 1.
B.4.8. Results
16 The sensitivity coefficients for the seven most influential model parameters at 1 and
17 4 hours of exposure are shown in Figure B-l 1. The three parameters with the highest sensitivity
18 coefficients in descending order are alveolar ventilation (QPC) (1.0), the blood:air partition
19 coefficient (PB) (0.65), and the slowly perfused tissue:air partition coefficient (PSA) (0.51). Not
20 surprisingly, these were the parameters that were doubled or given surrogate values in the Reitz
21 et al. (1990) model in order to achieve an adequate fit to the data. Because of the large influence
22 of these parameters on the model, it is important to assign values to these parameters in which
23 high confidence is placed, in order to reduce model uncertainty.
May 2009 B-14 DRAFT - DO NOT CITE OR QUOTE
-------
Sensitivity Coefficients: CV - 1hr
0.01 0.10 1.00
QPC
PB
PSA
£
-------
B.5.2. Defining Boundaries for Parameter Values
1 Given the possible 10-fold species differences in the apparent Vd for 1,4-dioxane in rats
2 and humans, boundary values for the partition coefficients were chosen to exercise the PBPK
3 model across its performance range to either minimize or maximize the simulated Vd. This was
4 accomplished by defining biologically plausible values for the partition coefficients as the
5 mean ± 2 standard deviations of the measured values. Thus, to minimize the simulated Vd for
6 1,4-dioxane, the selected blood:air partition coefficient was chosen to be the mean + 2 standard
7 deviations, while all of the other tissue:air partition coefficients were chosen to be the mean - 2
8 standard deviations. This created conditions that would sequester 1,4-dioxane in the blood, away
9 from other tissues. To maximize the simulated 1,4-dioxane Vd, the opposite selections were
10 made: blood and other tissue:air partition coefficients were chosen as the mean - 2 standard
11 deviations and mean + 2 standard deviations, respectively. Subsequently, Vmaxc, Km, and kme
12 were optimized to the empirical model output data as described in Section B.4.3. This procedure
13 was performed for both the Leung and Paustenbach (1990) and Sweeney et al. (2008) partition
14 coefficients (Table B-l). The two predicted time courses resulting from the re-calibrated model
15 with partition coefficients chosen to minimize or maximize the 1,4-dioxane Vd represent the
16 range of model performance as bounded by biologically plausible parameter values.
B.5.3. Results
17 The predicted time courses for a 6-hour, 50-ppm inhalation exposure for the re-calibrated
18 human PBPK model with mean (central tendency) and ± 2 standard deviations from the mean
19 values for partition coefficients are shown in Figure B-12 for the Leung and Paustenbach (1990)
20 values and Figure B-l3 for the Sweeney et al. (2008) values. The resulting fitted values for
21 Vmaxc, Km, and kme, are given in Table B-3. By bounding the tissue:air partition coefficients with
22 upper and lower limits on biologically plausible values from Leung and Paustenbach (1990) or
23 Sweeney et al. (2008), the model predictions are still at least six- to sevenfold lower than either
24 the empirical model output or the experimental observations. The range of possible urinary
25 HEAA predictions brackets the prediction of the empirical model, but this agreement is not
26 surprising, as the cumulative rate of excretion depends only on the rate of metabolism of
27 1,4-dioxane, and not on the apparent Vd for 1,4-dioxane. These data show that the PBPK model
28 cannot adequately reproduce the predictions of blood 1,4-dioxane concentrations of the Young
29 et al. (1977) human empirical model or the experimental observations when constrained by
30 biologically plausible values for physiological flow rates and tissue:air partition coefficients.
May 2009 B-16 DRAFT - DO NOT CITE OR QUOTE
-------
1,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure
Cumulative HEAA in human urine from a 6-hour, 50 ppm
exposure
700
25
Source: Leung and Paustenbach (1990)
Figure B-12. Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients with empirical model predictions and
experimental observations for blood 1,4-dioxane concentrations (left) and urinary
HEAA levels (right) from a 6-hour, 50-ppm inhalation exposure.
1,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure
c
o
•£
8 10.0 -
o
%%
5 i.
o
9 1.0-
T-"
1
m
0.1 -
_ _ Young et al. (1977) empirical model
Sweeney etal. (2008) PC -UCL
Sweeney et al. (2008) PC - Central
n n Sweeney et al. (2008) PC- LCL
<] * Young et al. (1977) observation
> ° data
/D
' V
'/ ' \ * ^
/^- \ . et
Sweeney etal. (2008) PC -LCL
;** n Young etal. (1977) observation da
_^S_
I
a
0 5 10 15 20 25
Time (hrs)
Source: Sweeney et al. (2008); Young et al. (1977).
Figure B-13. Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients with empirical model predictions and
experimental observations for blood 1,4-dioxane concentrations (left) and urinary
HEAA levels (right) from a 6-hour, 50-ppm inhalation exposure.
May 2009
B-17
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-3. PBPK metabolic and elimination parameter values resulting
from recalibration of the human model using biologically plausible values for
physiological flow rates3 and selected upper and lower boundary values for
tissuerair partition coefficients
Source of partition coefficients
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)b
Metabolic dissociation constant
(Km)c
HEAA urinary elimination rate
constant (kme)d
Leung and Pausenbach (1990)
For maximal Vd
14.95
5.97
0.18
For minimal Vd
18.24
0.0001
0.17
Sweeney et al. (2008)
For maximal Vd
17.37
4.88
0.26
For minimal Vd
21.75
0.0001
0.19
"Cardiac output = 17.0 L/hour/kg BW074' alveolar ventilation = 17.7 L/hour/kg BW074
bmg/hour/kg BW°75
cmg/L
dhour-1
B.5.4. Alternative Model Parameterization
1 Since the PBPK model does not predict the experimental observations of Young et al.
2 (1977) when parameterized by biologically plausible values, an exercise was performed to
3 explore alternative parameters and values capable of producing an adequate fit of the data. Since
4 the metabolism of 1,4-dioxane appears to be linear in humans for a 50-ppm exposure (Young
5 et al., 1977), the parameters Vmaxc and Km were replaced by a first-order, non-saturable
6 metabolism rate constant, kLc- This rate constant was fitted to the experimental blood
7 1,4-dioxane data using partition coefficient values of Sweeney et al. (2008) to minimize the Vd
8 (i.e., maximize the blood 1,4-dioxane levels). The resulting model predictions are shown in
9 Figure B-14. As before, the maximum blood 1,4-dioxane levels were approximately sevenfold
10 lower than the observed values.
May 2009
B-18
DRAFT - DO NOT CITE OR QUOTE
-------
1 ,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure: kLC(3.0) fitted to all observations
Blood 1,4-Dioxane Concentration
(mg/L) _^
ONJ-CiOCOONJ-CiC
D D
D
9 ' ' '
* *
i f
i »
•D '
/^" ^
Young et al. (1 977) empirical
model
l^c - fitted model
D Young etal. (1977)
observation data
%
D
0 2 4 6 8 10 12 14
Time (hrs)
Figure B-14. Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, RLC? to the experimental data.
1 A re-calibration was performed using only the data from the exposure phase of the
2 experiment, such that the elimination data did not influence the initial metabolism and tissue
3 distribution. The model predictions from this exercise are shown in Figure B-l 5. These
4 predictions are more similar to the observations made during the exposure phase of the
5 experiment; however, this is achieved at greatly reduced elimination rate (compare Figures B-10
6 and B-l 5).
May 2009
B-19
DRAFT - DO NOT CITE OR QUOTE
-------
16.1
14.1
12.1
10.1
u
"5> 8.1
_£_
6.1
4.1
2.1
0.1
1,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure: kLC (0.1) fitted 1 to 6-hour observations
Young et al. (1977) empirical
model
- fitted model
10
12
14
Time (hrs)
Figure B-15. Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, RLC? to only the exposure phase of the
experimental data.
1 Finally, the model was re-calibrated by simultaneously fitting kLc and the slowly
2 perfused tissue:air partition coefficient to the experimental data with no bounds on possible
3 values (except that they be non-zero). The fitted slowly perfused tissue:air partition coefficient
4 was an extremely low (and biologically unlikely) value of 0.0001 . The resulting model
5 predictions, however, were closer to the observations than even the empirical model predictions
6 (Figure B-16). These exercises show that better fits to the observed blood 1,4-dioxane kinetics
7 are achieved only when parameter values are adjusted in a way that corresponds to a substantial
8 decrease in apparent Vd of 1,4-dioxane in the human, relative to the rat (e.g., decreasing the
9 slowly perfused tissue:air partition coefficient to extremely low values, relative to observations).
10 Downward adjustment of the elimination parameters (e.g., decreasing kLc) increases the
1 1 predicted blood concentrations of 1,4-dioxane, achieving better agreement with observations
12 during the exposure phase of the experiment; however, it results in unacceptably slow
13 elimination kinetics, relative to observations following cessation of exposure. These
14 observations suggest that some other process not captured in the present PBPK model structure is
15 responsible for the species differences in 1,4-dioxane Vd and the inability to reproduce the
16 human experimental inhalation data with biologically plausible parameter values.
May 2009
B-20
DRAFT - DO NOT CITE OR QUOTE
-------
1,4-Dioxane in human blood from a 6-hour,
50 ppm exposure
rj Young etal. (1977)
observation data
10
12
Time (hrs)
14
Figure B-16. Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a first-order metabolism rate constant, kLC, and slowly perfused
tissuerair partition coefficient to the experimental data.
B.6. CONCLUSIONS
1 The rat and human empirical models of Young et al. (1978a, b, 1977) were successfully
2 implemented in acslXtreme and perform identically to the models reported in the published
3 papers (Figures 3-3 through 3-6), with the exception of the lower predicted HEAA
4 concentrations and early appearance of the peak HEAA levels in rat urine. The early appearance
5 of peak HEAA levels cannot presently be explained, but may result from manipulations of kme or
6 other parameters by Young et al. (1978a, b) that were not reported. The lower predictions of
7 HEAA levels are likely due to reliance on a standard urine volume production rate in the absence
8 of measured (but unreported) urine volumes. While the human urinary HEAA predictions were
9 lower than observations, this is due to parameter fitting of Young et al. (1977). No model output
10 was published in Young et al. (1977) for comparison. The empirical models were modified to
11 allow for user-defined inhalation exposure levels. However, no modifications were made to
12 model oral exposures because adequate data to parameterize such modifications do not exist for
13 rats or humans.
14 Several procedures were applied to the human PBPK model to determine if an adequate
15 fit of the model to the empirical model output or experimental observations could be attained
16 using biologically plausible values for the model parameters. The re-calibrated model
17 predictions for blood 1,4-dioxane levels do not come within 10-fold of the experimental values
18 using measured tissue:air partition coefficients from Leung and Paustenbach (1990) or Sweeney
19 et al. (2008) (Figures B-8 and B-9). Use of a slowly perfused tissue:air partition coefficient 10-
May 2009
B-21
DRAFT - DO NOT CITE OR QUOTE
-------
1 fold lower than measured values produces exposure-phase predictions that are much closer to
2 observations, but does not replicate the elimination kinetics (Figure B-10). Re-calibration of the
3 model with upper bounds on the tissue:air partition coefficients results in predictions that are still
4 six- to sevenfold lower than empirical model prediction or observations (Figures B-12 and B-13).
5 Exploration of the model space using an assumption of first-order metabolism (valid for the 50-
6 ppm inhalation exposure) showed that an adequate fit to the exposure and elimination data can
7 be achieved only when unrealistically low values are assumed for the slowly perfused tissue:air
8 partition coefficient (Figure B-16). Artificially low values for the other tissue:air partition
9 coefficients are not expected to improve the model fit, because the sensitivity analysis to exert
10 less influence on blood 1,4-dioxane than Vmaxc and Km. This suggests that the model structure is
11 insufficient to capture the apparent 10-fold species difference in the blood 1,4-dioxane Vd
12 between rats and humans. In the absence of actual measurements for the human slowly perfused
13 tissue:air partition coefficient, high uncertainty exists for this model parameter value.
14 Differences in the ability of rat and human blood to bind 1,4-dioxane may contribute to the
15 difference in Vd. However, this is expected to be evident in very different values for rat and
16 human blood:air partition coefficients, which is not the case (Table B-l). Therefore, some other,
17 as yet unknown, modification to model structure may be necessary.
B.7. RECOMMENDATIONS FOR UTILIZING EXISTING PBPK MODELS
18 The use of empirical or PBPK models to reduce uncertainty in extrapolation of dose-
19 responses (in terms of internal dosimetry) requires accurate representation of exposure and
20 biological reality. In the case of the empirical models of Young et al. (1978a, b, 1977), the
21 acslXtreme implementations are adequate for predicting blood 1,4-dioxane levels for a variety of
22 inhalation exposure levels in rats and up to 50 ppm in humans. However, the absence of data
23 with which to evaluate simulated oral absorption in either species precludes the inclusion of this
24 route of exposure in the models. Therefore, the empirical models may be useful for assessment
25 of toxicity by inhalation exposure, but not by oral exposure, and not for route-to-route
26 extrapolation. For the PBPK model, an apparent gap in the model structure exists such that
27 experimental observations of blood 1,4-dioxane levels in humans during and following
28 inhalation exposures to 1,4-dioxane cannot be reproduced under the constraints of biologically
29 plausible parameter values for all parameters. Therefore, the use of the PBPK model (in its
30 present form) is not recommended for application to the derivation of toxicity values for
31 1,4-dioxane.
May 2009 B-22 DRAFT - DO NOT CITE OR QUOTE
-------
B.8. ACSLXTREME CODE FOR THE YOUNG ET AL. (1978A, B) EMPIRCAL MODEL
FOR 1,4-DIOXANE IN RATS
1 PROGRAM: Young 1978 rat.csl
2 I
3 ! Created by Michael Lumpkin, Syracuse Research Corporation, 08/06
4 ! This program implements the 1-compartment empirical model for 1,4-dioxane
5 ! in rats, developed by Young et al. 1978a, b. Program was modified to run
6 ! in ACSL Xtreme and to include user-defined i.v. and inhalation concentrations
7 !(MLumpkin, 08/06)
9
10 INITIAL
11
12 !*****Timing and Integration Commands*****
13 ALGORITHM IALG=2 ! Gear integration algorithm for stiff systems
14 IMERROR %%%%=0.01 IRelative error for lead in plasma
15 NSTEPS NSTP=1000 INumber of integration steps per communication interval
16 CINTERVALCINT=0.1 ! Communication interval
17 CONSTANT TSTART=0. ! Start of simulation (hr)
18 CONSTANT TSTOP=70. !End of simulation (hr)
19
20 !*****MODEL PARAMETERS*****
21 CONSTANT BW=0.215 !Body weight (kg)
22 CONSTANT MINVOL=0.238 Irespiratory minute volume (L/min) estimated from Young et al.
23 (1978)
24 CONSTANT IVDOSE = 0. !IV dose (mg/kg)!
25 CONSTANT CONC = 0. !inhalation concentration (ppm)
26
27 CONSTANT MOLWT=88.105 !mol weight of 1,4-dioxane
28 CONSTANT TCHNG=6.0 lExposure pulse 1 width (hr)
29 CONSTANT TDUR=24.0 lExposure duration (hr)
30 CONSTANT TCHNG2=120.0 lExposure pulse 2 width (hr)
31 CONSTANT TDUR2=168.0 lExposure duration 2 (hr)
32
33 CONSTANT Vmax=4.008 !(mcg/mL/hr)
34 CONSTANT Km=6.308 !(mcg/mL)
35 CONSTANT Kinh=0.43 I pulmonary absorption constant (/hr)
36 CONSTANT Ke=0.0149 !(/hr)
37 CONSTANT Kme=0.2593 !(/hr)
38 CONSTANT Vd=0.3014 !(L)
39
40 IV = IVDOSE*BW
41 AmDIOXi=IV
42
43 END ! Of Initial Section
44
45 DYNAMIC
May 2009 B-23 DRAFT - DO NOT CITE OR QUOTE
-------
1 DERIVATIVE
2
3 ! * * * Dioxane inhalation concentration * * *
4 CIZONE=PULSE(0.0, TDUR, TCHNG) * PULSE(0.0, TDUR2, TCHNG2)
5 IFirst pulse is hours/day, second pulse is hours/week
6 CI=CONC*CIZONE*MOLWT/24450. ! Convert to mg/L
7
8 ! *** Dioxane metabolism/1 st order elimination ***
9 dAmDIOX=(Kinh*CI*(MINVOL*60))-((Vmax*(AmDIOX))/(Km+(AmDIOX)))-
10 (Ke*(AmDIOX))
11 AmDIOX=INTEG(dAmDIOX, AmDIOXi)
12 ConcDIOX=AmDIOX/Vd ! plasma dioxane concentration (mcg/mL)
13 AUCDIOX=INTEG(ConcDIOX,0) Iplasma dioxane AUC
14
15 ! * * * HEAA production and 1 st order metabolism * * *
16 dAmHEAA=((Vmax*(AmDIOX))/(Km+(AmDIOX)))-(Kme*(AmHEAA))
17 AmHEAA=INTEG(dAmHEAA,0.)
18 ConcHEAA=AmHEAA/Vd Iplasma HEAA concentration
19
20 ! * * * 1 st order dioxane elimination to urine * * *
21 dAmDIOXu=(Ke*(AmDIOX))*0.35
22 AmDIOXu=INTEG(dAmDIOXu,0.)
23 ConcDIOXu=Ke*AmDIOX*0.35/l .45e-3 lurine production approx 1.45e-3 L/hr in SD rats
24
25 ! * * * 1 st order dioxane exhaled * * *
26 dAmDIOXex=(Ke*(AmDIOX))*0.65
27 AmDIOXex=INTEG(dAmDIOXex,0.)
28
29 I * * * 1 st order HEAA elimination to urine * * *
30 dAmHEAAu=(Kme*(AmHEAA))
31 AmHEAAu=INTEG(dAmHEAAu,0.)
32 ConcHEAAu=Kme*AmHEAA/1.45e-3 lurine production approx 1.45e-3 L/hr in SD rats
33
34 END I of Derivative Section
35
36 DISCRETE
37
38 END I of Discrete Section
39
40 TERMT (T .GT. TSTOP)
41
42 END I of Dynamic Section
43
44 TERMINAL
45
46 END I of Terminal Section
47
48 END I of Program
May 2009 B-24 DRAFT - DO NOT CITE OR QUOTE
-------
B.9. ACSLXTREME CODE FOR THE YOUNG ET AL. (1977) EMPIRICAL MODEL
FOR 1,4-DIOXANE IN HUMANS
1 PROGRAM: Young 1977 human.csl
2 I
3 ! Created by Michael Lumpkin, Syracuse Research Corporation, 01/06
4 ! This program implements the 1-compartment model for 1,4-dioxane in humans,
5 ! developed by Young et al., 1977. Program was modified to run
6 ! in acslXtreme (MLumpkin, 08/06)
7 I
8
9 INITIAL
10
11 !*****Timing and Integration Commands*****
12 ALGORITHM IALG=2 ! Gear integration algorithm for stiff systems
13 !MERROR%%%%=0.01 !Relative error for lead in plasma
14 NSTEPS NSTP=1000 INumber of integration steps per communication interval
15 CINTERVALCINT=0.1 ! Communication interval
16 CONSTANT TSTART=0. ! Start of simulation (hr)
17 CONSTANT TSTOP=120. !End of simulation (hr)
18
19 I*****MODEL PARAMETERS*****
20 ! CONST ANT D AT A=l lOptimization dataset
21 CONSTANT MOLWT=88.105 !mol weight for 1,4-dioxane
22 CONSTANT DOSE=0. IDose (mg/kg
23 CONSTANT CONC=0. !Inhalation concentration (ppm)
24 CONSTANT BW=84.1 IBody weight (kg)
25 CONSTANT MINVOL=7.0 !pulmonary minute volume (L/min)
26 CONSTANT F=l .0 IFraction of dose absorbed
27 CONSTANT kinh=l .06 IRate constant for inhalation (mg/hr); optimized by MHL
28 CONSTANT ke=0.0033 IRate constant for dioxane elim to urine (hr-1)
29 CONSTANT km=0.7096 IRate constant for metab of dioxane to HEAA (hr-1)
30 CONSTANT kme=0.2593 IRate constant for transfer from rapid to blood (hr-1)
31 CONSTANT VdDkg=0.104 I Volume of distribution for dioxane (L/kg BW)
32
33 CONSTANT VdMkg=0.480 I Volume of distribution for HEAA (L/kg BW)
34 CONSTANT OStart=0. I Time of first oral dose (hr)
35 CONSTANT OPeriod=120. I Oral Dose pulse period (hr)
3 6 CONSTANT OWidth= 1. I Width (gavage/drink time) of oral dose (hr)
37
38 CONSTANT IStart=0. I Time of inhalation onset (hr)
39 CONSTANT IPeriod=120. I Inhalation pulse period (hr)
40 CONSTANT IWidth=6. I Width (duration) of inhalation exposure (hr)
41
42 END I Of Initial Section
43
44 DYNAMIC
45
46 DERIVATIVE
May 2009 B-25 DRAFT - DO NOT CITE OR QUOTE
-------
1 I * * * * VARIABLES and DEFINED VALUES *****
2 VdD=BW*VdDkg ! Volume of distribution for dioxane
3 VdM=BW* VdMkg ! Volume of distribution for HEAA
4
5 InhalePulse=PULSE(IStart,IPeriod,IWidth)
6 Inhale=CONC*InhalePulse*MOLWT/24450. !Convert to mg/L
7
8 ! * * * * *DIFFERENTIAL EQUATIONS FOR COMPARTMENTS* * * *
9
10 ! * * * Dioxane in the body (plasma) * * *
11 dAMTbD=(Kinh*Inhale*(MINVOL*60))-(AMTbD*km)-(AMTbD*ke)
12 AMTbD=INTEG(dAMTbD,0.)
13 CbD=AMTbD/VdD
14 AUCbD=INTEG(CbD,0)
15
16 ! * * * HEAA in the body (plasma)* * *
17 dAMTbM=AMTbD*km-AMTbM*kme
18 AMTbM=INTEG(dAMTbM,0.)
19 CbM=AMTbM/VdM
20
21 ! * * * Cumulative Dioxane in the urine * * *
22 dAMTuD=(AMTbD*ke)
23 AMTuD=INTEG(dAMTuD,0.)
24
25 ! * * * Cumulative HEAA in the urine * * *
26 dAMTuM=(AMTbM*kme)
27 AMTuM=INTEG(dAMTuM,0.)
28
29 END ! Of Derivative Section
30
31 DISCRETE
32
33 END !of Discrete Section
34
35 TERMT (T .GT. TSTOP)
36
37 END ! Of Dynamic Section
38
39 TERMINAL
40
41 END ! of Terminal Section
42
43 END ! of Program
May 2009 B-26 DRAFT - DO NOT CITE OR QUOTE
-------
B.IO. ACSLXTEME CODE FOR THE REITZ ET AL. (1990) PBPK MODEL FOR 1,4-
DIOXANE
1 PROGRAM: DIOXANE.CSL (Used in Risk Estimation Procedures)
2 ! Added a venous blood compartment and 1st order elim of metab.'
3 IMass Balance Checked OK for Inhal, IV, Oral, and Water RHR
4 IDefined Dose Surrogates for Risk Assessment 01/04/89'
5 [Modified the Inhal Route to use PULSE for exposure conditions'
6 Modifications by GLDiamond, Aug2004, marked as !**
7 !
8 IMetabolism of dioxane modified by MLumpkin, Oct2006, to include 1st order
9 lor saturable kinetics. For 1st order, set VmaxC=0; for M-Menten, set K1C=0.
10 !
11 INITIAL
12
13 INTEGER I
14 1=1
15 ! ARRAY TDATA(20) ! CONSTANT TDATA=999, 19*1.OE-6 !**
16 CONSTANT BW = 0.40 !'Body weight (kg)'
17 CONSTANT QPC =15. ! 'Alveolar ventilation rate (1/hr)'
18 CONSTANT QCC=15. !'Cardiac output (1/hr)'
19
20 IFlows to Tissue Compartments'
21 CONSTANT QLC = 0.25 !'Fractional blood flow to liver'
22 CONSTANT QFC = 0.05 !'Fractional blood flow to fat'
23 CONSTANT QSC = 0.18 !'Fractional blood flow to slow'
24 QRC =1.0- (QFC + QSC + QLC)
25 CONSTANT SPDC = 1.0 ! diffusion constant for slowly perfused tissues
26
27 ! Volumes of Tissue/Blood Compartments'
28 CONSTANT VLC = 0.04 !Traction liver tissue'
29 CONSTANT VFC = 0.07 !Traction fat tissue'
30 CONSTANT VRC = 0.05 !Traction Rapidly Perf tissue'
31 CONSTANT VBC = 0.05 ! Traction as Blood'
32 VSC = 0.91 - (VLC + VFC + VRC + VBC)
33
34 [Partition Coefficients'
35 CONSTANT PLA=1557. !'Liver/air partition coefficient'
36 CONSTANT PFA = 851. !'Fat/air partition coefficient'
37 CONSTANT PSA = 2065. !'Muscle/air (Slow Perf) partition'
38 CONSTANT PRA=1557. !'Richly perfused tissue/air partition'
39 CONSTANT PB = 1850. !'Blood/air partition coefficient'
40
41 ! Other Compound Specific Parameters'
42 CONSTANT MW = 88.1 !'Molecular weight (g/mol)'
43 CONSTANT KLC = 12.0 ! temp 1st order metab constant
44 CONSTANT VMAXC = 13.8 I'Maximum Velocity of Metabol.'
45 CONSTANT KM = 29.4 I'Michaelis Menten Constant'
46 CONSTANT ORAL = 0.0 !'Oral Bolus Dose (mg/kg)'
May 2009 B-27 DRAFT - DO NOT CITE OR QUOTE
-------
1 CONSTANT KA = 5.0 !'Oral uptake rate (/hr)'
2 CONSTANT WATER = 0.0 ! 'Cone in Water (mg/liter, ppm)'
3 CONSTANT WDOSE=0.0 !Water dose (mg/kg/day) **
4 CONSTANT IV = 0.0 !TV dose (mg/kg)'
5 CONSTANT CONC = 0.0 !'Inhaled concentration (ppm)'
6 CONSTANT KME = 0.276 !'Urinary Elim constant for met (hr-1)'
7
8 ! Timing commands'
9 CONSTANT TSTOP = 50 !'Length of experiment (hrs)'
10 CONSTANT TCHNG= 6 !'Length of inhalation exposure (hrs)'
11 CINTERVAL CINT=0.1
12 CONSTANT WIDD=24. !**
13 CONSTANT PERD=24. !**
14 CONSTANT PERW= 168. !**
15 CONST ANT WIDW= 168. !**
16 CONSTANT DAT=0.017 !**
17
18 ! Scaled parameters calculated in this section of Program'
19 QC=QCC*BW**0.74
20 QP=QPC*BW**0.74
21 QL=QLC*QC
22 QF=QFC*QC
23 QS=QSC*QC
24 QR=QRC*QC
25 VL=VLC*BW
26 VF=VFC*BW
27 VS=VSC*BW
28 VR=VRC*BW
29 VB=VBC*BW
30 PL=PLA/PB
31 PR=PRA/PB
32 PS=PSA/PB
33 PF=PFA/PB
34 KL = KLC*bw**0.7 ! 1st order metab constant
35 VMAX = VMAXC*BW**0.7
36 DOSE = ORAL*BW ITnitial Amount in Stomach'
37 ABO = IV*BW ITnitial Amount in Blood'
38 IDRINK = 0.102*BW**0.7*WATER/24 I'Input from water (mg/hr)' I**
39 IDRINKA = 0.102*BW**0.7*WATER/DAT ITnput from water (mg/hr)' I **
40 DRINKA=WDOSE*BW/DAT
41 CV = ABO/VB I'Initialize CV
42
43 END I'End of INITIAL'
44
45 DYNAMIC
46
47 ALGORITHM IALG = 2 I 'Gear method for stiff systems'
48 TERMT( T .GE. TSTOP )
49 CR = AR/VR
May 2009 B-28 DRAFT - DO NOT CITE OR QUOTE
-------
1 CS = AS/VS
2 CF = AF/VF
3 BODY = AL + AR + AS + AF + AB + TUMMY
4 BURDEN = AM + BODY
5 TMAS S = BURDEN + AX + AMEX
6
7 ! Calculate the Interval Excretion Data here:'
8 ! DAX = AMEX-AMEX2
9 ! IF(DOSE.LE. 0.0 .AND. IV .LE. 0.0 ) GO TO SKIP1
10 ! PCTAX= 100*(AX-AX2)/(DOSE + IV*BW)
11 ! PCTMX = 100*(AMEX - AMEX2)/(DOSE + IV*BW)
12 ! SKIP 1.. CONTINUE
13 ! IF( T .LT. TDATA(I) .OR. I .GE. 20 ) GO TO SKIP
14 ! AX2=AX
15 ! AMEX2=AMEX
16 ! 1=1+1
17 ! SKIP.. CONTINUE
18
19 IDISCRETE EXPOSE
20 ! CIZONE = 1.0 ! CALL LOGD(.TRUE.) Turns on inhalation exposure?
21 !END
22 IDISCRETE CLEAR
23 ! CIZONE = 0.0 ! CALL LOGD(.TRUE.)
24 !END
25
26 DERIVATIVE
27
28 !Use Zero-Crossing Form of DISCRETE Function Here'
29 ! SCHEDULE command must be in DERIVATIVE section'
30 ! DAILY = PULSE (0.0, PERI, TCHNG)
31 ! WEEKLY = PULSE ( 0.0, PER2, LEN2 )
32 ! SWITCHY = DAILY * WEEKLY
33 ! SCHEDULE EXPOSE .XP. SWITCHY - 0.995
34 ! SCHEDULE CLEAR .XN. SWITCHY - 0.005
35
36 DAILY=PULSE(0.0,PERD,WIDD)
37 WEEKLY=PULSE(0.0,PERW,WIDW)
3 8 SWITCHY = DAILY * WEEKLY
39
41 CI = CONC*MW724451.0* SWITCHY!**
42
43 !CA = Concentration in arterial blood (mg/1)'
44 CA = (QC*CV+QP*CI)/(QC+(QP/PB))
45 CX = CA/PB
46
47 DRINK=DRINKA* SWITCHY !**
48
49 ! TUMMY = Amount in stomach'
May 2009 B-29 DRAFT - DO NOT CITE OR QUOTE
-------
1 RTUMMY = -KA*TUMMY
2 TUMMY = INTEG(RTUMMY,DOSE)
3 !RAX = Rate of Elimination in Exhaled air'
4 RAX = QP*CX
5 AX = INTEG(RAX, 0.0)
6
7 ! AS = Amount in slowly perfused tissues (mg)'
8 RAS = SPDC*(CA-CVS) !now governed by diffusion-limited constant, SPDC, instead of QS
9 AS = INTEG(RAS,0.)
10 CVS = AS/(VS*PS)
11
12 ! AR = Amount in rapidly perfused tissues (mg)'
13 RAR = QR*(CA-CVR)
14 AR = INTEG(RAR,0.)
15 CVR = AR/(VR*PR)
16
17 ! AF = Amount in fat tissue (mg)'
18 RAF = QF*(CA-CVF)
19 AF = INTEG(RAF,0.)
20 CVF = AF/(VF*PF)
21
22 !AL = Amount in liver tissue (mg)'
23 RAL = QL*(CA-CVL) - KL*CVL - VMAX*CVL/(KM+CVL) + KA*TUMMY + DRINK
24 AL = INTEG(RAL,0.)
25 CVL = AL/(VL*PL)
26
27 ! AM = Amount metabolized (mg)'
28 RMEX = (KL*CVL)+(VMAX*CVL/(KM+CVL))
29 RAM = (KL*CVL)+(VMAX*CVL)/(KM+CVL) - KME*AM
30
31 AM = INTEG(RAM, 0.0) !'Amt Metabol'
32 CAM = AM/BW ! 'Cone Metabol in body'
33 AMEX = INTEG(KME*AM, 0.0) I'Amt Met Excret'
34
35 ! AB = Amount in Venous Blood'
36 RAB = QF*CVF + QL*CVL + QS*CVS + QR*CVR - QC*CV
37 AB = INTEG(RAB, ABO)
38 CV = AB/VB
39 AUCV = INTEG(CV, 0.0)
40
41 [Possible Dose Surrogates for Risk Assessment Defined Here'
42
43 CEX = 0.667*CX + 0.333*CI I'Conc inExhal Air'
44 AVECON = PLA * (CEX+CI)/2 !'Ave Cone in Nose Tissue'
45 AUCCON = INTEG(AVECON, 0.0) !'Area under Curve (Nose)'
46
47 AUCMET = INTEG(CAM, 0.0) I'Area under Curve (Metab)'
48
49 CL = AL/VL ! 'Cone Liver Tissue'
May 2009 B-30 DRAFT - DO NOT CITE OR QUOTE
-------
1 AUCL = INTEG(CL, 0.0) ! 'Area under Curve (Liver)'
2 AAUCL=AUCL/TIME
3
4 ! Dose Surrogates are Average Area under Time/Cone Curve per 24 hrs'
5 IF (T .GT. 0) TIME=T
6 DAYS = TIME/24.0
7 NOSE = AUCCON/DAYS !'Nasal Turbinates'
8 LIVER = AUCL/DAYS ! 'Liver Tissues'
9 METAB = AUCMET/DAYS !'Stable Metabolite'
10
11 END !'End of dynamic'
12
13 END ! End of TERMINAL
14
15 END !'End of PROGRAM
May 2009 B-31 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX C. DETAILS OF BMD ANALYSIS FOR ORAL RfD FOR 1,4-dioxane
1
2
3
4
C.I. CORTICAL TUBULE DEGENERATION.
All available dichotomous models in the Benchmark Dose Software (version 1.3.2) were
fit to the incidence data shown in Table C-l, for cortical tubule degeneration in male and female
Osborne-Mendel rats exposed to 1,4-dioxane in the drinking water (NCI, 1978). Doses
associated with a BMR of a 10% extra risk was calculated.
Table C-l. Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years
Males (mg/kg-day)
0
0/3 la
240
20/3 lb
(65%)
530
27/3 3b
(82%)
Females (mg/kg-day)
0
0/3 la
350
0/34
640
10/32b
(31%)
5
6
7
8
9
10
11
12
13
14
15
"Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this
review.
blncidence significantly elevated compared to control by Fisher's exact test (p < 0.05) performed for this review.
Source: NCI (1978).
As assessed by the %2 goodness-of-fit test, several models in the software provided
adequate fits to the data for the incidence of cortical tubule degeneration in male and female rats
(tf'p > 0.1) (Table C-2). Comparing across models, a better fit is indicated by a lower AIC
value (U.S. EPA, 2000b). As assessed by Akaike's Information Criterion (AIC), the log-probit
model provided the best fit to the cortical tubule degeneration incidence data for male rats (Table
C-2, Figure C-l) and the log-logistic and Weibull models provided the best fit to the data for
female rats (Table C-2, Figures C-2 and C-3). For those models that exhibit adequate fit, models
with the lower AIC values are preferred. Differences in AIC values of less than 1 are generally
not considered important. Benchmark doses (BMDs) and benchmark dose lower confidence
limits (BMDLs) associated with an extra risk of 10% were calculated for all models. These
values are also shown in Table C-2.
May 2009
C-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-2. Goodness-of-fit statistics and BMDio and BMDLio values from
models fit to incidence data for cortical tubule degeneration in male and
female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water
Model
X2 Goodness-of-Fit
Test/7-Valuea
AIC
BMD10
(mg/kg-day)
BMDL10
(mg/kg-day)
Male
Gammab
Logistic
Log-logistic0
MSe
Probit
Log-probif'd
Weibullb
0.65
0.001
1.00
0.65
0.001
0.75
0.65
74.46
89.01
75.62
74.46
88.78
74.17
74.46
28.80
88.48
20.85
28.80
87.10
51.41
28.80
22.27
65.84
8.59
22.27
66.32
38.53
22.27
Female
Gammab
Logistic
Log-logistic0'"1
MSe
Probit
Log-probif
Quanta! quadratic
Weibullb
0.95
0.9996
0.9999
0.03
0.9997
0.9997
0.14
0.9999
41.97
43.75
41.75
52.30
43.75
43.75
48.20
41.75
524.73
617.44
591.82
306.21
596.02
584.22
399.29
596.44
437.08
471.92
447.21
189.49
456.42
436.19
314.00
452.36
a p-Value from the %2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited
a statistically significant lack of fit, and thus a different model should be chosen.
bPower restricted to >1.
°Slope restricted to >1.
dBest-fitting model.
eBetas restricted to >0.
Source: NCI (1978).
May 2009
C-2
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
1 F
0.8
0.6
| 0.4
o
ro
0.2
Probit
MD
100 200 300
dose
400
500
20:2411/202006
Source: NCI (1978).
Figure C-l. BMD log probit model of cortical tubule degeneration incidence data
for male rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-2.
Log-Logistic Model with 0.95 Confidence Level
0.5
0.4
1 0.3
••8 °-2
£
LJ_
0.1
Log-Logistic
.3MD
iMD
100 200
300 400
dose
500
600
700
20:31 11/202006
Source: NCI (1978).
Figure C-2. BMD log-logistic model of cortical tubule degeneration incidence data
for female rats exposed to 1,4-dioxane in drinking water for 2 years to support
results Table C-2.
May 2009
C-3
DRAFT - DO NOT CITE OR QUOTE
-------
Weibull Model with 0.95 Confidence Level
0.5
0.4
2 0.3
I 0.2
0.1
Weibull
B.MP
BMD.
100
200
300 400
dose
500
600
700
20:3511/202006
Source: NCI (1978).
Figure C-3. BMD Weibull model of cortical tubule degeneration incidence data for
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-2.
1 An alternative BMR of 20% was calculated from the best fitting model for each data set.
2 The doses associated with the 20% extra risk (BMD20) and the 95% lower confidence limits
3 (BMDL20) for the incidence of cortical tubule degeneration in male rats are 79.81 and
4 59.82 mg/kg-day, respectively. The doses associated with the BMD20 and the BMDL20 for
5 female rats are 619.09 and 533.88 mg/kg-day for the log-logistic model, and 621.84 and
6 541.58 mg/kg-day for the Weibull model.
C.2. LIVER HYPERPLASIA.
7 All available dichotomous models in the Benchmark Dose Software (version 1.3.2) were
8 fit to the incidence data shown in Table C-1, for cortical tubule degeneration in male and female
9 Osborne-Mendel rats exposed to 1,4-dioxane in the drinking water (NCI, 1978). Doses
10 associated with a BMR of a 10% extra risk was calculated.
May 2009
C-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-3. Incidence of liver hyperplasia in F344/DuCrj rats exposed to
1,4-dioxane in drinking water
Males (mg/kg-day)
0
3/40
16
2/45
81
9/3 5a
398
12/22b
Females (mg/kg-day)
0
0/3 8a
21
0/37
103
1/38
514
14/24b
""Statistically significant compared to controls by the Dunnett's test (p < 0.05).
blncidence significantly elevated compared to control by %2 test (p < 0.01).
Source: JBRC(1998a).
1 For incidence of liver hyperplasia, the logistic and probit models both exhibited a
2 statistically significant lack of fit (i.e., x2/?-value < 0.1; see Table C-4), and thus should not be
3 considered further for identification of a POD. All of the remaining models exhibited adequate
4 fit, but the AIC values for the gamma, multistage, quantal-linear, and Weibull models were
5 significantly lower than the AIC values for the log-logistic and log-probit models. Thus, the log-
6 logistic and log-probit models should also be eliminated from further consideration for POD
7 determination. Finally, the AIC values for gamma, multistage, quantal-linear, and Weibull
8 models in Table C-4 are equivalent and, in this case, essentially represent the same model.
9 Therefore, consistent with the "Benchmark Dose Technical Guidance" (EPA, 2000b), any of the
10 highlighted models could be used to identify a POD for this endpoint of 34.7 mg/kg-day.
Table C-4. Benchmark dose modeling results based on the incidence of liver
hyperplasias in F344 male rats exposed to 1,4-dioxane in drinking water for 2
years
Fitted Dichotomous
Model3
Gamma
Logistic
Log-Logistic
MS (1-degree)
Probit
Log-Probit
Quantal-Linear
Weibull
X2 Goodness-of-Fit Test
/7-Valuea
0.35
0.07
0.19
0.35
0.09
0.15
0.35
0.35
AIC
114.13
116.99
115.73
114.13
116.61
115.47
114.13
114.13
BMD10
(mg/kg-day)
52.3
121.4
49.0
52.3
111.4
79.9
52.3
52.3
BMDL10
(mg/kg-day)
34.7
92.0
24.8
34.7
85.2
54.1
34.7
34.7
a/>-Value from the %2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model
exhibited a statistically significant lack of fit, and thus a different model should be chosen.
Source: JBRC(1998a).
May 2009
C-5
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
% 0-5
!t
=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0853659
Slope = 0.00334063
Power = 1.3
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Power, have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
Background
Slope
Background
1
-0.36
Slope
-0.36
1
May 2009
C-6
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Parameter Estimates
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
0.00230091 0.111184
0.000921605 0.00310987
NA
NA - Indicates that this parameter has hit a bound implied by some inequality
constraint and thus has no standard error.
Variable
Background
Slope
Power
Estimate
0.0567425
0.00201574
1
Std. Err.
0.0277768
0.000558241
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-53.9471
-55.0634
-67.6005
114.127
# Param' s
4
2
1
Deviance Test
2.23256
27.3066
d.f .
2
3
P-value
0.3275
<.0001
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0567
0.0867
0.1988
0.5771
Expected
2.270
3.901
6.959
12. 697
Observed
3
2
9
12
Size
40
45
35
22
Scaled
Residual
0.499
-1.007
0.864
-0.301
= 2.10 d.f. = 2
Benchmark Dose Computation
P-value = 0.3499
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
52.269
34.6825
May 2009
C-7
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
1 °-5
.g
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Variable
intercept
slope
Model
Full model
Fitted model
Reduced model
AIC:
Parameter Estimates
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
-2.29444 0.320777 -2.92315 -1.66572
0.00654235 0.00141544 0.00376813 0.00931657
Analysis of Deviance Table
Log(likelihood)
-53.9471
-56.4957
-67.6005
116.991
# Param' s
4
2
1
Deviance
5.09704
27.3066
Test d.f.
2
3
P-value
0.0782
<.0001
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0916
0.1007
0.1462
0.5767
Expected
3.663
4.530
5.118
12.688
Observed
3
2
9
12
Size
40
45
35
22
Scaled
Residual
-0.364
-1.254
1.857
-0.297
ChiA2 = 5.24
d.f. = 2
P-value = 0.0728
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
121.431
92.0138
May 2009
C-9
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
1 0.5
I
< 0.4
c
o
'•G 0.3
ro
LL
0.2
0.1
0
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
BMD
50
100
150
200
dose
250
300
350
400
13:1606/122008
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Figure C-6. BMD log-logistic model of liver hyperplasia incidence data for F344
male rats exposed to 1,4-dioxane in drinking water for 2 years to support the results
in Table C-4.
Logistic Model. (Version: 2.9; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\MALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\MALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 13:16:30 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = Response
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.075
intercept = -8.16843
slope = 1.41583
May 2009
C-10
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Asymptotic Correlation Matrix of Parameter Estimates
background intercept
background 1 -0.34
intercept -0.34 1
slope 0.28 -0.98
slope
0.28
-0.98
1
Parameter Estimates
Variable
background
intercept
slope
Estimate
0.0550676
-6.66232
1.1471
Std. Err.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-53.9471
-54.8671
-67.6005
115.734
Param' s
4
3
1
Deviance
1.83982
27.3066
Test d.f.
1
3
P-value
0.175
<.0001
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0551
0.0833
0.2110
0.5757
Expected
2.203
3.747
7.385
12.666
Observed
3
2
9
12
Size
40
45
35
22
Scaled
Residual
0.553
-0.942
0.669
-0.287
ChiA2 = 1.72
d.f. = 1
P-value = 0.1892
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
49.0334
24.8079
May 2009
C-ll
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
!t
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background
Beta(l)
Variable
Background
Beta(l)
1 -0.49
-0.49 1
Parameter Estimates
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0567425 * * *
0.00201574 * * *
* - Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0.
16.0000 0.
81.0000 0.
398.0000 0.
Chi^2 =2.10
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
BMDU
Analysis of Deviance Table
Log (likelihood) # Param's Deviance Test d.f. P-value
-53.9471 4
-55.0634 2 2.23256 2 0.3275
-67.6005 1 27.3066 3 <.0001
114.127
Goodness of Fit
Scaled
. Prob. Expected Observed Size Residual
0567 2.270 3 40 0.499
0867 3.901 2 45 -1.007
1988 6.959 9 35 0.864
5771 12.697 12 22 -0.301
d.f. = 2 P-value = 0.3499
Computation
0.1
= Extra risk
0.95
52.269
34.6825
88.4683
Taken together, (34.6825, 88.4683) is a 90% two-sided confidence interval for the BM:
May 2009
C-13
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
!t
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background, have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
intercept
slope
intercept
1
-0.62
slope
-0.62
1
Parameter Estimates
Variable
intercept
slope
Model
Full model
Fitted model
Reduced model
Estimate
-1.35669
0.00391559
Analysis
Log (likelihood)
-53.9471
-56.3042
-67.6005
Std. Err.
0.169086
0.000838003
Lower Conf. Limi
-1.68809
0.00227313
of Deviance Table
# Param' s
4
2
1
Deviance Test d.f.
4.71415 2
27.3066 3
95.0% Wald Confidence Interval
Upper Conf. Limit
-1.02528
0.00555805
P-value
0.0947
<.0001
AIC:
116.608
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0874
0.0978
0.1493
0.5799
Expected
3.498
4.402
5.225
12.758
Observed
3
2
9
12
Size
40
45
35
22
Scaled
Residual
-0.279
-1.205
1.791
-0.328
=4.84
d.f. = 2
P-value = 0.0887
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
111.437
85.1525
May 2009
C-15
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
| 0.5
fc
< 0.4
c
o
t> 0.3
5
LJ_
0.2
0.1
0
Probit Model with 0.95 Confidence Level
Probit
BMD
.BMP , , , ,,
50
100
150
200
dose
250
300
350 400
13:2406/122008
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Figure C-9. BMD probit model of liver hyperplasia incidence data for F344 male
rats exposed to 1,4-dioxane in drinking water for 2 years, accounting for
background incidence.
Probit Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\MALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\MALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.pit
Thu Jun 12 13:24:01 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = Background + (1-Background) * CumNorm(Intercept+Slope*Log(Dose)) ,
where CumNorm(.) is the cumulative normal distribution function
Dependent variable = Response
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial (and Specified) Parameter Values
background = 0.075
intercept = -5.64933
slope = 1
May 2009
C-16
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter (s) -slope, have been estimated at a boundary point,
been specified by the user, and do not appear in the correlation matrix )
background intercept
background 1 -0.45
intercept -0.45 1
Parameter Estimates
95.0% Wald Confidence Inte
Variable Estimate Std. Err. Lower Conf. Limit Upper Con
background 0.0712134 0.0303379 0.0117523 0.
intercept -5.66248 0.263846 -6.17961 -5.
slope 1 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log (likelihood) # Param's Deviance Test d.f. P-value
Full model -53.9471 4
Fitted model -55.7325 2 3.57078 2 0.1677
Reduced model -67.6005 1 27.3066 3 <.0001
AIC: 115.465
Goodness of Fit
Scaled
Dose Est. Prob. Expected Observed Size Residual
0.0000 0.0712 2.849 3 40 0.093
16.0000 0.0730 3.285 2 45 -0.736
81.0000 0.1663 5.821 9 35 1.443
398.0000 0.6536 14.379 12 22 -1.066
Chi^2 = 3.77 d.f. = 2 P-value = 0.1519
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 79.9119
BMDL = 54.0772
or have
rval
f. Limit
130675
14535
May 2009
C-17
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
% 0-5
!t
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Parameter Estimates
Variable
Background
Slope
Model
Full model
Fitted model
Reduced model
Estimate
0.0567426
0.00201574
Analysis
Log (likelihood)
-53.9471
-55.0634
-67.6005
Std. Err.
0.0277774
0.000558246
of Deviance
# Param' s
4
2
1
Lower Conf.
0.00229997
0.000921598
Table
Deviance Test
2.23256
27.3066
Limit
d.f .
2
3
95.0% Wald Confidence Interval
Upper Conf. Limit
0.111185
0.00310988
P-value
0.3275
<.0001
AIC:
114.127
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0567
0.0867
0.1988
0.5771
Expected
2.270
3.901
6.959
12. 697
Observed
3
2
9
12
Size
40
45
35
22
Scaled
Residual
0.499
-1.007
0.864
-0.301
=2.10
d.f. = 2
P-value = 0.3499
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
52.2689
34.6825
May 2009
C-19
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
% 0-5
!t
=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0853659
Slope = 0.00174595
Power = 1
May 2009
C-20
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Power, have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
Background
Slope
Background
1
-0.36
Slope
-0.36
1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0567426 0.0277773 0.00229996 0.111185
Slope 0.00201574 0.000558245 0.000921596 0.00310988
Power 1 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-53.9471
-55.0634
-67.6005
114.127
# Param's
4
2
1
Deviance Test d.f.
P-value
2.23256
27.3066
0.3275
<.0001
Goodness of Fit
Dose Est._Prob. Expected Observed Size
Scaled
Residual
0.0000
16.0000
81.0000
398.0000
0.0567
0.0867
0.1988
0.5771
2.270
3.901
6.959
12.697
3
2
9
12
40
45
35
22
0.499
-1.007
0.864
-0.301
ChiA2 =2.10
d.f. = 2
P-value = 0.3499
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
52.269
34.6825
For liver hyperplasias in F344 female rats exosed to 1,4-dioxane, the quantal-linear
model exhibited a statistically significant lack of fit (i.e., x2/>-value < o.l; See Table C-5), and
thus should not be considered further for identification of a POD. All of the remaining models
exhibited adequate fit, but the AIC values for the logistic, multistage, and probit models were
significantly lower than the AIC values for the gamma, log-logistic, log-probit, and Weibull
models. Thus, the gamma, log-logistic, log-probit, and Weibull models should also be
eliminated from further consideration for POD determination. Finally, the AIC values for the
three highlighted models in Table C-5 were essentially equivalent. Therefore, consistent with
the "Benchmark Dose Technical Guidance" (EPA, 2000b), the BMDLs from these three
highlighted models were averaged to yield a POD for this endpoint of 45.7 mg/kg-day.
May 2009
C-21
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-5. Benchmark dose modeling results based on the incidence of liver
hyperplasias in F344 female rats exposed to 1,4-dioxane in drinking water for
2 years
Fitted Dichotomous
Model
Gamma
Logistic
Log-Logistic
MS (2-degree)
Probit
Log-Probit
Quantal-Linear
Weibull
X2 Goodness-of-Fit Test
/7-Valuea
0.98
0.92
0.98
0.95
0.92
0.98
0.02
1.0
AIC
78.84
77.03
78.84
77.01
77.02
78.84
87.66
78.83
BMD10
(mg/kg-day)
88.7
67.6
96.7
69.0
64.8
92.8
26.2
81.7
BMDL10
(mg/kg-day)
51.2
50.8
64.2
38.7
47.5
64.0
19.0
45.3
a/>-Value from the %2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibits
a statistically significant lack of fit, and thus a different model should be chosen.
Source: JBRC(1998a).
Gamma Multi-Hit Model with 0.95 Confidence Level
0.8
0.6
c
'I 0.4
ro
0.2
0
Gamma Multi-Hit
[BMP
0 100
14:5206/122008
200 300
dose
400
500
Source: JBRC (1998a).
Figure C-12. BMD gamma model of liver hyperplasia incidence data for F344
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
May 2009
C-22
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
Gamma Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 14:52:14 2008
BMDS MODEL RUN
The form of the probability function is:
P[response]= background+(1-background)*CumGamma[slope*dose,power],
where CumGamma(.) is the cummulative Gamma distribution function
Dependent variable = Response
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0641026
Slope = 0.00375157
Power = 1.3
Asymptotic Correlation Matrix of Parameter Estimates
Background Slope Power
Background 1 0.021 0.021
Slope 0.021 1 1
Power 0.021 1 1
Parameter Estimates
Variable
Background
Slope
Power
Model
Full model
Fitted model
Reduced model
AIC:
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0533319 0.0259525 0.00246604 0.104198
0.0713984 3.22468 -6.24886 6.39166
10.1413 385.373 -745.176 765.459
Analysis of Deviance Table
Log(likelihood)
-36.4175
-36.4178
-79.9164
78.8357
# Param's
4
3
1
Deviance Test d.f.
P-value
0.000751059
86.9979
0.9781
<.0001
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0533
0.0533
0.2368
1.0000
Expected
2.027
1.973
8.999
24.000
Observed
2
2
9
24
Size
38
37
38
24
Scaled
Residual
-0.019
0.019
0.000
0.001
ChiA2 = 0.00 d.f. = 1
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 88.7086
BMDL = 51.1769
P-value = 0.9782
May 2009
C-23
DRAFT - DO NOT CITE OR QUOTE
-------
Logistic Model with 0.95 Confidence Level
0.8
o> 0.6
o
'•6 0.4
ro
0.2
0
Logistic
BMD
BMD
0 100
14:5506/122008
200 300
dose
400
500
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Source: JBRC (1998a).
Figure C-13. BMD logistic model of liver hyperplasia incidence data for F344
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
Logistic Model. (Version: 2.9; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 14:55:04 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = I/[1+EXP(-intercept-slope*dose)]
Dependent variable = Response
Independent variable = Dose
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
background = 0 Specified
intercept = -2.70218
slope = 0.0129047
May 2009
C-24
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background, have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
intercept
slope
intercept
1
-0.82
slope
-0.82
1
Parameter Estimates
Variable
intercept
slope
Model
Full model
Fitted model
Reduced model
AIC:
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
-3.1071 0.540423 -4.16631 -2.04789
0.01894 0.00613873 0.00690836 0.0309717
Analysis of Deviance Table
Log(likelihood)
-36.4175
-36.5147
-79.9164
77.0294
# Param's
4
2
1
Deviance Test d.f.
P-value
0.194506
86.9979
0.9073
<.0001
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0428
0.0624
0.2393
0.9987
Expected
1.627
2.310
9.095
23.968
Observed
2
2
9
24
Size
38
37
38
24
Scaled
Residual
0.299
-0.210
-0.036
0.178
=0.17
d.f. = 2
P-value = 0.9200
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
67.5596
50.8415
May 2009
C-25
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
o> 0.6
o
'•6 0.4
ro
0.2
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
100
200 300
dose
400
500
16:11 06/122008
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Source: JBRC (1998a).
Figure C-14. BMD log-logistic model of liver hyperplasia incidence data for F344
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
Logistic Model. (Version: 2.9; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 16:11:17 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose)
Dependent variable = Response
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.0526316
intercept = -16.4238
slope = 3.24981
May 2009
C-26
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Asymptotic Correlation Matrix of Parameter Estimates
background intercept slope
background 1 0.018 -0.018
intercept 0.018 1 -1
slope -0.018 -1 1
Parameter Estimates
Limit
Variable
background
intercept
slope
Estimate
0.0533333
-58.0881
12.2257
Std. Err.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf.
Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log (likelihood)
-36.4175
-36.4178
-79.9164
# Param' s
4
3 0.
1
Deviance
,000751795
86.9979
Test d.f.
1
3
P-value
0.9781
<.0001
78.8357
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0533
0.0533
0.2368
1.0000
Expected
2.027
1.973
9.000
24.000
Observed
2
2
9
24
Size
38
37
38
24
Scaled
Residual
-0.019
0.020
0.000
0.001
=0.00
d.f. = 1
P-value = 0.9781
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
96.6969
64.2472
May 2009
C-27
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
T3
CD
0.6
c
'I 0.4
ro
0.2
Multistage Model with 0.95 Confidence Level
Multistage
BMDI
BMP
0
16:1606/122008
100
200
300
400
500
dose
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Source: JBRC (1998a).
Figure C-15. BMD multistage model of liver hyperplasia incidence data for F344
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
Multistage Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.pit
Thu Jun 12 16:16:25 2008
BMDS MODEL RUN
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = Response
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 3.83316e+014
May 2009
C-28
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
0
o
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
Asymptotic
Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(l), have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
Background Beta (2)
Background
Beta(2)
Variable
Background
Beta(l)
Beta (2)
1 -0.37
-0.37 1
Parameter Estimates
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0480711 * * *
o * * *
2.21049e-005 * * *
* - Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0.
21.0000 0.
103.0000 0.
514.0000 0.
ChiA2 =0.11
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
BMDU
Analysis of Deviance Table
Log (likelihood) # Param's Deviance Test d.f. P-value
-36.4175 4
-36.5069 2 0.178792 2 0.9145
-79.9164 1 86.9979 3 <.0001
77.0137
Goodness of Fit
Scaled
. Prob. Expected Observed Size Residual
0481 1.827 2 38 0.131
0573 2.120 2 37 -0.085
2471 9.388 9 38 -0.146
9972 23.934 24 24 0.258
d.f. = 2 P-value = 0.9453
Computation
0.1
= Extra risk
0.95
69.0391
38.6809
92.891
Taken together, (38.6809, 92.891 ) is a 90% two-sided confidence interval for the BMD
May 2009
C-29
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
0.8
o> 0.6
o
'•6 0.4
ro
0.2
0
Probit
BMD
BMD,
0 100
16:1806/122008
200 300
dose
400
500
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Source: JBRC (1998a).
Figure C-16. BMD probit model of liver hyperplasia incidence data for F344 female
rats exposed to 1,4-dioxane in drinking water for 2 years to support the results in
Table C-5.
Probit Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 16:17:59 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = CumNorm(Intercept+Slope*Dose),
where CumNorm(.) is the cumulative normal distribution function
Dependent variable = Response
Independent variable = Dose
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
background = 0 Specified
intercept = -1.61994
slope = 0.00769493
May 2009
C-30
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
intercept
slope
intercept
1
-0.76
slope
-0.76
1
Parameter Estimates
Variable
intercept
slope
Estimate
-1.71967
0.00975802
Std. Err.
0.249073
0.00303637
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-2.20785 -1.2315
0.00380686
0.0157092
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-36.4175
-36.5097
-79.9164
77.0193
# Param's
4
2
1
Deviance Test d.f.
P-value
0.184385
86.9979
0.9119
<.0001
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0427
0.0649
0.2374
0.9995
Expected
1. 624
2.402
9.022
23.988
Observed
2
2
9
24
Size
38
37
38
24
Scaled
Residual
0.301
-0.268
-0.009
0.109
=0.17
d.f. = 2
P-value = 0.9164
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
64.8143
47.5376
May 2009
C-31
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
0.8
o> 0.6
o
'•6 0.4
ro
0.2
0
Probit
0 100
16:2006/122008
200 300
dose
400
500
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Source: JBRC (1998a).
Figure C-17. BMD log-probit model of liver hyperplasia incidence data for F344
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
Probit Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 16:20:01 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = Background + (1-Background) * CumNorm(Intercept+Slope*Log(Dose)
where CumNorm(.) is the cumulative normal distribution function
Dependent variable = Response
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
May 2009
C-32
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Default Initial (and Specified) Parameter Values
background = 0.0526316
intercept = -7.80926
slope = 1.55016
Asymptotic Correlation Matrix of Parameter Estimates
background intercept slope
background 1 -0.0044 0.0043
intercept -0.0044 1 -1
slope 0.0043 -1 1
Parameter Estimates
Variable
Limit
background
intercept
slope
Model
Full model
Fitted model
Reduced model
AIC:
Estimate
0.0533333
-19.3386
3.98616
Std. Err.
0.0259454
742.282
160.156
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf.
0.0024812
-1474.19
-309.915
0.104185
1435.51
317.887
Analysis of Deviance Table
Log(likelihood)
-36.4175
-36.4178
-79.9164
78.8357
# Param' s
4
3 0.
1
Deviance Test
,000751916
86.9979
d.f .
1
3
P-value
0.9781
<.0001
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0533
0.0533
0.2368
1.0000
Expected
2.027
1.973
9.000
24.000
Observed
2
2
9
24
Size
38
37
38
24
Scaled
Residual
-0.019
0.020
0.000
0.001
=0.00
d.f. = 1
P-value = 0.9781
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
92.7521
63.951
May 2009
C-33
DRAFT - DO NOT CITE OR QUOTE
-------
Quantal Linear Model with 0.95 Confidence Level
0.8
o> 0.6
o
'•6 0.4
ro
0.2
0
Quantal Linear
0 100
16:21 06/122008
200 300
dose
400
500
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Source: JBRC (1998a).
Figure C-18. BMD quantal linear model of liver hyperplasia incidence data for
F344 female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
Quantal Linear Model using Weibull Model (Version: 2.7; Date: 2/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.plt
Thu Jun 12 16:21:35 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-slope*dose)]
Dependent variable = Response
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0641026
Slope = 0.00748205
Power = 1 Specified
May 2009
C-34
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter (s) -Power, have been estimated at a boundary point, or have
been specified by
the user, and do not appear in the correlation matrix )
Background Slope
Background
Slope
Variable
Background
Slope
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0.
21.0000 0.
103.0000 0.
514.0000 0.
Chi^2 =7.62
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
1 -0.16
-0.16 1
Parameter Estimates
95.0% Wald Confidence Interval
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0302097 0.0203846 -0.00974332 0.0701627
0.00402312 0.000813792 0.00242812 0.00561813
Analysis of Deviance Table
Log (likelihood) # Param's Deviance Test d.f. P-value
-36.4175 4
-41.8322 2 10.8294 2 0.004451
-79.9164 1 86.9979 3 <.0001
87.6644
Goodness of Fit
Scaled
. Prob. Expected Observed Size Residual
0302 1.148 2 38 0.808
1088 4.025 2 37 -1.069
3592 13.650 9 38 -1.572
8774 21.057 24 24 1.832
d.f. = 2 P-value = 0.0221
Computation
0.1
= Extra risk
0.95
26.1887
19.0079
May 2009
C-35
DRAFT - DO NOT CITE OR QUOTE
-------
Weibull Model with 0.95 Confidence Level
0.8
T3
CD
0.6
c
'I 0.4
ro
0.2
0
Weibull
0 100
16:4406/122008
200 300
dose
400
500
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Source: JBRC (1998a).
Figure C-19. BMD Weibull model of liver hyperplasia incidence data for F344
female rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in Table C-5.
Weibull Model using Weibull Model (Version: 2.7; Date: 2/20/2007)
Input Data File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.(d)
Gnuplot Plotting File: M:\DIOXANE DOSE-RESPONSE
MODELING\FEMALE_RATS_LIVER_HYPERPLASIA_JBRC_1998.pit
Thu Jun 12 16:44:49 2008
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background) * [1-EXP (-slope*dose/xpower) ]
Dependent variable = Response
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0641026
Slope = 5.51356e-007
Power = 2.5244
May 2009
C-36
DRAFT - DO NOT CITE OR QUOTE
-------
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
Background
Slope
Power
1 -0
-0.42
0.42
.42
1
-1
0.42
-1
1
Parameter Estimates
Variable
Background
Slope
Power
Model
Full model
Fitted model
Reduced model
Estimate
0.052616
1.18324e-007
3.11095
Analysis
Log (likelihood)
-36.4175
-36.4175
-79.9164
Std. Err.
0.0286871
95.0% Wald
Lower Conf.
-0.00360976
Confidence Interval
Limit Upper Conf. Limit
0.108842
5.88858e-006 -1 . 14231e-005 1.16597e-005
10.7258
of Deviance
# Param' s
4
3 3,
1
-17.9112
Table
Deviance Test
.92191e-007
86.9979
24.1331
d.f. P-value
1 0.9995
3 <.0001
AIC:
78.8349
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0526
0.0541
0.2368
1.0000
Expected
1.999
2.001
9.000
24.000
Observed
2
2
9
24
Size
38
37
38
24
Scaled
Residual
0.000
-0.000
0.000
0.000
ChiA2 =0.00
d.f. = 1
P-value = 0.9995
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
81.747
45.2828
May 2009
C-37
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX D. DETAILS OF BMD ANALYSIS FOR ORAL CSF FOR 1,4-dioxane
1 The multistage (MS) model in the Benchmark Dose Software (BMDS) (version 1.3.2)
2 was fit to the incidence data for hepatocellular carcinoma and/or adenoma for mice and rats,
3 nasal cavity tumors, peritoneal mesothelioma, and mammary gland adenomas in rats and mice
4 exposed to 1,4-dioxane in the drinking water. Doses associated with a benchmark response
5 (BMR) of a 10% extra risk were calculated with the polynomial degree initially set at n-1 and
6 lower. BMDio and BMDLio values from the lowest degree polynomial models with an adequate
7 fit (%2' p > 0.1) were reported (U.S. EPA, 2000b). A summary of the model predictions for the
8 JBRC (1998a) study are shown in Table D-l. The data and BMD modeling results are presented
9 separately for each dataset as follows:
10 • Hepatic adenomas and carcinomas in female F344 rats (Tables D-2 and D-3; Figures D-l,
11 D-2, and D-3)
12 • Hepatic adenomas and carcinomas in male F344 rats (Tables D-4 and D-5; Figures D-4
13 and D-5)
14 • Significant tumor incidence data at sites other than the liver (i.e., nasal cavity, mammary
15 gland adenoma, and peritoneal mesothelioma) in male and female F344 rats (Table D-6)
16 o Nasal cavity tumors in female F344 rats (Table D-7; Figures D-6 and D-7)
17 o Nasal cavity tumors in male F344 rats (Table D-8)
18 o Mammary gland adenomas in female F344 rats (Table D-9; Figures D-8 and D-9)
19 o Peritoneal mesotheliomas in male F344 rats (Table D-10; Figures D-10)
20 • Hepatic adenomas and carcinomas in female BDFi mice (Tables D-l 1 and D-12; Figure
21 D-ll)
22 • Hepatic adenomas and carcinomas in male BDFi mice (Tables D-l3 and D-l4; Figure
23 D-12)
24 • MS models for male and female F344 rats (Table D-l5)
25 o MS-Combo analysis for F344 rats (Tables D-16 and D-17)
26 Data and BMD modeling results from the additional chronic bioassays (NCI, 1978; Kociba et al.,
27 1974) were evaluated in comparison to the JBRC (1998a) study. These results are presented as
28 follows:
29 • Calculation of HEDs for additional studies reporting the incidence of liver and nasal
30 cavity tumors in rats and mice exposed to 1,4-dioxane in the drinking water for 2 years
31 (Table D-l 8)
32 • Summary of BMD modeling estimates and CSF values associated with liver and nasal
33 tumor incidence data from chronic oral exposure to 1,4-dioxane in rats and mice (Table
34 D-19)
May 2009 D-1 DRAFT - DO NOT CITE OR QUOTE
-------
1 • Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in male and
2 female Sherman rats (combined) treated with 1,4-dioxane in the drinking water for
3 2 years (Table D-20)
4 o Goodness-of-fit statistics, BMDio HED and BMDLio HED values from MS models fit to
5 incidence data for hepatocellular carcinoma and nasal tumors in male and female
6 Sherman rats (combined) exposed to 1,4-dioxane in drinking water for 2 years (Table
7 D-21; Figures D-13 and D-14)
8 • Incidence of nasal cavity squamous cell carcinoma and hepatocellular adenoma in
9 Osborne-Mendel rats exposed to 1,4-dioxane in the drinking water (Table D-22)
10 o Goodness-of-fit statistics and BMDio HED and BMDLio HED values from MS models fit
11 to incidence data for hepatocellular adenoma and nasal tumors in male and female
12 Osborne-Mendel rats exposed to 1,4-dioxane in the drinking water for 2 years (Table
13 D-23; Figures D-15, D-16, and D-17)
14 • Incidence of hepatocellular adenoma or carcinoma in B6C3Fi mice exposed to
15 1,4-dioxane in drinking water (Table D-24)
16 o Goodness-of-fit statistics and BMDio HED and BMDLio HED values from MS models fit
17 to incidence data for hepatocellular adenoma or carcinoma in male and female
18 B6C3Fi mice exposed to 1,4-dioxane in the drinking water for 2 years (Table D-25)
D.I. GENERAL ISSUES AND APPROACHES TO BMDS MODELING
D.I.I. Combining Data on Adenomas and Carcinomas
19 The incidence of adenomas and the incidence of carcinomas within a dose group at a site
20 or tissue in rodents are sometimes combined. This practice is based upon the hypothesis that
21 adenomas are a severe endpoint by themselves and most would have developed into carcinomas
22 if exposure at the same dose was continued (U.S. EPA, 2005a). The incidence at high doses of
23 both tumors in rat and mouse liver is high in the key study (JBRC, 1998a). Consequently it is
24 necessary to add the incidence of hepatic adenomas and carcinomas without double-counting
25 them so as to calculate the combined incidence of either a hepatic carcinoma or a hepatic
26 adenoma or both in rodents.
May 2009 D-2 DRAFT - DO NOT CITE OR QUOTE
-------
1 The variable N is used to denote the total number of animals tested in the dose group.
2 The variable Y is used here to denote the number of rodents within a dose group that have
3 characteristic X, and the notation Y(X) is used to identify the number with a specific
4 characteristic X. Modeling was performed on the adenomas and carcinomas separately and the
5 following combinations of tumor types:
6 • Y(adenomas) = number of animals with adenomas, whether or not carcinomas are
7 present;
8 • Y(carcinomas) = number of animals with carcinomas, whether or not adenomas are also
9 present;
10 • Y(adenomas and carcinomas) = number of animals with both adenomas and carcinomas
11 present in the same animal;
12 • Y(either adenomas or carcinomas) = number of animals with adenomas, or carcinomas,
13 or both = Y(adenomas) + Y(carcinomas) - Y(both adenomas and carcinomas);
14 • Y(neither adenomas nor carcinomas) = number of animals with no adenomas and no
15 carcinomas = N - Y(either adenomas or carcinomas;
16 • Y(only carcinomas and not adenomas) = Y(carcinomas) - Y(adenomas and carcinomas);
17 • Y(only adenomas and not carcinomas) = Y(adenomas) - Y(adenomas and carcinomas).
D.1.2. Model Development Strategy
18 If the incidence data shown graphically appeared to have a monotone non-decreasing
19 dose-response function for adenomas, carcinomas, or both, the following sequence of models
20 were fit to the data: (1) A Weibull model was fit to evaluate the single power or exponent for
21 which the MS models might best fit. The MS models with a single polynomial term might be
22 considered as Weibull models whose exponents are integers (whole numbers); (2) MS models
23 might be considered a simplified special case of the clonal expansion model or MVK model of
24 carcinogenesis. Because of the limited number of dose groups (4 in the JBRC 1998a study, only
25 3 if the highest dose is dropped) and if a background parameter is included in the MS model,
26 then in order to have enough degrees of freedom to calculate a goodness-of-fit/>-value at most
27 two polynomial terms can be included within the exponential part of the MS model. This can be
28 done using the "Advanced" option in BMDS and specifying all but one or two polynomial
29 coefficients to be equal to 0. Additional considerations occur when there is no response at any
30 dose except for the highest dose, in which case BMDS will try to fit the highest-order
31 polynomial possible. This occurs in a number of models for carcinomas only shown in the
32 output. An additional ad hoc criterion was imposed to deal with this case: fit the lowest-order
33 polynomial whose AIC and/?-value are not measurably worse than those of the highest-order
34 polynomial that can be fit, with the assumption that it seems more plausible that the formation of
35 carcinomas does not require an extremely large number of distinct stages. In any case, by way of
36 sensitivity analyses, results of fitting both acceptable models with a small number of stages and a
May 2009 D-3 DRAFT - DO NOT CITE OR QUOTE
-------
1 model with the current maximum number of stages (8) are reported. Higher-order MS monomial
2 models with up to eighteen stages can be fit using the Weibull model with a user-specified power
3 or exponent, which is the reason that analysis started with the Weibull model.
4 Other BMDS quantal models were fit in the following sequence: gamma, (log)-logistic,
5 (log)-probit. In some cases where the power (slope in the log-transformed-dose models) is
6 estimated as less than one in the unconstrained case, it may be worthwhile to also fit a model
7 with the power or slope constrained to be greater than or equal to one, or more easily by setting
8 the exponent or power equal to one.
9 Section D.2 reports results for fitting hepatic adenomas and carcinomas in female F344
10 rats, Section D.3 reports results for fitting hepatic adenomas and carcinomas in male F344 rats,
11 Section D.4 reports results for fitting certain adenomas and carcinomas at other sites in female or
12 male F344 rats, and Section D.5 reports results for fitting the combined incidence of hepatic
13 adenomas or carcinomas and tumors at other sites in female and male rats using the best-fitting
14 models for hepatic and other tumors shown in Sections D.2 through D.4.
15 The mouse data had a profoundly non-monotonic dose-response function and thus the
16 combined incidence of hepatic adenomas and carcinomas retained some of this character due to
17 the much larger incidence of adenomas at the lower doses. This is discussed in Sections D.6 and
18 D.7.
19 Software for fitting combined tumor incidence in the liver or at one other site is currently
20 available in a program ("MS-combo") whose structure and output are virtually identical to that of
21 BMDS, but currently is limited to MS models at only two distinct sites. Results are presented in
22 Section D.7 so as to evaluate the sensitivity of the BMDL estimates to using as an adverse effect
23 the occurrence of tumors at one site, the other site, or tumors at both sites. In those cases where
24 there were strong dose-response relationships for tumors at other sites with other models,
25 analyses were restricted to the best-fitting MS models.
D.1.3. Model Selection Criteria
26 Multiple models were fit to each data set. The model selection criteria used in the BMDS
27 Benchmark Dose Technical Guidance Document (U.S. EPA, 2000b) were applied as follows:
28 • p-va\ue for goodness-of-fit > 0.10
29 • AIC smaller than other acceptable models
30 • % residuals as small as possible
31 • No systematic patterns of deviation of model from data
May 2009 D-4 DRAFT - DO NOT CITE OR QUOTE
-------
1 Additional criteria were applied to eliminate implausible dose-response functions:
2 • Monotonic dose-response functions, e.g. no negative coefficients of polynomials in MS
3 models
4 • No infinitely steep dose-response functions near 0 (control dose), achieved by requiring
5 the estimated parameters "power" in the Weibull and Gamma models and "slope" in the
6 log-logistic and log-probit models to have values >1.
7 • When combining risk estimates for different sites using the MS-Combo program, the
8 program automatically includes a linear term in the polynomial part of the MS model so
9 there is currently no ideal way to fit an optimal MS model in MS-Combo consistent with
10 the same model taken by itself if the optimal does not have a linear term.
11 Because no single set of criteria covers all contingencies, an extended list of preferred models are
12 presented below.
D.1.4. Summary
13 The BMDS models recommended to calculate rodent BMDio and BMDLio values and
14 corresponding human BMDio HED and BMDLio HED values are summarized in Table D-l.
May 2009 D-5 DRAFT - DO NOT CITE OR QUOTE
-------
Table D-l. Recommended models for rodents exposed to 1,4-dioxane in
drinking water (JBRC, 1998a)
Sex/
strain/
species
Female
F344
Rat
Male
F344
Rat
Female
BDFj
Mouse
Male
BDFj
Mouse
Endpoint
Hepatic
Tumors
Mammary
Gland
Tumors
Nasal
Cavity
Tumors
Hepatic
Tumors
Peritoneal
Mesothelioma
Nasal
Cavity
Tumors
Hepatic
Tumors
Hepatic
Tumors
Model
selection
criterion
Lowest
AIC
Lowest
AIC with
linear term
Lowest
AIC
Lowest
AIC
Low AIC
lowest
order non-
linear
Lowest
AIC
Low AIC,
lowest
order non-
linear
Lowest
AIC
Low AIC
Lowest
AIC
Low AIC,
lowest
order that
fits well
AlC-estim.
Power =1
AlC-estim.
Power =1
Model
type
MS
MS in
MS-
Combo
MS
MS
MS
MS
MS in
MS-
Combo
MS
MS in
MS-
Combo
MS
MS
Log-
logistic
Log-
logistic
Model
parameters
2
1,2
1
8
2
1,8
1,2
2
1,2
8
2
0.854
1
0.484
1
AIC
98.49
100.5
193.8
45.97
46.68
113.5
114.0
139.0
140.5
44.50
43.12
153.3
151.6
240.1
240.5
P-
value
0.473
0.449
0.845
1.000
0.947
0.574
0.368
0.760
0.809
1.000
0.956
0.635
0.749
0.984
0.284
BMD10
130
73.8
163
483
409
88.0
73.8
145
112
371
340
2.94
5.28
1.77
31.1
BMDL10
114
42.6
91.8
452
413
37.6
42.6
124
51.0
235
257
0.0864
3.47
0+
15.8
BMD10HED
31.11
17.66
39.00
115.6
97.87
23.88
20.03
39.35
30.39
100.7
92.27
0.4413
0.7926
0.2697
4.738
BMDL10
HED
27.30
10.19
21.97
108.2
98.83
10.20
11.56
33.65
13.84
63.77
69.74
0.01297
0.5209
0
2.4071
D.2. FEMALE F344 RATS: HEPATIC CARCINOMAS AND ADENOMAS
1 The data for hepatic carcinomas and adenomas in female F344 rats (JBRC, 1998a) are
2 shown in Table D-2.
May 2009
D-6
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-2. Data for hepatic adenomas and carcinomas in female F344 rats
(JBRC, 1998a)
Tumor type
Adenomas
Carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Both adenomas and carcinomas
Total number per group
Dose (mg/kg-day)
0
1
1
1
49
1
50
21
0
0
0
50
0
50
103
5
0
5
45
0
50
514
38
10
40
10
8
50
1
2
3
4
5
6
7
Source: JBRC(1998a).
Note that the incidence of rats with adenomas, with carcinomas, and with either
adenomas or carcinomas or both (combined incidence) are monotone non-decreasing functions
of dose except for 1 female rat in the control group. These data therefore appear to be
appropriate for dose-response modeling using BMDS.
The results of the BMDS modeling for the entire suite of models are presented in Table
D-3 in the order described in Section D. 1.
May 2009
D-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-3. Summary of BMDS dose-response modeling results for the
combined incidence of hepatic adenomas and carcinomas in female F344 rats
Model
Weibull
MS
Gamma
Log-
Logistic
Log-Probit
Power
Est.
1
T
3
4
l&2d
2&3
2&4
2&5
2&6
Est.
Est.
Est.
Estim.
1.837
2.484
2.337
1.355
Std.
Err.
0.320
0.648
0.382
0.204
AIC
100.262
110.201
98.492
102.598
104.046
100.463
100.157
100.196
100.136
100.135
100.020
99.993
99.773
P-
value
0.2638
0.0057
0.4734
0.0555
0.0302
0.2357
0.2736
0.2700
0.2753
0.2754
0.2883
0.2905
0.3130
BMD10
mg/kg-
day
116
45.8
130
207
261
126
112
111
111
111
113
111
108
BMDL10
mg/kg-
day
75.2
35.6
114
186
243
77.2
76.4
77.0
75.6
77.6
75.8
76.6
76.4
Max
f
0.873
2.016
0.861
2.007
2.141
0.892
0.847
0.857
0.841
0.841
0.803
0.796
0.721
Dose
maxb
mg/kg-
day
21
103
21
103
103
21
21
21
21
21
21
21
21
BMD10
HED
27.8
11.0
31.1
49.5
62.5
30.2
26.8
26.6
26.6
26.6
27.0
26.6
25.8
BMDL10
HED
18.0
8.52
27.3
44.5
58.2
18.5
18.3
18.4
18.1
18.6
18.1
18.3
18.3
aMaximum absolute %2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
"Best-fitting model.
dUtilized in combined analyses with tumors at other sites for consistency with male F344 rat model and because
MS-Combo software requires a linear term at this time; differs from optimal by < 2 AIC units.
May 2009
D-8
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
•d
I
o
CD
0.8
0.6
0.4
0.2
Multistage
BMD Lower Bound
BMQL BMP
0
16:3604/272008
100
200 300
dose
400
500
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-l. Multistage BMD model (2 degree) for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 16:36:47 2008
JBRC FEMALE carcino+adenomas multi deg 2 Table D-3
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/1
The parameter betas are not restricted
Dependent variable = EITHERRATF
Independent variable = DOSERATF
User specifies the following parameters:
Beta(l) = 0
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 7
May 2009
D-9
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
Degree of polynomial = 8
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User Inputs Initial Parameter Values
Background =
Beta(l) =
Beta(2) =
Beta(3) =
Beta(4) =
Beta(5) =
Beta(6) =
Beta(7) =
Beta(8) =
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(l), -Beta(3), -Beta(4), -Beta(5), -Beta(6),
-Beta(7), -Beta(8) have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix)
0.08
1
0
1
1
1
1
1
1
Specified
Specified
Specified
Specified
Specified
Specified
Specified
Background
Beta(2)
Background
1
-0.31
Beta(2)
-0.31
1
Parameter Estimates
Variable
Background
Beta (2)
Model
Full model
Fitted model
Reduced model
AIC:
Estimate
0.0119049
6.2293e-006
Std. Err.
0.0839009
1.29099e-006
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-0.152538 0.176348
3.699e-006 8.75959e-006
Analysis of Deviance Table
Log(likelihood)
-46.1762
-47.2461
-107.855
98.4922
# Param's Deviance
4
2 2.13973
1 123.358
Goodness of Fit
Test d.f.
2
3
P-value
0.3431
<.0001
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0119
0.0146
0.0751
0.8094
Expected
0.595
0.731
3.755
40.471
Observed
1
0
5
40
Size
50
50
50
50
Scaled
Residual
0.528
-0.861
0.668
-0.170
= 1.50
d.f. = 2
P-value = 0.4734
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
130.053
113.537
May 2009
D-10
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
"C
I
0.8
0.6
0.4
0.2
Multistage
BMD Lower Bound
BMDL BMD
0 100
16:4004/272008
200 300
dose
400
500
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-2. Multistage BMD model (1 & 2 degree) for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 16:40:47 2008
JBRC FEMALE carcino+adenomas multi deg 1&2 Table D-3
Observation # < parameter # for Multistage model.
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl-beta2*doseA2-beta3*doseA3-beta4*doseA4-beta5*doseA5-
beta6*dose/x6-beta7*dose/x7-beta8*dose/x8) ]
The parameter betas are not restricted
Dependent variable = EITHERRATF
Independent variable = DOSERATF
User specifies the following parameters:
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
May 2009
D-ll
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
Total number of specified parameters = 6
Degree of polynomial = 8
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User Inputs Initial Parameter Values
Background
Beta(l)
Beta (2)
Beta(3)
Beta(4)
Beta(5)
Beta(6)
Beta (7)
Beta(8)
0.08
0
0
1
1
1
1
1
1
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(3), -Beta(4), -Beta (5), -Beta (6), -Beta(7),
-Beta(8) 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) Beta(2)
Background 1 -0.61 0.5
Beta(l) -0.61 1 -0.95
Beta(2) 0.5 -0.95 1
Variable
Background
Beta(l)
Beta(2)
Model
Full model
Fitted model
Reduced model
Estimate
0.0113405
8.31025e-005
6.0135e-006
Analysis
Log (likelihood)
-46.1762
-47.2315
-107.855
Parameter Estimates
Std. Err. Low
0.106071
0.00204065
4.0241e-006
of Deviance Table
e
0
0
1
# Param's Deviance
4
3 2.110
1 123.3
95.0% Wald Confidence Interval
Conf. Limit Upper Conf. Limit
-0.196554 0.219235
-0.0039165 0.00408271
-1.87358e-006 1.39006e-005
Test d.f.
5 1
3 3
P-value
0.1463
<.0001
AIC:
100.463
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0113
0.0157
0.0803
0.8066
Expected
0.567
0.784
4.017
40.329
Observed
1
0
5
40
Size
50
50
50
50
Scaled
Residual
0.578
-0.892
0.511
-0.118
ChiA2 =1.41
d.f. = 1
P-value = 0.2357
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 125.636
BMDL = 77.1768
May 2009
D-12
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
"C
I
0.8
0.6
0.4
0.2
Log-Logistic
BMD Lower Bound
BMDL BMD
100
200 300
dose
400
500
16:5304/272008
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-3. Log-logistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female F344 rats.
Logistic Model. (Version: 2.5; Date: 09/24/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 16:53:34 2008
JBRC FEMALE carcino+adenomas log-logistic Table D-3
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = EITHERRATF
Independent variable = DOSERATF
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.02
intercept = -10.5681
slope = 1.86975
Asymptotic Correlation Matrix of Parameter Estimates
May 2009
D-13
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
background
background 1
intercept -0.18
slope 0.16
intercept
-0.18
1
-0.99
slope
0.16
-0.99
1
Parameter Estimates
Variable
background
intercept
slope
Model
Full model
Fitted model
Reduced model
Estimate
0.0102565
-13.1982
2.33662
Analysis
Log (likelihood)
-46.1762
-46.9966
-107.855
Std. Err.
0.01019
2.21197
0.381529
of Deviance
# Param' s
4
3
1
Lower
-0.0
-17.
1.5
Table
Deviance
1.64065
123.358
95.0% Wald Confidence Interval
E. Limit Upper Conf. Limit
L556 0.0302286
3 -8.86278
1 3.08441
Test d.f.
1
3
P-value
0.2002
<.0001
AIC:
99.9931
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0103
0.0125
0.0950
0.8023
Expected
0.513
0. 625
4.749
40.113
Observed
1
0
5
40
Size
50
50
50
50
Scaled
Residual
0.684
-0.796
0.121
-0.040
ChiA2 =1.12
d.f. = 1
P-value = 0.2905
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
110.835
76.5788
D.3. MALE F344 RATS: HEPATIC CARCINOMAS AND ADENOMAS
46 The data for hepatic adenomas and carcinomas in male F344 rats (JBRC, 1998a) are
47 shown in Table D-4.
May 2009
D-14
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-4. Data for hepatic adenomas and carcinomas in male F344 rats
(JBRC, 1998a)
Tumor type
Adenomas
Carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Both adenomas and carcinomas
Total number per group
Dose (mg/kg-day)
0
0
0
0
50
0
50
16
2
0
2
48
0
50
81
4
0
4
45
0
49
398
24
14
33
17
5
50
1 Note that the incidence of rats with hepatic adenomas, carcinomas, and with either
2 adenomas or carcinomas or both (combined incidence) are monotone non-decreasing functions
3 of dose. These data therefore appear to be appropriate for dose-response modeling using BMDS.
4 The results of the BMDS modeling for the entire suite of models tested using the data for
5 hepatic adenomas and carcinomas for male F344 rats are presented in Table D-5 in the order
6 described in Section D. 1.
Table D-5. Summary of BMDS dose-response modeling results for the
combined incidence of adenomas and carcinomas in livers of male F344 rats
Model
Weibull
MS
Gamma
Log-
Logistic
Log-Probit
Power
Est.
1
2
l&2d
1&3
1&4
1&8C
Est.
Est.
Est.
Estim.
1.427
1.831
1.956
1.200
Std.
Err.
0.746
1.200
0.552
0.251
AIC
116.801
114.636
115.532
113.973
113.623
113.553
113.535
117.075
117.134
117.394
/7-value
0.1424
0.3223
0.3678
0.4485
0.5448
0.5678
0.5738
0.1447
0.1555
0.1530
BMD10
mg/kg-
day
79.6
45.7
124
73.8
79.0
80.2
80.6
88.0
94.2
98.6
BMDL10
mg/kg-
day
39.0
35.2
108
42.6
43.8
44.0
44.0
37.6
42.6
51.8
Max
x2a
1.178
1.650
1.079
0.990
0.965
0.958
1.132
1.079
1.021
Dose
maxb
mg/kg-
day
16
81
16
16
16
16
16
16
16
BMD10
HED
21.60
12.40
33.65
20.03
21.44
21.76
21.87
23.88
25.56
26.76
BMDL10
HED
10.58
9.55
29.31
11.56
11.89
11.94
11.94
10.20
11.56
14.06
aMaximum absolute %2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
'Best-fitting model.
dThis model fits the data nearly as well as the optimal model. It is simpler and is at least as plausible, thus will be
carried forward into combined analyses with tumors at another site as a sensitivity analysis.
May 2009
D-15
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
"8
•d 0.5
!t
< 0.4
o
1 0.3
LJ_
0.2
0.1
0
Multistage Model with 0.95 Confidence Level
Multistage
BMD Lower Bound
BMDL BMD
0 50 100
150
200
dose
250
300
350
400
17:0404/272008
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-4. Multistage BMD model (1 & 2 degree) for the combined incidence of
hepatic adenomas and carcinomas in male F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 17:04:35 2008
JBRC MALE carcino+adenomas multi deg 1&2 Table D-5
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/x8)
The parameter betas are not restricted
Dependent variable = EITHERRATM
Independent variable = DOSERATM
User specifies the following parameters:
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 6
Degree of polynomial = 8
May 2009
D-16
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User Inputs Initial Parameter Values
Background
Beta (1)
Beta (2)
Beta(3)
Beta(4)
Beta (5)
Beta(6)
Beta(7)
Beta(8)
0.08
0
0
1
1
1
1
1
1
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background, -Beta (3), -Beta (4), -Beta (5), -Beta (6),
-Beta(7), -Beta(8) have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix)
Beta(l)
Beta(2)
Beta(l)
1
-0.97
Beta(2)
-0.97
1
Parameter Estimates
95.0% Wald Confidence Interval
Upper Conf. Limit
0.00609257
1.67575e-005
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Variable
Background
Beta(l)
Beta (2)
Estimate
0
0.00114469
3.82589e-006
Std. Err.
NA
0.00252448
6.59789e-006
Lower Conf. Li:
-0.0038032
-9.10573e-006
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-54.3032
-54.9865
-98.4609
113.973
# Param's Deviance Test d.f.
4
1.36669 2
P-value
2
i.3155
0.5049
<.0001
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0000
0.0191
0.1111
0.6541
Expected
0.000
0.955
5.446
32.705
Observed
0
2
4
33
Size
50
50
49
50
Scaled
Residual
0.000
1.079
-0.657
0.088
=1.60
d.f. = 2
P-value = 0.4485
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 73.8264
BMDL = 42.6043
May 2009
D-17
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.7
0.6
"8
•d 0.5
!t
< 0.4
o
1 0.3
LJ_
0.2
0.1
0
Multistage Model with 0.95 Confidence Level
Multistage
BMD Lower Bound
, BMDL BMP
0 50 100
150
200
dose
250
300
350
400
17:0904/272008
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-5. Multistage BMD model (1 & 8 degree) for the combined incidence of
hepatic adenomas and carcinomas in male F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 17:09:20 2008
JBRC MALE carcino+adenomas multi deg 1&8 Table D-5
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/x8)
The parameter betas are not restricted
Dependent variable = EITHERRATM
Independent variable = DOSERATM
User specifies the following parameters:
Beta(2) =
Beta(3) =
Beta(4) =
Beta(5) =
Beta(6) =
Beta(7) =
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 6
Degree of polynomial = 8
May 2009
D-18
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User Inputs Initial Parameter Values
Background
Beta (1)
Beta (2)
Beta(3)
Beta(4)
Beta (5)
Beta(6)
Beta(7)
Beta(8)
0.08
0
1
1
1
1
1
1
0
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background, -Beta (2), -Beta (3), -Beta (4), -Beta (5),
-Beta(6), -Beta(7) have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix )
Beta(l)
Beta(8)
Beta(l)
1
-0.96
Beta(8)
-0.96
1
Parameter Estimates
95.0% Wald Confidence Interval
Variable
Background
Beta(l)
Beta (8)
Estimate
0
0.00130773
8.86808e-022
Std. Err.
NA
0.00196467
1.30033e-021
Lower Conf. Limit
-0.00254296
-1. 6618e-021
Upper Conf. Lim
0.00515841
3.43541e-021
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-54.3032
-54.7675
-98.4609
113.535
# Param's Deviance Test d.f.
4
2 0.928571 2
1 88.3155 3
Goodness of Fit
P-value
0.6286
<.0001
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0000
0.0207
0.1005
0.6600
Expected
0.000
1.035
4.925
33.000
Observed
0
2
4
33
Size
50
50
49
50
Scaled
Residual
0.000
0.958
-0.439
0.000
=1.11
d.f. = 2
P-value = 0.5738
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 80.5665
BMDL = 44.0259
May 2009
D-19
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
D.4. F344 RATS: TUMORS AT OTHER SITES
The data for tumors at sites other than the liver in male and female F344 rats (JBRC,
1998a) are shown in Table D-6. Note that the incidence of rats with these endpoints are
monotone non-decreasing functions of dose whose estimated incidence at the highest dose is
greater than the BMR of 10% so that the BMD and BMDL are expected to be well within the
range of the observed data. These data therefore appear to be appropriate for dose-response
modeling using BMDS.
Table D-6. Data for significant tumors at other sites in male and female F344
rats
Tumor site and type
Nasal cavity carcinomas
Peritoneal mesothelioma
Mammary gland adenoma
Total Number per Group
Dose (mg/kg-day)
Female
0
0
21
0
103
0
514
8
Not available
6
50
7
50
10
50
16
50
Male
0
0
2
16
0
2
81
0
5
398
7
28
No relation for BMR 0. 10
50
50
49
50
Source: JBRC(1998a).
7 The results of the BMDS modeling for the entire suite of models are presented in Tables
8 D-7 through Table D-10 for tumors in the nasal cavity, mammary gland, and peritoneal cavity in
9 the order described in Section D. 1.
May 2009
D-20
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-7. Summary of BMDS dose-response modeling results for the
incidence of nasal cavity tumors in female F344 rats e
Model
Weibull
MS
Gamma
Log-
Logistic
Log-Probit
Power
Est.
1
2d
3
4
8C
Est.
Est.
Est.
Estim
10.98
14.74
10.92
2.942
Std
Err
NE
NE
NE
154
AIC
47.967
49.700
46.680
46.108
45.995
45.967
47.967
47.967
47.967
/7-value
1.0000
0.5488
0.9472
0.9951
0.9996
1.0000
1.0000
1.0000
1.0000
BMD10
mg/kg-
day
491
392
409
436
453
483
473
489
466
BMDL10
mg/kg-
day
312
231
313
365
397
452
306
305
283
Max
x23
0.000
1.184
0.580
0.264
0.118
0.005
0.001
0.000
0.001
Dose
maxb
mg/kg-
day
NA
103
103
103
103
103
103
NA
103
BMD10
HED
117.5
93.80
97.87
104.3
108.4
115.6
113.2
117.0
111.5
BMDL10
HED
74.66
55.28
74.90
87.34
95.00
108.2
73.22
72.98
67.72
aMaximum absolute y? residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
"Best-fitting model.
dThis model fits the data nearly as well as the optimal model. It is simpler and is at least as plausible, thus will be
carried forward into combined analyses with liver tumors as a sensitivity analysis.
eNasal cavity tumors in female F344 rats include squamous cell carcinoma and esthesioneuro-epithelioma.
May 2009
D-21
DRAFT - DO NOT CITE OR QUOTE
-------
0.3
0.25
| 0.2
< 0.15
o
I 0-1
0.05
0
Multistage Model with 0.95 Confidence Level
Multistage
BMD Lower Bound
BMDL
BMD
100
200
300
dose
400
500
600
17:4004/272008
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-6. Multistage BMD model (2 degree) for the nasal cavity tumors in female
F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 17:40:07 2008
JBRC FEMALE RAT Nasal Cavity Tumors multistage deg 2 Table D-7
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/1
The parameter betas are restricted to be positive
Dependent variable = NASALCAVITYF
Independent variable = DOSERATF
User specifies the following parameters:
Beta(l) = 0
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 7
May 2009
D-22
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Degree of polynomial = 8
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 =
Beta(l) =
Beta(2) =
Beta(3) =
Beta(4) =
Beta(5) =
Beta(6) =
Beta(7) =
Beta(8) =
0
0
0
0
0
0
9.47989e-018
0
0
Specified
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background, -Beta(l), -Beta(3), -Beta(4), -Beta(5),
-Beta(6), -Beta(7), -Beta(8) have been estimated at a boundary point, or have been
specified by the user, and do not appear in the correlation matrix)
Beta(2)
Beta(2)
1
Parameter Estimates
Variable
Background
Beta(2)
Estimate
0
6.31106e-007
Std. Err.
NA
5.7003e-007
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-4.86132e-007
1.74834e-006
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-21.9835
-22.34
-33.5888
46.6801
# Param's Deviance
4
1 0.713065
1 23.2107
Goodness of Fit
Test d.f.
3
3
P-value
0.8701
<.0001
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0000
0.0003
0.0067
0.1536
Expected
0.000
0.014
0.334
7. 679
Observed
0
0
0
8
Size
50
50
50
50
Scaled
Residual
0.000
-0.118
-0.580
0.126
= 0.37
d.f. = 3
P-value = 0.9472
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 408.59
BMDL = 313.309
May 2009
D-23
DRAFT - DO NOT CITE OR QUOTE
-------
0.3
0.25
1 0.2
£
0.15
c
o
0.1
0.05
0
Multistage Model with 0.95 Confidence Level
Multistage
BMD Lower Bound
BMDL BMD
100
200 300
dose
400
500
17:4204/272008
I
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-7. Multistage BMD model (8 degree) for the nasal cavity tumors in female
F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 17:42:03 2008
JBRC FEMALE RAT Nasal Cavity Tumors multistage deg 8 Table D-7
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/1
The parameter betas are restricted to be positive
Dependent variable = NASALCAVITYF
Independent variable = DOSERATF
User specifies the following parameters:
Beta(l) = 0
Beta(2) = 0
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 7
May 2009
D-24
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
Degree of polynomial = 8
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 =
Beta(l) =
Beta(2) =
Beta(3) =
Beta(4) =
Beta(5) =
Beta(6) =
Beta(7) =
Beta(8) =
0
0
0
0
0
0
9.47989e-018
0
0
Specified
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background, -Beta(l), -Beta(2), -Beta(3), -Beta(4),
-Beta(5), -Beta(6), -Beta(7) have been estimated at a boundary point, or have been
specified by the user, and do not appear in the correlation matrix)
Beta(8)
Beta(8)
1
Parameter Estimates
Variable Estimate Std. Err.
Background 0 NA
Beta(8) 3.57866e-023 3.16715e-023 -2.62883e-023
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
9.78616e-023
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log (likelihood) # Param's Deviance Test d.f. P-value
-21.9835 4
-21.9835
-33.5888
1 4.53344e-005
1 23.2107
<.0001
45.967
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.0000
0.0000
0.0000
0.1600
Expected
0.000
0.000
0.000
8.000
Observed
0
0
0
8
Size
50
50
50
50
Scaled
Residual
0.000
-0.000
-0.005
0.000
ChiA2 =0.00
d.f. = 3
P-value = 1.0000
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 482.636
BMDL = 451.634
May 2009
D-25
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-8. Summary of BMDS dose-response modeling results for the
incidence of nasal cavity tumors in male F344 rats
Model
Weibull
MS
Gamma
Log-
Logistic
Log-Probit
Power
Est.
1
2d
3
4
8C
Est.
Est.
Est.
Estim
10.74
14.65
11.09
2.919
Std.
Err
NE
NE
NE
174
AIC
44.496
45.711
43.119
43.571
42.522
42.496
44.496
44.496
44.496
/7-value
1.0000
0.6113
0.9564
0.9101
0.9996
1.0000
1.0000
1.0000
1.0000
BMD10
mg/kg-
day
385
350
340
379
364
378
375
489
371
BMDL10
mg/kg-
day
256
199
257
328
316
349
252
305
235
Max
x23
0.001
1.100
0.542
0.666
0.113
0.010
0.001
0.000
0.000
Dose
maxb
mg/kg-
day
81
81
81
81
81
81
81
NA
NA
BMD10
HED
104.48
94.98
92.27
102.85
98.78
102.58
101.76
132.70
100.67
BMDL10
HED
69.47
54.00
69.74
89.01
85.75
94.71
68.39
82.77
63.77
aMaximum absolute y? residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
"Best-fitting model.
dThis model fits the data nearly as well as the optimal model. It is simpler and is at least as plausible, thus will be
carried forward into combined analyses with liver tumors as a sensitivity analysis.
eNasal cavity tumors in male F344 rats include squamous cell carcinoma, Sarcoma: NOS, rhabdomyosarcoma, and
esthesioneuro-epithelioma.
May 2009
D-26
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-9. Summary of BMDS dose-response modeling results for the
incidence of mammary gland adenomas in female F344 rats
Model
Weibull
MS
Gamma
Log-
Logistic
Log-Probit
Power
Est.
lc
2
1&2
Est.
Fixed
Est.
Fixed
Est.
Fixed
Estim.
0.659
0.611
1
0.707
1
0.375
1
Std.
Err.
0.496
0.492
0.518
0.263
AIC
195.941
194.197
194.964
194.945
195.505
194.256
195.934
194.128
195.920
195.008
/7-value
0.8757
0.8671
0.5851
0.8652
0.8403
0.8388
0.8930
0.8982
0.9502
0.5718
BMD10
mg/kg-
day
129
211
349
270
97.0
201
127
193
122
323
BMDL10
mg/kg-
day
4.90
119
256
4.57
2.98
118
5.23
98.4
5.90
209
Max
x23
0.115
0.443
0.832
0.123
0.144
0.544
0.098
0.382
0.044
0.853
Dose
maxb
mg/kg-
day
21
103
103
21
21
103
21
103
21
103
BMD10
HED
30.87
50.49
83.51
64.61
23.21
48.10
30.39
46.18
29.19
77.29
BMDL10
HED
1.17
28.48
61.26
1.09
0.71
28.24
1.25
23.55
1.41
50.01
"Maximum absolute % residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
'Best-fitting model. It is simple and plausible, thus will be carried forward into combined analyses with liver
tumors as a sensitivity analysis.
May 2009
D-27
DRAFT - DO NOT CITE OR QUOTE
-------
"8
0.5
0.4
0.3
o
"C ,. 0
ro 0.2
0.1
Multistage Model with 0.95 Confidence Level
Multistage
BMD Lower Bound
BMP.L , BMP
0
18:3204/272008
100
200 300
dose
400
500
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-8. Multistage BMD model (1 degree) for mammary gland adenomas in
female F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 18:32:45 2008
JBRC FEMALE RAT Mammary Gland Adenoma multi deg 1 Table D-9
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-
beta8*dose/"8) ]
The parameter betas are not restricted
Dependent variable = MAMMGLANDF
Independent variable = DOSERATF
User specifies the following parameters:
Beta(2) = 0
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
May 2009
D-28
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Total number of parameters in model = 9
Total number of specified parameters = 7
Degree of polynomial = 8
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
Beta (1)
Beta (2)
Beta(3)
Beta(4)
Beta (5)
Beta (6)
Beta(7)
Beta(8)
0.0104878
0.00347368
= -4.98817e-006
= 5.12249e-009
= -8.5135e-012
= 4.0469e-014
= -7.50662e-017
= -1.84537e-019
= -3.25974e-023
Specified
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(2), -Beta(3), -Beta(4), -Beta(5), -Beta(6), -
Beta(7), -Beta(8) 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 -0.59
Beta(l) -0.59 1
Variable
Background
Beta(l)
Model
Full model
Fitted model
Reduced model
Estimate
0.132761
0.000498383
Analysis
Log (likelihood)
-94.958
-95.0985
-98.6785
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
0.0843828 -0.0326261 0.298148
0.000408202 -0.000301679 0.00129844
of Deviance Table
# Param's Deviance Test d.
4
2 0.280866 2
1 7.4409 3
f. P-value
0.869
0.0591
AIC:
194.197
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.1328
0.1418
0.1762
0.3287
Expected
6.638
7.090
8.808
16.437
Observed
6
7
10
16
Size
50
50
50
50
Scaled
Residual
-0.266
-0.036
0.443
-0.132
ChiA2 = 0.29
d.f. = 2
P-value = 0.8671
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 211.405
BMDL = 118.855
May 2009
D-29
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
"8
0.5
0.4
0.3
o
"C ,. 0
ro 0.2
0.1
Multistage
BMD Lower Bound
BMDL
BMD
0 100
18:4504/272008
200 300
dose
400
500
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-9. Multistage BMD model (2 degree) for mammary gland adenomas in
female F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 18:45:19 2008
JBRC FEMALE RAT Mammary Gland Adenoma multi deg 2 Table D-9
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/1
The parameter betas are restricted to be positive
Dependent variable = MAMMGLANDF
Independent variable = DOSERATF
User specifies the following parameters:
Beta(l) = 0
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 7
May 2009
D-30
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
Degree of polynomial = 8
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 =
Beta(l) =
Beta(2) =
Beta(3) =
Beta(4) =
Beta(5) =
Beta(6) =
Beta(7) =
Beta(8) =
0.135593
0.000477455
0
0
0
0
0
0
0
Specified
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(l), -Beta(3), -Beta(4), -Beta(5), -Beta(6), -
Beta(7), -Beta(8) have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix)
Background Beta (2)
Background 1 -0.49
Beta(2) -0.49 1
Parameter Estimates
Variable
Background
Beta (2)
Estimate
0.14939
8.66886e-007
Std. Err.
0.0766837
7.47534e-007
95
Lowe
-0
-5
Wald Confidence Interval
3nf. Limit Upper Conf. Limit
-0.000907529 0.299687
-5.98253e-007 2.33203e-006
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-94.958
-95.4821
-98.6785
194.964
# Param's Deviance
4
2 1.04822
1 7.4409
Goodness of Fit
Test d.f.
2
3
P-value
0.5921
0.0591
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.1494
0.1497
0.1572
0.3235
Expected
7.469
7.486
7.859
16.175
Observed
6
7
10
16
Size
50
50
50
50
Scaled
Residual
-0.583
-0.193
0.832
-0.053
ChiA2 =1.07
d.f. = 2
P-value = 0.5851
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 348.624
BMDL = 256.18
May 2009
D-31
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-10. Summary of BMDS dose-response modeling results for the
incidence of peritoneal mesotheliomas in male F344 rats
Model
Weibull
MS
Gamma
Log-
Logistic
Log-Probit
Power
Est.
1
2C
3
4
8
1&2
2&3
Est.
Est.
Est.
Estim.
1.537
1.796
1.776
1.017
Std.
Err.
0.505
0.810
0.537
0.267
AIC
140.497
140.557
138.996
140.149
140.472
140.559
140.537
140.482
140.488
140.487
140.477
P-
value
0.8885
0.3711
0.7602
0.4073
0.3432
0.3279
0.8808
0.9426
0.9164
0.8930
0.9848
BMD10
mg/kg-
day
107.9
59.5
145
205
243
311
112
103
106
127
102
BMDL10
mg/kg-
day
50.9
44.2
124
184
224
298
50.8
61.6
50.9
5.23
53.7
Max
x2a
0.108
1.049
0.651
1.119
1.049
1.223
0.191
0.053
0.080
0.098
0.014
Dose
maxb
mg/kg-
day
16
81
81
81
81
81
16
16
16
16
16
BMD10
HED
29.31
16.15
39.35
55.63
65.94
84.40
30.39
27.95
28.77
34.46
27.68
BMDL10
HED
13.81
11.99
33.65
49.93
60.79
80.87
13.79
16.72
13.81
1.42
14.57
"Maximum absolute % residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
'Best-fitting model will be carried forward into combined analyses with liver tumors.
May 2009
D-32
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
0.7
0.6
* °'5
t3
£ 0.4
<
I °-3
£
"- 0.2
0.1
0
Multistage
BMD Lower Bound
BMDL BMP
50
100
150
200
dose
250
300
350
400
20:38 04/27 2008
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Source: JBRC (1998a).
Figure D-10. Multistage BMD model (2 degree) for peritoneal mesotheliomas in
male F344 rats.
Multistage Model. (Version: 2.5; Date: 10/17/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 20:38:46 2008
JBRC MALE RAT Peritoneal Mesotheliama multi deg 2 Table D-10
Observation # < parameter # for Multistage model.
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*dose/x3-beta4*dose/x4-beta5*dose/x5-beta6*dose/x6-beta7*dose/x7-beta8*dose/1
The parameter betas are restricted to be positive
Dependent variable = PERITMESOTHELM
Independent variable = DOSERATM
User specifies the following parameters:
Beta(l) = 0
Beta(3) = 0
Beta(4) = 0
Beta(5) = 0
Beta(6) = 0
Beta(7) = 0
Beta(8) = 0
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 9
Total number of specified parameters = 7
May 2009
D-33
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Degree of polynomial = 8
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 =
Beta(l) =
Beta(2) =
Beta(3) =
Beta(4) =
Beta(5) =
Beta(6) =
Beta(7) =
Beta(8) =
0.035746
3.49814e-006
0
0
0
0
0
0
0 Speci
Specified
Specified
Specified
Specified
Specified
Specified
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(l), -Beta(3), -Beta(4), -Beta(5), -Beta(6),
-Beta(7), -Beta(8) have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix)
Background
Beta(2)
Background
1
-0.41
Beta(2)
-0.41
1
Parameter Estimates
Variable
Background
Beta (2)
Model
Full model
Fitted model
Reduced model
AIC:
Estimate
0.0465221
4.98704e-006
Std. Err.
0.0819877
1.49071e-006
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-0.114171 0.207215
2.06529e-006
7.90878e-006
Analysis of Deviance Table
Log(likelihood)
-67.2386
-67.4978
-95.5731
138.996
# Param's Deviance Test d.f.
4
2 0.518389 2
1 56.6691 3
Goodness of Fit
P-value
0.7717
<.0001
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0465
0.0477
0.0772
0.5673
Expected
2.326
2.387
3.784
28.363
Observed
2
2
5
28
Size
50
50
49
50
Scaled
Residual
-0.219
-0.257
0.651
-0.104
ChiA2 = 0.55
d.f. = 2
P-value = 0.7602
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 145.351
BMDL = 123.748
May 2009
D-34
DRAFT - DO NOT CITE OR QUOTE
-------
D.5. FEMALE BDFi MICE: HEPATIC CARCINOMAS AND ADENOMAS
1
2
3
4
5
6
7
8
9
10
Data for female BDFi mouse hepatic carcinomas and adenomas are shown in Table D-l 1.
Note that the incidence of carcinomas and the incidence of either adenomas or carcinomas or
both (combined incidence) are monotone non-decreasing functions of dose. These data therefore
appear to be appropriate for dose-response modeling using BMDS. However, the incidence of
adenomas clearly reaches a peak value at 77 mg/kg-day and then decreases sharply with
increasing dose. This cannot be modeled by a MS model using only non-negative coefficients.
To some extent the incidence of "either adenomas or carcinomas or both" retains some of the
inverted-U shaped dose-response of the adenomas, which dominate based on their high incidence
at the lowest dose groups (77 and 323 mg/kg-day), thus is not well characterized by any MS
model such as those that provided consistently good descriptions of the rat data.
Table D-ll. Data for hepatic adenomas and carcinomas in female BDFi mice
Tumor type
Adenomas
Carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Both adenomas and carcinomas
Total number per group
Dose(mg/kg-day)
0
4
0
4
46
0
50
77
30
6
34
16
2
50
323
20
30
41
7
9
48
1066
2
45
46
2
1
48
11
12
13
14
15
Source: JBRC(1998a).
The results of the BMDS modeling for the entire suite of models for hepatic adenomas
and carcinomas in female BDFi mice are presented in Table D-l2 in the order described in
Section D.I.
The graphical output from fitting these models suggested that a simpler model obtained
by dropping the data point for the highest dose (1,066 mg/kg-day) might also be adequate.
May 2009
D-35
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-12. Summary of BMDS dose-response modeling results for the
combined incidence of hepatic adenomas and carcinomas in female BDFi
mice
Model
Weibull
MS
Gamma
Log-
Logistic
Log-
Probit
Power
Est.
1
Est.
Est.
Fixedc
Est.
Fixed
Estim
0.407
0.195
0.854
1.000
0.478
1.000
Std.
Err.
0.112
0.083
0.248
0.134
AIC
153.087
172.372
153.092
153.298
151.629
153.181
164.002
P-
value
0.9001
0.0000
0.8846
0.6349
0.7494
0.7403
0.0000
BMD10
mg/kg-
day
0.274
19.3
0.0046
2.94
5.28
2.43
23.2
BMDL10
mg/kg-
day
0.0011
14.8
0.0004
0.0864
3.47
0.062
17.8
Max
*"
0.046
3.335
0.116
0.355
0.604
0.256
3.656
Dose
maxb
mg/kg-
day
77
77
323
323
323
323
1,066
BMD10
HED
0.04113
2.897
0.00069
0.4413
0.7926
0.3648
3.482
BMDL10
HED
0.000165
2.222
0.0006
0.01297
0.5209
0.009307
2.672
"Maximum absolute %2 residual deviation between observed and predicted count. Values much larger thanl are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
'Best-fitting model not supra-linear will be carried forward into combined analyses with tumors at another site.
Log-Logistic Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
Log-Logistic
BMD Lower Bound
0 200
21:02 04/27 2008
400 600
dose
800
1000
Source: JBRC (1998a).
Figure D-ll. Log-logistic BMD model (Fixed power=l) for the combined incidence
of hepatic adenomas and carcinomas in female BDFi mice.
May 2009
D-36
DRAFT - DO NOT CITE OR QUOTE
-------
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Logistic Model. (Version: 2.5; Date: 09/24/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 21:02:47 2008
JBRC FEMALE MOUSEcarcino+adenoma slope = 1 loglogit Tbl D-12
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = EITHERMUSF
Independent variable = DOSEMUSF
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.08
intercept = -3.96538
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
background
intercept
background
1
-0.25
intercept
-0.25
1
Parameter Estimates
Variable Estimate Std. Err.
background 0.0811252 0.0388749
intercept -3.86024 0.250749
slope 1
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
0.00493183 0.157319
-4.3517 -3.36878
NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-73.5356
-73.8143
-128.321
151.629
# Param's Deviance Test d.f.
4
2 0.557285 2
1 109.571 3
P-value
0.7568
<.0001
May 2009
D-37
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Dose
0.0000
77.0000
323.0000
1066.0000
Chi^2 = 0
Est
0.
0.
0.
0.
.58
Benchmark Dose
Specified
Risk Type
Confidence
effect
level
BMD
BMDL
Goodness of Fit
. Prob. Expected Observed Size
0811
6495
8822
9608
d.f. = 2
Computation
=
Extra
=
5.
4.056 4 50
32.477 34 50
42.348 41 48
46.119 46 48
P-value = 0.7494
0.1
risk
0.95
2752
Scaled
Residual
-0.029
0.452
-0.604
-0.089
3.46551
19
20
21
22
23
24
25
26
27
28
D.6. MALE BDFi MICE: HEPATIC CARCINOMAS AND ADENOMAS
Data for hepatic carcinomas and adenomas in male BDFi mice (JBRC, 1998a) are shown
in Table D-13. Note that the incidence of carcinomas and the incidence of either adenomas or
carcinomas or both (combined incidence) are monotone non-decreasing functions of dose. These
data therefore appear to be appropriate for dose-response modeling using BMDS. However, the
incidence of adenomas clearly reaches a peak value at 251 mg/kg-day and then decreases
sharply with increasing dose. This cannot be modeled by a MS model using only non-negative
coefficients. To some extent the incidence of "either adenomas or carcinomas or both" retains
some of the inverted-U shaped dose-response of the adenomas, which dominate based on their
high incidence at the lowest dose groups (66 and 251 mg/kg-day), thus is not well characterized
by any MS model such as those that provided consistently good descriptions of the rat data.
Table D-13. Data for hepatic adenomas and carcinomas in male BDFi mice
Tumor type
Adenomas
Carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Both adenomas and carcinomas
Total number per group
Dose (mg/kg-day)
0
7
15
21
29
1
50
66
18
20
31
17
7
48
251
22
23
37
13
8
50
768
8
36
39
9
5
48
29
30
31
32
33
Source: JBRC(1998a).
The results of the BMDS modeling for the entire suite of models for hepatic adenomas
and carcinomas in male BDFi mice are presented in Table D-14 in the order described in Section
D. 1. Fit in these models was also evaluated using a simpler model obtained by dropping the data
point for the highest dose (768 mg/kg-day). This did not appear to offer any great advantages
over using all four dose groups.
May 2009
D-38
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-14. Summary of BMDS dose-response modeling results for the
combined incidence of hepatic adenomas and carcinomas in male BDFi mice
Model
Weibull
MS
Gamma
Log-
Logistic
Log-
Probit
Power
Est
1
1&2
1&3
2&3
Est
Fixed
Est.
Fixed"
Est.
Fixed
Estim
0.334
0.228
1
0.484
1
0.301
1
Std.
Err
0.196
0.157
0.277
0.171
AIC
240.070
242.941
241.266
241.635
243.964
240.088
242.941
240.062
240.543
240.062
244.833
P-
value
0.9248
0.0885
0.2740
0.2119
0.0496
0.8701
0.0885
0.9845
0.2836
0.9801
0.0339
BMD10
mg/kg-
day
0.616
68.8
24.3
29.4
87.0
0.149
68.8
1.77
31.1
2.35
128
BMDL10
mg/kg-
day
0+
44.4
14.6
18.0
64.6
0+
44.4
0+
15.8
0+
77.6
Max
x23
0.020
1.120
0.879
1.014
1.494
0.130
1.120
0.015
1.059
0.020
1.701
Dose
maxb
mg/kg-
day
251
66
66
66
66
66
66
251
66
251
0
BMD10HED
0.09385
10.48
3.702
4.479
13.25
0.0227
10.48
0.2697
4.738
0.3580
19.50
BMDL10
HED
0
6.764
2.224
2.742
9.842
0
6.764
0
2.407
0
11.82
"Maximum absolute %2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bDose at which the maximum %2 residual deviation occurred.
"Best-fitting model not supra-linear.
May 2009
D-39
DRAFT - DO NOT CITE OR QUOTE
-------
o
0.9
0.8
0.6
0.5
0.4
0.3
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
BMD Lower Bound
$MPL BMP
0
23:05 04/27 2008
100
200
300
400
dose
500
600
700
800
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Source: JBRC (1998a).
Figure D-12. Log-logistic BMD model (Fixed power=l) for the combined incidence
of hepatic adenomas and carcinomas in male BDFi mice.
Logistic Model. (Version: 2.5; Date: 09/24/2005)
Input Data File: U:\DIOXANE\JBRCLIVER.(d)
Gnuplot Plotting File: U:\DIOXANE\JBRCLIVER.plt
Sun Apr 27 23:05:00 2008
JBRC MALE MOUSE carcino+adenoma slope = 1 loglogit Tbl D-14
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = EITHERMUSM
Independent variable = DOSEMUSM
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.42
intercept = -5.54398
slope = 1
May 2009
D-40
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
background
intercept
background
1
-0.69
intercept
-0.69
1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0.468776 0.0671164 0.33723 0.600322
intercept -5.63385 0.455627 -6.52686 -4.74084
slope 1 NA
NA - Indicates that this parameter has hit a bound
implied by some ineguality constraint and thus
has no standard error.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-117.031
-118.272
-126.524
240.543
Param's Deviance
4
2 2.48218
1 18.987
Goodness of Fit
Test d.f.
2
3
P-value
0.2891
0.0002751
Dose
0.0000
66.0000
251.0000
768.0000
Est. Prob.
0.4688
0.5702
0.7200
0.8582
Expected
23.439
27.369
36.000
41.192
Observed
21
31
37
39
Size
50
48
50
48
Scaled
Residual
-0.691
1.059
0.315
-0.907
= 2.52
d.f. = 2
P-value = 0.2836
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
31.0819
15.7951
D.7. COMBINING RISKS FOR DIFFERENT ENDPOINTS USING MULTISTAGE
MODELS
51 Analyses were restricted to those endpoints for which there were statistically significant
52 dose-response functions as determined by the criterion/? > 0.10. Table D-15 lists these for F344
53 female and male rats separately because there were sex differences.
May 2009
D-41
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-15. Statistically significant MS dose-response models for F344 rats
Endpoint
Liver adenomas and carcinomas
Nasal cavity tumors
Peritoneal mesotheliomas
Mammary gland adenomas
Female
Coefficients
used in model
1,2
1,2
/7-value
0.2357
0.5488
No significant
dose-response
1
0.8671
Male
Coefficients
used in model
1,2
1,2
1,2
^-value
0.4485
0.9564
0.8088
No significant
dose-response
1 The risks for the tumor type with the highest response at the largest dose (liver adenomas
2 and carcinomas, the key endpoint) were combined with the other statistically significant tumors
3 at other sites whose response at the highest dose also exceeded the BMR of 10% excess risk.
4 The MS-combo software was used. This software was recently developed by EPA to evaluate
5 the sensitivity of the BMDL for liver tumors to the inclusion of additional tumor sites. The risks
6 in the tables below were combined for liver tumors and an additional tumor type to better
7 evaluate how much the BMD and BMDL could be decreased when the excess risk is the
8 occurrence of a liver tumor or another significant tumor type, rather than just the occurrence of a
9 liver tumor. The sensitivity analysis was not extended to multiple tumor types including other
10 less significant or non-significant tumors occurring at other sites. The BMDLs in the following
11 tables are thus to be taken as upper bounds on the BMDL for the tumor types evaluated,
12 generally adenomas and carcinomas.
13 The effects can be examined on calculated human equivalent doses if several tumor types
14 are considered with the following example for F344 rats. The liver tumors are the most
15 significant tumor type for either sex. There are other sites for which a significant dose-response
16 relationship can be detected below the highest dose in the study: in male rats, peritoneal
17 mesotheliomas, and in female rats mammary gland mesotheliomas. A significant increase in
18 nasal cavity tumors occurs in the highest dose group only. One might hypothesize that when
19 risks are combined in the sense of finding a tumor either at one site or at the other, the greatest
20 decrease in BMD or BMDL occurs when the strength of the dose-response relationship at the
21 secondary site is nearly as large as at the primary site. The six pairwise combinations of tumors
22 have been ranked in order of increasing BMDL, and this hypothesis is demonstrated by the data
23 in the following tables.
24 Note that a very large reduction in BMDio HED and BMDLio HED occurs in Table D-17
25 relative to the separate BMDio HED and BMDLio HED for either the liver tumors or the peritoneal
26 mesotheliomas. When nasal cavity tumors are the secondary type, the reduction in BMDio HED
27 and BMDLio HED is relatively very small compared to the primary tumor type, either liver or
28 peritoneal mesotheliomas (Table D-17). For the female rats, one obtains similar findings with a
29 modest reduction in the BMDio HED and BMDLio HED when mammary gland adenomas are
May 2009
D-42
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
considered in addition to liver tumors (Table D-16), and relatively minor reductions in
BMDLio HED when nasal cavity tumors are combined with liver tumors or mammary gland
tumors (Table D-16). It therefore seems highly likely that considerably increasing the number of
rat tumor sites will yield substantial further reductions in BMDio HED and BMDLio HED.
Note that the smallest rat BMDio HED and BMDLio HED for combined liver tumors and
peritoneal mesotheliomas in male rats is 13.9 and 7.76 mg/kg-day respectively (Table D-17),
about 15-fold larger than the female mouse BMDio HED and BMDLio HED of 0.792 and 0.521
mg/kg-day respectively, and five- to sixfold larger than the male mouse BMDio HED and
BMDLio HED of 4.74 and 2.41 mg/kg-day respectively. If tumor data from other sites in the
mouse were available then one would expect the mouse BMDio HED and BMDLio HED values
combined over tumor sites would also be smaller. Therefore it is concluded that the mouse is the
more sensitive species and provides an appropriate basis for extrapolation to humans in a health
risk assessment.
Table D-16. MS-combo analysis of excess risks for liver adenomas/
carcinomas, mammary gland adenomas, or nasal cavity tumors in female
F344 rats using MS models
Tumor site
Liver
Mammary
Nasal cavity
Either
Coefficients
1,2
1
1
AIC
100.463
194.197
49.701
/7-value
0.2357
0.8671
0.5488
Liver or mammary
Liver or nasal cavity
Mammary or nasal cavity
BMD10
BMDL10
mg/kg-day
126
211
392
92.6
106
137
77.2
119
231
57.1
65.8
88.7
BMD10HED
BMDL10HED
mg/kg-day
30.2
50.5
93.8
22.2
25.4
32.8
18.5
28.5
55.3
13.7
15.7
21.25
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
MS_COMBO. (Version: 1.0; Date: 07/06/2007)
Input Data File: FLV2MMl.(d)
Gnuplot Plotting File: FLV2MMl.plt
Wed Apr 23 15:03:14 2008
Female Rat Liver Carcinomas or Adenomas AND Mammary Adenomas Degree 2, Tbl D-16
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = LIVCARAD
Independent variable = DOSE
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
May 2009
D-43
DRAFT - DO NOT CITE OR QUOTE
-------
1 Parameter Convergence has been set to: le-008
2
3 Default Initial Parameter Values
4 Background = 0.00712897
5 Beta(l) = 0.000360374
6 Beta(2) = 5.36416e-006
7
8 Asymptotic Correlation Matrix of Parameter Estimates
9
10 Background Beta(l) Beta(2)
11 Background 1 -0.61 0.5
12 Beta(l) -0.61 1 -0.95
13 Beta(2) 0.5 -0.95 1
14
15 Parameter Estimates
16 95.0% Wald Confidence Interval
17 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf.Limit
18 Background 0.0113405 * * *
19 Beta(l) 8.30975e-005 * * *
20 Beta(2) 6.01351e-006 * * *
21
22 * - Indicates that this value is not calculated.
23
24 Analysis of Deviance Table
25
26 Model Log(likelihood) # Param's Deviance Test d.f. P-value
27 Full model -46.1762 4
28 Fitted model -47.2315 3 2.11046 1 0.1463
29 Reduced model -107.855 1 123.358 3 <.0001
30
31 AIC: 100.463
32
33 Log-likelihood Constant 41.531079239232298
34 Goodness of Fit
35 Scaled
36 Dose Est. Prob. Expected Observed Size Residual
37
38
39
40
41
42
43 Chi^2 = 1.41 d.f. = 1 P-value = 0.2357
44
45 Benchmark Dose Computation
46 Specified effect = 0.1
47 Risk Type = Extra risk
48 Confidence level = 0.95
49 BMD = 125.636
50 BMDL = 77.1768
51 BMDU = 150.31
52
53 Taken together, (77.1768, 150.31 ) is a 90% two-sided confidence interval for the BMD
54
55 ====================================================================
56 MS_COMBO. (Version: 1.0; Date: 07/06/2007)
57 Input Data File: FLV2MMl.(d)
58 Gnuplot Plotting File: FLV2MMl.plt
59 Wed Apr 23 15:03:14 2008
60 ====================================================================
61 Female Rat Liver Carcinomas or Adenomas OR Mammary Adenomas Degree 2, Tbl D-16
f\~)
\J £*
63 The form of the probability function is:
64
65 P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]
66
67 The parameter betas are restricted to be positive
May 2009 D-44 DRAFT - DO NOT CITE OR QUOTE
0.0000
21.0000
103.0000
514.0000
0.0113
0.0157
0.0803
0.8066
0.567
0.784
4.017
40.329
1
0
5
40
50
50
50
50
0.578
-0.892
0.511
-0.118
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
Dependent variable = MAMMADEN
Independent variable = DOSE
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.135593
Beta(l) = 0.000477455
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.59
Beta(l) -0.59 1
Parameter Estimates
Variable
Background
Beta (1)
Estimate
0.132761
0.000498383
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-94.958
-95.0985
-98.6785
# Param's
4
2
1
Deviance Test d.f.
P-value
0.280866
7.4409
0.869
0.0591
194.197
Log-likelihood Constant
87.278562633109985
Goodness of Fit
Dose
0.0000
21.0000
103.0000
514.0000
Est. Prob.
0.1328
0.1418
0.1762
0.3287
Expected
6. 638
7.090
8.808
16.437
Observed
6
7
10
16
Size
50
50
50
50
Scaled
Residual
-0.266
-0.036
0.443
-0.132
= 0.29
d.f. = 2
P-value = 0.8671
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 211.405
BMDL = 118.855
BMDU = 597.799
Taken together, (118.855, 597.799) is a
interval for the BMD
90
% two-sided confidence
May 2009
D-45
DRAFT - DO NOT CITE OR QUOTE
-------
1 (Female Liver Carcinomas and Adenomas or Mammary Adenomas)
2 **** start of combined BMD and BMDL Calculations.****
3
4 Combined Log-Likelihood -142.32990579511375
5
6 Combined Log-likelihood Constant 128.8096418723423
7
8
9 Benchmark Dose Computation
10 Specified effect = 0.1
11 Risk Type = Extra risk
12 Confidence level = 0.95
13 BMD = 92.5711
14 BMDL = 57.0564
15 ====================================================================
16 MS_COMBO. (Version: 1.0; Date: 07/06/2007)
17 Input Data File: FLIV2NC2.(d)
18 Gnuplot Plotting File: FLIV2NC2.plt
19 Wed Apr 23 15:03:53 2008
20 ====================================================================
21 Female Rat Liver Carcinomas or Adenomas And Nasal Cavity Tumors Degree 2, Tbl D-16
22
23 The form of the probability function is:
24
25 P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]
26
27 The parameter betas are restricted to be positive
28
29 Dependent variable = NASALCAV
30 Independent variable = DOSE
31
32 Total number of observations = 4
33 Total number of records with missing values = 0
34 Total number of parameters in model = 2
35 Total number of specified parameters = 0
36 Degree of polynomial = 1
37 Maximum number of iterations =250
38 Relative Function Convergence has been set to: le-008
39 Parameter Convergence has been set to: le-008
40
41 Default Initial Parameter Values
42 Background = 0
43 Beta(l) = 0.000356274
44
45 Asymptotic Correlation Matrix of Parameter Estimates
46
47 (*** The model parameter(s) -Background, have been estimated at a boundary point, or
48 have been specified by the user, and do not appear in the correlation matrix)
49
50 Beta(l)
51 Beta(l) 1
52
53 Parameter Estimates
54
55 95.0% Wald Confidence Interval
56 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
57 Background 0 * * *
58 Beta(l) 0.000268486 * * *
59
60 * - Indicates that this value is not calculated.
61
62 Analysis of Deviance Table
63
64 Model Log(likelihood) # Param's Deviance Test d.f. P-value
65 Full model -21.9835 4
66 Fitted model -23.8503 1 3.73353 3 0.2917
67 Reduced model -33.5888 1 23.2107 3 <.0001
May 2009 D-46 DRAFT - DO NOT CITE OR QUOTE
-------
Dose
0.0000
21.0000
103.0000
514.0000
Chi^2 =2.12
Est. Prob.
0.0000
0.0056
0.0273
0.1289
d.f. = 3
Expected Ob:
0.000
0.281
1.364
6.445
P-value
served
0
0
0
8
= 0.5488
Size
50
50
50
50
Scaled
Residual
0.000
-0.532
-1.184
0.656
1
2 AIC: 49.7005
3
4 Log-likelihood Constant 20.101282649283011
5 Goodness of Fit
6
7
8
9
10
11
12
13
14 Benchmark Dose Computation
15 Specified effect = 0.1
16 Risk Type = Extra risk
17 Confidence level = 0.95
18 BMD = 392.425
19 BMDL = 230.801
20 BMDU = 747.347
21
22 Taken together, (230.801, 747.347) is a 90% two-sided confidence interval for the BMD
23
24 (Female Liver Carcinomas and Adenomas or Nasal Cavity Tumors)
25 **** start of combined BMD and BMDL Calculations.****
26
27 Combined Log-Likelihood -71.081712052481365
28
29 Combined Log-likelihood Constant 61.632361888515305
30
31 Benchmark Dose Computation
32 Specified effect = 0.1
33 Risk Type = Extra risk
34 Confidence level = 0.95
35 BMD = 106.322
36 BMDL = 65.8109
37
38 (Female Mammary Adenomas or Nasal Cavity Tumors)
39 **** start of combined BMD and BMDL Calculations.****
40
41 Combined Log-Likelihood -118.9487081361909
42
43 Combined Log-likelihood Constant 107.37984528239299
44
45 Benchmark Dose Computation
46 Specified effect = 0.1
47 Risk Type = Extra risk
48 Confidence level = 0.95
49 BMD = 137.391
50 BMDL = 88.7463
May 2009 D-47 DRAFT - DO NOT CITE OR QUOTE
-------
Table D-17. MS-combo analysis of excess risks for liver adenomas, liver
carcinomas, nasal cavity tumors, or peritoneal mesotheliomas in male F344
rats using MS models
Tumor site
Liver
Peritoneal
Mesothelioma
Nasal cavity
Either
Coefficients
1,2
2
2
AIC
113.973
140.537
43.119
/7-value
0.4485
0.8088
0.9564
Liver or peritoneal mesothelioma
Liver or nasal cavity
Peritoneal mesothelioma or nasal
cavity
BMD10
BMDL10
mg/kg-day
73.8
112.0
340
51.1
71.1
104
42.6
51.0
255
28.6
42.0
49.9
BMD10HED
BMDL10HED
mg/kg-day
20.0
30.4
92.3
13.9
19.3
28.2
11.6
13.8
69.7
7.76
11.4
13.5
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
MS_COMBO. (Version: 1.0; Date: 07/06/2007)
Input Data File: MLV12NC2.(d)
Gnuplot Plotting File: MLV12NC2.plt
Wed Apr 23 15:02:29 2008
Male Rat Liver Carcinomas or Adenomas AND and Nasal Cavity Degree 1&2, Tbl D-17
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = LIVCARAD
Independent variable = DOSE
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0136508
Beta(l) = 0.000489073
Beta(2) = 5.49397e-006
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background
have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix )
Beta(l)
Beta(2)
Beta(l)
1
-0.97
Beta(2)
-0.97
1
May 2009
D-48
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
Parameter Estimates
Variable
Background
Beta(l)
Beta (2)
Estimate
0
0.00114469
3.82589e-006
Std. Err.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
* *
* *
Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log (likelihood) # Param's Deviance Test d.f. P-value
-54.3032
-54.9865
-98.4609
1.36669
88.3155
0.5049
<.0001
113.973
Log-likelihood Constant
49.292679083903337
Goodness of Fit
Dose Est._Prob. Expected Observed Size
Scaled
Residual
0.0000
16.0000
81.0000
398.0000
0.0000
0.0191
0.1111
0.6541
0.000
0.955
5.446
32.705
0
2
4
33
50
50
49
50
0.000
1.079
-0.657
0.088
ChiA2 =1.60
d.f. = 2
P-value = 0.4485
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0.1
Extra risk
0.95
73.8264
42.6043
137.575
Taken together, (42.6043, 137.575) is a 90% two-sided confidence interval for the BMD
MS_COMBO. (Version: 1.0; Date: 07/06/2007)
Input Data File: MLV12NC2.(d)
Gnuplot Plotting File: MLV12NC2.plt
Wed Apr 23 15:02:29 2008
Male Rat Liver Carcinomas or Adenomas AND and Nasal Cavity Degree 1&2, Tbl D-17
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = NASALCAV
Independent variable = DOSE
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
May 2009
D-49
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
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
Beta(l) = 0
Beta(2) = 9.64541e-007
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background -Beta(l)
have been estimated at a boundary point, or have been specified by
the user, and do not appear in the correlation matrix )
Beta(2)
Beta(2)
Parameter Estimates
Variable
Background
Beta (1)
Beta (2)
Estimate
0
0
9.10658e-007
Std. Err.
*
*
*
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-20.2482
-20.5594
-30.3072
# Param' s
4
1
1
Deviance
0.622507
20.118
Test d.f.
3
3
P-value
0.8913
0.0001604
43.1189
Log-likelihood Constant
18.41952407526928
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0000
0.0002
0.0060
0.1343
Expected
0.000
0.012
0.292
6.717
Observed
0
0
0
7
Size
50
50
49
50
Scaled
Residual
0.000
-0.108
-0.542
0.118
ChiA2 = 0.32
d.f. = 3
P-value = 0.9564
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0.1
Extra risk
0.95
340.143
255.307
481.19
Taken together, (255.307, 481.19 ) is a 90% two-sided confidence interval for the BMD
May 2009
D-50
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
(Male rat liver adenomas and carcinomas or nasal cavity tumors)
**** Start of combined BMD and BMDL Calculations.****
Combined Log-Likelihood
Combined Log-likelihood Constant
Benchmark Dose Computation
-75.545958033937922
67.712203159172617
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0.1
Extra risk
0.95
71.116
41.9581
MS_COMBO. (Version: 1.0; Date: 07/06/2007)
Input Data File: MLV12PR2.(d)
Gnuplot Plotting File: MLV12PR2.plt
Wed Apr 23 15:01:06 2008
Male Rat Liver Carcinomas or Adenomas AND Peritonial Mesothelioma Degree 1&2, Tbl D-27
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = PERITON
Independent variable = DOSE
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.035746
Beta(l) = 0.000579165
Beta(2) = 3.49814e-006
Asymptotic Correlation Matrix of Parameter Estimates
Background
Beta(l)
Beta(2)
Background
1
-0.66
0.59
Beta(l)
-0.66
1
-0.98
Beta(2)
0.59
-0.98
1
Parameter Estimates
Variable
Background
Beta (1)
Beta (2)
Estimate
0.0365244
0.00053631
3.60862e-006
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
* - Indicates that this value is not calculated.
May 2009
D-51
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood) # Param's Deviance Test d.f.
-67.2386 4
-67.2683 3 0.0595644 1
-95.5731 1 56.6691 3
P-value
0.8072
<.0001
140.537
Log-likelihood Constant
60.799215762920063
Goodness of Fit
Dose
0.0000
16.0000
81.0000
398.0000
Est. Prob.
0.0365
0.0456
0.0991
0.5606
Expected
1.826
2.282
4.854
28.029
Observed
2
2
5
28
Size
50
50
49
50
Scaled
Residual
0.131
-0.191
0.070
-0.008
=0.06
d.f. = 1
P-value = 0.
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0.1
Extra risk
0.95
112.02
51.0435
171.695
Taken together, (51.0435, 171.695) is a 90% two-sided confidence interval for the BMD
(Male rat liver adenomas and carcinomas or peritoneal mesotheliomas)
**** Start of combined BMD and BMDL Calculations.****
-122.25487850925802
110.09189484682341
Combined Log-Likelihood
Combined Log-likelihood Constant
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 51.1199
BMDL = 28.5793
(Male rat peritoneal mesotheliomas or nasal cavity tumors)
**** Start of combined BMD and BMDL Calculations.****
Combined Log-Likelihood
Combined Log-likelihood Constant
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 104.476
BMDL = 49.9353
-87.827775784366935
79.21873983818935
May 2009
D-52
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-18. Calculation of HED values for additional studies reporting the
incidence of liver and nasal cavity tumors in rats and mice exposed to
1,4-dioxane in the drinking water for 2 years
Source
Kocibaetal., 1974
NCI, 1978
Species/strain/gender
Sherman rats, male and female
combined
Male Osborne-Mendel rats
Female Osborne-Mendel rats
Male B6C3FJ mice
Female B6C3F! mice
Animal BW (g)
TWAa
325
325
285b
470
470
310
310
32
32
30
30
Animal dose
(mg/kg-day)
14
121
1307
240
530
350
640
720
830
380
860
HED
(mg/kg-day)c
3.7
32
330
69
152
90
165
105
121
55
124
aTWA BWs were determined from the BW curve provided for control animals unless otherwise indicated.
bBWs of high dose male and female rats were significantly lower than controls throughout the study. TWA
represents the mean of TWA for male and females (calculated separately from growth curves).
°HEDs are calculated as HED = (animal dose) x (animal BW/human BW)%.
Table D-19. Summary of BMD modeling estimates and CSF values
associated with liver and nasal tumor incidence data resulting from chronic
oral exposure to 1,4-dioxane in rats and mice
Source
Species/strain/gender
BMD10HED
(mg/kg-day)
BMDL10HED
(mg/kg-day)
Liver tumors
Kocibaetal., 1974
NCI, 1978
Sherman rats, male and female combined3
Female Osborne-Mendel ratsb
Male B6C3FJ micec
Female B6C3FJ micec
238.9
30.19
51.68
23.47
148.4
21.44
18.40
9.87
Nasal cavity tumors
Kocibaetal., 1974
NCI, 1978
Sherman rats, male and female combined"1
Male Osborne-Mendel ratsd
Female Osborne-Mendel ratsd
880.8
18.75
36.90
387.8
13.85
25.57
"Incidence of hepatocellular carcinoma
blncidence of hepatocellular adenoma
Incidence of hepatocellular adenoma or carcinoma
Incidence of nasal squamous cell carcinoma
May 2009
D-53
DRAFT - DO NOT CITE OR QUOTE
-------
D.8. BMD MODELING RESULTS FROM ADDITIONAL CHRONIC BIOASSAYS (NCI,
1978; KOCIBA ET AL., 1974)
D.8.1. Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma (Kociba et al.,
1974)
1 The incidence data for hepatocellular carcinoma and nasal squamous cell carcinoma are
2 presented in Table D-20. As assessed by the j^ goodness-of-fit statistic, all degree MS
3 polynomial models (betas restricted > 0) provided adequate fit (y?p> 0.1) to the data for the
4 incidence of hepatocellular carcinoma and nasal squamous cell carcinoma (Table D-21). The
5 one-degree model was the lowest degree polynomial that provided an adequate fit to the data
6 (Figures D-13 and D-14). The predicted BMDio HED and BMDLio HED values are also presented
7 in Table D-21.
Table D-20. Incidence of hepatocellular carcinoma and nasal squamous cell
carcinoma in male and female Sherman rats (combined) treated with
1,4-dioxane in the drinking water for 2 years
HED (mg/kg-day)
(average of male and female dose)
0
3.7
32
330
Incidence of hepatocellular
carcinoma"
l/106b
0/110
1/106
10/66d
Incidence of nasal
squamous cell carcinoma3
0/106C
0/110
0/106
3/66d
aRats surviving until 12 months on study.
bp < 0.001; positive dose-related trend (Cochran-Armitage test).
°p < 0.01; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: Kociba etal. (1974).
May 2009
D-54
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-21. Goodness-of-fit statistics and BMDio HED and BMDLio HED from
multistage models fit to incidence data for hepatocellular carcinoma and
nasal tumors in male and female Sherman rats (combined) exposed to
1,4-dioxane in the drinking water for 2 years
Polynomial Degree
x2
^-value"
AIC
BMD10HED
(mg/kg-day)
BMDL10HED
(mg/kg-day)
Hepatocellular carcinoma
3
2
lb
0.31
0.31
0.37
86.28
86.29
85.20
263.56
263.56
238.92
161.11
161.11
148.35
Nasal squamous cell carcinoma
3
2
lb
1.00
1.00
0.91
26.42
26.50
27.39
433.59
500.51
880.84
329.84
332.09
387.79
aValues < 0.1 fail to meet conventional goodness-of-fit criteria.
bLowest degree polynomial with adequate fit.
Source: Kocibaetal. (1974).
Multistage Model with 0.95 Confidence Level
0.25
0.2
0.15
0.05
Multistage
BMDL
50
100
150 200
dose
250
300
350
22:41 11/29 2006
Source: Kociba et al. (1974).
Figure D-13. BMD multistage model (1-degree polynomial) of the incidence of
hepatocellular carcinoma in male and female Sherman rats exposed to 1,4-dioxane
in drinking water.
May 2009
D-55
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
0.1
3=
< 0.06
0.04
0.02
Multistage
BMD
MD
200
400
600
800
dose
22:51 11/29 2006
Source: Kociba et al. (1974).
Figure D-14. BMD multistage model (1-degree polynomial) of the incidence of nasal
squamous cell carcinoma in male and female Sherman rats exposed to 1,4-dioxane
in drinking water.
D.8.2. Nasal Cavity Squamous Cell Carcinoma and Liver Hepatocellular Adenoma in
Osborne-Mendel Rats (NCI, 1978)
1 The incidence data for hepatocellular adenoma (female rats) and nasal squamous cell
2 carcinoma (male and female rats) are presented in Table D-22. The one-degree multistage model
3 (betas restricted > 0) adequately fit both the male and female rat nasal squamous cell carcinoma
4 data (Figures D-15 to D-5). The predicted BMDio HED and BMDLio HED values are presented in
5 Table D-23.
May 2009
D-56
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-22. Incidence of nasal cavity squamous cell carcinoma and liver
hepatocellular adenoma in Osborne-Mendel rats exposed to 1,4-dioxane in
the drinking water
Male rat RED (mg/kg-day)a
Nasal cavity squamous cell carcinoma
0
0/33C
69b
12/26d
152
16733d
Female rat HED (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/34C
0/3 lc
90
10/30d
10/30d
165
8/29d
ll/29d
"Tumor incidence values were adjusted for mortality (animals surviving to 52 weeks, presented in text of
NCI, 1978).
bGroup not included in statistical analysis by NCI (1978) because the dose group was started a year earlier
without appropriate controls.
°p < 0.001; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: NCI (1978).
Table D-23. Goodness-of-fit statistics and BMDio HED and BMDLio HED from
multistage models fit to incidence data for hepatocellular adenoma and nasal
tumors in male and female Osborne-Mendel rats exposed to 1,4-dioxane in
the drinking water for 2 years
Degree
polynomial
X2
/7-valuea
AIC
BMD10HED
(mg/kg-day)
BMDL10HED
(mg/kg-day)
Males
Nasal cavity squamous cell carcinoma
lb
0.18
86.88
18.75
13.85
Females
Nasal cavity squamous cell carcinoma
lb
0.20
77.40
36.90
25.57
Hepatocellular adenoma
lb
0.60
79.69
30.19
21.44
"Values <0.1 fail to meet conventional goodness-of-fit criteria.
bLowest degree polynomial with adequate fit.
Source: NCI (1978).
May 2009
D-57
DRAFT - DO NOT CITE OR QUOTE
-------
0.7
0.6
0.5
3 0.4
0.3
0.2
0.1
Multistage
BMD
23:06 11/292006
Multistage Model with 0.95 Confidence Level
0 20 40 60
dose
100 120 140 160
Source: NCI (1978).
Figure D-15. BMD multistage model (1-degree polynomial) of the incidence of nasal
squamous cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
May 2009
D-58
DRAFT - DO NOT CITE OR QUOTE
-------
0.5
0.4
0.3
0.2
0.1
Multistage Model with 0.95 Confidence Level
Multistage
BMDL
BMD
20
40
60
100
120 140 160 180
dose
23:11 11/292006
Source: NCI (1978).
Figure D-16. BMD multistage model (1-degree polynomial) of the incidence of nasal
squamous cell carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
May 2009
D-59
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
0.6
0.5
0.4
0.3
0.2
0.1
Multistage
BMDL
BMD
20
40
60
80 100
dose
120 140 160 180
23:16 11/29 2006
Source: NCI (1978).
Figure D-17. BMD multistage model (1-degree polynomial) of the incidence of
hepatocellular adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
D.8.3. Hepatocellular Adenoma or Carcinoma in B6C3Fi Mice (NCI, 1978)
1 The incidence data for hepatocellular adenoma or carcinoma in male and female
2 mice are presented in Table D-24. As assessed by the j^ goodness-of-fit statistic, only the
3 2-degree polynomial models, provided adequate fit (%2 p > 0.1) to the data for the
4 incidence of hepatocellular carcinoma in both male and female mice (Table D-25). The
5 2-degree polynomial model (betas restricted > 0) was the lowest degree polynomial that
6 provided an adequate fit to both the male and female mouse data (Figures D-18 and D-19).
7 The predicted BMDio HED and BMDLio HED values are also presented in Table D-25.
May 2009
D-60
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-24. Incidence of hepatocellular adenoma or carcinoma in B6C3Fi
mice exposed to 1,4-dioxane in drinking water
Male mouse HED (mg/kg-day)a
0
8/49b
105
19/50d
121
28/47c
Female mouse HED (mg/kg-day)a
0
0/50b
55
21/48C
124
35/37c
"Tumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
°p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (1978).
Table D-25. Goodness-of-fit statistics and BMDiOHED and BMDLiOHED
values from multistage models fit to incidence data for hepatocellular
adenoma or carcinoma in male and female B6C3Fi mice exposed to
1,4-dioxane in the drinking water for 2 years
Degree
polynomial
x2
/7-valuea
AIC
BMD10HED
(mg/kg-day)
BMDL10HED
(mg/kg-day)
Males
2b
1
0.16
0.08
179.49
180.62
51.68
23.96
18.40
17.11
Females
2b
1
1.00
0.05
85.35
90.05
23.47
7.10
9.87
5.61
"Values < 0.1 fail to meet conventional goodness-of-fit criteria.
bLowest degree polynomial with adequate fit.
Source: NCI (1978).
May 2009
D-61
DRAFT - DO NOT CITE OR QUOTE
-------
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Model with 0.95 Confidence Level
Multistage
BMDI
MD
20
40
60
dose
100
120
08:32 12/01 2006
Source: NCI (1978).
Figure D-18. BMD multistage model (2-degree polynomial) of the incidence of
hepatocellular adenoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking
water.
May 2009
D-62
DRAFT - DO NOT CITE OR QUOTE
-------
0.8
0.6
0.4
Multistage Model with 0.95 Confidence Level
Multistage
BMDL
3MD
40
60
dose
100
120
08:37 12/01 2006
Source: NCI (1978).
Figure D-19. BMD multistage model (2-degree polynomial) of the incidence of
hepatocellular adenoma in female B6C3Fi mice exposed to 1,4-dioxane in drinking
water.
May 2009
D-63
DRAFT - DO NOT CITE OR QUOTE
------- |