DRAFT - DO NOT CITE OR QUOTE
EPA/63 5/R-09/005D
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 2010
This document is an Interagency Science Discussion draft. This information is distributed
solely for the purpose of pre-dissemination peer review under applicable information quality
guidelines. It has not been formally disseminated by EPA. It does not represent and should not
be construed to represent any Agency determination or policy. It is being circulated for review
of its technical accuracy and science policy implications.
U.S. Environmental Protection Agency
Washington, DC

-------
DISCLAIMER
This document is a preliminary draft for review purposes only and it has been reviewed
in accordance with U.S. Environmental Protection Agency policy and approved for publication..
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.
March 2010
11
DRAFT - DO NOT CITE OR QUOTE

-------
TABLE OF CONTENTS
LIST OF TABLES	vii
LIST OF FIGURES	x
LIST OF ABBREVIATIONS AND ACRONYMS	xiv
FOREWORD	 	xvii
AUTHORS, CONTRIBUTORS, AND REVIEWERS	xviii
1.	INTRODUCTION	 1
2.	CHEMICAL AM) 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	21
4.	HAZARD IDENTIFICATION	22
4.1.	STUDIES IN HUMANS - EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS	22
4.1.1.	Thiess etal. (1976)	 24
4.1.2.	Bufiler et al. (1978)	 25
4.2.	SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIO AS SAYS IN
ANIMALS - ORAL AM) INHALATION	26
4.2.1. Oral Toxicity	26
4.2.1.1.	Subchronic Oral Toxicity	26
4.2.1.1.1.	Stoner etal. (1986)	 26
4.2.1.1.2.	Stott et al. (1981)	27
4.2.1.1.3.	Kano etal. (2008)	 27
4.2.1.1.4.	Yamamoto et al. (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.	Kocibaet al. (1974)	 35
March 2010	iii	DRAFT - DO NOT CITE OR QUOTE

-------
4.2.1.2.5.	National Cancer Institute (NCI) (1978)	 37
4.2.1.2.6.	Kano et al. (2009); Japan Bioassay Research Center (JBRC) (1998a);
Yamazaki et al. (1994)	 41
4.2.2.	Inhalation Toxicity	51
4.2.2.1.	Subchronic Inhalation Toxicity	51
4.2.2.1.1. Faiiiey et al. (1934)	 51
4.2.2.2.	Chronic Inhalation Toxicity and Carcinogenicity	51
4.2.2.2.1. Torkelson etal. (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 et al. (1987)	 54
4.3.	REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION	54
4.3.1. Giavini et al. (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 et al. (1994)	 58
4.4.2.2.	Goldberg et al. (1964)	 59
4.4.2.3.	Kanada et al. (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	76
4.7.	EVALUATION OI CARCINOGENICITY	77
4.7.1.	Summary of Overall Weight of Evidence	77
4.7.2.	Synthesis of Human, Animal, and Other Supporting Evidence	78
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	81
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
March 2010	iv	DRAFT - DO NOT CITE OR QUOTE

-------
4.7.3.4.2. Nasal cavity	85
4.7.3.5.	Biological Plausibility and Coherence	85
4.7.3.5.1.	Liver	85
4.7.3.5.2.	Nasal cavity	86
4.7.3.6.	Other Possible Modes of Action	86
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	87
4.7.3.8.	Relevance of the Mode of Action to Humans	87
4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES	87
5.	DOSE-RESPONSE ASSESSMENTS	89
5.1.	ORAL REFERENCE DOSE (RID)	89
5.1.1.	Choice of Principal Studies and Critical Effect with Rationale and Justification	89
5.1.2.	Methods of Analysis—Including Models (PBPK, BMD, etc.)	90
5.1.3.	RfD Derivation - Including Application of Uncertainty Factors (UFs)	92
5.1.4.	RfD Comparison Information	93
5.1.5.	Previous RfD Assessment	98
5.2.	INHALATION REFERENCE CONCENTRATION (RfC)	98
5.3.	UNCERTAINTIES IN THE ORAL REFERENCE DOSE (RfD)	99
5.4.	CANCER ASSESSMENT	 101
5.4.1.	Choice of Study/Data - with Rationale and Justification	101
5.4.2.	Dose-Response Data	102
5.4.3.	Dose Adjustments and Extrapolation Method(s)	103
5.4.3.1.	Dose Adjustments	 103
5.4.3.2.	Extrapolation Method(s)	 105
5.4.4.	Oral Slope Factor and Inhalation Unit Risk	106
5.4.5.	Previous Cancer Assessment	108
5.5.	UNCERTAINTIES IN CANCER RISK VALUES	 108
5.5.1. Sources of Uncertainty	 108
5.5.1.1.	Choice of Low-Dose Extrapolation Approach	108
5.5.1.2.	Dose Metric	 110
5.5.1.3.	Cross-Species Scaling	110
5.5.1.4.	Statistical Uncertainty at the POD	 110
5.5.1.5.	Bioassay Selection	110
5.5.1.6.	Choice of Species/Gender	110
5.5.1.7.	Relevance to Humans	 Ill
5.5.1.8.	Human Population Variability	 Ill
6.	MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE	 113
6.1.	HUMAN HAZARD POTENTIAL	 113
6.2.	DOSE RESPONSE	 114
6.2.1.	Noncancer/Oral	 114
6.2.2.	Noncancer/Inhalation	 115
6.2.3.	Cancer/Oral	 115
6.2.3.1.	Choice of Low-Dose Extrapolation Approach	115
6.2.3.2.	Dose Metric	 117
6.2.3.3.	Cross-Species Scaling	117
March 2010
v
DRAFT - DO NOT CITE OR QUOTE

-------
6.2.3.4.	Statistical Uncertainty at the POD	 117
6.2.3.5.	Bioassay Selection	117
6.2.3.6.	Choice of Species/Gender	117
6.2.3.7.	Relevance to Humans	 118
6.2.3.8.	Human Population Variability	 118
6.2.4. Cancer/Inhalation	 118
7. REFERENCES	 119
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION	A-l
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
March 2010
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/DuCij 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/DuCij rats exposed to 1,4-dioxane in
drinking water for 2 years	47
Table 4-12. Incidence of histopathological lesions in male Cij :BDFi mice exposed to
1,4-dioxane in drinking water for 2 years	49
Table 4-13. Incidence of histopathological lesions in female Cij :BDFi mice exposed to
1,4-dioxane in drinking water for 2 years	49
Table 4-14. Incidence of liver tumors in Crj :BDF1 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; in vitro	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	83
Table 5-1. Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water for 2 years	91
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	92
Table 5-3. Incidence of liver hyperplasia in F344/DuCij rats exposed to 1,4-dioxane in
drinking water for 2 years	92
Table 5-4. BMD and BMDL values derived from BMD modeling of liver hyperplasia in
male and female F344/DuCij rats exposed to 1,4-dioxane in drinking water for
2 years	92
March 2010
vii
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)	 101
Table 5-6. Incidence of hepatocellular adenoma or carcinoma in rats and mice exposed to
1,4-dioxane in drinking water for 2 years	 103
Table 5-7. Calculated HEDs for the tumor incidence data used for dose-response modeling ..104
Table 5-8. BMDhed and BMDLhed 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	Error! Bookmark not defined.
Table 5-9. Summary of uncertainty in the 1,4-dioxane cancer risk assessment	112
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-l 1
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 (NCI, 1978) exposed to 1,4-dioxane in drinking water	C-2
Table C-3. Incidence of liver hyperplasia in F344/DuCij rats exposed to 1,4-dioxane in
drinking water	C-l
Table C-4. Benchmark dose modeling results based on the incidence of liver hyperplasias
in male and female F344 rats exposed to 1,4-dioxane in drinking water for 2
years	C-8
Table D-l. Recommended models for rodents exposed to 1,4-dioxane in drinking water
(Kano et al., 2009)	D-4
Table D-2. Data for hepatic adenomas and carcinomas in female F344 rats (Kano et al.,
2009)	D-4
Table D-3. BMDS dose-response modeling results for the combined incidence of hepatic
adenomas and carcinomas in female F344 rats (Kano et al., 2009)	D-5
Table D-4. Data for hepatic adenomas and carcinomas in male F344 rats (Kano et al.,
2009)	D-8
Table D-5. BMDS dose-response modeling results for the combined incidence of
adenomas and carcinomas in livers of male F344 rats (Kano et al., 2009)	D-9
Table D-6. Data for significant tumors at other sites in male and female F344 rats (Kano et
al., 2009)	D-14
Table D-7. BMDS dose-response modeling results for the incidence of nasal cavity tumors
in female F344 ratsa (Kano et al., 2009)	D-l5
Table D-8. BMDS dose-response modeling results for the incidence of nasal cavity tumors
in male F344 ratsa (Kano et al., 2009)	D-l8
Table D-9. BMDS dose-response modeling results for the incidence of mammary gland
adenomas in female F344 rats (Kano et al., 2009)	D-21
Table D-10. BMDS dose-response modeling results for the incidence of peritoneal
mesotheliomas in male F344 rats (Kano et al., 2009)	D-26
March 2010
viii
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-l 1. Data for hepatic adenomas and carcinomas in female BDF1 mice (Kano et al.,
2009)	D-31
Table D-12. BMDS dose-response modeling results for the combined incidence of hepatic
adenomas and carcinomas in female BDFi mice (Kano et al., 2009)	D-32
Table D-13. BMDS LogLogistic dose-response modeling results using BMRs of 10, 30,
and 50% for the combined incidence of hepatic adenomas and carcinomas in
female BDFi mice (Kano et al., 2009)	D-32
Table D-14. Data for hepatic adenomas and carcinomas in male BDFi mice (Kano et al.,
2009)	D-41
Table D-15. BMDS dose-response modeling results for the combined incidence of hepatic
adenomas and carcinomas in male BDFi mice (Kano et al., 2009)	D-42
Table D-16. Summary of BMDS dose-response modeling estimates associated with liver
and nasal tumor incidence data resulting from chronic oral exposure to
1,4-dioxane in rats and mice	D-48
Table D-17. Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in
male and female Sherman rats (combined) (Kociba et al., 1974) treated with
1,4-dioxane in the drinking water for 2 years	D-49
Table D-l 8. BMDS dose-response modeling results for the incidence of hepatocellular
carcinoma in male and female Sherman rats (combined) (Kociba et al., 1974)
exposed to 1,4-dioxane in the drinking water for 2 years	D-50
Table D-20. Incidence of nasal cavity squamous cell carcinoma and hepatocellular
adenoma in Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the
drinking water	D-58
Table D-21. BMDS dose-response modeling results for the incidence of hepatocellular
adenoma in female Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane
in the drinking water for 2 years	D-59
Table D-24. Incidence of hepatocellular adenoma or carcinoma in male and female
B6C3F1 mice (NCI, 1978) exposed to 1,4-dioxane in drinking water	D-74
Table D-25. BMDS dose-response modeling results for the combined incidence of
hepatocellular adenoma or carcinoma in female B6C3Fi mice (NCI, 1978)
exposed to 1,4-dioxane in the drinking water for 2 years	D-75
Table D-26. BMDS dose-response modeling results for the combined incidence of
hepatocellular adenoma or carcinoma in male B6C3Fi mice (NCI, 1978)
exposed to 1,4-dioxane in drinking water	D-78
March 2010
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	95
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	96
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	97
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	98
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-l3
March 2010	x	DRAFT - DO NOT CITE OR QUOTE

-------
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
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	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 HEAA 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
zero-order metabolism rate constant, Iclc, to the experimental data	B-19
Figure B-15. Predictions of blood 1,4-dioxane concentration following calibration of a
zero-order metabolism rate constant, Iclc, 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 zero-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 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-5
Figure C-3. 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-9
Figure C-4. BMD multistage (2 degree) 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-ll
Figure C-5. 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-l3
Figure C-6. 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-l5
Figure C-l. 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-l7
Figure D-l. Multistage BMD model (2 degree) for the combined incidence of hepatic
adenomas and carcinomas in female F344 rats	D-6
Figure D-2. Probit BMD model for the combined incidence of hepatic adenomas and
carcinomas in male F344 rats	D-10
Figure D-3. Multistage BMD model (3 degree) for the combined incidence of hepatic
adenomas and carcinomas in male F344 rats	D-12
Figure D-4. Multistage BMD model (3 degree) for nasal cavity tumors in female F344 rats. D-16
Figure D-5. Multistage BMD model (3 degree) for nasal cavity tumors in male F344 rats.... D-19
Figure D-6. LogLogistic BMD model for mammary gland adenomas in female F344 rats. .. D-22
March 2010	xi	DRAFT - DO NOT CITE OR QUOTE

-------
Figure D-7. Multistage BMD model (1 degree) for mammary gland adenomas in female
F344 rats	D-24
Figure D-8. Probit BMD model for peritoneal mesotheliomas in male F344 rats	D-27
Figure D-9. Multistage BMD (2 degree) model for peritoneal mesotheliomas in male F344
rats	D-29
Figure D-10. LogLogistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in female BDFi mice with a BMR of 10%	D-33
Figure D-l 1. LogLogistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in female BDFi mice with a BMR of 30%	D-35
Figure D-12. LogLogistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in female BDFi mice with a BMR of 50%	D-37
Figure D-13. Multistage BMD model (1 degree) for the combined incidence of hepatic
adenomas and carcinomas in female BDFI mice	D-39
Figure D-14. LogLogistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in male BDFi mice	D-43
Figure D-15. Multistage BMD model (1 degree) for the combined incidence of hepatic
adenomas and carcinomas in male BDFi mice	D-45
Figure D-16. Probit BMD model for the incidence of hepatocellular carcinoma in male and
female Sherman rats exposed to 1,4-dioxane in drinking water	D-51
Figure D-17. Multistage BMD model (1 degree) for the incidence of hepatocellular
carcinoma in male and female Sherman rats exposed to 1,4-dioxane in drinking
water	D-53
Table D-19. BMDS dose-response modeling results for the incidence of nasal squamous
cell carcinoma in male and female Sherman rats (combined) (Kociba et al.,
1974) exposed to 1,4-dioxane in the drinking water for 2 years	D-55
Figure D-l 8. Multistage BMD model (3 degree) for 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-19. LogLogistic BMD model for the incidence of hepatocellular adenoma in
female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water	D-60
Figure D-20. Multistage BMD model (1 degree) for the incidence of hepatocellular
adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water	D-62
Table D-22. BMDS dose-response modeling results for the incidence of nasal cavity
squamous cell carcinoma in female Osborne-Mendel rats (NCI, 1978) exposed
to 1,4-dioxane in the drinking water for 2 years	D-64
Figure D-21. LogLogistic BMD model for the incidence of nasal cavity squamous cell
carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water	D-65
Figure D-22. Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in female Osborne-Mendel rats exposed to
1,4-dioxane in drinking water	D-67
Table D-23. BMDS dose-response modeling results for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel rats (NCI, 1978) exposed to
1,4-dioxane in the drinking water for 2 years	D-69
Figure D-23. LogLogistic BMD model for the incidence of nasal cavity squamous cell
carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water	D-70
March 2010
Xll
DRAFT - DO NOT CITE OR QUOTE

-------
Figure D-24. Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane
in drinking water	D-72
Figure D-25. Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in female B6C3Fi mice exposed to 1,4-dioxane in
drinking water	D-76
Figure D-26. Gamma BMD model for the incidence of hepatocellular adenoma or
carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking water	D-79
Figure D-27. Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in
drinking water	D-81
March 2010
Xlll
DRAFT - DO NOT CITE OR QUOTE

-------
LIST OF ABBREVIATIONS AND ACRONYMS
AIC
Akaike's Information Criterion

ALP
alkaline phosphatase

ALT
alanine aminotransferase

AST
aspartate aminotransferase

ATSDR
Agency for Toxic Substances and Disease Registry

BMD
benchmark dose

BMDio
benchmark dose at 10% extra risk

BMD30
benchmark dose at 30% extra risk

BMD50
benchmark dose at 50% extra risk

BMDL
benchmark dose, lower 95% confidence limit

BMDL10
benchmark dose, lower 95% confidence limit at 10% extra
risk
BMDL30
benchmark dose, lower 95% confidence limit at 30% extra
risk
BMDL50
benchmark dose, lower 95% confidence limit at 50% extra
risk
BMDS
Benchmark Dose Software

BMR
benchmark response

BrdU
5 -b romo-2' -deoxyuri dine

BUN
blood urea nitrogen

BW(s)
body weight(s)

CASE
computer automated structure evaluator

CASRN
Chemical Abstracts Service Registry Number

CHO
Chinese hamster ovary (cells)

CI
confidence interval(s)

CNS
central nervous system

CPK
creatinine phosphokinase

CREST
antikinetochore

CSF
cancer slope factor

CV
concentration in venous blood

CYP450
cytochrome P450

DEN
diethylnitrosamine

FISH
fluorescence in situ hybridization

G-6-Pase
glucose-6-phosphatase

GC
gas chromatography

GGT
y-glutamyl transpeptidase

HEAA
P-hydroxyethoxy acetic acid

HED(s)
human equivalent dose(s)

HPLC
high-performance liquid chromatography

HSDB
Hazardous Substances Data Bank

Hz
Hertz

IARC
International Agency for Research on Cancer

i.p.
intraperitoneal

i.v.
intravenous

IRIS
Integrated Risk Information System

JBRC
Japan Bioassay Research Center

ke
1st order elimination rate of 1,4-dioxane

klNH
1st order 1,4-dioxane inhalation rate constant

kLc
1st order, non-saturable metabolism rate constant for 1,4-dioxane in the liver
March 2010
xiv
DRAFT - DO NOT CITE OR QUOTE

-------
Km
Michaelis constant for metabolism of 1,4-dioxane in the liver
kme
1st order elimination rate of HEAA (1,4-dioxane metabolite)
LAP
leucine aminopeptidase
LD50
median lethal dose
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
OECD
Organization for Economic Co-operation and Development
PB
blood:air partition coefficient
PBPK
physiologically based pharmacokinetic
PC
partition coefficient
PCB
polychlorinated biphenyl
PCE
polychromatic erthyrocyte
PFA
fat:air partition coefficient
PLA
liver:air 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
vd
volume of distribution
March 2010	xv	DRAFT - DO NOT CITE OR QUOTE

-------
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
%	Chi-squared
March 2010
xvi
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).
March 2010
xvii
DRAFT - DO NOT CITE OR QUOTE

-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGERS
EvaD. McLanahan, Ph.D. (current)
Lieutenant, U.S. Public Health Service
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Reeder Sams II, Ph.D. (former)
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
AUTHORS AND CONTRIBUTORS
J. Allen Davis, MSPH
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Hisham El-Masri, Ph.D.
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
JeffS. Gift, Ph.D.
National Center for Environmental Assessment
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, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
March 2010
xviii
DRAFT - DO NOT CITE OR QUOTE

-------
Allan Marcus, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Eva D. McLanahan, Ph.D.
Lieutenant, U.S. Public Health Service
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Marc Odin, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Susan Rieth
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
Andrew Rooney, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Reeder Sams II, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Paul Schlosser, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Julie Stickney, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
John Vandenberg, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
March 2010
xix
DRAFT - DO NOT CITE OR QUOTE

-------
REVIEWERS
This document has been provided for review to EPA scientists, interagency reviewers
from other federal agencies and White House offices, and the public, and peer reviewed by and
independent scientists external to EPA. 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.
INTERNAL EPA REVIEWERS
Anthony DeAngelo, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Nagu Keshava, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
Jason Lambert, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
Connie Meacham, M.S.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Debra Walsh, M.S.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Douglas Wolf, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
March 2010
xx
DRAFT - DO NOT CITE OR QUOTE

-------
EXTERNAL PEER REVIEWERS
George V. Alexeeff, Ph.D., DABT
Office of Environmental Health Hazard Assessment (OEHHA)
California EPA
Bruce C. Allen, M.S.
Bruce Allen Consulting
James V. Bruckner, Ph.D.
Department of Pharmaceutical and Biomedical Sciences
College of Pharmacy
The University of Georgia
Harvey J. Clewell III, Ph.D., DABT
Center for Human Health Assessment
The Hamner Institutes for Health Sciences
Lena Ernstgard, Ph.D.
Institute of Environmental Medicine
Karolinska Institutet
Frederick J. Kaskel, M.D., Ph.D.
Children's Hospital at Montefiore
Albert Einstein College of Medicine of Yeshiva University
Kannan Krishnan, Ph.D., DABT
Inter-University Toxicology Research Center (CIRTOX)
Universite de Montreal
RagubirP. Sharma, DVM, Ph.D.
Department of Physiology and Pharmacology
College of Veterinary Medicine {retired)
The University of Georgia
March 2010
xxi
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
1. INTRODUCTION
This document presents background information and justification for the Integrated Risk
Information System (IRIS) Summary of the hazard and dose-response assessment of
1,4-dioxane. IRIS Summaries may include oral reference dose (RfD) and inhalation reference
concentration (RfC) values for chronic and subchronic exposure durations, and a carcinogenicity
assessment.
The RfD and RfC, if derived, provide quantitative information for use in risk assessments
for health effects known or assumed to be produced through a nonlinear (presumed threshold)
mode of action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with
uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
"3
deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m ) is
analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for both the respiratory system (portal-of-entry) and for
effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
acute (< 24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
lifetime) exposure durations, all of which are derived based on an assumption of continuous
exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
derived for chronic exposure durations.
The carcinogenicity assessment provides information on the carcinogenic hazard
potential of the substance in question and quantitative estimates of risk from oral and inhalation
exposure may be derived. The information includes a weight-of-evidence judgment of the
likelihood that the agent is a human carcinogen and the conditions under which the carcinogenic
effects may be expressed. Quantitative risk estimates may be derived from the application of a
low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
the estimate of risk per mg/kg-day of oral exposure. Similarly, an inhalation unit risk is a
"3
plausible upper bound on the estimate of risk per (j,g/m air breathed.
Development of these hazard identification and dose-response assessments for
1,4-dioxane has followed the general guidelines for risk assessment as set forth by the National
Research Council (NRC, 1983). EPA guidelines and Risk Assessment Forum Technical Panel
Reports that may have been used in the development of this assessment include the following:
Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S. EPA, 1986a), Guidelines
for Mutagenicity Risk Assessment (U.S. EPA, 1986b), Recommendations for and Documentation
of Biological Values for Use in Risk Assessment (U.S. EPA, 1988), Guidelines for
Developmental Toxicity Risk Assessment (U.S. EPA, 1991), Interim Policy for Particle Size and
Limit Concentration Issues in Inhalation Toxicity (U.S. EPA, 1994a), Methods for Derivation of
1	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
Inhalation Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA,
1994b), Use of the Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995),
Guidelines for Reproductive Toxicity Risk Assessment (U.S. EPA, 1996), Guidelines for
Neurotoxicity Risk Assessment (U.S. EPA, 1998), Science Policy Council Handbook. Risk
Characterization (U.S. EPA, 2000a), Benchmark Dose Technical Guidance Document (U.S.
EPA, 2000b), Supplementary Guidance for Conducting Health Risk Assessment of Chemical
Mixtures (U.S. EPA, 2000c), A Review of the Reference Dose and Reference Concentration
Processes (U.S. EPA, 2002a), Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a),
Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens
(U.S. EPA, 2005b), Science Policy Council Handbook. Peer Review (U.S. EPA, 2006a), and A
Framework for Assessing Health Risks of Environmental Exposures to Children (U.S. EPA,
2006b).
The literature search strategy employed for this compound was based on the Chemical
Abstracts Service Registry Number (CASRN) and at least one common name. Any pertinent
scientific information submitted by the public to the IRIS Submission Desk was also considered
in the development of this document. The relevant literature was reviewed through September
2009. Note that during the development of this assessment, new data regarding the toxicity of
1,4-dioxane through the inhalation route of exposure became available. These data have not been
included in the current assessment and will be evaluated in a separate IRIS assessment.
2
DRAFT - DO NOT CITE OR QUOTE

-------
2. CHEMICAL AND PHYSICAL INFORMATION
1,4-Dioxane, a volatile organic compound (VOC), is a colorless liquid with a pleasant
odor (Lewis, 2001, 2000). Synonyms include diethylene ether, 1,4-diethylene dioxide,
diethylene oxide, dioxyethylene ether, and dioxane (Lewis, 2001). The chemical structure of
1,4-dioxane is shown in Figure 2-1. Selected chemical and physical properties of this substance
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)
C4H802 (O'Neil, 2001)
101.1°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-m3/molecule at 25°C (Park et al., 1987)
1.09 x 10"11 cmVmolecule sec at 25°C (Atkinson, 1989)
17 (estimated using log Kow) (Lyman et al., 1990)
0.4 (estimated using log Kow) (Meylan et al., 1999)
3	3
1 ppm = 3.6 mg/m ; 1 mg/m = 0.278 ppm
(25°C and 1 atm) (HSDB, 2007)
March 2010
3
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
1,4-Dioxane is produced commercially through the dehydration and ring closure of
diethylene glycol (Surprenant, 2002). Concentrated sulfuric acid is used as a catalyst
(Surprenant, 2002). This is a continuous distillation process with operating temperatures and
pressures of 130-200°C and 188-825 mmHg, respectively (Surprenant, 2002). During the years
1986 and 1990, the U.S. production of 1,4-dioxane reported by manufacturers was within the
range of 10-50 million pounds (U.S. EPA, 2002b). The production volume reported during the
years 1994, 1998, and 2002 was within the range of 1-10 million pounds (U.S. EPA, 2002b).
Historically, 1,4-dioxane has been used as a stabilizer for the solvent 1,1,1-trichloro-
ethane (Suprenant, 2002). However, this use is no longer expected to be important due to the
1990 Amendments to the Clean Air Act and the Montreal Protocol, which mandate the eventual
phase-out of 1,1,1-trichloroethane production in the U.S. (ATSDR, 2007; 2006; UNEP, 2000;
U.S. EPA, 1990). 1,4-Dioxane is a contaminant of some ingredients used in the manufacture of
personal care products and cosmetics. 1,4-Dioxane is also used as a solvent for cellulosics,
organic products, lacquers, paints, varnishes, paint and varnish removers, resins, oils, waxes,
dyes, cements, fumigants, emulsions, and polishing compositions (Lewis, 2001; O'Neil, 2001;
IARC, 1999). 1,4-Dioxane has been used as a solvent in the formulation of inks, coatings, and
adhesives and in the extraction of animal and vegetable oil (Suprenant, 2002). Reaction products
of 1,4-dioxane are used in the manufacture of insecticides, herbicides, plasticizers, and
monomers (Suprenant, 2002).
When 1,4-dioxane enters the air, it will exist as a vapor, as indicated by its vapor pressure
(HSDB, 2007). It is expected to be degraded in the atmosphere through photooxidation with
hydroxyl radicals (HSDB, 2007; Suprenant, 2002). The estimated half-life for this reaction is
6.7 hours (HSDB, 2007). It may also be broken down by reaction with nitrate radicals, although
this removal process is not expected to compete with hydroxyl radical photooxidation (Grosjean,
1990). 1,4-Dioxane is not expected to undergo direct photolysis (Wolfe and Jeffers, 2000).
1,4-Dioxane is primarily photooxidized to 2-oxodioxane and through reactions with nitrogen
oxides (NOx) results in the formation of ethylene glycol diformate (Platz et al., 1997).
1,4-Dioxane is expected to be highly mobile in soil based on its estimated Koc and is expected to
leach to lower soil horizons and groundwater (ATSDR, 2007; Lyman et al., 1990). This
substance may volatilize from dry soil surfaces based on its vapor pressure (HSDB, 2007). The
estimated bioconcentration factor value indicates that 1,4-dioxane will not bioconcentrate in
aquatic or marine organisms (Meylan et al., 1999; Franke et al., 1994). 1,4-Dioxane is not
expected to undergo hydrolysis or to biodegrade readily in the environment (HSDB, 2007;
ATSDR, 2007). Therefore, volatilization is expected to be the dominant removal process for
moist soil and surface water. Based on a Henry's Law constant of 4.8x 10"6 atm-m3/mole, the
half-life for volatilization of 1,4-dioxane from a model river is 5 days and that from a model lake
is 56 days (HSDB, 2007; Lyman et al., 1990; Park et al., 1987). 1,4-Dioxane may be more
persistent in groundwater where volatilization is hindered.
March 2010
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).
March 2010
5
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
3. TOXICOKINETICS
Data for the toxicokinetics of 1,4-dioxane in humans are very limited. However,
absorption, distribution, metabolism, and elimination of 1,4-dioxane are well described in rats
exposed via the oral, inhalation, or intravenous (i.v.) routes. 1,4-Dioxane is extensively absorbed
and metabolized in humans and rats to P-hydroxyethoxy acetic acid (HEAA), which is
predominantly excreted in the urine. Saturation of 1,4-dioxane metabolism has been observed in
rats and would be expected in humans; however, human exposure levels associated with
nonlinear toxicokinetics are not known.
Important data elements that have contributed to our current understanding of the
toxicokinetics of 1,4-dioxane are summarized in the following sections.
3.1. ABSORPTION
Absorption of 1,4-dioxane following inhalation exposure has been qualitatively
demonstrated in workers and volunteers. Workers exposed to a time-weighted average (TWA)
of 1.6 parts per million (ppm) of 1,4-dioxane in air for 7.5 hours showed a HEAA/1,4-dioxane
ratio of 118:1 in urine (Young et al., 1976). The authors assumed lung absorption to be 100%
and calculated an average absorbed dose of 0.37 mg/kg, although no exhaled breath
measurements were taken. In a study with four healthy male volunteers, Young et al. (1977)
reported 6-hour inhalation exposures of adult volunteers to 50 ppm of 1,4-dioxane in a chamber,
followed by blood and urine analysis for 1,4-dioxane and HEAA. The study protocol was
approved by a seven-member Human Research Review Committee of the Dow Chemical
Company, and written informed consent of study participants was obtained. At a concentration
of 50 ppm, uptake of 1,4-dioxane into plasma was rapid and approached steady-state conditions
by 6 hours. The authors reported a calculated absorbed dose of 5.4 mg/kg. However, the
exposure chamber atmosphere was kept at a constant concentration of 50 ppm and exhaled
breath was not analyzed. Accordingly, gas uptake could not be measured. As a result, the
absorbed fraction of inhaled 1,4-dioxane could not be accurately determined in humans. Rats
inhaling 50 ppm for 6 hours exhibited 1,4-dioxane and HEAA in urine with an HEAA to
1,4-dioxane ratio of over 3,100:1 (Young et al., 1978a, b). Plasma concentrations at the end of
the 6-hour exposure period averaged 7.3 [j,g/mL. The authors calculated an absorbed 1,4-dioxane
dose of 71.9 mg/kg; however, the lack of exhaled breath data and dynamic exposure chamber
precluded the accurate determination of the absorbed fraction of inhaled 1,4-dioxane.
No human data are available to evaluate the oral absorption of 1,4-dioxane.
Gastrointestinal absorption was nearly complete in male Sprague Dawley rats orally dosed with
10-1,000 mg/kg of [14C]-l,4-dioxane given as a single dose or as 17 consecutive daily doses
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
(Young et al., 1978a, b). Cumulative recovery of radiolabel in the feces was <1-2% of
administered dose regardless of dose level or frequency.
No human data are available to evaluate the dermal absorption of 1,4-dioxane; however,
Bronaugh (1982) reported an in vitro study in which 1,4-dioxane penetrated excised human skin
10 times more under occluded conditions (3.2% of applied dose) than unoccluded conditions
(0.3%) of applied dose). [14C]-l,4-dioxane was dissolved in lotion, applied to the excised skin in
occluded and unoccluded diffusion cells, and absorption of the dose was recorded 205 minutes
after application. Bronaugh (1982) also reported observing rapid evaporation, which further
decreased the small amount available for skin absorption.
Dermal absorption data in animals are also limited. Dermal absorption in animals was
reported to be low following exposure of forearm skin of monkeys (Marzulli, 1981). In this
study, Rhesus monkeys were exposed to [14C]-l,4-dioxane in methanol or skin lotion vehicle for
24 hours (skin was uncovered/unoccluded). Only 2—3% of the original radiolabel was
cumulatively recovered in urine over a 5-day period.
3.2. DISTRIBUTION
No data are available for the distribution of 1,4-dioxane in human tissues. No data are
available for the distribution of 1,4-dioxane in animals following oral or inhalation exposures.
Mikheev et al. (1990) studied the distribution of [14C]-l,4-dioxane in the blood, liver,
kidney, brain, and testes of rats (strain not reported) for up to 6 hours following intraperitoneal
(i.p.) injection of approximately one-tenth the median lethanl dose (LD50) (actual dose not
reported). While actual tissue concentrations were not reported, tissue:blood ratios were given
for each tissue at six time points ranging from 5 minutes to 6 hours. The time to reach maximum
accumulation of radiolabel was shorter for liver and kidney than for blood or the other tissues,
which the authors suggested was indicative of selective membrane transport. Tissue:blood ratios
were less than one for all tissues except testes, which had a ratio greater than one at the 6-hour
time point. The significance of these findings is questionable since the contribution of residual
blood in the tissues was unknown (though saline perfusion may serve to clear tissues of highly
water-soluble 1,4-dioxane), the tissue concentrations of radiolabel were not reported, and data
were collected from so few time points.
"3
Woo et al. (1977b) administered i.p. doses of [ H]-1,4-dioxane (5 mCi/kg body weight
[BW]) to male Sprague Dawley rats with and without pretreatment using mixed-function oxidase
inducers (phenobarbital, 3-methylcholanthrene, or poly chlorinated biphenyls [PCBs]). Liver,
kidney, spleen, lung, colon, and skeletal muscle tissues were collected from 1, 2, 6, and 12 hours
after dosing. Distribution was generally uniform across tissues, with blood concentrations higher
than tissues at all times except for 1 hour post dosing, when kidney levels were approximately
20% higher than blood. Since tissues were not perfused prior to analysis, the contribution of
residual blood to radiolabel measurements is unknown, though loss of 1,4-dioxane from tissues
7	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
would be unknown had saline perfusion been performed. Covalent binding reached peak
percentages at 6 hours after dosing in liver (18.5%), spleen (22.6%), and colon (19.5%). At
16 hours after dosing, peak covalent binding percentages were observed in whole blood (3.1%),
kidney (9.5%), lung (11.2%), and skeletal muscle (11.2%). Within hepatocytes, radiolabel
distribution at 6 hours after dosing was greatest in the cytosolic fraction (43.8%) followed by the
microsomal (27.9%), mitochondrial (16.6%), and nuclear (11.7%) fractions. While little
covalent binding of radiolabel was measured in the hepatic cytosol (4.6%), greater binding was
observed at 16 hours after dosing in the nuclear (64.8%), mitochondrial (45.7%), and
microsomal (33.4%) fractions. Pretreatment with inducers of mixed-function oxidase activity
did not significantly change the extent of covalent binding in subcellular fractions.
3.3. METABOLISM
The major product of 1,4-dioxane metabolism appears to be HEAA, although there is
one report that identified l,4-dioxane-2-one as a major metabolite (Woo et al., 1977b).
However, the presence of this compound in the sample was believed to result from the acidic
conditions (pH of 4.0-4.5) of the analytical procedures. The reversible conversion of HEAA and
p-l,4-dioxane-2-one is pH-dependent (Braun and Young, 1977). Braun and Young (1977)
identified HEAA (85%) as the major metabolite, with most of the remaining dose excreted as
unchanged 1,4-dioxane in the urine of Sprague Dawley rats dosed with 1,000 mg/kg of
uniformly labeled l,4-[14C]dioxane. In fact, toxicokinetic studies of 1,4-dioxane in humans and
rats (Young et al., 1978a, b, 1977) employed an analytical technique that converted HEAA to the
more volatile dioxanone prior to gas chromatography (GC).
A proposed metabolic scheme for 1,4-dioxane metabolism (Woo et al., 1977b) in
Sprague Dawley rats is shown in Figure 3-1. Oxidation of 1,4-dioxane to diethylene glycol
(pathway a), l,4-dioxane-2-ol (pathway c), or directly to l,4-dioxane-2-one (pathway b) could
result in the production of HEAA. 1,4-Dioxane oxidation appears to be cytochrome P450
(CYP450)-mediated, as CYP450 induction with phenobarbital or Aroclor 1254 (a commercial
PCB mixture) and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine or cobaltous
chloride were effective in significantly increasing and decreasing, respectively, the appearance of
HEAA in the urine of male Sprague Dawley rats following 3 g/kg i.p. dose (Woo et al., 1978,
1977c). 1,4-Dioxane itself induced CYP450-mediated metabolism of several barbiturates in
Hindustan mice given i.p. injections of 25 and 50 mg/kg 1,4-dioxane (Mungikar and Pawar,
1978). Of the three possible pathways proposed in this scheme, oxidation to diethylene glycol
and HEAA appears to be the most likely, because diethylene glycol was found as a minor
metabolite in Sprague Dawley rat urine following a single 1,000 mg/kg gavage dose of
1,4-dioxane (Braun and Young, 1977). Additionally, i.p. injection of 100-400 mg/kg diethylene
glycol in Sprague Dawley rats resulted in urinary elimination of HEAA (Woo et al., 1977a).
8
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
,0. OH
(c)
r O
[V]
,OH O
r''-
O
[VI]
(a ) HOH2C wi i2'
o
[I]
CH.OH
V
[II]
HOH2C COOH

(b)
+h2o
-h2o
-°V°
o
[IV]
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 (CO2)
identified in expired air after labeled 1,4-dioxane exposure.
Metabolism of 1,4-dioxane in humans is extensive. In a survey of 1,4-dioxane plant
workers exposed to a TWA of 1.6 ppm of 1,4-dioxane for 7.5 hours, Young et al. (1976) found
HEAA and 1,4-dioxane in the worker's urine at a ratio of 118:1. Similarly, in adult male
volunteers exposed to 50 ppm for 6 hours (Young et al., 1977), over 99% of inhaled 1,4-dioxane
(assuming negligible exhaled excretion) appeared in the urine as HEAA. The linear elimination
of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism was a
nonsaturated, first-order process at this exposure level.
Like humans, rats extensively metabolize inhaled 1,4-dioxane, as HEAA content in urine
was over 3,000-fold higher than that of 1,4-dioxane following exposure to 50 ppm for 6 hours
(Young et al., 1978a, b). 1,4-Dioxane metabolism in rats was a saturable process, as exhibited
by oral and i.v. exposures to various doses of [14C]-l,4-dioxane (Young et al., 1978a, b). Plasma
9
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
data from Sprague Dawley rats given single i.v. doses of 3, 10, 30, 100, 300, or 1,000 mg
[14C]-l,4-dioxane/kg demonstrated a dose-related shift from linear, first-order to nonlinear,
saturable metabolism of 1,4-dioxane between plasma 1,4-dioxane levels of 30 and 100 |ig/mL
(Figure 3-2). Similarly, in rats given, via gavage in distilled water, 10, 100, or 1,000 mg
[14C]-l,4-dioxane/kg singly or 10 or 1,000 mg [14C]-l,4-dioxane/kg in 17 daily doses, the
percent urinary excretion of the radiolabel decreased significantly with dose while radiolabel in
expired air increased. Specifically, with single [14C]-l,4-dioxane/kg doses, urinary radiolabel
decreased from 99 to 76% and expired 1,4-dioxane increased from <1 to 25% as dose increased
from 10 to 1,000 mg/kg. Likewise, with multiple daily doses 10 or 1,000 mg
[14C]-l,4-dioxane/kg, urinary radiolabel decreased from 99 to 82% and expired 1,4-dioxane
increased from 1 to 9% as dose increased. The differences between single and multiple doses in
urinary and expired radiolabel support the notion that 1,4-dioxane may induce its own
metabolism.
ooo
I CO
SO *£ CIO
ft
Hi
Source: Young et al. (1978a).
Figure 3-2. Plasma 1,4-dioxane levels in rats following i.v. doses of 3-5,600 mg/kg.
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
1,4-Dioxane has been shown to induce several isoforms of CYP450 in various tissues
following acute oral administration by gavage or drinking water (Nannelli et al., 2005). Male
Sprague Dawley rats were exposed to either 2,000 mg/kg 1,4-dioxane via gavage for
2 consecutive days or by ingestion of a 1.5% 1,4-dioxane drinking water solution for 10 days.
Both exposures resulted in significantly increased CYP2B1/2, CYP2C11, and CYP2E1 activities
in hepatic microsomes. The gavage exposure alone resulted in increased CYP3A activity. The
increase in 2C11 activity was unexpected, as that isoform has been observed to be under
hormonal control and was typically suppressed in the presence of 2B1/2 and 2E1 induction. In
the male rat, hepatic 2C11 induction is associated with masculine pulsatile plasma profiles of
growth hormone (compared to the constant plasma levels in the female), resulting in
masculinization of hepatocyte function (Waxman et al., 1991). The authors postulated that
1,4-dioxane may alter plasma growth hormone levels, resulting in the observed 2C11 induction.
However, growth hormone induction of 2C11 is primarily dependent on the duration between
growth hormone pulses and secondarily on growth hormone plasma levels (Agrawal and
Shapiro, 2000; Waxman et al., 1991). Thus, the induction of 2C11 by 1,4-dioxane may be
mediated by changes in the time interval between growth hormone pulses rather than changes
in growth hormone levels. This may be accomplished by 1,4-dioxane temporarily influencing
the presence of growth hormone cell surface binding sites (Agrawal and Shapiro, 2000).
However, no studies are available to confirm the influence of 1,4-dioxane on either growth
hormone levels or changes in growth hormone pulse interval.
In nasal and renal mucosal cell microsomes, CYP2E1 activity, but not CYP2B1/2
activity, was increased. Pulmonary mucosal CYP450 activity levels were not significantly
altered. Observed increases in 2E1 mRNA in rats exposed by gavage and i.p. injection suggest
that 2E1 induction in kidney and nasal mucosa is controlled by a transcriptional activation of
2E1 genes. The lack of increased mRNA in hepatocytes suggests that induction is regulated via
a post-transcriptional mechanism. Differences in 2E1 induction mechanisms in liver, kidney,
and nasal mucosa suggest that induction is controlled in a tissue-specific manner.
3.4. ELIMINATION
In workers exposed to a TWA of 1.6 ppm for 7.5 hours, 99% of 1,4-dioxane eliminated in
urine was in the form of HEAA (Young et al., 1976). The elimination half-life was 59 minutes
in adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours, with 90% of urinary
1,4-dioxane and 47% of urinary HEAA excreted within 6 hours of onset of exposure (Young
et al., 1977). There are no data for 1,4-dioxane elimination in humans from oral exposures.
Elimination of 1,4-dioxane in rats (Young et al., 1978a, b) was primarily via urine. Like
humans, the elimination half-life in rats exposed to 50 ppm 1,4-dioxane for 6 hours was
calculated to be 1.01 hours. In Sprague Dawley rats given single daily doses of 10, 100, or
1,000 mg [14C]-l,4-dioxane/kg or multiple doses of 10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary
11	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
radiolabel ranged from 99% down to 76% of total radiolabel. Fecal elimination was less than
2% for all doses. The effect of saturable metabolism on expired 1,4-dioxane was apparent, as
expired 1,4-dioxane in singly dosed rats increased with dose from 0.4 to 25% while expired
14C02 changed little (between 2 and 3%) across doses. The same relationship was seen in
Sprague Dawley rats dosed i.v. with 10 or 1,000 mg [14C]-l,4-dioxane/kg. Higher levels of
14C02 relative to 1,4-dioxane were measured in expired air of the 10 mg/kg group, while higher
levels of expired 1,4-dioxane relative to 14C02 were measured in the 1,000 mg/kg group.
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS
PBPK models have been developed for 1,4-dioxane in rats and humans (Leung and
Paustenbach, 1990; Reitz et al., 1990) and lactating women (Fisher et al., 1997). Each of the
models simulates the body as a series of compartments representing tissues or tissue groups that
receive blood from the central vascular compartment (Figure 3-3). Modeling was conducted
under the premise that transfers of 1,4-dioxane between blood and tissues occur sufficiently fast
to be effectively blood flow-limited, which is consistent with the available data (Ramsey and
Andersen, 1984). Blood time course and metabolite production data in rats and humans suggest
that absorption and metabolism are accomplished through common mechanisms in both species
(Young et al., 1978a, b, 1977), allowing identical model structures to be used for both species
(and by extension, for mice as well). In all three models, physiologically relevant, species-
specific parameter values for tissue volume, blood flow, and metabolism and elimination are
used. The models and supporting data are reviewed below, from the perspective of assessing
their utility for predicting internal dosimetry and for cross-species extrapolation of exposure-
response relationships for critical neoplastic and non-neoplastic endpoints (also see Appendix B).
12
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IV
infusion
"O
o
o
~~
to
Z3
o
(D
>
inhalation
absorption
Fat
Slowly
perfused
Rapidly
perfused
Liver
-a
o
~
~~
QJ
t:
<
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
Animal and human data sets available for model calibration derive from Young et al.
(1978a, b, 1977), Mikheev et al. (1990), and Woo et al. (1977a, b). Young et al. (1978a, b)
studied the disposition of radiolabeled [14C]-l,4-dioxane in adult male Sprague Dawley rats
following i.v., inhalation, and single and multiple oral gavage exposures. Plasma concentration-
time profiles were reported for i.v. doses of 3, 10, 30, 100, and 1,000 mg/kg. In addition,
exhaled 14CC>2 and urinary 1,4-dioxane and HEAA profiles were reported following i.v. doses of
10 and 1,000 mg/kg. The plasma 1,4-dioxane concentration-time course, cumulative urinary
1,4-dioxane and cumulative urinary HEAA concentrations were reported following a 6-hour
inhalation exposure to 50 ppm. Following oral gavage doses of 10-1,000 mg/kg, percentages of
total orally administered radiolabel were measured in urine, feces, expired air, and the whole
body.
Oral absorption of 1,4-dioxane was extensive, as only approximately 1% of the
administered dose appeared in the feces within 72 hours of dosing (Young et al., 1978a, b).
Although it may be concluded that the rate of oral absorption was high enough to ensure nearly
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
complete absorption by 72 hours, a more quantitative estimate of the rate of oral absorption is
not possible due to the absence of plasma time course data by oral exposure.
Saturable metabolism of 1,4-dioxane was observed in rats exposed by either the i.v. or
oral routes (Young et al., 1978a, b). Elimination of 1,4-dioxane from plasma appeared to be
linear following i.v. doses of 3-30 mg/kg, but was nonlinear following doses of 100-
1,000 mg/kg. Accordingly, 10 mg/kg i.v. doses resulted in higher concentrations of 14CC>2 (from
metabolized 1,4-dioxane) in expired air relative to unchanged 1,4-dioxane, while 1,000 mg/kg
i.v. doses resulted in higher concentrations of expired 1,4-dioxane relative to 14CC>2. Thus, at
higher i.v. doses, a higher proportion of unmetabolized 1,4-dioxane is available for exhalation.
Taken together, the i.v. plasma and expired air data from Young et al. (1978a, b) corroborate
previous studies describing the saturable nature of 1,4-dioxane metabolism in rats (Woo et al.
1977a, b) and are useful for optimizing metabolic parameters (Vmax and Km) in a PBPK model.
Similarly, increasing single or multiple oral doses of 10-1,000 mg/kg resulted in
increasing percentage of 1,4-dioxane in exhaled air and decreasing percentage of radiolabel
(either as 1,4-dioxane or a metabolite) in the urine, with significant differences in both metrics
being observed between doses of 10 and 100 mg/kg (Young et al., 1978a, b). These data identify
the region (10-100 mg/kg) in which oral exposures will result in nonlinear metabolism of
1,4-dioxane and can be used to test whether metabolic parameter value estimates derived from
i.v. dosing data are adequate for modeling oral exposures.
Post-exposure plasma data from a single 6-hour, 50 ppm inhalation exposure in rats were
reported (Young et al., 1978a, b). The observed linear elimination of 1,4-dioxane after
inhalation exposure suggests that, via this route, metabolism is in the linear region at this
exposure level.
The only human data adequate for use in PBPK model development (Young et al., 1977)
come from adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours. Plasma
1,4-dioxane and HEAA concentrations were measured both during and after the exposure period,
and urine concentrations were measured following exposure. Plasma levels of 1,4-dioxane
approached steady-state at 6 hours. HEAA data were insufficient to describe the appearance or
elimination of HEAA in plasma. Data on elimination of 1,4-dioxane and HEAA in the urine up
to 24 hours from the beginning of exposure were reported. At 6 hours from onset of exposure,
approximately 90% and 47% of the cumulative (0-24 hours) urinary 1,4-dioxane and HEAA,
respectively, were measured in the urine. The ratio of HEAA to 1,4-dioxane in urine 24 hours
after onset of exposure was 192:1 (similar to the ratio of 118:1 observed by Young et al. [1976]
in workers exposed to 1.6 ppm for 7.5 hours), indicating extensive metabolism of 1,4-dioxane
As with Sprague Dawley rats, the elimination of 1,4-dioxane from plasma was linear across all
observations (6 hours following end of exposure), suggesting that human metabolism of
1,4-dioxane is linear for a 50 ppm inhalation exposure to steady-state. Thus, estimation of
14
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
human Vmax and Km from these data will introduce uncertainty into internal dosimetry performed
in the nonlinear region of metabolism.
Further data were reported for the tissue distribution of 1,4-dioxane in rats. Mikheev
et al. (1990) administered i.p. doses of [14C]-l,4-dioxane to rats (strain not reported) and reported
time-to-peak blood, liver, kidney, and testes concentrations. They also reported ratios of tissue
to blood concentrations at various time points after dosing. Woo et al. (1977a, b) administered
i.p. doses of [14C]-l,4-dioxane to Sprague Dawley rats and measured radioactivity levels in
urine. However, since i.p. dosing is not relevant to human exposures, these data are of limited
use for PBPK model development.
3.5.2. Published PBPK Models for 1,4-Dioxane
3.5.2.1. Leung andPaustenbac/i (1990
Leung and Paustenbach (1990) developed a PBPK model for 1,4-dioxane and its primary
metabolite, HEAA, in rats and humans. The model, based on the structure of a PBPK model for
styrene (Ramsey and Andersen, 1984), consists of a central blood compartment and four tissue
compartments: liver, fat, slowly perfused tissues (mainly muscle and skin), and richly perfused
tissues (brain, kidney, and viscera other than the liver). Tissue volumes were calculated as
percentages of total BW, and blood flow rates to each compartment were calculated as
percentages of cardiac output. Equivalent cardiac output and alveolar ventilation rates were
allometrically scaled to a power (0.74) of BW for each species. The concentration of
1,4-dioxane in alveolar blood was assumed to be in equilibrium with alveolar air at a ratio equal
to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane between
blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
between blood and tissue, governed by tissue:blood equilibrium partition coefficients. The latter
were derived from the quotient of blood:air and tissue:air partition coefficients, which were
measured in vitro (Leung and Paustenbach, 1990) for blood, liver, fat, and skeletal muscle
(slowly perfused tissue). Blood:air partition coefficients were measured for both humans and
rats. Rat tissue:air partition coefficients were used as surrogate values for humans, with the
exception of slowly perfused tissue:blood, which was estimated by optimization to the plasma
time-course data. Portals of entry included i.v. infusion (over a period of 36 seconds) into the
venous blood, inhalation by diffusion from the alveolar air into the lung blood at the rate of
alveolar ventilation, and oral administration via zero-order absorption from the gastrointestinal
tract to the liver. Elimination of 1,4-dioxane was accomplished through pulmonary exhalation
and saturable hepatic metabolism. Urinary excretion of HEAA was assumed to be instantaneous
with the generation of HEAA from the hepatic metabolism of 1,4-dioxane.
The parameter values for hepatic metabolism of 1,4-dioxane, Vmax and Km, were
optimized and validated against plasma and/or urine time course data for 1,4-dioxane and HEAA
15
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
in rats following i.v. and inhalation exposures and humans following inhalation exposure (Young
et al., 1978a, b, 1977); the exact data (i.e., i.v., inhalation, or both) used for the optimization and
calibration were not reported. Although the liver and fat were represented by tissue-specific
compartments, no tissue-specific concentration data were available for model development,
raising uncertainty as the model's ability to adequately predict exposure to these tissues. The
human inhalation exposure of 50 ppm for 6 hours (Young et al., 1977) was reported to be in the
linear range for metabolism; thus, uncertainty exists in the ability of the allometrically-scaled
value for the human metabolic Vmax to accurately describe 1,4-dioxane metabolism from
exposures resulting in metabolic saturation. Nevertheless, these values resulted in the model
producing good fits to the data. For rats, the values for Vmax had to be adjusted upwards by a
factor of 1.8 to reasonably simulate exposures greater than 300 mg/kg. The model authors
attributed this to metabolic enzyme induction by high doses of 1,4-dioxane.
3.5.2.2. Jieitz et al (1990
Reitz et al. (1990) developed a model for 1,4-dioxane and HEAA in the mouse, rat, and
human. This model, also based on the styrene model of Ramsey and Andersen (1984), included
a central blood compartment and compartments for liver, fat, and rapidly and slowly perfused
tissues. Tissue volumes and blood flow rates were defined as percentages of total BW and
cardiac output, respectively. Physiological parameter values were similar to those used by
Andersen et al. (1987), except that flow rates for cardiac output and alveolar ventilation were
doubled in order to produce a better fit of the model to human blood level data (Young et al.,
1977). Portals of entry included i.v. injection into the venous blood, inhalation, oral bolus
dosing, and oral dosing via drinking water. Oral absorption of 1,4-dioxane was simulated, in all
three species, as a first-order transfer to liver (halftime approximately 8 minutes).
Alveolar blood levels of 1,4-dioxane were assumed to be in equilibrium with alveolar air
at a ratio equal to the experimentally measured blood:air partition coefficient. Transfers of
1,4-dioxane between blood and tissues were assumed to be blood flow-limited and to achieve
rapid equilibrium between blood and tissue, governed by tissue:blood equilibrium partition
coefficients. These coefficients were derived by dividing experimentally measured (Leung and
Paustenbach, 1990) in vitro blood:air and tissue:air partition coefficients for blood, liver, fat.
Blood:air partition coefficients were measured for both humans and rats. The mouse blood:air
partition coefficient was different from rat or human values; the source of the partition
coefficient for blood in mice was not reported. Rat tissue:air partition coefficients were used as
surrogate values for humans. Rat tissue partition coefficient values were the same values as used
in the Leung and Paustenbach (1990) model (with the exception of slowly perfused tissues) and
were used in the models for all three species. The liver value was used for the rapidly perfused
tissues, as well as slowly perfused tissues. Although slowly perfused tissue:air partition
coefficients for rats were measured, the authors suggested that 1,4-dioxane in the muscle and air
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
may not have reached equilibrium in the highly gelatinous tissue homogenate (Reitz et al., 1990).
Substitution of the liver value provided much closer agreement to the plasma data than when the
muscle value was used. Further, doubling of the measured human blood:air partition coefficient
improved the fit of the model to the human blood level data compared to the fit resulting from
the measured value (Reitz et al., 1990). The Reitz et al. (1990) model simulated three routes of
1,4-dioxane elimination: pulmonary exhalation, hepatic metabolism to HEAA, and urinary
excretion of HEAA. The elimination of HEAA was modeled as a first-order transfer of
1,4-dioxane metabolite to urine.
Values for the metabolic rate constants, Vmax and Km, were optimized to achieve
agreement with various observations. Reitz et al. (1990) optimized values for human Vmax and
Km against the experimental human 1,4-dioxane inhalation data (Young et al., 1977). As noted
previously, because the human exposures were below the level needed to exhibit nonlinear
kinetics, uncertainty exists in the ability of the optimized value of Vmax to simulate human
1,4-dioxane metabolism above the concentration that would result in saturation of metabolism.
Rat metabolic rate constants were obtained by optimization to simulated data from a
two-compartment empirical pharmacokinetic model, which was fitted to i.v. exposure data
(Young et al., 1978a, b). As with the Leung and Paustenbach (1990) model, the Reitz et al.
(1990) model included compartments for the liver and fat, although no tissue-specific
concentration data were available to validate dosimetry for these organs. The derivations of
human and rat HEAA elimination rate constants were not reported. Since no pharmacokinetics
data for 1,4-dioxane in mice were available, mouse metabolic rate constants were allometrically
scaled from rat and human values.
3.5.2.3. fis/ier ef al (7997)
A PBPK model was developed by Fisher et al. (1997) to simulate a variety of volatile
organic compounds (VOCs, including 1,4-dioxane) in lactating humans. This model was similar
in structure to those of Leung and Paustenbach (1990) and Reitz et al. (1990) with the addition of
elimination of 1,4-dioxane to breast milk. Experimental measurements were made for blood:air
and milk:air partition coefficients. Other partition coefficient values were taken from Reitz et al.
(1990). The model was not optimized, nor was performance tested against experimental
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
As previously described, several pharmacokinetic models have been developed to predict
the absorption, distribution, metabolism, and elimination of 1,4-dioxane in rats and humans.
Single compartment, empirical models for rats (Young et al., 1978a, b) and humans (Young
et al., 1977) were developed to predict blood levels of 1,4-dioxane and urine levels of the
primary metabolite, HEAA. PBPK models that describe the kinetics of 1,4-dioxane using
17
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
biologically realistic flow rates, tissue volumes, enzyme affinities, metabolic processes, and
elimination behaviors were also developed (Sweeney et al, 2008; Fisher et al., 1997; Leung and
Paustenbach, 1990; Reitz et al., 1990).
In developing updated toxicity values for 1,4-dioxane the available PBPK models were
evaluated for their ability to predict observations made in experimental studies of rat and human
exposures to 1,4-dioxane (Appendix B). The Reitz et al. (1990) and Leung and Paustenbach
(1990) PBPK models were both developed from a PBPK model of styrene (Ramsey and
Anderson, 1984), with the exception of minor differences in the use of partition coefficients and
biological parameters. The model code for Leung and Paustenbach (1990) was unavailable in
contrast to Reitz et al. (1990). The model of Reitz et al. (1990) was identified for further
consideration to assist in the derivation of toxicity values, and the Sweeney et al. (2008) PBPK
model was also evaluated.
The biological plausibility of parameter values in the Reitz et al. (1990) human model
were examined. The model published by Reitz et al. (1990) was able to predict the only
available human inhalation data (50 ppm 1,4-dioxane for 6 hours; Young et al., 1977) by
increasing (i.e., approximately doubling) the parameter values for human alveolar ventilation (30
L/hour/kg0'74), cardiac output (30 L/hour/kg0'74), and the blood:air partition coefficient (3,650)
above the measured values of 13 L/minute/kg0'74 (Brown et al., 1997), 14 L/hour/kg0'74 (Brown et
al., 1997), and 1,825 (Leung and Paustenbach, 1990), respectively. Furthermore, Reitz et al.
(1990) replaced the measured value for the slowly perfused tissue:air partition coefficient (i.e.,
muscle—value not reported in manuscript) with the measured liver value (1,557) to improve the
fit. Analysis of the Young et al. (1977) human data suggested that the apparent volume of
distribution (Vd) for 1,4-dioxane was approximately 10-fold higher in rats than humans,
presumably due to species differences in tissue partitioning or other process not represented in
the model. Based upon these observations, several model parameters (e.g.,
metabolism/elimination parameters) were re-calibrated using biologically plausible values for
flow rates and tissue:air partition coefficients.
Appendix B describes all activities that were conducted in the evaluation of the empirical
models and the re-calibration and evaluation of the Reitz et al. (1990) PBPK model to determine
the adequacy and preference for the potential use of the models.
The evaluation consisted of implementation of the Young et al. (1978a, b, 1977)
empirical rat and human models using the acslXtreme simulation software, re-calibration of the
Reitz et al. (1990) human PBPK model, and evaluation of the model parameters published by
Sweeney et al. (2008). Using the model descriptions and equations given in Young et al. (1978a,
b, 1977), model code was developed for the empirical models and executed, simulating the
reported experimental conditions. The model output was then compared with the model output
reported in Young et al. (1978a, b, 1977).
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
The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for
cardiac and alveolar flow rates and tissue:air partition coefficients. The predictions of blood and
urine levels of 1,4-dioxane and HEAA, respectively, from the re-calibrated model were
compared with the empirical model predictions of the same dosimeters to determine whether the
re-calibrated PBPK model could perform similarly to the empirical model. As part of the PBPK
model evaluation, EPA performed a sensitivity analysis to identify the model parameters having
the greatest influence on the primary dosimeter of interest, the blood level of 1,4-dioxane.
Variability data for the experimental measurements of the tissue:air partition coefficients were
incorporated to determine a range of model outputs bounded by biologically plausible values for
these parameters. Model parameters from Sweeney et al. (2008) were also tested to evaluate the
ability of the PBPK model to predict human data following exposure to 1,4-dioxane.
The rat and human empirical models of Young et al. (1978a, b, 1977) were successfully
implemented in acslXtreme and perform identically to the models reported in the published
papers (Figures B-3 through B-6), with the exception of the lower predicted HEAA
concentrations and early appearance of the peak HEAA levels in rat urine. The early appearance
of peak HEAA levels cannot presently be explained, but may result from manipulations of kme or
other parameters by Young et al. (1978a, b) that were not reported. The lower predictions of
HEAA levels are likely due to reliance on a standard urine volume production rate in the absence
of measured (but unreported) urine volumes. While the human urinary HEAA predictions were
lower than observations, this is due to parameter fitting of Young et al. (1977). No model output
was published in Young et al. (1977) for comparison. The empirical models were modified to
allow for user-defined inhalation exposure levels. However, no modifications were made to
model oral exposures as adequate data to parameterize such modifications do not exist for rats or
humans.
Several procedures were applied to the Reitz et al. (1990) human PBPK model to
determine if an adequate fit of the model to the empirical model output or experimental
observations could be attained using biologically plausible values for the model parameters. The
re-calibrated model predictions for blood 1,4-dioxane levels do not come within 10-fold of the
experimental values using measured tissue:air partition coefficients from Leung and Paustenbach
(1990) or Sweeney et al. (2008) (Figures B-8 and B-9). The utilization of a slowly perfused
tissue:air partition coefficient 10-fold lower than measured values produces exposure-phase
predictions that are much closer to observations, but does not replicate the elimination kinetics
(Figure B-10). Recalibration of the model with upper bounds on the tissue:air partition
coefficients results in predictions that are still six- to sevenfold lower than empirical model
prediction or observations (Figures B-12 and B-13). Exploration of the model space using an
assumption of zero-order metabolism (valid for the 50 ppm inhalation exposure) showed that an
adequate fit to the exposure and elimination data can be achieved only when unrealistically low
values are assumed for the slowly perfused tissue:air partition coefficient (Figure B-16).
19
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
Artificially low values for the other tissue:air partition coefficients are not expected to improve
the model fit, as these parameters are shown in the sensitivity analysis to exert less influence on
blood 1,4-dioxane than Vmaxc and Km. In the absence of actual measurements for the human
slowly perfused tissue:air partition coefficient, high uncertainty exists for this model parameter
value. Differences in the ability of rat and human blood to bind 1,4-dioxane may contribute to
the difference in Vd. However, this is expected to be evident in very different values for rat and
human blood:air partition coefficients, which is not the case (Table B-l). Therefore, some other,
as yet unknown, modification to model structure may be necessary.
Similarly, Sweeney et al. (2008) also evaluated the available PBPK models (Leung and
Paustenbach, 1990; Reitz et al., 1990) for 1,4-dioxane. To address uncertainties and deficiencies
in these models, the investigators conducted studies to fill data gaps and reduce uncertainties
pertaining to the pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. The
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.
The studies conducted by Sweeney et al. (2008) resulted in partition coefficients that
were consistent with previously measured values and those used in the Leung and Paustenbach
(1990) model. Of noteworthy significance, the laboratory results of Sweeney et al. (2008) did
not confirm the human blood:air partition coefficient Reitz et al. (1990) reported. Furthermore,
Sweeney et al. (2008) estimated metabolic rate constants (VmaxC and Km) within the range
used in the previous models (Leung and Paustenback, 1990; Reitz et al., 1990). Overall, the
Sweeney et al. (2008) model utilized more rodent in vivo and in vitro data in model
parameterization and refinement; however, the model was still unable to adequately predict the
human blood data from Young et al. (1977).
Updated PBPK models were developed based on these new data and data from previous
kinetic studies in rats, workers, and human volunteers reported by Young et al. (1978a, b, 1977,
1976). The optimized rate of metabolism for the mouse was significantly higher than the value
previously estimated. The optimized rat kinetic parameters were similar to those in the 1990
models. Of the two available human studies (Young et al., 1977, 1976), model predictions were
consistent with one study, but did not fit the second as well.
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
3.6. RAT NASAL EXPOSURE VIA DRINKING WATER
Sweeney et al. (2008) conducted a rat nasal exposure study to explore the potential for
direct contact of nasal tissues with 1,4-dioxane-containing drinking water under bioassay
conditions. Two groups of male Sprague Dawley rats (5/group) received drinking water in
45-mL drinking water bottles containing a fluorescent dye mixture (Cell Tracker
Red/FluoSpheres). The drinking water for one of these two groups also contained 0.5%
1,4-dioxane, a concentration within the range used in chronic toxicity studies. A third group of
five rats received tap water alone (controls). Water was provided to the rats overnight. The next
morning, the water bottles were weighed to estimate the amounts of water consumed. Rats were
sacrificed and heads were split along the midline for evaluation by fluorescence microscopy.
One additional rat was dosed twice by gavage with 2 mL of drinking water containing
fluorescent dye (the second dose was 30 minutes after the first dose; total of 4 mL administered)
and sacrificed 5 hours later to evaluate the potential for systemic delivery of fluorescent dye to
the nasal tissues.
The presence of the fluorescent dye mixture had no measurable impact on water
consumption; however, 0.5% 1,4-dioxane reduced water consumption by an average of 62% of
controls following a single, overnight exposure. Fluorescent dye was detected in the oral cavity
and nasal airways of each animal exposed to the Cell Tracker Red/FluoSpheres mixture in their
drinking water, including numerous areas of the anterior third of the nose along the nasal
vestibule, maxillary turbinates, and dorsal nasoturbinates. Fluorescent dye was occasionally
detected in the ethmoid turbinate region and nasopharynx. 1,4-Dioxane had no effect on the
detection of the dye. Little or no fluorescence at the wavelength associated with the dye mixture
was detected in control animals or in the single animal that received the dye mixture by oral
gavage. The investigators concluded that the findings indicate rat nasal tissues are exposed by
direct contact with drinking water under bioassay conditions.
21
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
4. HAZARD IDENTIFICATION
4.1. STUDIES IN HUMANS - EPIDEMIOLOGY, CASE REPORTS, CLINICAL
CONTROLS
Case reports of acute occupational poisoning with 1,4-dioxane indicated that exposure to
high concentrations resulted in liver, kidney, and central nervous system (CNS) toxicity
(Johnstone, 1959; Barber, 1934). Barber (1934) described four fatal cases of hemorrhagic
nephritis and centrilobular necrosis of the liver attributed to acute inhalation exposure to high
(unspecified) concentrations of 1,4-dioxane. Death occurred within 5-8 days of the onset of
illness. Autopsy findings suggested that the kidney toxicity may have been responsible for
lethality, while the liver effects may have been compatible with recovery. Jaundice was not
observed in subjects and fatty change was not apparent in the liver. Johnstone (1959) presented
the fatal case of one worker exposed to high concentrations of 1,4-dioxane through both
inhalation and dermal exposure for a 1 week exposure duration. Measured air concentrations in
the work environment of this subject were 208-650 ppm, with a mean value of 470 ppm.
Clinical signs that were observed following hospital admission included severe epigastric pain,
renal failure, headache, elevation in blood pressure, agitation and restlessness, and coma.
Autopsy findings revealed significant changes in the liver, kidney, and brain. These included
centrilobular necrosis of the liver and hemorrhagic necrosis of the kidney cortex. Perivascular
widening was observed in the brain with small foci of demyelination in several regions (e.g.,
cortex, basal nuclei). It was suggested that these neurological changes may have been secondary
to anoxia and cerebral edema.
Several studies examined the effects of acute inhalation exposure in volunteers. In a
study performed at the Pittsburgh Experimental Station of the U.S. Bureau of Mines, eye
irritation and a burning sensation in the nose and throat were reported in five men exposed to
5,500 ppm of 1,4-dioxane vapor for 1 minute (Yant et al., 1930). Slight vertigo was also
reported by three of these men. Exposure to 1,600 ppm of 1,4-dioxane vapor for 10 minutes
resulted in similar symptoms with a reduced intensity of effect. In a study conducted by the
Government Experimental Establishment at Proton, England (Fairley et al., 1934), four men
were exposed to 1,000 ppm of 1,4-dioxane for 5 minutes. Odor was detected immediately and
one volunteer noted a constriction in the throat. Exposure of six volunteers to 2,000 ppm for 3
minutes resulted in no symptoms of discomfort. Wirth and Klimmer (1936), of the Institute of
Pharmacology, University of Wurzburg, reported slight mucous membrane irritation in the nose
and throat of several human subjects exposed to concentrations greater than 280 ppm for several
minutes. Exposure to approximately 1,400 ppm for several minutes caused a prickling sensation
in the nose and a dry and scratchy throat. Silverman et al. (1946) exposed 12 male and 12
female subjects to varying air concentrations of 1,4-dioxane for 15 minutes. A 200 ppm
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
concentration was reported to be tolerable, while a concentration of 300 ppm caused irritation to
the eyes, nose, and throat. The study conducted by Silverman et al. (1946) was conducted by the
Department of Industrial Hygiene, Harvard School of Public Health, and was sponsored and
supported by a grant from the Shell Development Company. These volunteer studies published
in the 1930s and 1940s (Silverman et al., 1946; Wirth and Klimmer, 1936; Fairley et al., 1934;
Yant et al., 1930) do not provide information on the human subjects research ethics procedures
undertaken in these study; however, there is no evidence that the conduct of the research was
fundamentally unethical or significantly deficient relative to the ethical standards prevailing at
the time the research was conducted.
Young et al. (1977) exposed four healthy adult male volunteers to a 50-ppm
concentration of 1,4-dioxane for 6 hours. The investigators reported that the protocol of this
study was approved by a seven-member Human Research Review Committee of the Dow
Chemical Company and was followed rigorously. Perception of the odor of 1,4-dioxane
appeared to diminish over time, with two of the four subjects reporting inability to detect the
odor at the end of the exposure period. Eye irritation was the only clinical sign reported in this
study. The pharmacokinetics and metabolism of 1,4-dioxane in humans were also evaluated in
this study (see Section 3.3). Clinical findings were not reported in four workers exposed in the
workplace to a TWA concentration of 1.6 ppm for 7.5 hours (Young et al., 1976).
Ernstgard et al. (2006) examined the acute effects of 1,4-dioxane vapor in male and
female volunteers. The study protocol was approved by the Regional Ethics Review Board in
Stockholm, and performed following informed consent and according to the Helsinki
declaration. In a screening study by these investigators, no self-reported symptoms (based on a
visual analogue scale (VAS) that included ratings for discomfort in eyes, nose, and throat,
breathing difficulty, headache, fatigue, nausea, dizziness, or feeling of intoxication) were
observed at concentrations up to 20 ppm; this concentration was selected as a tentative no-
observed-adverse-effect-level (NOAEL) in the main study. In the main study, six male and six
female healthy volunteers were exposed to 0 or 20 ppm 1,4-dioxane, at rest, for 2 hours. This
exposure did not significantly affect symptom VAS ratings, blink frequency, pulmonary function
or nasal swelling (measured before and at 0 and 3 hours after exposure), or inflammatory
markers in the plasma (C-reactive protein and interleukin-6) of the volunteers. Only ratings for
"solvent smell" were significantly increased during exposure.
Only two well documented epidemiology studies were available for occupational workers
exposed to 1,4-dioxane (Buffler et al., 1978; Thiess et al., 1976). These studies did not provide
evidence of effects in humans; however, the cohort size and number of reported cases were
small.
23
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
4.1.1. Thiess et al. (1976)
A cross-sectional survey was conducted in German workers exposed to 1,4-dioxane. The
study evaluated health effects in 74 workers, including 24 who were still actively employed in
1,4-dioxane production at the time of the investigation, 23 previously exposed workers who were
still employed by the manufacturer, and 27 retired or deceased workers. The actively employed
workers were between 32 and 62 years of age and had been employed in 1,4-dioxane production
for 5-41 years. Former workers (age range not given) had been exposed to 1,4-dioxane for 3-
38 years and retirees (age range not given) had been exposed for 12-41 years. Air
concentrations in the plant at the time of the study were 0.06-0.69 ppm. A simulation of
previous exposure conditions (prior to 1969) resulted in air measurements between 0.06 and
7.2 ppm.
Active and previously employed workers underwent a thorough clinical examination and
X-ray, and hematological and serum biochemistry parameters were evaluated. The examination
did not indicate pathological findings for any of the workers and no indication of malignant
disease was noted. Hematology results were generally normal. Serum transaminase levels were
elevated in 16 of the 47 workers studied; however, this finding was consistent with chronic
consumption of more than 80 g of alcohol per day, as reported for these workers. No liver
enlargement or jaundice was found. Renal function tests and urinalysis were normal in exposed
workers. Medical records of the 27 retired workers (15 living at the time of the study) were
reviewed. No symptoms of liver or kidney disease were reported and no cancer was detected.
Medical reasons for retirement did not appear related to 1,4-dioxane exposure (e.g., emphysema,
arthritis).
Chromosome analysis was performed on six actively employed workers and six control
persons (not characterized). Lymphocyte cultures were prepared and chromosomal aberrations
were evaluated. No differences were noted in the percent of cells with gaps or other
chromosome aberrations. Mortality statistics were calculated for 74 workers of different ages
and varying exposure periods. The proportional contribution of each of the exposed workers to
the total time of observation was calculated as the sum of man-years per 10-year age group.
Each person contributed one man-year per calendar year to the specific age group in which he
was included at the time. The expected number of deaths for this population was calculated from
the age-specific mortality statistics for the German Federal Republic for the years 1970-1973.
From the total of 1,840.5 person-years, 14.5 deaths were expected; however, only 12 deaths were
observed in exposed workers between 1964 and 1974. Two cases of cancer were reported,
including one case of lamellar epithelial carcinoma and one case of myelofibrotic leukemia.
These cancers were not considered to be the cause of death in these cases and other severe
illnesses were present. Standardized mortality ratios (SMRs) for cancer did not significantly
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
differ from the control population (SMR for overall population = 0.83; SMR for 65-75-year-old
men = 1.61; confidence intervals (CIs) not provided).
4.1.2. Buffler et al. (1978)
Buffler et al. (1978) conducted a mortality study on workers exposed to 1,4-dioxane at a
chemical manufacturing facility in Texas. 1,4-Dioxane exposure was known to occur in a
manufacturing area and in a processing unit located 5 miles from the manufacturing plant.
Employees who worked between April 1, 1954, and June 30, 1975, were separated into two
cohorts based on at least 1 month of exposure in either the manufacturing plant (100 workers) or
the processing area (65 workers). Company records and follow-up techniques were used to
compile information on name, date of birth, gender, ethnicity, job assignment and duration, and
employment status at the time of the study. Date and cause of death were obtained from copies
of death certificates and autopsy reports (if available). Exposure levels for each job category
were estimated using the 1974 Threshold Limit Value for 1,4-dioxane (i.e., 50 ppm) and
information from area and personal monitoring. Exposure levels were classified as low
(<25 ppm), intermediate (50-75 ppm), and high (>75 ppm). Monitoring was not conducted prior
to 1968 in the manufacturing areas or prior to 1974 in the processing area; however, the study
authors assumed that exposures would be comparable, considering that little change had been
made to the physical plant or the manufacturing process during that time. Exposure to
1,4-dioxane was estimated to be below 25 ppm for all individuals in both cohorts.
Manufacturing area workers were exposed to several other additional chemicals and processing
area workers were exposed to vinyl chloride.
Seven deaths were identified in the manufacturing cohort and five deaths were noted for
the processing cohort. The average exposure duration was not greater for those workers who
died, as compared to those still living at the time of the study. Cancer was the underlying cause
of death for two cases from the manufacturing area (carcinoma of the stomach, alveolar cell
carcinoma) and one case from the processing area (malignant mediastinal tumor). The workers
from the manufacturing area were exposed for 28 or 38 months and both had a positive smoking
history (>1 pack/day). Smoking history was not available for processing area workers. The
single case of cancer in this area occurred in a 21-year-old worker exposed to 1,4-dioxane for
1 year. The mortality data for both industrial cohorts were compared to age-race-sex specific
death rates for Texas (1960-1969). Person-years of observation contributed by workers were
determined over five age ranges with each worker contributing one person-year for each year of
observation in a specific age group. The expected number of deaths was determined by applying
the Texas 1960-1969 death rate statistics to the number of person years calculated for each
cohort. The observed and expected number of deaths for overall mortality (i.e., all causes) was
comparable for both the manufacturing area (7 observed versus 4.9 expected) and the processing
area (5 observed versus 4.9 expected). No significant excess in cancer-related deaths was
25
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
identified for both areas of the facility combined (3 observed versus 1.7 expected). A separate
analysis was performed to evaluate mortality in manufacturing area workers exposed to
1,4-dioxane for more than 2 years. Six deaths occurred in this group as compared to
4.1 expected deaths. The use of a conditional Poisson distribution indicated no apparent excess
in mortality or death due to malignant neoplasms in this study. It is important to note that the
cohorts evaluated were limited in size. In addition, the mean exposure duration was less than
5 years (<2 years for 43% of workers) and the latency period for evaluation was less than
10 years for 59% of workers. The study authors recommended a follow-up investigation to
allow for a longer latency period; however, no follow-up study of these workers has been
published.
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS - ORAL AND INHALATION
The majority of the subchronic (>30 days) and chronic (>1 year) studies conducted for
1,4-dioxane were oral drinking water studies. Longer-term inhalation studies consisted of only
one subchronic study (Fairley et al., 1934) and one chronic study (Torkelson et al., 1974). These
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. Sufic/iron/c OraiToxicity
Six rats and six mice (unspecified strains) were given drinking water containing 1.25%
1,4-dioxane for up to 67 days (Fairley et al. 1934). Using reference BWs and drinking water
ingestion rates for rats and mice (U.S. EPA, 1988), it can be estimated that these rats and mice
received doses of approximately 1,900 and 3,300 mg/kg-day, respectively. Gross pathology and
histopathology were evaluated in all animals. Five of the six rats in the study died or were
sacrificed in extremis prior to day 34 of the study. Mortality was lower in mice, with five of six
mice surviving up to 60 days. Kidney enlargement was noted in 5/6 rats and 2/5 mice. Renal
cortical degeneration was observed in all rats and 3/6 mice. Large areas of necrosis were
observed in the cortex, while cell degeneration in the medulla was slight or absent. Tubular casts
were observed and vascular congestion and hemorrhage were present throughout the kidney.
Hepatocellular degeneration with vascular congestion was also noted in five rats and three mice.
For this assessment, EPA identified the tested doses of 1,900 mg/kg-day in rats and 3,300 mg/kg-
day in mice as the lowest-observed-adverse-effect-levels (LOAELs) for liver and kidney
degeneration in this study.
4.2.1.1.1. Stoner et al (1986). 1,4-Dioxane was evaluated for its ability to induce lung adenoma
formation in A/J mice. Six- to 8-week-old male and female A/J mice (16/sex/group) were given
1,4-dioxane by gavage or i.p. injection, 3 times/week for 8 weeks. Total cumulative dose levels
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
were given as 24,000 mg/kg (oral), and 4,800, 12,000, or 24,000 mg/kg (i.p.). Average daily
dose estimates were calculated to be 430 mg/kg-day (oral), and 86, 210, or 430 mg/kg-day (i.p.)
by assuming an exposure duration of 56 days. The authors indicated that i.p. doses represent the
maximum tolerated dose (MTD), 0.5 times the MTD, and 0.2 times the MTD. Mice were killed
24 weeks after initiation of the bioassay, and lungs, liver, kidney, spleen, intestines, stomach,
thymus, salivary, and endocrine glands were examined for gross lesions. Histopathology
examination was performed if gross lesions were detected. 1,4-Dioxane did not induce lung
tumors in male or female A/J mice in this study.
4.2.1.1.2.	Stott et a/. (1981). Male Sprague Dawley rats (4-6/group) were given average doses
of 0, 10, or 1,000 mg/kg-day 1,4-dioxane (>99% pure) in their drinking water, 7 days/week for
11 weeks. It should be noted that the methods description in this report stated that the high dose
was 100 mg/kg-day, while the abstract, results, and discussion sections indicated that the high
dose was 1,000 mg/kg-day. Rats were implanted with a [6"3H]thymidine loaded osmotic pump
7 days prior to sacrifice. Animals were sacrificed by cervical dislocation and livers were
"3
removed, weighed, and prepared for histopathology evaluation. [ H]-Thymidine incorporation
was measured by liquid scintillation spectroscopy.
An increase in the liver to BW ratio was observed in rats from the high dose group
(assumed to be 1,000 mg/kg-day). Histopathological alterations, characterized as minimal
centrilobular swelling, were also seen in rats from this dose group (incidence values were not
"3
reported). Hepatic DNA synthesis, measured by [ H]-thymidine incorporation, was increased
1.5-fold in high-dose rats. No changes relative to control were observed for rats exposed to
10 mg/kg-day. EPA found a NOAEL value of 10 mg/kg-day and a LOAEL value of
1,000 mg/kg-day for this study based on histopathological changes in the liver.
Stott et al. (1981) also performed several acute experiments designed to evaluate
potential mechanisms for the carcinogenicity of 1,4-dioxane. These experiments are discussed
separately in Section 4.5.2 (Mechanistic Studies).
4.2.1.1.3.	HCanoet a/. (2008). Groups of 6-week-old F344/DuCij rats (10/sex/group) and
Crj:BDFi mice (10/sex/group) were administered 1,4-dioxane (>99% pure) in the drinking water
for 13 weeks. The animals were observed daily for clinical signs of toxicity. Food consumption
and BWs were measured once per week and water consumption was measured twice weekly.
Food and water were available ad libitum. The concentrations of 1,4-dioxane in the water for
rats and mice were 0, 640, 1,600, 4,000, 10,000, or 25,000 ppm. The investigators used data
from water consumption and BW changes to calculate a daily intake of 1,4-dioxane by the male
and female animals. Thus, male rats received doses of approximately 0, 52, 126, 274, 657, and
1,554 mg 1,4-dioxane/kg-day and female rats received 0, 83, 185, 427, 756, and
1,614 mg/kg-day. Male mice received 0, 86, 231, 585, 882, or 1,570 mg/kg-day and female mice
received 0, 170, 387, 898, 1,620, or 2,669 mg/kg-day.
27	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
No information was provided as to when the blood and urine samples were collected.
Hematology analysis included red blood cell (RBC) count, hemoglobin, hematocrit, mean
corpuscular volume (MCV), platelet count, white blood cell (WBC) count, and differential
WBCs. Serum biochemistry included total protein, albumin, bilirubin, glucose, cholesterol,
triglyceride (rat only), alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate
dehydrogenase (LDH), leucine aminopeptidase (LAP), alkaline phosphatase (ALP), creatinine
phosphokinase (CPK) (rat only), urea nitrogen, creatinine (rat only), sodium, potassium,
chloride, calcium (rat only), and inorganic phosphorous (rat only). Urinalysis parameters were
pH, protein, glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ
weights (brain, lung, liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and
gross necropsy and histopathologic examination of tissues and organs were performed on all
animals (skin, nasal cavity, trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart,
tongue, salivary glands, esophagus, stomach, small and large intestine, liver, pancreas, kidney,
urinary bladder, pituitary thyroid adrenal, testes, epididymis, seminal vesicle, prostate, ovary,
uterus, vagina, mammary gland, brain, spinal cord, sciatic nerve, eye, Harderian gland, muscle,
bone, and parathyroid). Dunnett's test and % test were used to assess the statistical significance
of changes in continuous and discrete variables, respectively.
Clinical signs of toxicity in rats were not discussed in the study report. One female rat in
the high dose group (1,614 mg/kg-day) group died, but cause and time of death were not
specified. Final BWs were reduced at the two highest dose levels in females (12 and 21%) and
males (7 and 21%), respectively. Food consumption was reduced 13% in females at
1,614 mg/kg-day and 8% in 1,554 mg/kg-day males. A dose-related decrease in water
consumption was observed in male rats starting at 52 mg/kg-day (15%) and in females starting at
185 mg/kg-day (12%). Increases in RBCs, hemoglobin, hematocrit, and neutrophils, and a
decrease in lymphocytes were observed in males at 1554 mg/kg-day. In females, MCV was
decreased at doses > 756 mg/kg and platelets were decreased at 1,614 mg/kg-day. With the
exception of the 30% increase in neutrophils in high-dose male rats, hematological changes were
within 2-15% of control values. Total serum protein and albumin were significantly decreased
in males at doses > 274 mg/kg-day and in females at doses > 427 mg/kg-day. Additional
changes in high-dose male and female rats included decreases in glucose, total cholesterol,
triglycerides, and sodium (and calcium in females), and increases in ALT (males only), AST,
ALP, and LAP. Serum biochemistry parameters in treated rats did not differ more than twofold
from control values. Urine pH was decreased in males at > 274 mg/kg-day and in females at
>756 mg/kg-day.
Kidney weights were increased in females at >185 mg/kg-day with a maximum increase
of 15%) and 44% at 1,614 mg/kg-day for absolute and relative kidney weight, respectively. No
organ weight changes were noted in male rats. Histopathology findings in rats that were related
to exposure included nuclear enlargement of the respiratory epithelium, nuclear enlargement of
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
Male dose (mg/kg-day)a
0
52
126
274
657
1,554
Nuclear enlargement; nasal respiratory epithelium
0/10
0/10
9/10b
10/10b
9/10b
10/10b
Nuclear enlargement; nasal olfactory epithelium
0/10
0/10
0/10
10/10b
9/10b
10/10b
Nuclear enlargement; tracheal epithelium
0/10
0/10
0/10
10/10b
10/10b
10/10b
Hepatocyte swelling
0/10
0/10
9/10b
10/10b
10/10b
10/10b
Vacuolic change; liver
0/10
0/10
0/10
0/10
10/10b
10/10b
Granular change; liver
0/10
0/10
0/10
5/10°
2/10
10/10b
Single cell necrosis; liver
0/10
0/10
0/10
5/10°
2/10
10/10b
Nuclear enlargement; renal proximal tubule
0/10
0/10
0/10
1/10
5/10°
9/10b
Hydropic change; renal proximal tubule
0/10
0/10
0/10
0/10
0/10
7/10b
Vacuolic change; brain
0/10
0/10
0/10
0/10
0/10
10/10b

Female dose (mg/kg-day)a
0
83
185
427
756
1,614
Nuclear enlargement; nasal respiratory epithelium
0/10
0/10
5/10°
10/10b
10/10b
8/9b
Nuclear enlargement; nasal olfactory epithelium
0/10
0/10
0/10
9/10b
10/10b
8/9b
Nuclear enlargement; tracheal epithelium
0/10
0/10
0/10
9/10b
10/10b
9/9b
Hepatocyte swelling
0/10
0/10
0/10
0/10
9/10b
9/9b
Vacuolic change; liver
0/10
0/10
0/10
0/10
0/10
9/9b
Granular change; liver
2/10
0/10
1/10
5/10°
5/10°
8/9b
Single cell necrosis; liver
2/10
0/10
1/10
5/10
5/10
8/9b
Nuclear enlargement; proximal tubule
0/10
0/10
0/10
0/10
8/10b
9/9b
Hydropic change; proximal tubule
0/10
0/10
0/10
0/10
0/10
5/9°
Vacuolic change; brain
0/10
0/10
0/10
0/10
0/10
9/9b
"Data are presented for sacrificed animals.
hp < 0.01 by x2 test.
cp < 0.05.
Source: Kano et al. (2008).
29
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
Clinical signs of toxicity in mice were not discussed in the study report One male mouse
in the high-dose group (1,570 mg/kg-day) died, but no information was provided regarding cause
or time of death. Final BWs were decreased 29% in male mice at 1,570 mg/kg-day, but changed
less than 10% relative to controls in the other male dose groups and in female mice. Food
consumption was not significantly reduced in any exposure group. Water consumption was
reduced 14—18% in male mice exposed to 86, 231, or 585 mg/kg-day. Water consumption was
further decreased by 48 and 70% in male mice exposed to 882 and 1,570 mg/kg-day,
respectively. Water consumption was also decreased 31 and 57% in female mice treated with
1,620 and 2,669 mg/kg-day, respectively. An increase in MCV was observed in the two highest
dose groups in both male (882 and 1,570 mg/kg-day) and female mice (1,620 and
2,669 mg/kg-day). Increases in RBCs, hemoglobin, and hematocrit were also observed in high
dose males (1570 mg/kg-day). Hematological changes were within 2—15% of control values.
Serum biochemistry changes in exposed mice included decreased total protein (at 1,570
mg/kg-day in males, >1,620 mg/kg-day in females), decreased glucose (at 1,570 mg/kg-day in
males, >1,620 mg/kg-day in females), decreased albumin (at 1,570 mg/kg-day in males, 2,669
mg/ kg-day in females), decreased total cholesterol (> 585 mg/kg-day in males, >1,620
mg/kg-day in females), increased serum ALT (at 1,570 mg/kg-day in males, > 620 mg/kg-day in
females), increased AST (at 1,570 mg/kg-day in males, 2,669 mg/kg-day in females), increased
ALP (> 585 mg/kg-day in males, 2,669 mg/kg-day in females), and increased LDH (in females
only at doses > 1,620 mg/kg-day). With the exception of a threefold increase in ALT in male
and female mice, serum biochemistry parameters in treated rats did not differ more than twofold
from control values. Urinary pH was decreased in males at > 882 mg/kg-day and in females at
> 1,620 mg/kg-day.
Absolute and relative lung weights were increased in males at 1,570 mg/kg-day and in
females at 1,620 and 2,669 mg/kg-day. Absolute kidney weights were also increased in females
at 1,620 and 2,669 mg/kg-day and relative kidney weight was elevated at 2,669 mg/kg-day.
Histopathology findings in mice that were related to exposure included nuclear enlargement of
the respiratory epithelium, nuclear enlargement of the olfactory epithelium, eosinophilic change
in the olfactory epithelium, vacuolic change in the olfactory nerve, nuclear enlargement of the
tracheal epithelium, accumulation of foamy cells in the lung and bronchi, nuclear enlargement
and degeneration of the bronchial epithelium, hepatocyte swelling of the centrilobular area of the
liver, and single cell necrosis in the liver. The incidence data for histopathological lesions in
mice are presented in Table 4-2. Based on the changes in the bronchial epithelium in female
mice, the authors identified the dose level of 387 mg/kg-day as the LOAEL for mice; the
NOAEL was 170 mg/kg-day (Kano et al., 2008).
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
Male dose (mg/kg-day)a
0
86
231
585
882
1,570
Nuclear enlargement; nasal respiratory epithelium
0/10
0/10
0/10
2/10
5/10b
0/9
Eosinophilic change; nasal respiratory epithelium
0/10
0/10
0/10
0/10
0/10
5/9b
Nuclear enlargement; nasal olfactory epithelium
0/10
0/10
0/10
9/10°
10/10°
9/9°
Eosinophilic change; nasal olfactory epithelium
0/10
0/10
0/10
0/10
0/10
6/9°
Vacuolic change; olfactory nerve
0/10
0/10
0/10
0/10
0/10
9/9°
Nuclear enlargement; tracheal epithelium
0/10
0/10
0/10
7/10°
9/10°
9/9°
Accumulation of foamy cells; lung/bronchi
0/10
0/10
0/10
0/10
0/10
6/9°
Nuclear enlargement; bronchial epithelium
0/10
0/10
0/10
9/10°
9/10°
9/9°
Degeneration; bronchial epithelium
0/10
0/10
0/10
0/10
0/10
8/9°
Hepatocyte swelling
0/10
0/10
0/10
10/10°
10/10°
9/9°
Single cell necrosis; liver
0/10
0/10
0/10
5/10b
10/10°
9/9°

Female dose (mg/kg-day)a
0
170
387
898
1,620
2,669
Nuclear enlargement; nasal respiratory epithelium
0/10
0/10
0/10
3/10
3/10
7/10°
Eosinophilic change; nasal respiratory epithelium
0/10
0/10
1/10
1/10
5/10b
9/10°
Nuclear enlargement; nasal olfactory epithelium
0/10
0/10
0/10
6/10b
10/10°
10/10°
Eosinophilic change; nasal olfactory epithelium
0/10
0/10
0/10
1/10°
6/10b
6/10b
Vacuolic change; olfactory nerve
0/10
0/10
0/10
0/10
2/10
8/10°
Nuclear enlargement; tracheal epithelium
0/10
0/10
2/10
9/10°
10/10°
10/10°
Accumulation of foamy cells; lung/bronchi
0/10
0/10
0/10
0/10
10/10°
10/10°
Nuclear enlargement; bronchial epithelium
0/10
0/10
10/10°
10/10°
10/10°
10/10°
Degeneration; bronchial epithelium
0/10
0/10
0/10
0/10
7/10°
10/10°
Hepatocyte swelling
0/10
1/10
1/10
10/10°
10/10°
9/10b
Single cell necrosis; liver
0/10
0/10
0/10
7/10°
10/10°
9/10°
aData are presented for sacrificed animals.
hp < 0.01 by x2 test.
cp < 0.05.
Source: Kano et al (2008).
1	4.2.1.1.4. Famamoto et a/. (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 F1 offspring of transgenic male C57BLr6J and
5	normal female BALBrcByJ mice. CB6F1 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
31	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
compared to treated nontransgenic mice. The tumor incidence in transgenic animals, however,
was not greater than that observed in vehicle-treated transgenic mouse controls. Further study
details were not provided.
4.2.1.2. Chronic OralToxicity and Carcinogenicity
4.2.1.2.1.	Argus et a/. (1965). Twenty-six adult male Wistar rats weighing between 150 and
200 g were exposed to 1,4-dioxane (purity not reported) in the drinking water at a concentration
of 1% for 64.5 weeks. A group of nine untreated rats served as control. Food and water were
available ad libitum. The drinking water intake for treated animals was reported to be
30 mL/day, resulting in a dose/rat of 300 mg/day. Using a reference BW of 0.462 kg for chronic
exposure to male Wistar rats (U.S. EPA, 1988), it can be estimated that these rats received daily
doses of approximately 640 mg/kg-day. All animals that died or were killed during the study
underwent a complete necropsy. A list of specific tissues examined microscopically was not
provided; however, it is apparent that the liver, kidneys, lungs, lymphatic tissue, and spleen were
examined. No statistical analysis of the results was conducted.
Six of the 26 treated rats developed hepatocellular carcinomas, and these rats had been
treated for an average of 452 days (range, 448-455 days). No liver tumors were observed in
control rats. In two rats that died after 21.5 weeks of treatment, histological changes appeared to
involve the entire liver. Groups of cells were found that had enlarged hyperchromic nuclei. Rats
that died or were killed at longer intervals showed similar changes, in addition to large cells with
reduced cytoplasmic basophilia. Animals killed after 60 weeks of treatment showed small
neoplastic nodules or multifocal hepatocellular carcinomas. No cirrhosis was observed in this
study. Many rats had extensive changes in the kidneys often resembling glomerulonephritis,
however, incidence data was not reported for these findings. This effect progressed from
increased cellularity to thickening of the glomerular capsule followed by obliteration of the
glomeruli. One treated rat had an early transitional cell carcinoma in the kidney's pelvis; this rat
also had a large tumor in the liver. The lungs from many treated and control rats (incidence not
reported) showed severe bronchitis with epithelial hyperplasia and marked peribronchial
infiltration, as well as multiple abscesses. One rat treated with 1,4-dioxane developed leukemia
with infiltration of all organs, particularly the liver and spleen, with large, round, isolated
neoplastic cells. In the liver, the distribution of cells in the sinusoids was suggestive of myeloid
leukemia. The dose of 640 mg/kg-day tested in this study was a free-standing LOAEL,
identified by EPA, for glomerulonephritis in the kidney and histological changes in the liver
(hepatocytes with enlarged hyperchromic nuclei, large cells with reduced cytoplasmic
basophilia).
4.2.1.2.2.	Argus et a/. (1973);Hoch-Ligetiet a/ (1970). Groups of 2-3-month-old male
Sprague Dawley rats (28-32/dose group) weighing 110-230 g at the beginning of the experiment
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
were administered 1,4-dioxane (purity not reported) in the drinking water for up to 13 months at
concentrations of 0, 0.75, 1.0, 1.4, or 1.8%. The drinking water intake was determined for each
group over a 3-day measurement period conducted at the beginning of the study and twice during
the study (weeks were not specified). The rats were killed with ether at 16 months or earlier if
nasal tumors were clearly observable. Complete autopsies were apparently performed on all
animals, but only data from the nasal cavity and liver were presented and discussed. The nasal
cavity was studied histologically only from rats in which gross tumors in these locations were
present; therefore, early tumors may have been missed and pre-neoplastic changes were not
studied. No statistical analysis of the results was conducted. Assuming a BW of 0.523 kg for an
adult male Sprague Dawley rat (U.S. EPA, 1988) and a drinking water intake of 30 mL/day as
reported by the study authors, dose estimates were 0, 430, 574, 803, and 1,032 mg/kg-day. The
progression of liver tumorigenesis was evaluated by an additional group of 10 male rats
administered 1% 1,4-dioxane in the drinking water (574 mg/kg-day), 5 of which were sacrificed
after 8 months of treatment and 5 were killed after 13 months of treatment. Liver tissue from
these rats and control rats was processed for electron microscopy examination.
Nasal cavity tumors were observed upon gross examination in six rats (1/30 in the 0.75%
group, 1/30 in the 1.0% group, 2/30 in the 1.4% group, and 2/30 in the 1.8% group). Gross
observation showed the tumors visible either at the tip of the nose, bulging out of the nasal
cavity, or on the back of the nose covered by intact or later ulcerated skin. As the tumors
obstructed the nasal passages, the rats had difficulty breathing and lost weight rapidly. No
neurological signs or compression of the brain were observed. In all cases, the tumors were
squamous cell carcinomas with marked keratinization and formation of keratin pearls. Bony
structure was extensively destroyed in some animals with tumors, but there was no invasion into
the brain. In addition to the squamous carcinoma, two adenocarcinomatous areas were present.
One control rat had a small, firm, well-circumscribed tumor on the back of the nose, which
proved to be subcutaneous fibroma. The latency period for tumor onset was 329-487 days.
Evaluation of the latent periods and doses received did not suggest an inverse relationship
between these two parameters.
Argus et al. (1973) studied the progression of liver tumorigenesis by electron microscopy
of liver tissues obtained following interim sacrifice at 8 and 13 months of exposure (5 rats/group,
574 mg/kg-day). The first change observed in the liver was an increase in the size of the nucleus
of the hepatocytes, mostly in the periportal area. Precancerous changes were characterized by
disorganization of the rough endoplasmic reticulum, an increase in smooth endoplasmic
reticulum, and a decrease in glycogen and increase in lipid droplets in hepatocytes. These
changes increased in severity in the hepatocellular carcinomas in rats exposed to 1,4-dioxane for
13 months.
Three types of liver nodules were observed in exposed rats at 13-16 months. The first
consisted of groups of cells with reduced cytoplasmic basophilia and a slightly nodular
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
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
Incipient tumors
Hepatomas
Total
430
4
0
4
574
9
0
9
803
13
3
16
1,032
11
12
23
aPrecise 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 et al. (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. Hocfi-LigetiandArgus (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
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
between 944 and 1,019 mg 1,4-dioxane/kg-day. A group of ten untreated guinea pigs served as
controls. All animals were sacrificed within 28 months, but the scope of the postmortem
examination was not provided.
Nine treated guinea pigs showed peri- or intrabronchial epithelial hyperplasia and nodular
mononuclear infiltration in the lungs. Also, two guinea pigs had carcinoma of the gallbladder,
three had early hepatomas, and one had an adenoma of the kidney. Among the controls, four
guinea pigs had peripheral mononuclear cell accumulation in the lungs, and only one had
hyperplasia of the bronchial epithelium. One control had formation of bone in the bronchus. No
further information was presented in the brief narrative of this study. Given the limited reporting
of the results, a NOAEL or LOAEL value was not provided for this study.
4.2.1.2.4. Xociba et aL (1974). Groups of 6-8-week-old Sherman rats (60/sex/dose level) were
administered 1,4-dioxane (purity not reported) in the drinking water at levels of 0 (controls),
0.01, 0.1, or 1.0% for up to 716 days. The drinking water was prepared twice weekly during the
first year of the study and weekly during the second year of the study. Water samples were
collected periodically and analyzed for 1,4-dioxane content by routine gas liquid
chromatography. Food and water were available ad libitum. Rats were observed daily for
clinical signs of toxicity, and BWs were measured twice weekly during the first month, weekly
during months 2-7, and biweekly thereafter. Water consumption was recorded at three different
time periods during the study: days 1-113, 114-198, and 446-460. Blood samples were
collected from a minimum of five male and five female control and high-dose rats during the 4th,
6th, 12th, and 18th months of the study and at termination. Each sample was analyzed for
packed cell volume, total erythrocyte count, hemoglobin, and total and differential WBC counts.
Additional endpoints evaluated included organ weights (brain, liver, kidney, testes, spleen, and
heart) and gross and microscopic examination of major tissues and organs (brain, bone and bone
marrow, ovaries, pituitary, uterus, mesenteric lymph nodes, heart, liver, pancreas, spleen,
stomach, prostate, colon, trachea, duodenum, kidneys, esophagus, jejunum, testes, lungs, spinal
cord, adrenals, thyroid, parathyroid, nasal turbinates, and urinary bladder). The number of rats
with tumors, hepatic tumors, hepatocellular carcinomas, and nasal carcinomas were analyzed for
statistical significance with Fisher's Exact test (one-tailed), comparing each treatment group
against the respective control group. Survival rates were compared using % Contingency Tables
and Fisher's Exact test. Student's t test was used to compare hematological parameters, body
and organ weights, and water consumption of each treatment group with the respective control
group.
Male and female rats in the high-dose group (1% in drinking water) consumed slightly
less water than controls. BW gain was depressed in the high-dose groups relative to the other
groups almost from the beginning of the study (food consumption data were not provided).
Based on water consumption and BW data for specific exposure groups, Kociba et al. (1974)
35
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
calculated mean daily doses of 9.6, 94, and 1,015 mg/kg-day for male rats and 19, 148, and
1,599 mg/kg-day for female rats during days 114-198 for the 0.01, 0.1, and 1.0% concentration
levels, respectively. Treatment with 1,4-dioxane significantly increased mortality among high-
dose males and females beginning at about 2-4 months of treatment. These rats showed
degenerative changes in both the liver and kidneys. From the 5th month on, mortality rates of
control and treated groups were essentially the same. There were no treatment-related alterations
in hematological parameters. At termination, the only alteration in organ weights noted by the
authors was a significant increase in absolute and relative liver weights in male and female high-
dose rats (data not shown). Histopathological lesions were restricted to the liver and kidney from
the mid- and high-dose groups and consisted of variable degrees of renal tubular epithelial and
hepatocellular degeneration and necrosis (no quantitative incidence data were provided). Rats
from these groups also showed evidence of hepatic regeneration, as indicated by hepatocellular
hyperplastic nodule formation and evidence of renal tubular epithelial regenerative activity
(observed after 2 years of exposure). These changes were not seen in controls or in low-dose
rats. The authors determined a LOAEL of 94 mg/kg-day based on the liver and kidney effects in
male rats. The corresponding NOAEL value was 9.6 mg/kg-day.
Histopathological examination of all the rats in the study revealed a total of 132 tumors in
114 rats. Treatment with 1% 1,4-dioxane in the drinking water resulted in a significant increase
in the incidence of hepatic tumors (hepatocellular carcinomas in six males and four females). In
addition, nasal carcinomas (squamous cell carcinoma of the nasal turbinates) occurred in one
high-dose male and two high-dose females. Since 128 out of 132 tumors occurred in rats from
the 12th to the 24th month, Kociba et al. (1974) assumed that the effective number of rats was
the number surviving at 12 months, which was also when the first hepatic tumor was noticed.
The incidences of liver and nasal tumors from Kociba et al. (1974) are presented in Table 4-4.
Tumors in other organs were not elevated when compared to control incidence and did not
appear to be related to 1,4-dioxane administration.
36
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
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)
Effective
number of
animals"
Number of tumor-
bearing animals
Number of animals
Hepatic tumors
(all types)
Hepatocellular
carcinomas
Nasal
carcinomas
0
106
31
2
1
0
14
110
34
0
0
0
121
106
28
1
1
0
1307
66
21
12b
10c
3d
aRats surviving until 12 months on study.
hp = 0.00022 by one-tailed Fisher's Exact test.
°p = 0.00033 by one-tailed Fisher's Exact test.
dp = 0.05491 by one-tailed Fisher's Exact test.
Source: Kociba et al. (1974).
The only dose level that increased the formation of liver tumors over control (average
dose for male and female rats, 1,307 mg/kg-day) was also demonstrated to cause significant liver
and kidney toxicity in these animals. The mid-dose group (average dose for male and female
rats, 121 mg/kg-day) experienced hepatic and renal degeneration and necrosis, as well as
regenerative hyperplasia in hepatocytes and renal tubule epithelial cells. No increase in tumor
formation was seen in the mid-dose group. No toxicity or tumor formation was observed in the
low-dose group of rats (average dose for male and female rats, 14 mg/kg-day).
4.2.1.2.5. National Cancer Institute (NCI) (1978). Groups of Osborne-Mendel rats
(35/sex/dose) and B6C3Fi mice (50/sex/dose) were administered 1,4-dioxane (> 99.95% pure) in
the drinking water for 110 or 90 weeks, respectively, at levels of 0 (matched controls), 0.5, or
1%. Solutions of 1,4-dioxane were prepared with tap water. The report indicated that at
105 weeks from the earliest starting date, a new necropsy protocol was instituted. This affected
the male controls and high-dose rats, which were started a year later than the original groups of
rats and mice. Food and water were available ad libitum. Endpoints monitored in this bioassay
included clinical signs (twice daily), BWs (once every 2 weeks for the first 12 weeks and every
month during the rest of the study), food and water consumption (once per month in 20% of the
animals in each group during the second year of the study), and gross and microscopic
appearance of all major organs and tissues (mammary gland, trachea, lungs and bronchi, heart,
bone marrow, liver, bile duct, spleen, thymus, lymph nodes, salivary gland, pancreas, kidney,
esophagus, thyroid, parathyroid, adrenal, gonads, brain, spinal cord, sciatic nerve, skeletal
muscle, stomach, duodenum, colon, urinary bladder, nasal septum, and skin). Based on the
measurements of water consumption and BWs, the investigators calculated average daily intakes
of 1,4-dioxane of 0, 240, and 530 mg/kg-day in male rats, 0, 350, and 640 mg/kg-day in female
rats, 0, 720, and 830 mg/kg-day in male mice, and 0, 380, and 860 mg/kg-day in female mice.
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
19
20
21
22
According to the report, the doses of 1,4-dioxane in high-dose male mice were only slightly
higher than those of the low-dose group due to decreased fluid consumption in high-dose male
mice.
During the second year of the study, the BWs of high-dose rats were lower than controls,
those of low-dose males were higher than controls, and those of low-dose females were
comparable to controls. The fluctuations in the growth curves were attributed to mortality by the
investigators; quantitative analysis of BW changes was not done. Mortality was significantly
increased in treated rats, beginning at approximately 1 year of study. Analysis of Kaplan-Meier
curves (plots of the statistical estimates of the survival probability function) revealed significant
positive dose-related trends (p < 0.001, Tarone test). In male rats, 33/35 (94%) in the control
group, 26/35 (74%) in the mid-dose group, and 33/35 (94%) in the high-dose group were alive
on week 52 of the study. The corresponding numbers for females were 35/35 (100%), 30/35
(86%>), and 29/35 (83%). Nonneoplastic lesions associated with treatment with 1,4-dioxane were
seen in the kidneys (males and females), liver (females only), and stomach (males only). Kidney
lesions consisted of vacuolar degeneration and/or focal tubular epithelial regeneration in the
proximal cortical tubules and occasional hyaline casts. Elevated incidence of hepatocytomegaly
also occurred in treated female rats. Gastric ulcers occurred in treated males, but none were seen
in controls. The incidence of pneumonia was increased above controls in high-dose female rats.
The incidence of nonneoplastic lesions in rats following drinking water exposure to 1,4-dioxane
is presented in Table 4-5. EPA identified the LOAEL in rats from this study as 240 mg/kg-day
for increased incidence of gastric ulcer and cortical tubular degeneration in the kidney in males;
a NOAEL was not established.
Table 4-5. Incidence of nonneoplastic lesions in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water

Males (mg/kg-day)
Females (mg/kg-day)

0
240
530
0
350
640
Cortical tubule degeneration
0/3 r
20/3 lb
27/3 3b
0/3 r
0/34
10/32b


(65%)
(82%)


(31%)
Hepatocytomegaly
5/31
3/32
11/33
7/3 r
11/33
17/32b

(16%)
(9%)
(33%)
(23%)
(33%)
(53%)
Gastric ulcer
0/3 0a
5/28b
5/3 0b
0/31
1/33
1/30


(18%)
(17%)

(3%)
(3%)
Pneumonia
8/30
15/31
14/33
6/3 0a
5/34
25/32b

(27%)
(48%)
(42%)
(20%)
(15%)
(78%)
aStatistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this
review.
incidence significantly elevated compared to control by Fisher's Exact test (p < 0.05) performed for this review.
Source: NCI (1978).
38
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
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

0
240b
530
Nasal cavity squamous cell carcinoma
0/33 (0%)
12/33 (36%)
16/34 (47%)°
Hepatocellular adenoma
2/31 (6%)
2/32 (6%)
1/33 (3%)
Females (mg/kg-day)a

0
350
640
Nasal cavity squamous cell carcinoma
0/34 (0%)d
10/35 (29%)e
8/35 (23%)°
Hepatocellular adenoma
0/31 (0%)f
10/33 (30%)e
11/32 (34%)e
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.
cp < 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.
fp = 0.001 by Cochran-Armitage test.
Source: NCI (1978).
The incidence of hepatocellular adenomas in male and female rats is presented in
Table 4-6. 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
39
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
high-dose animals, respectively). The difference between the treated groups and controls was
not statistically significant. These tumors were characterized as rounded and papillary
projections of mesothelial cells, each supported by a core of fibrous tissue. Other reported
neoplasms were considered spontaneous lesions not related to treatment with 1,4-dioxane.
In mice, mean BWs of high-dose female mice were lower than controls during the second
year of the study, while those of low-dose females were higher than controls. In males, mean
BWs of high-dose animals were higher than controls during the second year of the study.
According to the investigators, these fluctuations could have been due to mortality; no
quantitative analysis of BWs was done. No other clinical signs were reported. Mortality was
significantly increased in female mice (p < 0.001, Tarone test), beginning at approximately
80 weeks on study. The numbers of female mice that survived to 91 weeks were 45/50 (90%) in
the control group, 39/50 (78%) in the low-dose group, and 28/50 (56%) in the high-dose group.
In males, at least 90% of the mice in each group were still alive at week 91. Nonneoplastic
lesions that increased significantly due to treatment with 1,4-dioxane were pneumonia in males
and females and rhinitis in females. The incidences of pneumonia were 1/49 (2%), 9/50 (18%),
and 17/47 (36%) in control, low-dose, and high-dose males, respectively; the corresponding
incidences in females were 2/50 (4%), 33/47 (70%), and 32/36 (89%). The incidences of rhinitis
in female mice were 0/50, 7/48 (14%), and 8/39 (21%) in control, low-dose, and high-dose
groups, respectively. Pair-wise comparisons of low-dose and high-dose incidences with controls
for incidences of pneumonia and rhinitis in females using Fisher's Exact test (done for this
review) yielded ^-values < 0.001 in all cases. Incidences of other lesions were considered to be
similar to those seen in aging mice. The authors stated that hepatocytomegaly was commonly
found in dosed mice, but the incidences were not significantly different from controls and
showed no dose-response trend. EPA concluded the LOAEL for 1,4-dioxane in mice was
380 mg/kg-day based on the increased incidence of pneumonia and rhinitis in female mice; a
NOAEL was not established in this study.
As shown in Table 4-7, treatment with 1,4-dioxane significantly increased the incidence
of hepatocellular carcinomas or adenomas in male and female mice in a dose-related manner.
Tumors were first observed on week 81 in high-dose females and in week 58 in high-dose males.
Tumors were characterized by parenchymal cells of irregular size and arrangement, and were
often hypertrophic with hyperchromatic nuclei. Mitoses were seldom seen. Neoplasms were
locally invasive within the liver, but metastasis to the lungs was rarely observed.
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
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

0
720
830
Hepatocellular carcinoma
2/49 (4%)b
18/50 (36%)°
24/47 (51%)°
Hepatocellular adenoma or carcinoma
8/49 (16%)b
19/50 (38%)d
28/47 (60%)°
Females (mg/kg-day)a

0
380
860
Hepatocellular carcinoma
0/50 (0%)b
12/48 (25%)°
29/37 (78%)°
Hepatocellular adenoma or carcinoma
0/50 (0%)b
21/48 (44%)°
35/37 (95%)°
aTumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
cp < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (1978).
In addition to liver tumors, a variety of other benign and malignant neoplasms occurred.
However, the report (NCI, 1978) indicated that each type had been encountered previously as a
spontaneous lesion in the B6C3Fi mouse. The report further stated that the incidences of these
neoplasms were unrelated by type, site, group, or sex of the animal, and hence, not attributable to
exposure to 1,4-dioxane. There were a few nasal adenocarcinomas (1/48 in low-dose females
and 1/49 in high-dose males) that arose from proliferating respiratory epithelium lining of the
nasal turbinates. These growths extended into the nasal cavity, but there was minimal local
tissue infiltration. Nasal mucosal polyps were rarely observed. The polyps were derived from
mucus-secreting epithelium and were otherwise unremarkable. There was a significant negative
trend for alveolar/bronchiolar adenomas or carcinomas of the lung in male mice, such that the
incidence in the matched controls was higher than in the dosed groups. The report (NCI, 1978)
indicated that the probable reason for this occurrence was that the dosed animals did not live as
long as the controls, thus diminishing the possibility of the development of tumors in the dosed
groups.
4.2.1.2.6. /{(/no e/ a/. (2009// Japan fiioassay Hesearch Center (JBJtC) (1998a); Va/nazaA/'
et al. (1994). Groups of F344/DuCrj rats (50/sex/dose level) were exposed to 1,4-dioxane
(>99% pure) in the drinking water at levels of 0, 200, 1,000, or 5,000 ppm for 2 years. Groups of
Crj :BDFi mice (50/sex/dose level) were similarly exposed to 0, 500, 2,000, or 8,000 ppm of
1,4-dioxane in the drinking water. The high dose was selected based on results from the Kano et
al. (2008) 13-week drinking water study so as not to exceed the maximum tolerated dose (MTD)
1 Data from Kano et al. (2009) was previously published as Yamazaki et al. (1994). Kano et al. (2009) results differ
from those reported previously (Yamazaki et al., 1994) because Kano et al. (2009) reported data using an improved
diagnosis of pre- and neoplastic lesions in the liver according to the current diagnostic criteria (see references in
Kano et al., 2009).
41
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
in that study. Both rats and mice were 6 weeks old at the beginning of the study. Food and
water were available ad libitum. The animals were observed daily for clinical signs of toxicity,
and BWs were measured once per week for 14 weeks and once every 2 weeks until the end of
the study. Food consumption was measured once a week for 14 weeks and once every 4 weeks
for the remainder of the study. The investigators used data from water consumption and BW to
calculate the daily intake of 1,4-dioxane by the male and female animals. Kano et al. (2009)
reported mean estimated daily doses of 1,4-dioxane for the duration of the study. Male rats
received doses of approximately 0, 11, 55, or 274 mg/kg-day and female rats received 0, 18, 83,
or 429 mg/kg-day. Male mice received doses of 0, 49, 191, or 677 mg/kg-day and female mice
received 0, 66, 278, or 967 mg/kg-day. The Kano et al. (2009) study was conducted in
accordance with the Organization for Economic Co-operation and Development (OECD)
Principles for Good Laboratory Practice (GLP).
Growth and mortality rates were reported in Kano et al. (2009) for the duration of the
study. Both male and female rats in the high dose groups (274 and 429 mg/kg-day, respectively)
both exhibited slower growth rates and terminal body weights that were significantly different (p
<0.05) compared to controls. Similarly in mice, male and female mice growth rates were slower
than controls and terminal body weights were lower for the mid (p<0.01 for males administered
191 mg/kg-day and p<0.05 for females administered 278 mg/kg-day) and high doses (p<0.05 for
males and females administered 677 and 967 mg/kg-day, respectively).
Survival rates of the male and female rats in the high dose groups (274 and 429 mg/kg-
day, respectively) were approximately 50%, which was significantly different compared to
controls. The authors attributed these early deaths to the increased incidence in nasal tumors and
peritoneal mesotheliomas in male rats and nasal and hepatic tumors in female rats. There were
no differences in survival rates between control and treated male mice; however, survival rates
were significantly decreased compared to controls for female mice in the mid (278 mg/kg-day,
approximately 40% survival) and high (967 mg/kg-day, approximately 20% survival) dose
groups. The study authors attributed these early female mouse deaths to the significant incidence
of hepatic tumors, and they reported tumor incidence for all animals in the study (N=50),
including animals that became moribund or died before the end of the study.
No information was provided as to when urine samples were collected. Blood samples
were collected only at the end of the 2-year study (email from Dr. Kazunori Yamazaki, JBRC, to
Dr. Julie Stickney, SRC, dated 12/18/06). Hematology analysis included RBCs, hemoglobin,
hematocrit, MCV, platelets, WBCs and differential WBCs. Serum biochemistry included total
protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only), phospholipid, ALT, AST,
LDH, LAP, ALP, y-glutamyl transpeptidase (GGT), CPK, urea nitrogen, creatinine (rat only),
sodium, potassium, chloride, calcium, and inorganic phosphorous. Urinalysis parameters were
pH, protein, glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ
weights (brain, lung, liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and
42
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
gross necropsy and histopathologic examination of tissues and organs were performed on all
animals (skin, nasal cavity, trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart,
tongue, salivary glands, esophagus, stomach, small and large intestine, liver, pancreas, kidney,
urinary bladder, pituitary, thyroid, adrenal, testes, epididymis, seminal vesicle, prostate, ovary,
uterus, vagina, mammary gland, brain, spinal cord, sciatic nerve, eye, Harderian gland, muscle,
bone, and parathyroid). Dunnett's test and % test were used to assess the statistical significance
of changes in continuous and discrete variables, respectively.
Survival was significantly decreased in the rat high-dose groups (80% in control males
versus 44% in high-dose males; 76% in control females versus 48% in high-dose females). The
effect on survival in high-dose rats occurred in the second year of the study, as all control and
exposed rats lived at least 12 months following study initiation (email from Dr. Kazunori
Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). The extra mortality in the high-
dose groups was primarily related to tumors in these groups (peritoneal mesothelioma, liver and
nasal tumors) (email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated
12/18/06). Food consumption was not significantly affected by treatment in male or female rats;
however, water consumption in female rats administered 18 mg/kg-day was significantly greater
(p<0.05) . A statistically significant reduction in terminal BWs was observed in high-dose male
rats (5%>, p<0.01) and in high-dose female rats (18%, p<0.01) (Kano et al., 2009). RBC (male
rats only), hemoglobin, hematocrit, and MCV were decreased, and platelets were increased in
high-dose groups (JBRC, 1998a). These changes (except for MCV) also occurred in mid-dose
males. With the exception of a 23% decrease in hemoglobin in high-dose male rats and a 27%
increase in platelets in high-dose female rats, hematological changes were within 15% of control
values. Significant changes in serum chemistry parameters occurred only in high-dose rats
(males: increased phospholipids, AST, ALT, LDH, ALP, GGT, CPK, potassium, and inorganic
phosphorus and decreased total protein, albumin, and glucose; females: increased total bilirubin,
cholesterol, phospholipids, AST, ALT, LDH, GGT, ALP, CPK, and potassium, and decreased
blood glucose) (JBRC, 1998a). Increases in serum enzyme activities ranged from <2- to 17-fold
above control values, with the largest increases seen for ALT, AST, and GGT. Urine pH was
significantly decreased at 274 mg/kg-day in male rats (not tested at other dose levels) and at
83 and 429 mg/kg-day in female rats (JBRC, 1998a). Also, blood in the urine was seen in
female rats at 83 and 429 mg/kg-day (JBRC, 1998a). In male rats, relative liver weights were
increased at 55 and 274 mg/kg-day (Kano et al., 2009). In female rats, relative liver weight was
increased at 429 mg/kg-day (Kano et al., 2009).
Microscopic examination of the tissues showed nonneoplastic alterations in the nasal
cavity, liver, and kidneys mainly in high-dose rats and, in a few cases, in mid-dose rats (Tables
4-8 and 4-9). Alterations in high-dose (274 mg/kg-day) male rats consisted of nuclear
enlargement and metaplasia of the olfactory and respiratory epithelia, atrophy of the olfactory
epithelium, hydropic changes and sclerosis of the lamina propria, adhesion, and inflammation.
43
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
Table 4-8. Incidence of histopathological lesions in male F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years

Dose (mg/kg-day)a
0
11
55
274
Nuclear enlargement; nasal respiratory epithelium
0/50
0/50
0/50
26/50b
Squamous cell metaplasia; nasal respiratory epithelium
0/50
0/50
0/50
31/50b
Squamous cell hyperplasia; nasal respiratory epithelium
0/50
0/50
0/50
2/50
Nuclear enlargement; nasal olfactory epithelium
0/50
0/50
5/50°
38/50b
Respiratory metaplasia; nasal olfactory epithelium
12/50
11/50
20/50
43/50b
Atrophy; nasal olfactory epithelium
0/50
0/50
0/50
36/50b
Hydropic change; lamina propria
0/50
0/50
0/50
46/50b
Sclerosis; lamina propria
0/50
0/50
1/50
44/50b
Adhesion; nasal cavity
0/50
0/50
0/50
48/50b
Inflammation; nasal cavity
0/50
0/50
0/50
13/50b
Hyperplasia; liver
3/50
2/50
10/50
24/50b
Spongiosis hepatis; liver
12/50
20/50
25/50°
40/50
Clear cell foci; liver
3/50
3/50
9/50
8/50
Acidophilic cell foci; liver
12/50
8/50
7/50
5/50
Basophilic cell foci; liver
7/50
11/50
8d/50
16/50°
Mixed-cell foci; liver
2/50
8/50
14/50b
13/50b
Nuclear enlargement; kidney proximal tubule
0/50
0/50
0/50
50/50b
aData presented for all animals, including animals that became moribund or died before the end of the study.
hp < 0.01 by x2 test.
°p < 0.05 by x2 test.
dReported in JBRC (1998a) as 6/50 and in Kano et al. (2009) as 8/50. The Kano et al. (2009) value is
reported in the table.
Sources: Kano et al. (2009) and JBRC (1998a).
44	DRAFT - DO NOT CITE OR QUOTE
In female rats, nuclear enlargement of the olfactory epithelium occurred at doses >83 mg/kg-day,
and nuclear enlargement and metaplasia of the respiratory epithelium, squamous cell
hyperplasia, respiratory metaplasia of the olfactory epithelium, hydropic changes and sclerosis of
the lamina propria, adhesion, inflammation, and proliferation of the nasal gland occurred at
429 mg/kg-day. Alterations were seen in the liver at >55 mg/kg-day in male rats (spongiosis
hepatis, hyperplasia, and clear and mixed cell foci) and at 429 mg/kg-day in female rats
(hyperplasia, spongiosis hepatis, cyst formation, and mixed cell foci). Nuclear enlargement of
the renal proximal tubule occurred in males at 274 mg/kg-day and in females at > 83 mg/kg-day
(JBRC, 1998a).

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Table 4-9. Incidence of histopathological lesions in female F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years

Dose (mg/kg-day)a
0
18
83
429
Nuclear enlargement; nasal respiratory epithelium
0/50
0/50
0/50
13/50b
Squamous cell metaplasia; nasal respiratory epithelium
0/50
0/50
0/50
35/50b
Squamous cell hyperplasia; nasal cavity
0/50
0/50
0/50
5/50
Nuclear enlargement; nasal olfactory epithelium
0/50
0/50
28/50b
39/50b
Respiratory metaplasia; nasal olfactory epithelium
2/50
0/50
2/50
42/50b
Atrophy; nasal olfactory epithelium
0/50
0/50
1/50
40/50b
Hydropic change; lamina propria
0/50
0/50
0/50
46/50b
Sclerosis; lamina propria
0/50
0/50
0/50
48/50b
Adhesion; nasal cavity
0/50
0/50
0/50
46/50b
Inflammation; nasal cavity
0/50
0/50
1/50
15/50b
Proliferation; nasal gland
0/50
0/50
0/50
1 l/50b
Hyperplasia; liver
3/50
2/50
1 l/50b
47/50b
Spongiosis hepatis; liver
0/50
0/50
1/50
20/50b
Cyst formation; liver
0/50
1/50
1/50
8/5 0b
Acidophilic cell foci; liver
1/50
1/50
1/50
1/50
Basophilic cell foci; liver
23/50
27/50
31/50
8/5 0b
Clear cell foci; liver
1/50
1/50
5/50
4/50
Mixed-cell foci; liver
1/50
1/50
3/50
1 l/50b
Nuclear enlargement; kidney proximal tubule
0/50
0/50
6/50°
39/50b
aData presented for all animals, including animals that became moribund or died before the end of the study.
hp < 0.01 by x2 test.
°p < 0.05 by x2 test.
Sources: Kano et al. (2009) and JBRC (1998a).
NOAEL and LOAEL values for rats in this study were identified by EPA as 55 and
274 mg/kg-day, respectively, based on toxicity observed in nasal tissue of male rats (i.e., atrophy
of olfactory epithelium, adhesion, and inflammation). Metaplasia and hyperplasia of the nasal
epithelium were also observed in high-dose male and female rats. These effects are likely to be
associated with the formation of nasal cavity tumors in these dose groups. Nuclear enlargement
was observed in the nasal olfactory epithelium and the kidney proximal tubule at a dose of
83 mg/kg-day in female rats; however, it is unclear whether these alterations represent adverse
toxicological effects. Hematological effects noted in male rats given 55 and 274 mg/kg-day
(decreased RBCs, hemoglobin, hematocrit, increased platelets) were within 20% of control
values. In female rats decreases in hematological effects were observed in the high dose group
(429 mg/kg-day). A reference range database for hematological effects in laboratory animals
(Wolford et al., 1986) indicates that a 20% change in these parameters may fall within a normal
range (10th-90th percentile values) and may not represent a treatment-related effect of concern.
Liver lesions were also seen at a dose of 55 mg/kg-day in male rats; these changes are likely to
45	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
be associated with liver tumorigenesis. Clear and mixed-cell foci are commonly considered
preneoplastic changes and would not be considered evidence of noncancer toxicity. The nature
of spongiosis hepatis as a preneoplastic change is less well understood (Bannasch, 2003; Karbe
and Kerlin, 2002; Stroebel et al., 1995). Spongiosis hepatis is a cyst-like lesion that arises from
the peri sinusoidal Ito cells of the liver. It is commonly seen in aging rats, but has been shown to
increase in incidence following exposure to hepatocarcinogens. Spongiosis hepatis can be seen
in combination with preneoplastic foci in the liver or with hepatocellular adenoma or carcinoma
and has been considered a preneoplastic lesion (Bannasch et al., 2003; Stroebel et al., 1995).
This change can also be associated with hepatocellular hypertrophy and liver toxicity and has
been regarded as a secondary effect of some liver carcinogens (Karbe and Kerlin, 2002). In the
case of the JBRC (1998a) study, spongiosis hepatis was associated with other preneoplastic
changes in the liver (clear and mixed-cell foci). No other lesions indicative of liver toxicity were
seen in this study; therefore, spongiosis hepatis was not considered indicative of noncancer
effects. Serum chemistry changes (increases in total protein, albumin, and glucose; decreases in
AST, ALT, LDH, and ALP, potassium, and inorganic phosphorous) were observed in both male
and female rats (JBRC, 1998a) in the high dose groups, 274 and 429 mg/kg-day, respectively.
These serum chemistry changes seen in terminal blood samples from high-dose male and female
rats are likely related to tumor formation in these dose groups.
Significantly increased incidences of liver tumors (adenomas and carcinomas) and tumors
of the nasal cavity occurred in high-dose male and female rats (Tables 4-10 and 4-11) treated
with 1,4-dioxane for 2 years. The first liver tumor was seen at 85 weeks in high-dose male rats
and 73 weeks in high-dose female rats (vs. 101-104 weeks in lower dose groups and controls)
(email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). In
addition, a significant increase (p < 0.01, Fisher's Exact test) in mesotheliomas of the
peritoneum was seen in high-dose males (28/50 versus 2/50 in controls). Mesotheliomas were
the single largest cause of death among high-dose male rats, accounting for 12 of 28
preterminal on deaths (email from Dr. Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, SRC,
dated 12/18/06). Also, in males, there were increasing trends in mammary gland fibroadenoma
and fibroma of the subcutis, both statistically significant (p < 0.01) by the Peto test of dose-
response trend. Females showed a significant increasing trend in mammary gland adenomas (p <
0.01 by Peto's test). The tumor incidence values presented in Tables 4-10 and 4-11 were not
adjusted for survival because all rats lived longer than 12 months on study.
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

Males
Females
Dose (mg/kg-day)
0
11
55
274
0
18
83
429
Nasal Cavity
Squamous cell carcinoma
0/50
0/50
0/50
3/50a
0/50
0/50
0/50
7/50a'°
Sarcoma
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
Rhabdomyosarcoma
0/50
0/50
0/50
1/50
0/50
0/50
0/50
0/50
Esthesioneuroepithelioma
0/50
0/50
0/50
1/50
0/50
0/50
0/50
1/50
Peritoneum
Mesothelioma
2/50
2/50
5/50
28/50a'°
1/50
0/50
0/50
0/50
Mammary Gland
Fibroadenoma
1/50
1/50
0/50
4/50a
3/50
2/50
1/50
3/50
Adenoma
0/50
1/50
2/50
2/50
6/50
7/50
10/50
16/50a'd
Either Adenoma or Fibroadenoma
1/50
2/50
2/50
6/50a
8/50
8/50
11/50
18/50a'd
a/> < 0.01 by Peto's test for trend.
hp < 0.05 by Peto's test for trend.
cp < 0.01 by Fisher's exact test.
dp < 0.05 by Fisher's exact test.
Source: Kano et al. (2009).
Table 4-11. Incidence of liver tumors in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years

Males
Females
Dose (mg/kg-day)
0
11
55
274
0
18
83
429
Hepatocellular adenoma
3/50a
4/50
7/50
32/50a'b
3/50
1/50
6/50
48/50a'b
Hepatocellular carcinoma
0/50a
0/50
0/50
14/50a'b
0/50
0/50
0/50
lo/so3-13
Adenoma or carcinoma
3/50a
4/50
7/50
39/50a'b
3/50
1/50
6/50
48/50a'b
ap < 0.01 by Fisher's Exact test.
hp < 0.01 by Peto test for trend.
Source: Kano et al. (2009).
1	In the mouse study, survival rates did not differ between the control male mice and the
2	1,4-dioxane-dosed male mice; however, decreased survival rates were seen in the female mice
3	given 278 and 967 mg/kg-day (29/50, 29/50, 17/50, and 5/50 in control, 66, 278, and 967 mg/kg-
4	day dose groups, respectively). Deaths occurred primarily during the second year of the study.
5	Survival at 12 months in male mice was 50/50, 48/50, 50/50, and 48/50 in control, low-, mid-,
6	and high-dose groups, respectively. Female mouse survival at 12 months was 50/50, 50/50,
7	48/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively (email from Dr.
8	Kazunori Yamazaki, JBRC, to Dr. Julie Stickney, Syracuse Research Corporation (SRC), dated
47	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
12/18/06). The deaths were primarily tumor-related (e.g., liver tumors were listed as the cause of
death for 31 of the 45 pretermination deaths in high-dose female rats) (email from Dr. Kazunori
Yamazaki, JBRC, to Dr. Julie Stickney, SRC, dated 12/18/06). Food consumption was not
significantly affected, but water consumption was reduced 26% in high-dose male mice and 28%
in high-dose female mice. Final BWs were reduced 43% in high-dose male mice and 15 and
45% in mid- and high-dose female mice, respectively. Male mice showed increases in RBC
counts, hemoglobin, and hematocrit, whereas in female mice, there was a decrease in platelets in
mid- and high-dose rats. With the exception of a 60% decrease in platelets in high-dose female
mice, hematological changes were within 15% of control values. Serum AST, ALT, LDH, and
ALP activities were significantly increased in mid- and high-dose male mice, whereas LAP and
CPK were increased only in high-dose male mice. AST, ALT, LDH, and ALP activities were
increased in mid- and high-dose female mice, but CPK activity was increased only in high-dose
female mice. Increases in serum enzyme activities ranged from less than two- to sevenfold
above control values. Glucose and triglycerides were decreased in high-dose males and in mid-
and high-dose females. High-dose female mice also showed decreases in serum phospholipid
and albumin concentrations (not reported in males). Blood calcium was lower in high-dose
females and was not reported in males. Urinary pH was decreased in high-dose males, whereas
urinary protein, glucose, and occult blood were increased in mid- and high-dose female mice.
Relative and absolute lung weights were increased in high-dose males and in mid- and high-dose
females (JBRC, 1998a). Microscopic examination of the tissues for nonneoplastic lesions
showed significant alterations in the epithelium of the respiratory tract, mainly in high-dose
animals, although some changes occurred in mid-dose mice (Tables 4-12 and 4-13). Commonly
seen alterations included nuclear enlargement, atrophy, and inflammation of the epithelium.
Other notable changes observed included nuclear enlargement of the proximal tubule of the
kidney and angiectasis in the liver in high-dose male mice.
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

Dose (mg/kg-day)a
0
49
191
677
Nuclear enlargement; nasal respiratory epithelium
0/50
0/50
0/50
31/50b
Nuclear enlargement; nasal olfactory epithelium
0/50
0/50
9/5 0b
49/50b
Atrophy; nasal olfactory epithelium
0/50
0/50
1/50
48/50b
Inflammation; nasal cavity
1/50
2/50
1/50
25/50b
Atrophy; tracheal epithelium
0/50
0/50
0/50
42/50b
Nuclear enlargement; tracheal epithelium
0/50
0/50
0/50
17/50b
Nuclear enlargement; bronchial epithelium
0/50
0/50
0/50
41/50b
Atrophy; lung/bronchial epithelium
0/50
0/50
0/50
43/50b
Accumulation of foamy cells; lung
1/50
0/50
0/50
27/50b
Angiectasis; liver
2/50
3/50
4/50
16/50b
Nuclear enlargement; kidney proximal tubule
0/50
0/50
0/50
39/50b
aData presented for all animals, including animals that became moribund or died before the end of the study.
hp < 0.01 by x2 test.
°p < 0.05 by x2 test.
Source: Kano et al. (2009) and JBRC (1998a).
Table 4-13. Incidence of histopathological lesions in female CrjrBDFi mice
exposed to 1,4-dioxane in drinking water for 2 years

Dose (mg/kg-day)a
0
66
278
967
Nuclear enlargement; nasal respiratory epithelium
0/50
0/50
0/50
41/50b
Nuclear enlargement; nasal olfactory epithelium
0/50
0/50
41/50b
33/50b
Atrophy; nasal olfactory epithelium
0/50
0/50
1/50
42/50b
Inflammation; nasal cavity
2/50
0/50
7/50
42/50b
Atrophy; tracheal epithelium
0/50
0/50
2/50
49/50b
Nuclear enlargement; bronchial epithelium
0/50
1/50
22/50b
48/50b
Atrophy; lung/bronchial epithelium
0/50
0/50
7/50°
50/50b
Accumulation of foamy cells; lung
0/50
1/50
4/50
45/50b
aData presented for all animals, including animals that became moribund or died before the end of the study.
hp < 0.01 by x2 test.
°p < 0.05 by x2 test.
Source: Kano et al. (2009) and JBRC (1998a).
1	NOAEL and LOAEL values for mice in this study were identified by EPA as 66 and
2	278 mg/kg-day, respectively, based on nasal inflammation observed in female mice. Nuclear
3	enlargement of the nasal olfactory epithelium and bronchial epithelium was also observed at a
4	dose of 278 mg/kg-day in female mice; however, it is unclear whether these alterations represent
5	adverse toxicological effects. The serum chemistry changes seen in terminal blood samples from
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
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 677 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 hepatocellular carcinoma in all treatment groups, while an
increase in hepatocellular adenoma incidence was only seen in the 66 and 278 mg/kg-day dose
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 in Table 4-14 are based on reanalyzed sample data
presented in Kano et al. (2009) that included lesions in animals that became moribund or died
prior to the completion of the 2-year study.
Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and
carcinomas in control male and female BDFi mice from ten 2-year bioassays at the JBRC. For
female mice, out of 499 control mice, the incidence rates were 4.4% for hepatocellular adenomas
and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported a 10% incidence rate for
hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in control female
BDFi.
Table 4-14. Incidence of liver tumors in CrjrBDFi mice exposed to
1,4-dioxane in drinking water for 2 years

Males
Females
Dose (mg/kg-day)
0
49
191
677
0
66
278
967
Hepatocellular adenoma
9/50
17/50
23/503
11/50
5/50
31/503
20/503
3/50
Hepatocellular carcinoma
15/503
20/50
23/50
36/50^
0/50a
6/50°
30/503
45/50^
Hepatocellular adenoma or
carcinoma
23/50
31/50°
37/50°
40/50^
5/50a
35/503
41/503
46/50a'b
< 0.01 by Fisher's Exact test.
hp < 0.01; positive dose-related trend (Peto's test)
°p < 0.05 by Fisher's Exact test.
Sources: Kano et al. (2009).
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
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.	Subc/ironic Inhalation Toxicity
4.2.2.1.1. Fair/ey 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 two-times a day for 1.5
hours (3 hours/day) for 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 animal (rat) survived a 7-day exposure. The rest of the animals (six
guinea pigs, three mice, and two rats) died within the first five exposures. Severe liver and
kidney damage and acute vascular congestion of the lungs were observed in these animals.
Kidney damage was described as patchy degeneration of cortical tubules with vascular
congestion and hemorrhage. Liver lesions varied from cloudy hepatocyte swelling to large areas
of necrosis. At 5,000 ppm, mortality was observed in two mice and one guinea pig following
15-34 exposures. The remaining animals were sacrificed following 49.5 hours (3 weeks) of
exposure (three rabbits) or 94.5 hours (5 weeks) of exposure (three guinea pigs). Liver and
kidney damage in both dead and surviving animals was similar to that described for the
10,000 ppm concentration. Animals (four rabbits, four guinea pigs, six rats, and five mice) were
exposed to 2,000 ppm for 45-102 total exposure hours (approximately 2-6 weeks). Kidney and
liver damage was still apparent in animals exposed to this concentration. Animals exposed to
1,000 ppm were killed at intervals with the total exposure duration ranging between 78 and
202.5 hours (approximately 4-12 weeks). Cortical kidney degeneration and hepatocyte
degeneration and liver necrosis were observed in these animals (two rabbits, three guinea pigs,
three rats, and four mice). The low concentration of 1,000 ppm was identified by EPA as a
LOAEL for liver and kidney degeneration in rats, mice, rabbits, and guinea pigs in this study.
4.2.2.2.	Chronic Inhalation Toxicity and Carcinogenicity
4.2.2.2.1. Torke/son et a/. (1974). Whole body exposures of male and female Wistar rats
(288/sex) to 1,4-dioxane vapors (99.9% pure) at a concentration of 0.4 mg/L (111 ppm), were
carried out 7 hours/day, 5 days/week for 2 years. The age of the animals at the beginning of the
study was not provided. The concentration of 1,4-dioxane vapor during exposures was
determined with infrared analyzers. Food and water were available ad libitum except during
exposures. Endpoints examined included clinical signs, eye and nasal irritation, skin condition,
respiratory distress, and tumor formation. BWs were determined weekly. Standard
hematological parameters were determined on all surviving animals after 16 and 23 months of
exposure. Blood collected at termination was used also for determination of clinical chemistry
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
parameters (serum AST and ALP activities, blood urea nitrogen [BUN], and total protein).
Liver, kidneys, and spleen were weighed and the major tissues and organs were processed for
microscopic examination (lungs, trachea, thoracic lymph nodes, heart, liver, pancreas, stomach,
intestine, spleen, thyroid, mesenteric lymph nodes, kidneys, urinary bladder, pituitary, adrenals,
testes, ovaries, oviduct, uterus, mammary gland, lacrimal gland, lymph nodes, brain, vagina, and
bone marrow, and any abnormal growths). Nasal tissues were not obtained for histopathological
evaluation. Control and experimental groups were compared statistically using Student's t test,
Yates corrected % test, or Fisher's Exact test.
Exposure to 1,4-dioxane vapors had no significant effect on mortality or BW gain and
induced no signs of eye or nasal irritation or respiratory distress. Slight, but statistically
significant, changes in hematological and clinical chemistry parameters were within the normal
physiological limits and were considered to be of no toxicological importance by the
investigators. Altered hematological parameters included decreases in packed cell volume, RBC
count, and hemoglobin, and an increase in WBC count in male rats. Clinical chemistry changes
consisted of a slight decrease in both BUN (control—23 ± 9.9; 111-ppm 1,4-dioxane—19.8 ±
8.8) and ALP activity (control—34.4 ± 12.1; 111-ppm 1,4-dioxane—29.9 ± 9.2) and a small
increase in total protein (control—7.5 ± 0.37; 111-ppm 1,4-dioxane—7.9 ± 0.53) in male rats
(values are mean ± standard deviation). Organ weights were not significantly affected.
Microscopic examination of organs and tissues did not reveal any treatment-related effects.
Based of the lack of significant effects on several endpoints, EPA identified the exposure
concentration of 0.4 mg/L (111 ppm) as a free standingNOAEL. The true NOAEL was likely to
be higher.
Tumors, observed in all groups including controls, were characteristic of the rat strain
used and were considered unrelated to 1,4-dioxane inhalation. The most common tumors were
reticulum cell sarcomas and mammary tumors. Using Fisher's Exact test and a significance level
ofp < 0.05, no one type of tumor occurred more frequently in treated rats than in controls. No
hepatic or nasal cavity tumors were seen in any rat.
4.2.3. Initiation/Promotion Studies
4.2.3.1. Bu//eta/. (1986)
Bull et al. (1986) tested 1,4-dioxane as a cancer initiator in mice using oral,
subcutaneous, and topical routes of exposure. A group of 40 female SENCAR mice (6-8 weeks
old) was administered a single dose of 1,000 mg/kg 1,4-dioxane (purity >99%) by gavage,
subcutaneous injection, or topical administration (vehicle was not specified). A group of rats
was used as a vehicle control (number of animals not specified). Food and water were provided
ad libitum. Two weeks after administration of 1,4-dioxane, 12-0-tetradecanoylphorbol-13-
acetate (TPA) (1.0 |ig in 0.2 mL of acetone) was applied to the shaved back of mice
52
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
3 times/week for a period of 20 weeks. The yield of papillomas at 24 weeks was selected as a
potential predictor of carcinoma yields at 52 weeks following the start of the promotion
schedule. Acetone was used instead of TPA in an additional group of 20 mice in order to
determine whether a single dose of 1,4-dioxane could induce tumors in the absence of TPA
promotion.
1,4-Dioxane did not increase the formation of papillomas compared to mice initiated with
vehicle and promoted with TPA, indicating lack of initiating activity under the conditions of the
study. Negative results were obtained for all three exposure routes. A single dose of
1,4-dioxane did not induce tumors in the absence of TPA promotion.
4.2.3.2. King et a/. (7973)
1,4-Dioxane was evaluated for complete carcinogenicity and tumor promotion activity in
mouse skin. In the complete carcinogenicity study, 0.2 mL of a solution of 1,4-dioxane (purity
not specified) in acetone was applied to the shaved skin of the back of Swiss Webster mice
(30/sex) 3 times/week for 78 weeks. Acetone was applied to the backs of control mice (30/sex)
for the same time period. In the promotion study, each animal was treated with 50 jag of
dimethylbenzanthracene 1 week prior to the topical application of the 1,4-dioxane solution
described above (0.2 mL, 3 times/week, 78 weeks) (30 mice/sex). Acetone vehicle was used in
negative control mice (30/sex). Croton oil was used as a positive control in the promotion study
(30/sex). Weekly counts of papillomas and suspect carcinomas were made by gross
examination. 1,4-Dioxane was also administered in the drinking water (0.5 and 1%) to groups of
Osborne-Mendel rats (35/sex/group) and B6C3Fi mice for 42 weeks (control findings were only
reported for 34 weeks).
1,4-Dioxane was negative in the complete skin carcinogenicity test using dermal
exposure. One treated female mouse had malignant lymphoma; however, no papillomas were
observed in male or female mice by 60 weeks. Neoplastic lesions of the skin, lungs, and kidney
were observed in mice given the promotional treatment with 1,4-dioxane. In addition, the
percentage of mice with skin tumors increased sharply after approximately 10 weeks of
promotion treatment. Significant mortality was observed when 1,4-dioxane was administered as
a promoter (only 4 male and 5 female mice survived for 60 weeks), but not as a complete
carcinogen (22 male and 25 female mice survived until 60 weeks). The survival of acetone-
treated control mice in the promotion study was not affected (29 male and 26 female mice
survived until 60 weeks); however, the mice treated with croton oil as a positive control
experienced significant mortality (0 male and 1 female mouse survived for 60 weeks). The
incidence of mice with papillomas was similar for croton oil and 1,4-dioxane; however, the
tumor multiplicity (i.e., number of tumors/mouse) was higher for the croton oil treatment.
Oral administration of 1,4-dioxane in drinking water caused appreciable mortality in rats,
but not mice, and increased weight gain in surviving rats and male mice. Histopathological
53
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
lesions (i.e., unspecified liver and kidney effects) were also reported in exposed male and female
rats; however, no histopathological changes were indicated for mice.
1,4-Dioxane was demonstrated to be a tumor promoter, but not a complete carcinogen in
mouse skin, in this study. Topical administration for 78 weeks following initiation with
dimethylbenzanthracene caused an increase in the incidence and multiplicity of skin tumors in
mice. Tumors were also observed at remote sites (i.e., kidney and lung), and survival was
affected. Topical application of 1,4-dioxane for 60 weeks in the absence of the initiating
treatment produced no effects on skin tumor formation or mortality in mice.
4.2.3.3. Lundberg et al. (1987)
Lundberg et al. (1987) evaluated the tumor promoting activity of 1,4-dioxane in rat liver.
Male Sprague Dawley rats (8/dose group, 19 for control group) weighing 200 g underwent a
partial hepatectomy followed 24 hours later by an i.p. injection of 30 mg/kg diethylnitrosamine
(DEN) (initiation treatment). 1,4-Dioxane (99.5% pure with 25 ppm butylated hydroxytoluene
as a stabilizer) was then administered daily by gavage (in saline vehicle) at doses of 0, 100, or
1,000 mg/kg-day, 5 days/week for 7 weeks. Control rats were administered saline daily by
gavage, following DEN initiation. 1,4-Dioxane was also administered to groups of rats that were
not given the DEN initiating treatment (saline used instead of DEN). Ten days after the last
dose, animals were sacrificed and liver sections were stained for GGT. The number and total
volume of GGT-positive foci were determined.
1,4-Dioxane did not increase the number or volume of GGT-foci in rats that were not
given the DEN initiation treatment. The high dose of 1,4-dioxane (1,000 mg/kg-day) given as a
promoting treatment (i.e., following DEN injection) produced an increase in the number of
GGT-positive foci and the total foci volume. Histopathological changes were noted in the livers
of high-dose rats. Enlarged, foamy hepatocytes were observed in the midzonal region of the
liver, with the foamy appearance due to the presence of numerous fat-containing cytoplasmic
vacuoles. These results suggest that cytotoxic doses of 1,4-dioxane may be associated with
tumor promotion of 1,4-dioxane in rat liver.
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION
4.3.1. Giavini et al. (1985)
Pregnant female Sprague Dawley rats (18-20 per dose group) were given 1,4-dioxane
(99% pure, 0.7% acetal) by gavage in water at concentrations of 0, 0.25, 0.5, or 1 mL/kg-day,
corresponding to dose estimates of 0, 250, 500, or 1,000 mg/kg-day (density of 1,4-dioxane is
approximately 1.03 g/mL). The chemical was administered at a constant volume of 3 mL/kg on
days 6-15 of gestation. Food consumption was determined daily and BWs were measured every
3 days. The dams were sacrificed with chloroform on gestation day 21 and the numbers of
54
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
corpora lutea, implantations, resorptions, and live fetuses were recorded. Fetuses were weighed
and examined for external malformations prior to the evaluation of visceral and skeletal
malformations (Wilson's free-hand section method and staining with Alizarin red) and a
determination of the degree of ossification.
Maternal weight gain was reduced by 10% in the high-dose group (1,000 mg/kg-day).
Food consumption for this group was 5% lower during the dosing period, but exceeded control
levels for the remainder of the study. No change from control was observed in the number of
implantations, live fetuses, or resorptions; however, fetal birth weight was 5% lower in the
highest dose group (p < 0.01). 1,4-Dioxane exposure did not increase the frequency of major
malformations or minor anomalies and variants. Ossification of the sternebrae was reduced in
the 1,000 mg/kg-day dose group (p < 0.05). The study authors suggested that the observed delay
in sternebrae ossification combined with the decrease in fetal birth weight indicated a
developmental delay related to 1,4-dioxane treatment. NOAEL and LOAEL values of 500 and
1,000 mg/kg-day were identified from this study by EPA and based on delayed ossification of
the sternebrae and reduced fetal BWs.
4.4. OTHER DURATION OR ENDPOINT-SPECIFIC STUDIES
4.4.1. Acute and Short-term Toxicity
The acute (< 24 hours) and short-term toxicity studies (<30 days) of 1,4-dioxane in
laboratory animals are summarized in Table 4-15. Several exposure routes were employed in
these studies, including dermal application, drinking water exposure, gavage, vapor inhalation,
and i.v. or i.p. injection.
4.4.1.1.	Oral Toxicity
Mortality was observed in many acute high-dose studies, and LD50 values for
1,4-dioxane were calculated for rats, mice, and guinea pigs (see Table 4-15; Pozzani et al., 1959;
Smyth et al., 1941; Laug et al., 1939). Clinical signs of CNS depression were observed,
including staggered gait, narcosis, paralysis, coma, and death (Nelson, 1951; Laug et al., 1939;
Schrenk and Yant, 1936; de Navasquez, 1935). Severe liver and kidney degeneration and
necrosis were often seen in acute studies (JBRC, 1998b; David, 1964; Kesten et al., 1939; Laug
et al., 1939; Schrenk and Yant, 1936; de Navasquez, 1935). JBRC (1998b) additionally reported
histopathological lesions in the nasal cavity and the brain of rats following 2 weeks of exposure
to 1,4-dioxane in the drinking water.
4.4.1.2.	fn/ia/ation Toxicity
Acute and short-term toxicity studies (all routes) are summarized in Table 4-15.
Mortality occurred in many high-concentration studies (Pozzani et al., 1959; Nelson, 1951;
Wirth and Klimmer, 1936). Inhalation of 1,4-dioxane caused eye and nasal irritation, altered
55	DRAFT - DO NOT CITE OR QUOTE

-------
respiration, and pulmonary edema and congestion (Yant et al., 1930). Clinical signs of CNS
depression were observed, including staggered gait, narcosis, paralysis, coma, and death (Nelson,
1951; Wirth and Klimmer, 1936). Liver and kidney degeneration and necrosis were also seen in
acute and short-term inhalation studies (Drew et al., 1978; Fairley et al., 1934).
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)
Oral via
drinking water
1-10 Days of
exposure
Ultrastructural
changes in the
kidney, degenerative
nephrosis, hyaline
droplet accumulation,
crystal formation in
mitochondria
11,000 mg/kg-day
(5%)
David, 1964
Rat (strain and
gender unspecified)
Oral via
drinking water
5-12 Days of
exposure
Extensive
degeneration of the
kidney, liver damage,
mortality in
8/10 animals by
12 days
11,000 mg/kg-day
(5%)
Kesten et al.,
1939
F344/DuCrj rat
Oral via
drinking water
14-Day exposure
Mortality, decreased
BWs,
histopathological
lesions in the nasal
cavity, liver, kidney,
and brain
2,500 mg/kg-day
(nuclear
enlargement of
olfactory epithelial
cells),
>7,500 mg/kg-day
for all other effects
JBRC, 1998b
Female
Sprague Dawley rat
Gavage
0, 168, 840, 2550,
or 4,200 mg/kg by
gavage, 21 and
4 hours prior to
sacrifice
Increased ODC
activity, hepatic
CYP450 content, and
DNA single-strand
breaks
840 mg/kg (ODC
activity only)
Kitchin and
Brown, 1990
Female Carworth
Farms-Nelson rat
Gavage
Determination of a
single dose LD50
Lethality
LD50= 6,400 mg/kg
(14,200 ppm)
Pozzani et al.,
1959
Male Wistar rat,
guinea pig
Gavage
Single dose,
LD50 determination
Lethality
LD50 (mg/kg):
rat = 7,120
guinea pig = 3,150
Smyth et al.,
1941
Rat, mouse, guinea
Pig
Gavage
Single dose;
several dose
groups
Clinical signs of CNS
depression, stomach
hemorrhage, kidney
enlargement, and
liver and kidney
degeneration
LD50 (mg/kg):
mouse = 5,900
rat = 5,400
guinea pig = 4,030
Laug et al.,
1939
Rabbit
Gavage
Single gavage dose
of 0, 207, 1,034, or
2,068 mg/kg-day
Clinical signs of CNS
depression, mortality
at 2068 mg/kg, renal
toxicity (polyuria
followed by anuria),
histopathological
changes in liver and
kidneys
1,034 mg/kg-day
de
Navasquez,
1935
56	DRAFT - DO NOT CITE OR QUOTE

-------
Animal
Exposure route
Test conditions
Results
Dose3
Reference
Rat, rabbit
Gavage
Single dose;
mortality after
2 weeks
Mortality and
narcosis
3,160 mg/kg
Nelson, 1951
Crj:BDFi mouse
Oral via
drinking water
14-Day exposure
Mortality, decreased
BWs,
histopathological
lesions in the nasal
cavity, liver, kidney,
and brain
10,800 mg/kg-day;
hepatocellular
swelling
JBRC, 1998b
Dog
Drinking water
ingestion
3-10 Days of
exposure
Clinical signs of CNS
depression, and liver
and kidney
degeneration
11,000 mg/kg-day
(5%)
Schrenk and
Yant, 1936
Inhalation studies
Male CD1 rat
Vapor
inhalation
Serum enzymes
measured before
and after a single
4 hour exposure
Increase in ALT,
AST, and OCT; no
change in G-6-Pase
1,000 ppm
Drew et al.,
1978
Rat
Vapor
inhalation
5 Hours of
exposure
Mortality and
narcosis
6,000 ppm
Nelson, 1951
Female Carworth
Farms-Nelson rat
Vapor
inhalation
Determination of a
4-hour inhalation
LC50
Lethality
LC50 — 51.3 mg/L
Pozzani et al.,
1959
Mouse, cat
Vapor
inhalation
8 Hours/day for
17 days
Paralysis and death
8,400 ppm
Wirth and
Klimmer,
1936
Guinea pig
Vapor
inhalation
8-Hour exposure to
0.1-3% by volume
Eye and nasal
irritation, retching
movements, altered
respiration, narcosis,
pulmonary edema
and congestion,
hyperemia of the
brain
0.5% by volume
Yant et al.,
1930
Rabbit, guinea pig,
rat, mouse
Vapor
inhalation
3 Hours exposure,
for 5 days;
1.5 hour exposure
for 1 day
Degeneration and
necrosis in the kidney
and liver, vascular
congestion in the
lungs
10,000 ppm
Fairley et al.,
1934
Other routes
Male COBS/Wistar
rat
Dermal
Nonoccluded
technique using
shaved areas of the
back and flank;
single application,
14-day observation
Negative; no effects
noted
8,300 mg/kg
Clark et al.,
1984
Rabbit, cat
i.v. injection
Single injection of
0, 207, 1,034,
1,600 mg/kg-day
Clinical signs of CNS
depression, narcosis
at 1,034 mg/kg,
mortality at 1,600
mg/kg
1,034 mg/kg-day
de
Navasquez,
1935
57	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
Animal
Exposure route
Test conditions
Results
Dose3
Reference
Female
Sprague Dawley rat
i.p. injection
Single dose;
LD50 values
determined
24 hours and
14 days after
injection
Increased serum SDH
activity at 1/16th of
the LD50 dose; no
change at higher or
lower doses
LD50 (mg/kg):
24 hours = 4,848
14 days = 799
Lundberg
et al., 1986
CBA/J mouse
i.p. injection
Daily injection for
7 days, 0,0.1, 1,5,
and 10%
Slightly lower
lymphocyte response
to mitogens
2,000 mg/kg-day
(10%)
Thurman
et al., 1978
"Lowest 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
Clinical signs of CNS depression have been reported in humans and laboratory animals
following high dose exposure to 1,4-dioxane (see Sections 4.1 and 4.2.1.1). Neurological
symptoms were reported in the fatal case of a worker exposed to high concentrations of
1,4-dioxane through both inhalation and dermal exposure (Johnstone, 1959). These symptoms
included headache, elevation in blood pressure, agitation and restlessness, and coma. Autopsy
findings demonstrated perivascular widening in the brain, with small foci of demyelination in
several regions (e.g., cortex, basal nuclei). It was suggested that these neurological changes may
have been secondary to anoxia and cerebral edema. In laboratory animals, the neurological
effects of acute high-dose exposure included staggered gait, narcosis, paralysis, coma, and death
(Nelson, 1951; Laug et al., 1939; Schrenk and Yant, 1936; deNavasquez, 1935; Yant et al.,
1930). The neurotoxicity of 1,4-dioxane was further investigated in several studies described
below (Frantik et al., 1994; Kanada et al., 1994; Goldberg et al., 1964; Knoefel, 1935).
4.4.2.1. Fran//A e/ a/. (1994)
The acute neurotoxicity of 1,4-dioxane was evaluated following a 4-hour inhalation
exposure to male Wistar rats (four per dose group) and a 2-hour inhalation exposure to female
H-strain mice (eight per dose group). Three exposure groups and a control group were used in
this study. Exposure concentrations were not specified, but apparently were chosen from the
linear portion of the concentration-effect curve. The neurotoxicity endpoint measured in this
study was the inhibition of the propagation and maintenance of an electrically-evoked seizure
discharge. This endpoint has been correlated with the behavioral effects and narcosis that occur
following acute exposure to higher concentrations of organic solvents. Immediately following
1,4-dioxane exposure, a short electrical impulse was applied through ear electrodes (0.2 seconds,
50 hertz (Hz), 180 volts (V) in rats, 90 V in mice). Several time characteristics of the response
were recorded; the most sensitive and reproducible measures of chemically-induced effects were
58
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
determined to be the duration of tonic hind limb extension in rats and the velocity of tonic
extension in mice.
Linear regression analysis of the concentration-effect data was used to calculate an
isoeffective air concentration that corresponds to the concentration producing a 30% decrease in
the maximal response to an electrically-evoked seizure. The isoeffective air concentrations for
1,4-dioxane were 1,860 ± 200 ppm in rats and 2,400 ± 420 ppm in mice. A NOAEL value was
not identified from this study.
4.4.2.2.	Goldberg et a/. (1964)
Goldberg et al. (1964) evaluated the effect of solvent inhalation on pole climb
performance in rats. Female rats (Carworth Farms Elias strain) (eight per dose group) were
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
10 exposure days. Conditioned avoidance and escape behaviors were evaluated using a pole
climb methodology. Prior to exposure, rats were trained to respond to a buzzer or shock stimulus
by using avoidance/escape behavior within 2 seconds. Behavioral criteria were the abolishment
or significant deferment (>6 seconds) of the avoidance response (conditioned or buzzer response)
or the escape response (buzzer plus shock response). Behavioral tests were administered on day
1, 2, 3, 4, 5, and 10 of the exposure period. Rat BWs were also measured on test days.
1,4-Dioxane exposure produced a dose-related effect on conditioned avoidance behavior
in female rats, while escape behavior was generally not affected. In the 1,500 ppm group, only
one of eight rats had a decreased avoidance response, and this only occurred on days 2 and 5 of
exposure. A larger number of rats exposed to 3,000 ppm (two or three of eight) experienced a
decrease in the avoidance response, and this response was observed on each day of the exposure
period. The maximal decrease in the avoidance response was observed in the 6,000 ppm group
during the first 2 days of exposure (75-100% of the animals were inhibited in this response). For
exposure days 3-10, the percent of rats in the 6,000 ppm group with significant inhibition of the
avoidance response ranged from 37-62%. At the end of the exposure period (day 10), the BWs
for rats in the high exposure group were lower than controls.
4.4.2.3.	/Canada et a/. (7994)
Kanada et al. (1994) evaluated the effect of oral exposure to 1,4-dioxane on the regional
neurochemistry of the rat brain. 1,4-Dioxane was administered by gavage to male
Sprague Dawley rats (5/group) at a dose of 1,050 mg/kg, approximately equal to one-fourth the
oral LD50. Rats were sacrificed by microwave irradiation to the head 2 hours after dosing, and
brains were dissected into small brain areas. Each brain region was analyzed for the content of
biogenic amine neurotransmitters and their metabolites using high-performance liquid
chromatography (HPLC) or GC methods. 1,4-Dioxane exposure was shown to reduce the
dopamine and serotonin content of the hypothalamus. The neurochemical profile of all other
brain regions in exposed rats was similar to control rats.
59
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
4.4.2.4. K/toefe/(/9S5)
The narcotic potency of 1,4-dioxane was evaluated following i.p. injection in rats and
gavage administration in rabbits. Rats were given i.p. doses of 20, 30, or 50 mmol/kg. No
narcotic effect was seen at the lowest dose; however, rats given 30 mmol/kg were observed to
sleep approximately 8-10 minutes. Rats given the high dose of 50 mmol/kg died during the
study. Rabbits were given 1,4-dioxane at oral doses of 10, 20, 50, 75, or 100 mmol/kg. No
effect on the normal erect animal posture was observed in rabbits treated with less than
50 mmol/kg. At 50 and 75 mmol/kg, a semi-erect or staggering posture was observed; lethality
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
The genotoxicity data for 1,4-dioxane are presented in Table 4-16. 1,4-Dioxane has been
tested for genotoxic potential using in vitro assay systems with prokaryotic organisms, non-
mammalian eukaryotic organisms, and mammalian cells, and in vivo assay systems using several
strains of rats and mice. In the large majority of in vitro systems, 1,4-dioxane was not genotoxic.
Where a positive genotoxic response was observed, it was generally observed in the presence of
toxicity. Similarly, 1,4-dioxane was not genotoxic in the majority of available in vivo studies.
1,4-Dioxane did not bind covalently to DNA in a single study with calf thymus DNA. Several
investigators have reported that 1,4-dioxane caused increased DNA synthesis indicative of cell
proliferation. Overall, the available literature indicates that 1,4-dioxane is nongenotoxic or
weakly genotoxic.
Negative findings were reported for mutagenicity in in vitro assays with the prokaryotic
organisms Salmonella typhimurium, Escherichia coli, and Photobacterium phosphoreum
(Mutatox assay) (Morita and Hayashi, 1998; Hellmer and Bolcsfoldi, 1992; Kwan et al., 1990;
Khudoley et al., 1987; Nestmann et al., 1984; Haworth et al., 1983; Stott et al., 1981). In in vitro
assays with nonmammalian eukaryotic organisms, negative results were obtained for the
induction of aneuploidy in yeast (Saccharomyces cerevisiae) and in the sex-linked recessive
lethal test in Drosophila melanogaster (Yoon et al., 1985; Zimmerman et al., 1985). In the
presence of toxicity, positive results were reported for meiotic nondisjunction in Drosophila
(Munoz and Barnett, 2002).
The ability of 1,4-dioxane to induce genotoxic effects in mammalian cells in vitro has
been examined in model test systems with and without exogenous metabolic activation and in
hepatocytes that retain their xenobiotic-metabolizing capabilities. 1,4-Dioxane was reported as
negative in the mouse lymphoma cell forward mutation assay (Morita and Hayashi, 1998;
McGregor et al., 1991). 1,4-Dioxane did not produce chromosomal aberrations or micronucleus
60
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
formation in Chinese hamster ovary (CHO) cells (Morita and Hayashi, 1998; Galloway et al.,
1987). Results were negative in one assay for sister chromatid exchange (SCE) in CHO (Morita
and Hayashi, 1998) and were weakly positive in the absence of metabolic activation in another
(Galloway et al., 1987). In rat hepatocytes, 1,4-dioxane exposure in vitro caused single-strand
breaks in DNA at concentrations also toxic to the hepatocytes (Sina et al., 1983) and produced a
positive genotoxic response in a cell transformation assay with BALB/3T3 cells also in the
presence of toxicity (Sheu et al., 1988).
1,4-Dioxane was not genotoxic in the majority of available in vivo mammalian assays.
Studies of micronucleus formation following in vivo exposure to 1,4-dioxane produced mostly
negative results, including studies of bone marrow micronucleus formation in B6C3Fi, BALB/c,
CBA, and C57BL6 mice (McFee et al., 1994; Mirkova, 1994; Tinwell and Ashby, 1994) and
micronucleus formation in peripheral blood of CD1 mice (Morita and Hayashi, 1998; Morita,
1994). Mirkova (1994) reported a dose-related increase in the incidence of bone marrow
micronuclei in male and female C57BL6 mice 24 or 48 hours after administration of
1,4-dioxane. At a sampling time of 24 hours, a dose of 450 mg/kg produced no change relative
to control, while doses of 900, 1,800, and 3,600 mg/kg increased the incidence of bone marrow
micronuclei by approximately two-, three-, and fourfold, respectively. A dose of 5,000 mg/kg
also increased the incidence of micronuclei by approximately fourfold at 48 hours. This
compares with the negative results for BALB/c male mice tested in the same study at a dose of
5,000 mg/kg and sampling time of 24 hours. Tinwell and Ashby (1994) could not explain the
difference in response in the mouse bone marrow micronucleus assay with C57BL6 mice
obtained in their laboratory (i.e., nonsignificant 1.6-fold increase over control) with the dose-
related positive findings reported by Mirkova (1994) using the same mouse strain, 1,4-dioxane
dose (3,600 mg/kg) and sampling time (24 hours). Morita and Hayashi (1998) demonstrated an
increase in micronucleus formation in hepatocytes following 1,4-dioxane dosing and partial
hepatectomy to induce cellular mitosis. DNA single-strand breaks were demonstrated in
hepatocytes following gavage exposure to female rats (Kitchin and Brown, 1990).
Roy et al. (2005) examined micronucleus formation in male CD1 mice exposed to
1,4-dioxane to confirm the mixed findings from earlier mouse micronucleus studies and to
identify the origin of the induced micronuclei. Mice were administered 1,4-dioxane by gavage at
doses of 0, 1,500, 2,500, and 3,500 mg/kg-day for 5 days. The mice were also implanted with
5-bromo-2-deoxyuridine (BrdU)-releasing osmotic pumps to measure cell proliferation in the
liver and to increase the sensitivity of the hepatocyte assay. The frequency of micronuclei in the
bone marrow erythrocytes and in the proliferating BrdU-labeled hepatocytes was determined
24 hours after the final dose. Significant dose-related increases in micronuclei were seen in the
bone-marrow at all the tested doses (> 1,500 mg/kg-day). In the high-dose (3,500-mg/kg) mice,
the frequency of bone marrow erythrocyte micronuclei was about 10-fold greater than the control
frequency. Significant dose-related increases in micronuclei were also observed at the two
61
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
highest doses (> 2,500 mg/kg-day) in the liver. Antikinetochore (CREST) staining or
pancentromeric fluorescence in situ hybridization (FISH) was used to determine the origin of the
induced micronuclei. The investigators determined that 80-90% of the micronuclei in both
tissues originated from chromosomal breakage; small increase in micronuclei originating from
chromosome loss was seen in hepatocytes. Dose-related statistically significant decreases in the
ratio of bone marrow polychromatic erythrocytes (PCE):normochromatic erthyrocytes (NCE), an
indirect measure of bone marrow toxicity, were observed. Decreases in hepatocyte proliferation
were also observed. Based on these results, the authors concluded that at high doses 1,4-dioxane
exerts genotoxic effects in both the mouse bone marrow and liver; the induced micronuclei are
formed primarily from chromosomal breakage; and 1,4-dioxane can interfere with cell
proliferation in both the liver and bone marrow. The authors noted that reasons for the
discrepant micronucleus assay results among various investigators was unclear, but could be
related to the inherent variability present when detecting moderate to weak responses using small
numbers of animals, as well as differences in strain, dosing regimen, or scoring criteria.
1,4-Dioxane did not affect in vitro or in vivo DNA repair in hepatocytes or in vivo DNA
repair in the nasal cavity (Goldsworthy et al., 1991; Stott et al., 1981), but increased hepatocyte
DNA synthesis indicative of cell proliferation in several in vivo studies (Miyagawa et al., 1999;
Uno et al., 1994; Goldsworthy et al., 1991; Stott et al., 1981). 1,4-Dioxane caused a transient
inhibition of RNA polymerase A and B in the rat liver (Kurl et al., 1981), indicating a negative
impact on the synthesis of ribosomal and messenger RNA (DNA transcription). Intravenous
administration of 1,4-dioxane at doses of 10 or 100 mg/rat produced inhibition of both
polymerase enzymes, with a quicker and more complete recovery of activity for RNA
polymerase A, the polymerase for ribosomal RNA synthesis.
1,4-Dioxane did not covalently bind to DNA under in vitro study conditions (Woo et al.,
1977a). DNA alkylation was also not detected in the liver 4 hours following a single gavage
exposure (1,000 mg/kg) in male Sprague Dawley rats (Stott et al., 1981).
Rosenkranz and Klopman (1992) analyzed 1,4-dioxane using the computer automated
structure evaluator (CASE) structure activity method to predict its potential genotoxicity and
carcinogenicity. The CASE analysis is based on information contained in the structures of
approximately 3,000 chemicals tested for endpoints related to mutagenic/genotoxic and
carcinogenic potential. CASE selects descriptors (activating [biophore] or inactivating
[biophobe] structural fragments) from a learning set of active and inactive molecules. Using the
CASE methodology, Rosenkranz and Klopman (1992) predicted that 1,4-dioxane would be
inactive for mutagenicity in several in vitro systems, including Salmonella, induction of
chromosomal aberrations in CHO cells, and unscheduled DNA synthesis in rat hepatocytes.
1,4-Dioxane was predicted to induce SCE in cultured CHO cells, micronuclei formation in rat
bone marrow, and carcinogenicity in rodents.
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. As reported
6	by the authors (Van Delft et al., 2004), the gene expression profile for 1,4-dioxane indicated a
7	classification of this chemical as a "nongenotoxic" carcinogen, and thus, 1,4-dioxane was
8	included in the training set as a "nongenotoxic" carcinogen. The accuracy for carcinogen
9	classification using this method ranged from 33 to 100%, depending on which chemical data sets
10	and gene expression signals were included in the analysis.
Table 4-16a. Genotoxicity studies of 1,4-dioxane; in vitro
Test system
Endpoint
Test conditions
Results"
Doseb
Source
Without
activation
With
activation
Prokaryotic organisms in vitro
S. typhimurium
strains TA98, TA100,
TA1535, TA1537
Reverse
mutation
Plate incorporation
assay
—
—
10,000 (xg/plate
Haworth
et al., 1983
S. typhimurium
strains TA98, TA100,
TA1530, TA1535,
TA1537
Reverse
mutation
Plate incorporation
assay


ND
Khudoley
et al., 1987
S. typhimurium
strains TA98, TA100,
TA1535, TA1537
Reverse
mutation
Plate incorporation
and preincubation
assays
—
—
5,000 (xg/plate
Morita and
Hayashi,
1998
S. typhimurium
strains TA100,
TA1535
Reverse
mutation
Preincubation
assay
—
—
103 mg
Nestmann
et al., 1984
S. typhimurium
strains TA98, TA100,
TA1535, TA1537,
TA1538
Reverse
mutation
Plate incorporation
assay


103 mg
Stott et al.,
1981
E. coli K-12
uvrB/recA
DNA repair
Host mediated
assay
—
—
1,150 mmol/L
Hellmer and
Bolcsfoldi,
1992
E. coli
WP2/WP2uvrA
Reverse
mutation
Plate incorporation
and preincubation
assays
—
—
5,000 (xg/plate
Morita and
Hayashi,
1998
P. phosphoreum
M169
Mutagenicity,
DNA damage
Mutatox assay
-
ND
ND
Kwan et al.,
1990
63
DRAFT - DO NOT CITE OR QUOTE

-------
Test system
Endpoint
Test conditions
Results"
Doseb
Source
Without
activation
With
activation
Nonmammalian eukaryotic organisms in vitro
S. cerevisiae D61.M
Aneuploidy
Standard 16-hour
incubation or cold-
interruption
regimen
-T
ND
4.75%
Zimmerman
etal., 1985
D. melanogaster
Meiotic
nondisjunction
Oocytes were
obtained for
evaluation 24 and
48 hours after
mating
+T°
NDd
2% in sucrose
media
Munoz and
Barnett, 2002
D. melanogaster
Sex-linked
recessive lethal
test
Exposure by
feeding and
injection

NDd
35,000 ppm in
feed, 7 days or
50,000 ppm
(5% in water)
by injection
Yoon et al.,
1985
Mammalian cells in vitro
Rat hepatocytes
DNA damage;
single-strand
breaks measured
by alkaline
elution
3-Hour exposure
to isolated primary
hepatocytes
+Te
NDd
0.3 mM
Sina et al.,
1983
Primary hepatocyte
culture from male
F344 rats
DNA repair
Autoradiography
—
NDd
1 mM
Goldsworthy
etal., 1991
L5178Y mouse
lymphoma cells
Forward
mutation assay
Thymidine kinase
mutagenicity assay
(trifluorothymidin
e resistance)


5,000 ng/mL
McGregor
etal., 1991
L5178Y mouse
lymphoma cells
Forward
mutation assay
Thymidine kinase
mutagenicity assay
(trifluorothymidin
e resistance)

-T
5,000 (xg/mL
Morita and
Hayashi,
1998
BALB/3T3 cells
Cell
transformation
48-Hour exposure
followed by
4 weeks
incubation; 13 day
exposure followed
by 2.5 weeks
incubation
+Tf
NDd
0.5 mg/mL
Sheu et al.,
1988
64
DRAFT - DO NOT CITE OR QUOTE

-------
Test system
Endpoint
Test conditions
Results"
Doseb
Source
Without
activation
With
activation
CHO cells
SCE
BrdU was added
2 hours after
1,4-dioxane
addition; chemical
treatment was
2 hours with S9
and 25 hours
without S9
±g

10,520 (xg/mL
Galloway
etal., 1987
CHO cells
Chromosomal
aberration
Cells were
harvested 8-
12 hours or 18-
26 hours after
treatment (time of
first mitosis)


10,520 (xg/mL
Galloway
etal., 1987
CHO cells
SCE
3 Hour pulse
treatment;
followed by
continuous
treatment of BrdU
for 23 or 26 hours


5,000 (xg/mL
Morita and
Hayashi,
1998
CHO cells
Chromosomal
aberration
5 Hour pulse
treatment, 20 hour
pulse and
continuous
treatments, or 44
hour continuous
treatment; cells
were harvested 20
or 44 hours
following
exposure


5,000 (xg/mL
Morita and
Hayashi,
1998
65
DRAFT - DO NOT CITE OR QUOTE

-------
Test system
Endpoint
Test conditions
Results"
Doseb
Source
Without
activation
With
activation
CHO cells
Micronucleus
formation
5 Hour pulse
treatment or 44
hour continuous
treatment; cells
were harvested 42
hours following
exposure


5,000 (xg/mL
Morita and
Hayashi,
1998
Calf thymus DNA
Covalent
binding to DNA
Incubation with
microsomes from
3 -methy lcholanthr
ene treated rats


0.04 pmol/mg
DNA (bound)
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.
eCell 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).
fFor 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 ng/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 ng/L) in the absence of S9 mix
or at any concentration level (1,050, 3,500, 10,500 (ig/L) tested in the presence of S9.
Table 4-16b. Genotoxicity studies of 1,4-dioxane; mammalian in vivo
Test system
Endpoint
Test Conditions
Results
Dose
Source
Female
Sprague Dawley
Rat
DNA damage;
single-strand breaks
measured by alkaline
elution
Two gavage doses given 21
and 4 hours prior to
sacrifice
+h
2,550 mg/kg
Kitchin and
Brown, 1990
Male
Sprague Dawley
Rat
DNA alkylation in
hepatocytes
Gavage; DNA isolation and
HPLC analysis 4 hours after
dosing

1,000 mg/kg
Stott et al.,
1981
66
DRAFT - DO NOT CITE OR QUOTE

-------
Test system
Endpoint
Test Conditions
Results
Dose
Source
Male
B6C3Fi
Mouse
Micronucleus
formation in bone
marrow
i.p. injection; analysis of
polychromatic erythrocytes
24 or 48 hours after dosing

Single dose of
4,000 mg/kg;
3 daily doses of
2,000
McFee et al.,
1994
Male and female
C57BL6
Mouse;
male BALB/c
Mouse
Micronucleus
formation in bone
marrow
Gavage; analysis of
polychromatic erythrocytes
24 or 48 hours after dosing
+ (C57BL6)1
- (BALB/c)
900 mg/kg
(C57BL6);
5,000 mg/kg
(BALB/c)
Mirkova,
1994
Male
CD1
Mouse
Micronucleus
formation in
peripheral blood
Two i.p. injections (1/day);
micronucleated
reticulocytes measured 24,
48, and 72 hours after the
2nd dose

3,200 mg/kg
Morita, 1994
Male
CD1
Mouse
Micronucleus
formation in
hepatocytes
Gavage, partial
hepatectomy 24 hours after
dosing, hepatocytes
analyzed 5 days after
hepatectomy
+J
2,000 mg/kg
Morita and
Hayashi,
1998
Male
CD1
Mouse
Micronucleus
formation in
peripheral blood
Gavage, partial
hepatectomy 24 hours after
dosing, peripheral blood
obtained from tail vein 24
hours after hepatectomy

3,000 mg/kg
Morita and
Hayashi,
1998
Male
CBA and
C57BL6 Mouse
Micronucleus
formation in bone
marrow
Gavage; analysis of
polychromatic erythrocytes
from specimens prepared
24 hours after dosing

3,600 mg/kg
Tinwell and
Ashby, 1994
Male
CD1
Mouse
Micronuclei
formation in bone
marrow
Gavage; analysis for
micronucleated erythrocytes
24 hours after dosing
+k
1,500 mg/kg-day
for 5 days
Roy et al.,
2005
Male
CD1
Mouse
Micronuclei
formation in
hepatocytes
Gavage; analysis for
micronuclei 24 hours after
dosing
+1
2,500 mg/kg-day
for 5 days
Roy et al.,
2005
Male
Sprague Dawley
Rat
DNA repair in
hepatocytes
Drinking water; thymidine
incorporation with
hydroxyurea to repress
normal DNA synthesis

1,000 mg/kg-day
for 11 weeks
Stott et al.,
1981
Male
F344
Rat
DNA repair in
hepatocytes
(autoradiography)
Gavage and drinking water
exposure; thymidine
incorporation

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
Goldsworthy
et al., 1991
Male
F344
Rat
DNA repair in nasal
epithelial cells from
the nasoturbinate or
maxilloturbinate
Gavage and drinking water
exposure; thymidine
incorporation

1,500 mg/kg-day
for 8 days +
1,000 mg/kg
gavage dose
12 hours prior to
sacrifice
Goldsworthy
et al., 1991
Male
F344
Rat
Replicative DNA
synthesis (i.e., cell
proliferation) in
hepatocytes
Gavage and drinking water
exposure; thymidine
incorporation
+m
(1-2-week
exposure)
1,000 mg/kg for
24 or 48 hours;
1,500 mg/kg-day
for 1 or 2 weeks
Goldsworthy
et al., 1991
67	DRAFT - DO NOT CITE OR QUOTE

-------
Test system
Endpoint
Test Conditions
Results
Dose
Source
Male
F344
Rat
Replicative DNA
synthesis (i.e., cell
proliferation) in nasal
epithelial cells
Drinking water exposure;
thymidine incorporation

1,500 mg/kg-day
for 2 weeks
Goldsworthy
etal., 1991
Male
Sprague Dawley
Rat
RNA synthesis;
inhibition of RNA
polymerase A and B
i.v. injection; activity
measured in isolated
hepatocytes
+n
10 mg/rat
Kurl et al.,
1981
Male
F344
Rat
DNA synthesis in
hepatocytes
Gavage; thymidine and
BrdU incorporation
+°
1,000 mg/kg
Miyagawa
et al., 1999
Male
F344
Rat
DNA synthesis in
hepatocytes
Thymidine incorporation
±p
2,000 mg/kg
Uno et al.,
1994
Male
Sprague Dawley
Rat
DNA synthesis in
hepatocytes
Drinking water; thymidine
incorporation
+q
1,000 mg/kg-day
for 11 weeks
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.
mNo 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.
68
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
4.5.2. Mechanistic Studies
4.5.2.1.	Free HadicaiGeneration
Burmistrov et al. (2001) investigated the effect of 1,4-dioxane inhalation on free radical
processes in the rat ovary and brain. Female rats (6-9/group, unspecified strain) were exposed to
0, 10, or 100 mg/m of 1,4-dioxane vapor for 4 hours/day, 5 days/week, for 1 month. Rats were
sacrificed during the morning or evening following exposure and the ovaries and brain cortex
were removed and frozen. Tissue preparations were analyzed for catalase activity, glutathione
-3
peroxidase activity, and protein peroxidation. Inhalation of 100 mg/m of 1,4-dioxane resulted in
a significant increase (p<0.05) in glutathione peroxidase activity, and activation of free radical
processes were apparent in both the rat ovary and brain cortex. No change in catalase activity or
protein peroxidation was observed at either concentration. A circadian rhythm for glutathione
peroxidase activity was absent in control rats, but occurred in rat brain and ovary following
1,4-dioxane exposure.
4.5.2.2.	Induction ofMetabo/ism
The metabolism of 1,4-dioxane is discussed in detail in Section 3.3. 1,4-Dioxane has
been shown to induce its own metabolism (Young et al., 1978a, b). Nannelli et al. (2005) (study
details provided in Section 3.3) characterized the CYP450 isozymes that were induced by
1,4-dioxane in the liver, kidney, and nasal mucosa of the rat. In the liver, the activities of several
CYP450 isozymes were increased (i.e., CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1
was inducible in the kidney and nasal mucosa. CYP2E1 mRNA was increased approximately
two- to threefold in the kidney and nasal mucosa, but mRNA levels were not increased in the
liver, suggesting that regulation of CYP2E1 is organ-specific. Induction of hepatic CYPB1/2
and CYP2E1 levels by phenobarbital or fasting did not increase the liver toxicity of 1,4-dioxane,
as measured by hepatic glutathione content or serum ALT activity. This result suggested that
highly reactive and toxic intermediates did not play a large role in the liver toxicity of
1,4-dioxane, even under conditions where metabolism was enhanced. This finding is similar to
an earlier conclusion by Kociba et al. (1975) who evaluated toxicity from a chronic drinking
water study alongside data providing a pharmacokinetic profile for 1,4-dioxane. Kociba et al.
(1975) concluded that liver toxicity and eventual tumor formation occurred only at doses where
clearance pathways were saturated and elimination of 1,4-dioxane from the blood was reduced.
Nannelli et al. (2005) further suggested that a sustained induction of CYP2E1 may lead to
generation of reactive oxygen species contributing to target organ toxicity and regenerative cell
proliferation; however, no data were provided to support this hypothesis.
4.5.2.3.	Mechanisms of Tumor Induction
Several studies have been performed to evaluate potential mechanisms for the
carcinogenicity of 1,4-dioxane (Goldsworthy et al., 1991; Kitchin and Brown, 1990; Stott et al.,
69
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
1981). Stott et al. (1981) evaluated 1,4-dioxane in several test systems, including salmonella
mutagenicity in vitro, rat hepatocyte DNA repair activity in vitro, DNA synthesis determination
in male Sprague Dawley rats following acute gavage dosing or an 11-week drinking water
exposure (described in Section 4.2.1), and hepatocyte DNA alkylation and DNA repair following
a single gavage dose. This study used doses of 0, 10, 100, or 1,000 mg/kg-day, with the highest
dose considered to be a tumorigenic dose level. Liver histopathology and liver to BW ratios
were also evaluated in rats from acute gavage or repeated dose drinking water experiments.
The histopathology evaluation indicated that liver cytotoxicity (i.e., centrilobular
hepatocyte swelling) was present in rats from the 1,000 mg/kg-day dose group that received
1,4-dioxane in the drinking water for 11 weeks (Stott et al., 1981). An increase in the liver to
BW ratio accompanied by an increase in hepatic DNA synthesis was also seen in this group of
animals. No effect on histopathology, liver weight, or DNA synthesis was observed in acutely
exposed rats or rats that were exposed to a lower dose of 10 mg/kg-day for 11 weeks.
1,4-Dioxane produced negative findings in the remaining genotoxicity assays conducted as part
of this study (i.e., Salmonella mutagenicity, in vitro and in vivo rat hepatocyte DNA repair, and
DNA alkylation in rat liver). The study authors suggested that the observed lack of genotoxicity
at tumorigenic and cytotoxic dose levels indicates an epigenetic mechanism for 1,4-dioxane
hepatocellular carcinoma in rats.
Goldsworthy et al. (1991) evaluated potential mechanisms for the nasal and liver
carcinogenicity of 1,4-dioxane in the rat. DNA repair activity was evaluated as a measure of
DNA reactivity and DNA synthesis was measured as an indicator of cell proliferation or
promotional activity. In vitro DNA repair was evaluated in primary hepatocyte cultures from
control and 1,4-dioxane-treated rats (1 or 2% in the drinking water for 1 week). DNA repair and
DNA synthesis were also measured in vivo following a single gavage dose of 1,000 mg/kg, a
drinking water exposure of 1% (1,500 mg/kg-day) for 1 week, or a drinking water exposure of
2% (3,000 mg/kg-day) for 2 weeks. Liver to BW ratios and palmitoyl CoA oxidase activity were
measured in the rat liver to determine whether peroxisome proliferation played a role in the liver
carcinogenesis of 1,4-dioxane. In vivo DNA repair was evaluated in rat nasal epithelial cells
derived from either the nasoturbinate or the maxilloturbinate of 1,4-dioxane-treated rats. These
rats received 1% 1,4-dioxane (1,500 mg/kg-day) in the drinking water for 8 days, followed by a
single gavage dose of 10, 100, or 1,000 mg/kg 12 hours prior to sacrifice. Archived tissues from
the NCI (1978) bioassay were reexamined to determine the primary sites for tumor formation in
the nasal cavity following chronic exposure in rats. Histopathology and cell proliferation were
determined for specific sites in the nasal cavity that were related to tumor formation. This
evaluation was performed in rats that were exposed to drinking water containing 1% 1,4-dioxane
(1,500 mg/kg-day) for 2 weeks.
1,4-Dioxane and its metabolite l,4-dioxane-2-one did not affect in vitro DNA repair in
primary hepatocyte cultures (Goldsworthy et al., 1991). In vivo DNA repair was also unaffected
70
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
by acute gavage exposure or ingestion of 1,4-dioxane in the drinking water for a 1- or 2-week
period. Hepatocyte cell proliferation was not affected by acute gavage exposure, but was
increased approximately twofold following a 1-2-week drinking water exposure. A 5-day
drinking water exposure to 1% 1,4-dioxane (1,500 mg/kg-day) did not increase the activity of
palmitoyl coenzyme A or the liver to BW ratio, suggesting that peroxisome proliferation did not
play a role in the hepatocarcinogenesis of 1,4-dioxane. Nannelli et al. (2005) also reported a lack
of hepatic palmitoyl CoA induction following 10 days of exposure to 1.5% 1,4-dioxane in the
drinking water (2,100 mg/kg-day).
Treatment of rats with 1% (1,500 mg/kg-day) 1,4-dioxane for 8 days did not alter DNA
repair in nasal epithelial cells (Goldsworthy et al., 1991). The addition of a single gavage dose
of up to 1,000 mg/kg 12 hours prior to sacrifice also did not induce DNA repair. Reexamination
of tissue sections from the NCI (1978) bioassay suggested that the majority of nasal tumors were
located in the dorsal nasal septum or the nasoturbinate of the anterior portion of the dorsal
meatus (Goldsworthy et al., 1991). No histopathological lesions were observed in nasal section
of rats exposed to drinking water containing 1% 1,4-dioxane (1,500 mg/kg-day) for 2 weeks and
no increase was observed in cell proliferation at the sites of highest tumor formation in the nasal
cavity.
Female Sprague Dawley rats (three to nine per group) were given 0, 168, 840, 2,550, or
4,200 mg/kg 1,4-dioxane (99% purity) by corn oil gavage in two doses at 21 and 4 hours prior to
sacrifice (Kitchin and Brown, 1990). DNA damage (single-strand breaks measured by alkaline
elution), ODC activity, reduced glutathione content, and CYP450 content were measured in the
liver. Serum ALT activity and liver histopathology were also evaluated. No changes were
observed in hepatic reduced glutathione content or ALT activity. Light microscopy revealed
minimal to mild vacuolar degeneration in the cytoplasm of hepatocytes from three of five rats
from the 2,550 mg/kg dose group. No histopathological lesions were seen in any other dose
group, including rats given a higher dose of 4,200 mg/kg. 1,4-Dioxane caused 43 and 50%
increases in DNA single-strand breaks at dose levels of 2,550 and 4,200 mg/kg, respectively.
CYP450 content was also increased at the two highest dose levels (25 and 66% respectively).
ODC activity was increased approximately two-, five-, and eightfold above control values at
doses of 840, 2,550, and 4,200 mg/kg, respectively. The results of this study demonstrated that
hepatic DNA damage can occur in the absence of significant cytotoxicity. Parameters associated
with tumor promotion (i.e., ODC activity, CYP450 content) were also elevated, suggesting that
promotion may play a role in the carcinogenesis of 1,4-dioxane.
4.6. SYNTHESIS OF MAJOR NONCANCER EFFECTS
Liver and kidney toxicity were the primary noncancer health effects associated with
exposure to 1,4-dioxane in humans and laboratory animals. Several fatal cases of hemorrhagic
nephritis and centrilobular necrosis of the liver were related to occupational exposure (i.e.,
71
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
inhalation and dermal contact) to 1,4-dioxane (Johnstone, 1959; Barber, 1934). Neurological
changes were also reported in one case; including, headache, elevation in blood pressure,
agitation and restlessness, and coma (Johnstone, 1959). Perivascular widening was observed in
the brain of this worker, with small foci of demyelination in several regions (e.g., cortex, basal
nuclei). Liver and kidney degeneration and necrosis were observed in acute oral and inhalation
studies (JBRC, 1998b; Drew et al., 1978; David, 1964; Kesten et al., 1939; Laug et al., 1939;
Schrenk and Yant, 1936; de Navasquez, 1935; Fairley et al., 1934). The results of subchronic
and chronic studies are discussed below.
4.6.1. Oral
Table 4-17 presents a summary of the noncancer results for the subchronic and chronic
oral studies of 1,4-dioxane toxicity in experimental animals. Liver and kidney toxicity were the
primary noncancer health effects of oral exposure to 1,4-dioxane in animals. Kidney damage at
high doses was characterized by degeneration of the cortical tubule cells, necrosis with
hemorrhage, and glomerulonephritis (NCI, 1978; Kociba et al., 1974; Argus et al., 1973, 1965;
Fairley et al., 1934). Renal cell degeneration generally began with cloudy swelling of cells in the
cortex (Fairley et al., 1934). Nuclear enlargement of proximal tubule cells was observed at doses
below those producing renal necrosis (Kano et al., 2008; JBRC, 1998a), but is of uncertain
toxicological significance. The lowest dose reported to produce kidney damage was 94 mg/kg-
day, which produced renal degeneration and necrosis of tubule epithelial cells in male rats in the
Kociba et al. (1974) study. Cortical tubule degeneration was seen at higher doses in the NCI
(1978) bioassay (240 mg/kg-day, male rats), and glomerulonephritis was reported for rats given
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
Rats 0 or 1,900 mg/kg-
day; mice 0 or
3,300 mg/kg-day for
67 days
NA
1,900 rats
3,300 mice
Renal cortical degeneration
and necrosis, hemorrhage;
hepatocellular degeneration
Fairley et al.,
1934
Male
Sprague Dawley
Rat
(4-6/group)
0, 10, or 1,000 mg/kg-day
for 11 weeks
10
1,000
Minimal centrilobular
hepatocyte swelling;
increased DNA synthesis
Stott et al.,
1981
F344/DuCrj rat
(10/sex/group)
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
52
126
Nuclear enlargement of
nasal respiratory
epithelium; hepatocyte
swelling
Kano et al.,
2008
72
DRAFT - DO NOT CITE OR QUOTE

-------
Species
Dose/duration
NOAEL
(mg/kg-day)
LOAEL
(mg/kg-day)
Effect
Reference
Crj:BDFi Mouse
(10/sex/group)
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
170
387
Nuclear enlargement of
bronchial epithelium
Kano et al.,
2008
Chronic studies
Male
Wistar
Rat (26 treated,
9 controls)
0 or 640 mg/kg-day for
63 weeks
NA
640
Hepatocytes with enlarged
hyperchromic nuclei;
glomerulonephritis
Argus et al.,
1965
Male
Sprague Dawley
rats (30/group)
0, 430, 574, 803, or
1,032 mg/kg-day for
13 months
NA
430
Hepatocytomegaly;
glomerulonephritis
Argus et al.,
1973
Sherman rat
(60/sex/dose
group)
Males 0, 9.6, 94, or
1,015 mg/kg-day; females
0, 19, 148, or
1,599 mg/kg-day for
2 years
9.6
94
Degeneration and necrosis
of renal tubular cells and
hepatocytes
Kociba et al.,
1974
Osborne-Mendel
rat (35/sex/dose
level)
Males 0, 240, or
530 mg/kg-day; females
0, 350, or 640 mg/kg-day
for 110 weeks
NA
240
Pneumonia, gastric ulcers,
and cortical tubular
degeneration in the kidney
NCI, 1978
B6C3Fi mouse
(50/sex/dose
level)
Males 0, 720, or
830 mg/kg-day; females
0, 380, or 860 mg/kg-day
for 90 weeks
NA
380
Pneumonia and rhinitis
NCI, 1978
F344/DuCrj rat
(50/sex/dose
level)
Males 0, 11, 55, or
274 mg/kg-day; females
0, 18, 83, or 429 mg/kg-
day for 2 years
55
274
Atrophy of nasal olfactory
epithelium; nasal adhesion
and inflammation
Kano et al.,
2009;
JBRC, 1998a
F344/DuCrj rat
(50/sex/dose
level)
Males 0, 11, 55, or
274 mg/kg-day; females
0, 18, 83, or 429 mg/kg-
day for 2 years
11
55
Liver hyperplasia
Kano et al.,
2009; JBRC,
1998a
F344/DuCrj rat
(50/sex/dose
level)
Males 0, 11, 55, or
274 mg/kg-day; females
0, 18, 83, or 429 mg/kg-
day for 2 years
55
274
Increases in serum liver
enzymes (GOT, GPT, LDH,
and ALP)
Kano et al.,
2009; JBRC,
1998a
Crj:BDFi mouse
(50/sex/dose
level)
Males 0, 49, 191 or
677 mg/kg-day; females
0, 66, 278, or 967 mg/kg-
day for 2 years
66
278
Nasal inflammation
Kano et al.,
2009; JBRC,
1998a
Crj:BDFi mouse
(50/sex/dose
level)
Males 0, 49, 191 or
677 mg/kg-day; females
0, 66, 278, or 967 mg/kg-
day for 2 years
49
191
Increases in serum liver
enzymes (GOT, GPT, LDH,
and ALP)
Kano et al.,
2009; 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
73	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
Liver effects included degeneration and necrosis, hepatocyte swelling, cells with
hyperchromic nuclei, spongiosis hepatis, hyperplasia, and clear and mixed cell foci of the liver
(Kano et al., 2008; NCI, 1978; Kociba et al., 1974; Argus et al., 1965, 1973; Fairley et al., 1934).
Hepatocellular degeneration and necrosis were seen at high doses in a subchronic study
(1,900 mg/kg-day in rats) (Fairley et al., 1934) and at lower doses in a chronic study
(94 mg/kg-day, male rats) (Kociba et al., 1974). Argus et al. (1973) described a progression of
preneoplastic effects in the liver of rats exposed to a dose of 575 mg/kg-day. Early changes
(8 months exposure) were described as an increased nuclear size of hepatocytes, disorganization
of the rough endoplasmic reticulum, an increase in smooth endoplasmic reticulum, a decrease in
glycogen, an increase in lipid droplets in hepatocytes, and formation of liver nodules.
Spongiosis hepatis, hyperplasia, and clear and mixed-cell foci were also observed in the liver of
rats (doses > 55 mg/kg-day in male rats) (Kano et al., 2009; JBRC, 1998a). Clear and mixed-cell
foci are commonly considered preneoplastic changes and would not be considered evidence of
noncancer toxicity when observed in conjunction with tumor formation. If exposure to
1,4-dioxane had not resulted in tumor formation, these lesions could represent potential
noncancer toxicity. The nature of spongiosis hepatis as a preneoplastic change is less well
understood (Bannash, 2003; Karbe and Kerlin, 2002; Stroebel et al., 1995). Spongiosis hepatis is
a cyst-like lesion that arises from the perisinusoidal Ito cells of the liver. This change is
sometimes associated with hepatocellular hypertrophy and liver toxicity (Karbe and Kerlin,
2002), but may also occur in combination with preneoplastic foci, or hepatocellular adenoma or
carcinoma (Bannash et al., 2003; Stroebel et al., 1995). In the case of the JBRC (1998a) study,
spongiosis hepatis was associated with other preneoplastic changes in the liver (hyperplasia,
clear and mixed-cell foci). No other lesions indicative of liver toxicity were seen in this study;
therefore, spongiosis hepatis was not considered indicative of noncancer effects. The activity of
serum enzymes (i.e., AST, ALT, LDH, and ALP) was increased in rats and mice exposed to
1,4-dioxane, although only in groups with high incidence of liver tumors. Blood samples were
collected only at the end of the 2-year study, so altered serum chemistry may be associated with
the tumorigenic changes in the liver.
Hematological changes were reported in the JBRC (1998a) study only. Mean doses are
reported based on information provided in Kano et al. (2009). Observed increases in RBCs,
hematocrit, hemoglobin in high-dose male mice (677 mg/kg-day) may be related to lower
drinking water consumption (74% of control drinking water intake). Hematological effects
noted in male rats given 55 mg/kg-day (decreased RBCs, hemoglobin, hematocrit, increased
platelets) were within 20% of control values. A reference range database for hematological
effects in laboratory animals (Wolford et al., 1986) indicates that a 20% change in these
parameters may fall within a normal range (10th-90th percentile values) and may not represent a
treatment-related effect of concern.
74
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
Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice
only, dose >380 mg/kg-day) and JBRC (1998a) studies (>274 mg/kg-day in rats, >278 mg/kg-
day in mice). The JBRC (1998a) study also demonstrates atrophy of the nasal epithelium and
adhesion in rats and mice. Nasal inflammation may be a response to direct contact of the nasal
mucosa with drinking water containing 1,4-dioxane (Sweeney et al., 2008; Goldsworthy et al.,
1991) or could result from systemic exposure. Regardless, inflammation may indicate toxicity
due to 1,4-dioxane exposure. A significant increase in the incidence of pneumonia was reported
in mice from the NCI (1978) study. The significance of this effect is unclear, as it was not
observed in other studies that evaluated lung histopathology (Kano et al., 2008; JBRC, 1998a;
Kociba et al., 1974). No studies were available regarding the potential for 1,4-dioxane to cause
immunological effects. Metaplasia and hyperplasia of the nasal epithelium were also observed in
high-dose male and female rats (JBRC, 1998a); however, these effects are likely to be associated
with the formation of nasal cavity tumors in these dose groups. Nuclear enlargement of the nasal
olfactory epithelium was observed at a dose of 83 mg/kg-day in female rats (Kano et al., 2009);
however, it is unclear whether this alteration represents an adverse toxicological effect. Nuclear
enlargement of the tracheal and bronchial epithelium and an accumulation of foamy cells in the
lung were also seen in male and female mice give 1,4-dioxane at doses of >278 mg/kg for
2 years (JBRC, 1998a).
4.6.2. Inhalation
Only one subchronic study (Fairley et al., 1934) and one chronic inhalation study
(Torkelson et al., 1974) were identified. In the subchronic study, 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 1.5 hours two times a day for 5 days, 1.5 hours for one day, and no exposure on the
seventh day. Animals were exposed until death occurred or were sacrificed after various
durations of exposure (3-202.5 hours). Detailed dose-response information was not provided;
however, severe liver and kidney damage and acute vascular congestion of the lungs were noted
for all exposure concentrations tested. Kidney damage was described as patchy degeneration of
cortical tubules with vascular congestion and hemorrhage. Liver lesions varied from cloudy
hepatocyte swelling to large areas of necrosis. Torkelson et al. (1974) performed a chronic
inhalation study in which male and female Wistar rats (288/sex) were exposed to 111 ppm
1,4-dioxane vapor for 7 hours/day, 5 days/week for 2 years. Control rats (192/sex) were exposed
to filtered air. No significant effects were observed on BWs, survival, organ weights,
hematology, clinical chemistry, or histopathology. These studies were not sufficient to
characterize the inhalation risks of 1,4-dioxane, due to the nature of the available data (i.e., free-
standing LOAEL and NOAEL values).
75
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
4.6.3. Mode of Action Information
The metabolism of 1,4-dioxane in humans was extensive at low doses (<50 ppm). The
linear elimination of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism
was a nonsaturated, first-order process at this exposure level (Young et al., 1977, 1976). Like
humans, rats extensively metabolized inhaled 1,4-dioxane; however, plasma data from rats given
single i.v. doses of 3, 10, 30, 100, or 1,000 mg [14C]-l,4-dioxane/kg demonstrated a dose-related
shift from linear, first-order to nonlinear, saturable metabolism of 1,4-dioxane (Young et al.,
1978a, b).
1,4-Dioxane oxidation appeared to be CYP450-mediated, as CYP450 induction with
phenobarbital or Aroclor 1254 and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine
or cobaltous chloride was effective in significantly increasing and decreasing, respectively, the
appearance of HEAA in the urine of rats (Woo et al., 1978, 1977c). 1,4-Dioxane itself induced
CYP450-mediated metabolism of several barbiturates in Hindustan mice given i.p. injections of
25 and 50 mg/kg of 1,4-dioxane (Mungikar and Pawar, 1978). The differences between single
and multiple doses in urinary and expired radiolabel support the notion that 1,4-dioxane may
induce its own metabolism. 1,4-Dioxane has been shown to induce several isoforms of CYP450
in various tissues following acute oral administration by gavage or drinking water (Nannelli
et al., 2005). In the liver, the activity of several CYP450 isozymes was increased (i.e.,
CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was inducible in the kidney and nasal
mucosa. CYP2E1 mRNA was increased approximately two- to threefold in the kidney and nasal
mucosa, but mRNA levels were not increased in the liver, suggesting that regulation of CYP2E1
was organ-specific.
Nannelli et al. (2005) investigated the role of CYP450 isozymes in the liver toxicity of
1,4-dioxane. Hepatic CYPB1/2 and CYP2E1 levels were induced by phenobarbital or fasting
and liver toxicity was measured as hepatic glutathione content or serum ALT activity. No
increase in glutathione content or ALT activity was observed, suggesting that highly reactive and
toxic intermediates did not play a large role in the liver toxicity of 1,4-dioxane, even under
conditions where metabolism was enhanced. Pretreatment with inducers of mixed-function
oxidases also did not significantly change the extent of covalent binding in subcellular fractions
(Woo et al., 1977a). Covalent binding was measured in liver, kidney, spleen, lung, colon, and
skeletal muscle 1-12 hours after i.p. dosing with 1,4-dioxane. Covalent binding was highest in
liver, spleen, and colon. Within hepatocytes, 1,4-dioxane distribution was greatest in the
cytosolic fraction, followed by the microsomal, mitochondrial, and nuclear fractions.
The absence of an increase in toxicity following an increase in metabolism suggests that
accumulation of the parent compound may be related to 1,4-dioxane toxicity. This hypothesis is
supported by a comparison of the pharmacokinetic profile of 1,4-dioxane with the toxicology
data from a chronic drinking water study (Kociba et al., 1975). This analysis indicated that liver
76
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
toxicity did not occur unless clearance pathways were saturated and elimination of 1,4-dioxane
from the blood was reduced. Alternative metabolic pathways (i.e., not CYP450 mediated) may
be present at high doses of 1,4-dioxane; however, the available studies have not characterized
these pathways or identified any possible reactive intermediates. The mechanism by which
1,4-dioxane induces tissue damage is not known, nor is it known whether the toxic moiety is 1,4-
dioxane or a transient or terminal metabolite.
4.7. EVALUATION OF CARCINOGENICITY
4.7.1. Summary of Overall Weight of Evidence
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), 1,4-dioxane
can be described as "likely to be carcinogenic to humans," based on evidence of liver
carcinogenicity in several 2-year bioassays conducted in three strains of rats, two strains of mice,
and in guinea pigs (Kano et al., 2009; JBRC, 1998a; Yamazaki, et al., 1994; NCI, 1978; Kociba
et al., 1974; Argus et al., 1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus
et al., 1965). Additionally, mesothiolomas of the peritoneum (Kano et al., 2009; JBRC, 1998a;
Yamazaki et al., 1994), mammary (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994), and
nasal tumors (Kano et al., 2009; JBRC, 1998a; Yamazaki, et al., 1994; NCI, 1978; Kociba et al.,
1974; Argus et al., 1973; Hoch-Ligeti et al., 1970) have been observed in rats due to exposure to
1,4-dioxane. Studies in humans are inconclusive regarding evidence for a causal link between
occupational exposure to 1,4-dioxane and increased risk for cancer; however, only two studies
were available and these were limited by small cohort size and a small number of reported cancer
cases (Buffler et al., 1978; Thiess et al., 1976).
The available evidence is inadequate to establish a mode of action (MOA) by which
1,4-dioxane induces liver tumors in rats and mice. A MO A hypothesis involving sustained
proliferation of spontaneously transformed liver cells has some support from data indicating that
1,4-dioxane acts as a tumor promoter in mouse skin and rat liver bioassays (Lundberg
et al.,1987; King et al., 1973). Dose-response and temporal data support the occurrence of cell
proliferation and hyperplasia prior to the development of liver tumors (JBRC, 1998a; Kociba
et al., 1974) in the rat model. However, the dose-response relationship for induction of hepatic
cell proliferation has not been characterized, and it is unknown if it would reflect the dose-
response relationship for liver tumors in the 2-year rat and mouse studies. Conflicting data from
rat and mouse bioassays (JBRC, 1998a; Kociba et al., 1974) suggest that cytotoxicity may not be
a required precursor event for 1,4-dioxane-induced cell proliferation. Data regarding a plausible
dose response and temporal progression (see Table 4-18) from cytotoxicity and cell proliferation
to eventual liver tumor formation are not available.
77
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
The MO A by which 1,4-dioxane produces liver, nasal, peritoneal (mesothiolomas), and
mammary gland tumors is unknown, and the available data do not support any hypothesized
carcinogenic MO A for 1,4-dioxane.
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence
Human studies of occupational exposure to 1,4-dioxane were inconclusive; in each case,
the cohort size and number of reported cases were of limited size (Buffler et al., 1978; Thiess
etal., 1976).
Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and
guinea pigs (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994; NCI, 1978; Kociba et al.,
1974; Torkelson et al., 1974; Argus et al., 1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti
et al., 1970; Argus et al., 1965). Liver tumors have been observed following drinking water
exposure in male Wistar rats (Argus et al., 1965), male guinea pigs (Hoch-Ligeti and Argus,
1970), male Sprague Dawley rats (Argus et al., 1973; Hoch-Ligeti et al., 1970), male and female
Sherman rats (Kociba et al., 1974), female Osborne-Mendel rats (NCI, 1978), male and female
F344/DuCij rats (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994), male and female
B6C3Fi mice (NCI, 1978), and male and female Crj:BDFi mice (Kano et al., 2009; JBRC,
1998a, Yamazaki et al., 1994). In the earliest cancer bioassays, the liver tumors were described
as hepatomas (Argus et al., 1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus
et al., 1965); however, later studies made a distinction between hepatocellular carcinoma and
hepatocellular adenoma (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994; NCI, 1978;
Kociba et al., 1974). Both tumor types have been seen in rats and mice exposed to 1,4-dioxane.
Kociba et al. (1974) noted evidence of liver toxicity at or below the dose levels that produced
liver tumors but did not report incidence data for these effects. Hepatocellular degeneration and
necrosis were observed in the mid- and high-dose groups of male and female Sherman rats
exposed to 1,4-dioxane, while tumors were only observed at the highest dose. Hepatic
regeneration was indicated in the mid- and high-dose groups by the formation of hepatocellular
hyperplastic nodules. Findings from JBRC (1998a) also provided evidence of liver hyperplasia
in male F344/DuCrj rats at a dose level below the dose that induced a statistically significant
increase in tumor formation.
Nasal cavity tumors were also observed in Sprague Dawley rats (Argus et al., 1973;
Hoch-Ligeti et al., 1970), Osborne-Mendel rats (NCI, 1978), Sherman rats (Kociba et al., 1974),
and F344/DuCij rats (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994). Most tumors
were characterized as squamous cell carcinomas. Nasal tumors were not elevated in B6C3Fi or
Cij :BDFi mice. JBRC (1998a) was the only study that evaluated nonneoplastic changes in nasal
cavity tissue following prolonged exposure to 1,4-dioxane in the drinking water.
Histopathological lesions in female F344/DuCrj rats were suggestive of toxicity and regeneration
in this tissue (i.e., atrophy, adhesion, inflammation, nuclear enlargement, and hyperplasia and
78	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
metaplasia of respiratory and olfactory epithelium). Some of these effects occurred at a lower
dose (83 mg/kg-day) than that shown to produce nasal cavity tumors (429 mg/kg-day) in female
rats. Reexamination of tissue sections from the NCI (1978) bioassay suggested that the majority
of nasal tumors were located in the dorsal nasal septum or the nasoturbinate of the anterior
portion of the dorsal meatus. Nasal tumors were not observed in an inhalation study in Wistar
rats exposed to 111 ppm for 5 days/week for 2 years (Torkelson et al., 1974).
Tumor initiation and promotion studies in mouse skin and rat liver suggested that
1,4-dioxane does not initiate the carcinogenic process, but instead acts as a tumor promoter
(Lundberg et al., 1987; Bull et al., 1986; King et al., 1973) (see Section 4.2.3).
In addition to the liver and nasal tumors observed in several studies, a statistically
significant increase in mesotheliomas of the peritoneum was seen in male rats from the Kano et
al. (2009) study (also JBRC, 1998a; Yamazaki et al., 1994). Female rats dosed with 429 mg/kg-
day in drinking water for 2 years also showed a statistically significant increase in mammary
gland adenomas (Kano et al., 2009; JBRC, 199a; Yamazaki, et al., 1994). A significant increase
in the incidence of these tumors was not observed in other chronic oral bioassays of 1,4-dioxane
(NCI, 1978; Kociba et al., 1974).
4.7.3. Mode of Action Information
The MO A by which 1,4-dioxane produces liver, nasal, peritoneal (mesothiolomas), and
mammary gland tumors is unknown, and the available data do not support any hypothesized
mode of carcinogenic action for 1,4-dioxane. Available data also do not clearly identify whether
1,4-dioxane or one of its metabolites is responsible for the observed effects. The hypothesized
MO As for 1,4-dioxane carcinogenicity are discussed below within the context of the modified
Hill criteria of causality as recommended in the most recent Agency guidelines (U.S. EPA,
2005a). MOA analyses were not conducted for peritoneal or mammary gland tumors due to the
absence of any chemical specific information for these tumor types.
4.7.3.1. Identification ofKey Events for Carcinogenicity
4.7.3.1.1. Liver. A key event in this MOA hypothesis is sustained proliferation of
spontaneously transformed liver cells, resulting in the eventual formation of liver tumors.
Precursor events in which 1,4-dioxane may promote proliferation of transformed liver cells are
uncertain. One study suggests that induced liver cytotoxicity may be a key precursor event to
cell proliferation leading to the formation of liver tumors (Kociba et al., 1974), however, this
study did not report incidence data for these effects. Other studies suggest that cell proliferation
can occur in the absence of liver cytotoxicity. Liver tumors were observed in female rats and
female mice in the absence of lesions indicative of cytotoxicity (Kano et al., 2008; JBRC, 1998a;
NCI, 1978). Figure 4-1 presents a schematic representation of possible key events in the MOA
for 1,4-dioxane liver carcinogenicity. These include: (1) oxidation by CYP2E1 and CYP2B1/2
79	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
(i.e., detoxification pathway for 1,4-dioxane), (2) saturation of metabolism/clearance leading to
accumulation of the parent 1,4-dioxane, (3) liver damage followed by regenerative cell
proliferation, or (4) cell proliferation in the absence of cytotoxicity (i.e., mitogenesis),
(5) hyperplasia, and (6) tumor formation. It is suggested that liver toxicity is related to the
accumulation of the parent compound following metabolic saturation at high doses (Kociba
et al., 1975); however, no in vivo or in vitro assays have examined the toxicity of metabolites
resulting from 1,4-dioxane to support this hypothesis. Nanelli et al. (2005) demonstrated that an
increase in the oxidative metabolism of 1,4-dioxane via CYP450 induction using phenobarbital
or fasting does not result in an increase in liver toxicity. This result suggested that highly
reactive and toxic intermediates did not play a large role in the liver toxicity of 1,4-dioxane, even
under conditions where metabolism was enhanced. Alternative metabolic pathways (e.g., not
CYP450 mediated) may be present at high doses of 1,4-dioxane; although the available studies
have not characterized these pathways nor identified any possible reactive intermediates. Tumor
promotion studies in mouse skin and rat liver suggest that 1,4-dioxane may enhance the growth
of previously initiated cells (Lundberg et al.,1987; King et al., 1973). This is consistent with the
increase in hepatocyte cell proliferation observed in several studies (Miyagawa et al., 1999; Uno
et al., 1994; Goldsworthy et al., 1991; Stott et al., 1981). These mechanistic studies provide
evidence of cell proliferation, but do not indicate whether mitogenesis or cytotoxicity is
responsible for increased cell turnover.
Hyperplasia
Hyperplasia
Tumor promotion
Tumor formation
Toxicokinetics
Oral absorption of
1,4-dioxane
Hepatocellular
cytotoxicity
Regenerative cell
proliferation
HEAA elimination
in the urine
MOA for Liver Tumors
Cell proliferation in
absence of
cytotoxicity
Metabolism by
CYP2E1 and
CYP2B1/2
Metabolic
saturation and
accumulation of
1,4-dioxane in the
blood
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
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
4.7.3.1.2. Nasal cavity. A possible key event in the MO A hypothesis for nasal tumors is
sustained proliferation of spontaneously transformed nasal epithelial cells, resulting in the
eventual formation of nasal cavity tumors. Precursor events in which 1,4-dioxane may promote
proliferation of transformed nasal cells are highly uncertain. Figure 4-2 presents a schematic
representation of possible key events leading to the formation of nasal cavity tumors.
Histopathological lesions in female rats were suggestive of toxicity and regeneration in this
tissue (i.e., atrophy, adhesion, inflammation, nuclear enlargement, and hyperplasia and
metaplasia of respiratory and olfactory epithelium) (Kano et al., 2009; JBRC, 1998a).
Hyperplasia
Tumor formation
Inhalation of
water droplets
Toxicokinetics
Oral
absorption of
1,4-dioxane
Cytotoxicity to
nasal cell
epithelium
Regenerative
cell
proliferation
HEAA
elimination in
the urine
Metabolism
by CYP2E1
and
CYP2B1/2
Chronic
irritation due to
direct contact
with nasal
epithelium
MOA for Nasal Cavity
Tumors
Metabolic
saturation and
accumulation of
1,4-dioxane in the
blood, exhalation
of 1,4-dioxane in
breath
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
4.7.3.2.1. Liver. The plausibility of a MOA that would include liver cytotoxicity, with
subsequent reparative cell proliferation, as precursor events to liver tumor formation is
minimally supported by findings that nonneoplastic liver lesions occurred at exposure levels
lower than those resulting in significantly increased incidences of hepatocellular tumors (Kociba
et al., 1974) and the demonstration of nonneoplastic liver lesions in subchronic (Kano et al.,
2008) and acute and short-term oral studies (see Table 4-15). Because the incidence of
nonneoplastic lesions was not reported by Kociba et al. (1974), it is difficult to know whether the
incidence of liver lesions increased with increasing 1,4-dioxane concentration. Contradicting the
observations by Kociba et al. (1974), liver tumors were observed in female rats and female mice
in the absence of lesions indicative of cytotoxicity (Kano et al., 2008; JBRC, 1998a; NCI, 1978).
81	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
This suggests that cytotoxicity may not be a requisite step in the MOA for liver cancer.
Mechanistic and tumor promotion studies suggest that enhanced cell proliferation without
cytotoxicity may be a key event; however, data showing a plausible dose response and temporal
progression from cell proliferation to eventual liver tumor formation are not available (see
Sections 4.7.3.3 and 4.7.3.4). Mechanistic studies that demonstrated cell proliferation after
short-term exposure did not evaluate liver cytotoxicity (Miyagawa et al., 1999; Uno et al., 1994;
Goldsworthy et al., 1991). Studies have not investigated possible precursor events that may lead
to cell proliferation in the absence of cytotoxicity (i.e., genetic regulation of mitogenesis).
4.7.3.2.2. Nasal cavity. Nasal cavity tumors have been demonstrated in several rat strains
(Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994; NCI, 1978; Kociba et al., 1974), but
were not elevated in two strains of mice (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 2994;
NCI, 1978). Chronic irritation was indicated by the observation of rhinitis and inflammation of
the nasal cavity in rats from the JBRC (1998a) study. This study also showed atrophy of the
nasal epithelium and adhesion in rats. Regeneration of the nasal epithelium is demonstrated by
metaplasia and hyperplasia observed in rats exposed to 1,4-dioxane (Kano et al., 2009; JBRC,
1998a; Yamazaki et al., 1994).
4.7.3.3. Dose-Jiesponse Jie/aiions/tip
4.7.3.3.1. Liver. Table 4-18 presents the temporal sequence and dose-response relationship for
possible key events in the liver carcinogenesis of 1,4-dioxane. Dose-response information
provides some support for enhanced cell proliferation as a key event in the liver tumorigenesis of
1,4-dioxane; however, the role of cytotoxicity as a required precursor event is not supported by
data from more than one study. Kociba et al. (1974) demonstrated that liver toxicity and
hepatocellular regeneration occurred at a lower dose level than tumor formation. Hepatocellular
degeneration and necrosis were observed in the mid- and high-dose groups of Sherman rats
exposed to 1,4-dioxane, although it is not possible to discern whether this effect was observed in
both genders due to the lack of incidence data (Kociba et al., 1974). Hepatic tumors were only
observed at the highest dose (Kociba et al., 1974). Hepatic regeneration was indicated in the
mid- and high-dose group by the formation of hepatocellular hyperplastic nodules. Liver
hyperplasia was also seen in rats from the JBRC (1998a) study, at or below the dose level that
resulted in tumor formation (Kano et al., 2009); however, hepatocellular degeneration and
necrosis were not observed. These results suggest that hepatic cell proliferation and hyperplasia
may occur in the absence of significant cytotoxicity. Liver angiectasis (i.e., dilation of blood or
lymphatic vessels) was observed in male mice at the same dose that produced liver tumors;
however, the relationship between this vascular abnormality and tumor formation is unclear.
82
DRAFT - DO NOT CITE OR QUOTE

-------
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
a



a
14
+b
a
a
a
a
121
+b
+c
a
+c
a
1,307
+b
+c
a
+c
+c
NCI, 1978—female Osborne-Mendel rats
0
a
a
a
a
a
350
+b
a
a
a
+c
640
+b
a
a
a
+c
NCI, 1978—male B6C3F! mice
0
a
a
a
a
a
720
+b
a
a
a
+c
830
+b
a
a
a
+c
NCI, 1978—female B6C3F! mice
0
a
a
a
a
a
380
+b
a
a
a
+c
860
+b
a
a
a
+c
Kano et al., 2009; JBRC, 1998a—male F344/DuCrj rats
0
a
a
a
a
a
11
+b
a
a
a
a
55
+b
a
a
+c
a
274
+b
+C'd
a
+c
+c
83	DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
Dose (mg/kg-day)
Key event (time —>)
Metabolism
1,4-dioxane
Liver damage
Cell proliferation
Hyperplasia
Adenomas
and/or
carcinomas
Kano et al., 2009; JBRC, 1998a—female F344/DuCrj rats
0
a
a
a
a
a
18
+b
a
a
a
a
83
+b
a
a
a
a
429
+b
a
a
+c
+c
Kano et al., 2009; JBRC, 1998a—male Crj: BDF, mice
0
a
a
a
a
a
49
+b
a
a
a
+c
191
+b
a
a
a
+c
677
+b
+C,d
a
a
+c
Kano et al., 2009; JBRC, 1998a—female Crj:BDFi mice
0
a
a
a
a
a
66
+b
a
a
a
+c
278
+b
a
a
a
+c
967
+b
+C,d
a
a
+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 (Kano et al., 2008).
4.7.3.3.2. Nasalcavity. Toxicity and regeneration in nasal epithelium (i.e., atrophy, adhesion,
inflammation, and hyperplasia and metaplasia of respiratory and olfactory epithelium) was
evident in one study at the same dose levels that produced nasal cavity tumors (Kano et al, 2009;
see alsoJBRC, 1998a).
4.7.3.4. TemporalJie/aiions/tip
4.7.3.4.1. Liver. Available information regarding temporal relationships between the key event
(sustained proliferation of spontaneously transformed liver cells) and the eventual formation of
liver tumors is limited. A comparison of 13-week and 2-year studies conducted in F344/DuCrj
rats and Crj:BDFi mice at the same laboratory revealed that tumorigenic doses of 1,4-dioxane
produced liver toxicity by 13 weeks of exposure (Kano et al., 2009; Kano et al., 2008; JBRC,
1998a). Hepatocyte swelling of the centrilobular area of the liver, vacuolar changes in the liver,
granular changes in the liver, and single cell necrosis in the liver were observed in mice and rats
given 1,4-dioxane in the drinking water for 13 weeks. Sustained liver damage could presumably
lead to regenerative hyperplasia and tumor formation following chronic exposure. As discussed
84
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
above, histopathological evidence of regenerative hyperplasia has been seen following long-term
exposure to 1,4-dioxane (JBRC, 1998a; Kociba et al., 1974). Tumors occurred earlier at high
doses in both mice and rats from this study (email from Dr. Kazunori Yamazaki, JBRC, to Dr.
Julie Stickney, SRC, dated 12/18/06); however, temporal information regarding hyperplasia or
other possible key events was not available (i.e., interim blood samples not collected, interim
sacrifices were not performed). Argus et al. (1973) studied the progression of tumorigenesis by
electron microscopy of liver tissues obtained following interim sacrifices at 8 and 13 months of
exposure (five rats/group, 574 mg/kg-day). The first change observed was an increase in the size
of the nuclei of the hepatocytes, mostly in the periportal area. Precancerous changes were
characterized by disorganization of the rough endoplasmic reticulum, increase in smooth
endoplasmic reticulum, and decrease in glycogen and increase in lipid droplets in hepatocytes.
These changes increased in severity in the hepatocellular carcinomas in rats exposed to
1,4-dioxane for 13 months.
Three types of liver nodules were observed in exposed rats at 13-16 months. The first
consisted of groups of these cells with reduced cytoplasmic basophilia and a slightly nodular
appearance as viewed by light microscopy. The second type of 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.
4.7.3.4.2. Nasal cavity. No information was available regarding the temporal relationship
between toxicity in the nasal epithelium and the formation of nasal cavity tumors.
4.7.3.5. Bio/ogicaiP/ausibi/ity and Co/terence
4.7.3.5.1. Liver. The hypothesis that sustained proliferation of spontaneously transformed liver
cells is a key event within a MOA is possible based on supporting evidence indicating that
1,4-dioxane is a tumor promoter of mouse skin and rat liver tumors (Lundberg et al., 1987; Bull
et al., 1986; King et al., 1973). Further support for this hypothesis is provided by studies
demonstrating that 1,4-dioxane increased hepatocyte DNA synthesis, indicative of cell
proliferation (Miyagawa et al., 1999; Uno et al., 1994; Goldsworthy et al., 1991; Stott et al.,
1981). In addition, the generally negative results for 1,4-dioxane in a number of genotoxicity
assays indicates the carcinogenicity of 1,4-dioxane may not be mediated by a mutagenic MOA.
The importance of cytotoxicity as a necessary precursor to sustained cell proliferation is
biologically plausible, but is not supported by the dose-response in the majority of studies of
1,4-dioxane carcinogenicity.
85
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
4.7.3.5.2. Nasa/cavity. Sustained cell proliferation in response to cell death from toxicity may
be related to the formation of nasal cavity tumors; however, this MOA is also not established .
Nasal carcinogens are generally characterized as potent genotoxins (Ashby, 1994); however,
other MO As have been proposed for nasal carcinogens that induce effects through other
mechanisms (Kasper et al. 2007; Green et al. 2000).
The National Toxicological Program (NTP) database identified 12 chemicals from
approximately 500 bioassays as nasal carcinogens and 1,4-dioxane was the only identified nasal
carcinogen that showed little evidence of genotoxicity (Haseman and Hailey, 1997). Nasal
tumors were not observed in an inhalation study in Wistar rats exposed to 111 ppm for
5 days/week for 2 years (Torkelson et al., 1974).
4.7.3.6.	Ot/ier Possib/eModes ofAction
An alternate MOA could be hypothesized that 1,4-dioxane alters DNA, either directly or
indirectly, which causes mutations in critical genes for tumor initiation, such as oncogenes or
tumor suppressor genes. Following these events, tumor growth may be promoted by a number of
molecular processes leading to enhanced cell proliferation or inhibition of programmed cell
death. The results from in vitro and in vivo assays do not provide overwhelming support for the
hypothesis of a genotoxic MOA for 1,4-dioxane carcinogenicity. The genotoxicity data for
1,4-dioxane were reviewed in Section 4.5.1 and were summarized in Table 4-16. Negative
findings were reported for mutagenicity in Salmonella typhimurium, Escherichia coli, and
Photobacterium phosphoreum (Mutatox assay) (Morita and Hayashi, 1998; Hellmer and
Bolcsfoldi, 1992; Kwan et al., 1990; Khudoley et al., 1987; Nestmann et al., 1984; Haworth
et al., 1983; Stott et al., 1981). Negative results were also indicated for the induction of
aneuploidy in yeast (Saccharomyces cerevisiae) and the sex-linked recessive lethal test in
Drosophila melanogaster (Zimmerman et al., 1985). In contrast, positive results were reported in
assays for sister chromatid exchange (Galloway et al., 1987), DNA damage (Kitchin and Brown,
1990), and in in vivo micronucleus formation in bone marrow (Roy et al., 2005; Mirkova, 1994),
and liver (Roy et al., 2005; Morita and Hayashi, 1998). Lastly, in the presence of toxicity,
positive results were reported for meiotic nondisjunction in drosophila (Munoz and Barnett,
2002), DNA damage (Sina et al., 1983), and cell transformation (Sheu et al., 1988).
Additionally, 1,4-dioxane metabolism did not produce reactive intermediates that
covalently bound to DNA (Stott et al., 1981; Woo et al., 1977a) and DNA repair assays were
generally negative (Goldsworthy et al., 1991; Stott et al., 1981). No studies were available to
assess the ability of 1,4-dioxane or its metabolites to induce oxidative damage to DNA.
4.7.3.7.	Conclusions About t/ie HypothesizedMode of Action
4.7.3.7.1. Liver. The MOA by which 1,4-dioxane produces liver tumors is unknown, and
available evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
86
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
inconclusive. A MOA hypothesis involving 1,4-dioxane induced cell proliferation is possible
but data are not available to support this hypothesis. Pharmacokinetic data suggest that
clearance pathways were saturable and target organ toxicity occurs after metabolic saturation.
Liver toxicity preceded tumor formation in one study (Kociba et al., 1974) and a regenerative
response to tissue injury was demonstrated by histopathology. Liver hyperplasia and tumor
formation have also been observed in the absence of cytotoxicity (Kano et al., 2009; see also
JBRC, 1998a). Cell proliferation and tumor promotion have been shown to occur after
prolonged exposure to 1,4-dioxane (Miyagawa et al., 1999; Uno et al., 1994; Goldsworthy et al.,
1991; Lundberg et al., 1987; Bull et al., 1986; Stott et al., 1981; King et al., 1973).
4.7.3.7.2. Nasalcavity. The MOA for the formation of nasal cavity tumors is unknown, and
evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
inconclusive.
4.7.3.8. Jie/evance of the Mode of Action to Humans
Several hypothesized MO As for 1,4-dioxane induced tumors in laboratory animals have
been discussed along with the supporting evidence for each. As was stated, the MOA by which
1,4-dioxane produces liver, nasal, peritoneal, and mammary gland tumors is unknown. Some
mechanistic information is available to inform the MOA of the liver and nasal tumors but no
information exists to inform the MOA of the observed peritoneal or mammary gland tumors
(Kano et al., 2009; see also JBRC, 1998a; Yamazaki et al., 1994).
4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES
There is no direct evidence to establish that certain populations and lifestages may be
potentially susceptible to 1,4-dioxane. Changes in susceptibility with lifestage as a function of
the presence of microsomal enzymes that metabolize and detoxify this compound (i.e., CYP2E1
present in liver, kidney, and nasal mucosa can be hypothesized). Vieira et al. (1996) reported
that large increases in hepatic CYP2E1 protein occur postnatally between 1 and 3 months in
humans. Adult hepatic concentrations of CYP2E1 are achieved sometime between 1 and
10 years. To the extent that hepatic CYP2E1 levels are lower, children may be more susceptible
to liver toxicity from 1,4-dioxane than adults. CYP2E1 has been shown to be inducible in the rat
fetus. The level of CYP2E1 protein was increased by 1.4-fold in the maternal liver and 2.4-fold
in the fetal liver following ethanol treatment, as compared to the untreated or pair-fed groups
(Carpenter et al., 1996). Pre- and postnatal induction of microsomal enzymes resulting from
exposure to 1,4-dioxane or other drugs or chemicals may reduce overall toxicity following
sustained exposure to 1,4-dioxane.
Genetic polymorphisms have been identified for the human CYP2E1 gene (Watanabe
et al., 1994; Hayashi et al., 1991) and were considered to be possible factors in the abnormal
liver function seen in workers exposed to vinyl chloride (Huang et al., 1997). Individuals with a
87	DRAFT - DO NOT CITE OR QUOTE

-------
1	CYP2E1 genetic polymorphism resulting in increased expression of this enzyme may be less
2	susceptible to toxicity following exposure to 1,4-dioxane.
3	Gender differences were noted in subchronic and chronic toxicity studies of 1,4-dioxane
4	in mice and rats (see Sections 4.6 and 4.7). No consistent pattern of gender sensitivity was
5	identified across studies.
88
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
5. DOSE-RESPONSE ASSESSMENTS
5.1. ORAL REFERENCE DOSE (RFD)
5.1.1. Choice of Principal Studies and Critical Effect with Rationale and Justification
Liver and kidney toxicity were the primary noncancer health effects associated with
exposure to 1,4-dioxane in humans and laboratory animals. Occupational exposure to
1,4-dioxane has resulted in hemorrhagic nephritis and centrilobular necrosis of the liver
(Johnstone, 1959; Barber, 1934). In animals, liver and kidney degeneration and necrosis were
observed frequently in acute oral and inhalation studies (JBRC, 1998b; Drew et al., 1978; David,
1964; Kesten et al., 1939; Laug et al., 1939; Schrenk and Yant, 1936; deNavasquez, 1935;
Fairley et al., 1934). Liver and kidney effects were also observed following chronic oral
exposure to 1,4-dioxane in animals (Kano et al., 2009; JBRC, 1998a; Yamazaki et al., 1994;
NCI, 1978; Kociba et al., 1974; Argus et al., 1973, 1965) (see Table 4-17).
Liver toxicity in the available chronic studies was characterized by necrosis, spongiosis
hepatic, hyperplasia, cyst formation, clear foci, and mixed cell foci. Kociba et al. (1974)
demonstrated hepatocellular degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in
male rats) or greater. The NOAEL for liver toxicity was 9.6 mg/kg-day and 19 mg/kg-day in
male and female rats, respectively. No quantitative incidence data were provided in this study.
Argus et al. (1973) described early preneoplastic changes in the liver and JBRC (1998a)
demonstrated liver lesions that are primarily associated with the carcinogenic process. Clear and
mixed-cell foci in the liver are commonly considered preneoplastic changes and would not be
considered evidence of noncancer toxicity. In the JBRC (1998a) study, spongiosis hepatis was
associated with other preneoplastic changes in the liver (clear and mixed-cell foci) and no other
lesions indicative of liver toxicity were seen. Spongiosis hepatis was therefore not considered
indicative of noncancer effects in this study. The activity of serum enzymes (i.e., AST, ALT,
LDH, and ALP) was increased in mice and rats chronically exposed to 1,4-dioxane (JBRC,
1998a); however, these increases were seen only at tumorigenic dose levels. Blood samples
were collected at study termination and elevated serum enzymes may reflect changes associated
with tumor formation. Histopathological evidence of liver toxicity was not seen in rats from the
JBRC (1998a) study. The highest non-tumorigenic dose levels for this study approximated the
LOAEL derived from the Kociba et al. (1974) study (94 and 148 mg/kg-day for male and female
rats, respectively).
Kidney damage in chronic toxicity studies was characterized by degeneration of the
cortical tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCI, 1978; Kociba
et al., 1974; Argus et al., 1965, 1973; Fairley et al., 1934). Kociba et al. (1974) described renal
89
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
tubule epithelial cell degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats)
or greater, with a NOAEL of 9.6 mg/kg-day. No quantitative incidence data were provided in
this study. Doses of > 430 mg/kg-day 1,4-dioxane induced marked kidney alterations (Argus
et al., 1973). The observed changes included glomerulonephritis and pyelonephritis, with
characteristic epithelial proliferation of Bowman's capsule, periglomerular fibrosis, and
distension of tubules. Quantitative incidence data were not provided in this study. In the NCI
(1978) study, kidney lesions in rats consisted of vacuolar degeneration and/or focal tubular
epithelial regeneration in the proximal cortical tubules and occasional hyaline casts. Kidney
toxicity was not seen in rats from the JBRC (1998a) study at any dose level (highest dose was
274 mg/kg-day in male rats and 429 mg/kg-day in female rats).
Kociba et al. (1974) was chosen as the principal study for derivation of the RfD because
the liver and kidney effects in this study are considered adverse and represent the most sensitive
effects identified in the database (NOAEL 9.6 mg/kg-day, LOAEL 94 mg/kg-day in male rats).
Kociba et al. (1974) reported degenerative effects in the liver, while liver lesions reported in
other studies (JBRC, 1998a; Argus et al., 1973) appeared to be related to the carcinogenic
process. Kociba et al. (1974) also reported degenerative changes in the kidney. NCI (1978) and
Argus et al. (1973) provided supporting data for this endpoint; however, kidney toxicity was
observed in these studies at higher doses. JBRC (1998a) reported nasal inflammation in rats
(NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and mice (NOAEL 66 mg/kg-day, LOAEL
278 mg/kg-day).
5.1.2. Methods of Analysis—Including Models (PBPK, BMD, etc.)
Several procedures were applied to the human PBPK model to determine if an adequate
fit of the model to the empirical model output or experimental observations could be attained
using biologically plausible values for the model parameters. The re-calibrated model
predictions for blood 1,4-dioxane levels did not come within 10-fold of the experimental values
using measured tissue:air partition coefficients of Leung and Paustenbach (1990) or Sweeney
et al. (2008) (Figures B-8 and B-9). The utilization of a slowly perfused tissue:air partition
coefficient 10-fold lower than measured values produces exposure-phase predictions that are
much closer to observations, but does not replicate the elimination kinetics (Figure B-10). Re-
calibration of the model with upper bounds on the tissue:air partition coefficients results in
predictions that are still six- to sevenfold lower than empirical model prediction or observations
(Figures B-12 and B-13). Exploration of the model space using an assumption of zero-order
metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate fit to the
exposure and elimination data can be achieved only when unrealistically low values are assumed
for the slowly perfused tissue:air partition coefficient (Figure B-16). Artificially low values for
the other tissue:air partition coefficients are not expected to improve the model fit, as these
parameters are shown in the sensitivity analysis to exert less influence on blood 1,4-dioxane than
90
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
Vmaxc and Km. This suggests that the model structure is insufficient to capture the apparent 10-
fold species difference in the blood 1,4-dioxane between rats and humans. In the absence of
actual measurements for the human slowly perfused tissue:air partition coefficient, high
uncertainty exists for this model parameter value. Differences in the ability of rat and human
blood to bind 1,4-dioxane may contribute to the difference in Vj. However, this is expected to
be evident in very different values for rat and human blood:air partition coefficients, which is not
the case (Table B-l). Therefore, some other, as yet unknown, modification to model structure
may be necessary.
Kociba et al. (1974) did not provide quantitative incidence or severity data for liver and
kidney degeneration and necrosis. Benchmark dose (BMD) modeling could not be performed
for this study and the NOAEL for liver and kidney degeneration (9.6 mg/kg-day in male rats)
was used as the point of departure (POD) in deriving the RfD for 1,4-dioxane.
Alternative PODs were calculated using incidence data reported for cortical tubule
degeneration in male and female rats (NCI, 1978) and liver hyperplasia (JBRC, 1998a). The
incidence data for cortical tubule cell degeneration in male and female rats exposed to
1,4-dioxane in the drinking water for 2 years are presented in Table 5-1. Details of the BMD
analysis of these data are presented in Appendix C. Male rats were more sensitive to the kidney
effects of 1,4-dioxane than females and the male rat data provided the lowest POD for cortical
tubule degeneration in the NCI (1978) study (BMDLio of 22.3 mg/kg-day) (see Table 5-2).
Incidence data (Kano et al., 2009; JBRC, 1998a) for liver hyperplasia in male and female rats
exposed to 1,4-dioxane in the drinking water for 2 years are presented in Table 5-3. Details of
the BMD analysis of these data are presented in Appendix C. Male rats were more sensitive to
developing liver hyperplasia due to exposure to 1,4-dioxane than females and the male rat data
provided the lowest POD for hyperplasia in the JBRC (1998a) study (BMDLio of 23.8 mg/kg-
day) (see Table 5-4). The BMDLio values of 22.3 mg/kg-day and 23.8 mg/kg-day from the NCI
(1978) and JBRC (1998a) studies, respectively, are within a factor of two of the NOAEL
(9.6 mg/kg-day) observed by Kociba et al. (1974).
Table 5-1. Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years
Males (mg/kg-day)
Females (mg/kg-day)
0
240
530
0
350
640
0/3 r
20/3 lb
27/3 3b
0/3 r
0/34
10/32b
aStatistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this
review.
incidence significantly elevated compared to control by Fisher's Exact test (p < 0.001) performed for this review.
Source: NCI (1978).
91
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

BMD10 (mg/kg-day)
BMDL10 (mg/kg-day)
Male rats
28.8
22.3
Female rats
596.4
452.4
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)
Females (mg/kg-day)
0
11
55
274
0
18
83
429
3/40
2/45
9/3 5a
12/22b
0/38a
0/37
1/38
14/24b
aStatistically significant compared to controls by the Dunnett's test (p < 0.05).
incidence significantly elevated compared to control by x2 test (p < 0.01).
Sources: Kano et al. (2009); 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

BMD10 (mg/kg-day)
BMDLio (mg/kg-day)
Male rats
35.9
23.8
Female rats
137.3
88.5
Source: Kano et al. (2009); JBRC (1998a).
5.1.3. RfD Derivation - Including Application of Uncertainty Factors (UFs)
_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:
92
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
RfD = NOAEL / UF
= 9.6 mg/kg-day / 300
= 0.03 or 3 x 10-2 mg/kg-day
The composite UF of 300 includes factors of 10 for animal-to-human extrapolation and
for interindividual variability, and an UF of 3 for database deficiencies.
A default interspecies UF of 10 was used to account for pharmacokinetic and
pharmacodynamic differences across species. Existing PBPK models could not be used to derive
an oral RfD for 1,4-dioxane (see Appendix B).
A default interindividual variability UF of 10 is used to account for variation in
sensitivity within human populations because there is limited information on the degree to which
humans of varying gender, age, health status, or genetic makeup might vary in the disposition of,
or response to, 1,4-dioxane.
An UF of 3 for database deficiencies is applied due to the lack of a multigeneration
reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was
available for 1,4-dioxane (Giavini et al., 1985). This developmental study indicates that the
developing fetus may be a target of toxicity.
An UF to extrapolate from a subchronic to a chronic exposure duration was not necessary
because the RfD was derived from a study using a chronic exposure protocol.
An UF to extrapolate from a LOAEL to a NOAEL was not necessary because the RfD
was based on a NOAEL. Kociba et al. (1974) was a well-conducted, chronic drinking water
study with an adequate number of animals. Histopathological examination was performed for
many organs and tissues, but clinical chemistry analysis was not performed. NOAEL and
LOAEL values were derived from the study based on liver and kidney toxicity. Several
additional oral studies (acute/short-term, subchronic, and chronic durations) were available that
support liver and kidney toxicity as the critical effect (Kano et al., 2008; JBRC, 1998a; NCI,
1978; Argus et al., 1973, see Tables 4-15 and 4-17). Although degenerative liver and kidney
toxicity was not observed in rats from the JBRC (1998a) study at doses at or below the LOAEL
in the Kociba et al. (1974) study, other endpoints such as metaplasia and hyperplasia of the nasal
epithelium, nuclear enlargement, and hematological effects, were noted.
5.1.4. RfD Comparison Information
PODs and sample oral RfDs based on selected studies included in Table 4-17 are arrayed
in Figures 5-1 to 5-3, and provide perspective on the RfD supported by Kociba et al. (1974).
These figures should be interpreted with caution because the PODs across studies are not
necessarily comparable, nor is the confidence in the data sets from which the PODs were derived
the same. PODs in these figures may be based on a NOAEL, LOAEL, or BMDL (as indicated),
and the nature, severity, and incidence of effects occurring at a LOAEL are likely to vary. To
some extent, the confidence associated with the resulting sample RfD is reflected in the
93	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
magnitude of the total UF applied to the POD (i.e., the size of the bar); however, the text of
Sections 5.1.1 and 5.1.2 should be consulted for a more complete understanding of the issues
associated with each data set and the rationale for the selection of the critical effect and principal
study used to derive the RfD.
The predominant noncancer effect of chronic oral exposure to 1,4-dioxane is
degenerative effects in the liver and kidney. Figure 5-1 provides a graphical display of effects
that were observed in the liver following chronic oral exposure to 1,4-dioxane. Information
presented includes the PODs and UFs that could be considered in deriving the oral RfD. As
discussed in Sections 5.1.1 and 5.1.2, among those studies that demonstrated liver toxicity, the
study by Kociba et al. (1974) provided the data set most appropriate for deriving the RfD. For
degenerative liver effects resulting from 1,4-dioxane exposure, the Kociba et al. (1974) study
represents the most sensitive effect and dataset observed in a chronic bioassay (Figure 5-1).
Kidney toxicity as evidenced by glomerulonephritis (Argus et al., 1973; 1965) and
degeneration of the cortical tubule (NCI, 1978; Kociba et al., 1974) has also been observed in
response to chronic exposure to 1,4-dioxane. As was discussed in Sections 5.1 and 5.2,
degenerative effects were observed in the kidney at the same dose level as effects in the liver
(Kociba et al., 1974). A comparison of the available datasets from which an RfD could
potentially be derived is presented in Figure 5-2.
Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice
only, dose > 380 mg/kg-day) and JBRC (1998a) studies (> 274 mg/kg-day in rats, >278 mg/kg-
day in mice). JBRC (1998a) reported nasal inflammation in rats (NOAEL 55 mg/kg-day,
LOAEL 274 mg/kg-day) and mice (NOAEL 66 mg/kg-day, LOAEL 278 mg/kg-day). A
comparison of the available datasets from which an RfD could potentially be derived is presented
in Figure 5-3.
Figure 5-4 displays PODs for the major targets of toxicity associated with oral exposure
to 1,4-dioxane. Studies in experimental animals have also found that relatively high doses of
1,4-dioxane (1,000 mg/kg-day) during gestation can produce delayed ossification of the
sternebrae and reduced fetal BWs (Giavini et al., 1985). This graphical display (Figure 5-4)
compares organ specific toxicity for 1,4-dioxane, including a single developmental study. The
most sensitive measures of degenerative liver are and kidney effects. The sample RfDs for
degenerative liver and kidney effects are identical since they were derived from the same study
and dataset (Kociba et al., 1974) and are presented for completeness.
94
DRAFT - DO NOT CITE OR QUOTE

-------
100
Rat
Rat
Mouse
Rat
Rat
• POD
OAnimal-to-human
~Human variation
E3LOAEL to NOAEL
~Subchronic to Chronic
|Database deficiencies
ORfD
0.01
Liver hyperplasia;
NOAEL; 2 yr rat
drinking water study
(JBRC, 1998a)
Hepatocellular
degeneration and
necrosis; NOAEL; 2 yr
rat drinking water study
(Kociba et al., 1974)
Increase in serum liver Increase in serum liver	Liver hyperplasia;
enzymes; NOAEL; 2 yr enzymes; NOAEL; 2 yr	BMDL10; 2 yr rat
mouse drinking water	rat drinking water study	drinking water study
study (JBRO, 1998a)	(JBRC, 1998a)	(JBRC, 1998a)
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.
95
DRAFT - DO NOT CITE OR QUOTE

-------
Rat	Rat	Rat
4
>
i

1
	
>
<
>
1
1
¦ ^
o
Glomerulonephritis; LOAEL; 13 month Degeneration and necrosis of tubular Cortical tubule degeneration; BMDL10; 2
rat drinking water study (Argus et al., epithelium; NOAEL; 2 yr rat drinking yr rat drinking water study (NCI, 1978)
1973)	water study (Kociba et al., 1974)
ITU An i mal-to-hu man
~ Human variation
E23LOAEL to NOAEL
QSubchronic to Chronic
H Database deficiencies
ORfD
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.
96
DRAFT - DO NOT CITE OR QUOTE

-------
Mouse
Rat
100
10
o
o
0.1
I
o
• POD
U|Animal-to-human
QHuman variation
E^LOAELto NOAEL
dSubchronic to Chronic
I Database deficiencies
ORfD
I
o
Nasal inflammation; NOAEL; 2 yr mouse drinking water
study (JBRC, 1998a)
Nasal inflammation; NOAEL; 2 yr rat drinking water study
(JBRC, 1998a)
Figure 5-3. Potential points of departure (POD) for nasal inflammation with
corresponding applied uncertainty factors and derived sample RfDs following oral
exposure to 1,4-dioxane.
97
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
1000
100
Rat
Rat
Rat
Mouse
7
ro
E
o
Q
0.01
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)
• POD
IDAnimal-to-human
~Human variation
0LOAELto NOAEL
~Subchronic to Chronic
^Database deficiencies
ORfD
Figure 5-4. Potential points of departure (POD) for organ specific toxicity endpoints
with corresponding applied uncertainty factors and derived sample RfDs following
oral exposure to 1,4-dioxane.
5.1.5. Previous RfD Assessment
An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
oral RfD was not developed as part of the 1988 assessment.
5.2. INHALATION REFERENCE CONCENTRATION (RFC)
NOTE: During the development of this assessment, new data regarding the toxicity of
1,4-dioxane through the inhalation route of exposure became available. The IRIS Program will
evaluate the more recently published 1,4-dioxane inhalation data for the potential to derive an
RfC in a separate assessment. A description of the studies that were available at the time that this
assessment was under development are described below..
Inhalation studies for 1,4-dioxane evaluated in this assessment were not adequate for the
determination of an RfC value. Only one subchronic study (Fairley et al., 1934) and one chronic
inhalation study (Torkelson et al., 1974) were identified. In the subchronic study, rabbits, guinea
pigs, rats, and mice (3-6/species/group) were exposed to 1,000, 2,000, 5,000, or 10,000 ppm of
98	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
1,4-dioxane vapor for 1.5 hours two times a day for 5 days, 1.5 hours for one day, and no
exposure on the seventh day. Animals were exposed until death occurred or were sacrificed after
various durations of exposure (3-202.5 hours). Detailed dose-response information was not
provided; however, severe liver and kidney damage and acute vascular congestion of the lungs
were observed at concentrations > 1,000 ppm. Kidney damage was described as patchy
degeneration of cortical tubules with vascular congestion and hemorrhage. Liver lesions varied
from cloudy hepatocyte swelling to large areas of necrosis.
Torkelson et al. (1974) performed a chronic inhalation study in which male and female
Wistar rats (288/sex) were exposed to 111 ppm 1,4-dioxane vapor for 7 hours/day, 5 days/week
for 2 years. Control rats (192/sex) were exposed to filtered air. No significant effects were
observed on BWs, survival, organ weights, hematology, clinical chemistry, or histopathology.
Because Fairley et al. (1934) identified a free-standing LOAEL only, and Torkelson et al. (1974)
identified a free-standing NOAEL only, neither study was sufficient to characterize the
inhalation risks of 1,4-dioxane. A route extrapolation from oral toxicity data was not performed
because 1,4-dioxane inhalation causes direct effects on the respiratory tract (i.e., respiratory
irritation in humans, pulmonary congestion in animals) (Wirth and Klimmer, 1936; Fairley et al.,
1934; Yant et al., 1930), which would not be accounted for in a cross-route extrapolation. In
addition, available kinetic models are not suitable for this purpose (see Appendix B).
An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
inhalation RfC was not developed as part of the 1988 assessment.
5.3. UNCERTAINTIES IN THE ORAL REFERENCE DOSE (RfD)
Risk assessments need to portray associated uncertainty. The following discussion
identifies uncertainties associated with the RfD for 1,4-dioxane. As presented earlier in this
section (5.1.2 and 5.1.3), the uncertainty factor approach (U.S. EPA, 2002a, 1994b), was applied
to a POD. Factors accounting for uncertainties associated with a number of steps in the analyses
were adopted to account for extrapolating from an animal bioassay to human exposure, a diverse
population of varying susceptibilities, and to account for database deficiencies. These
extrapolations are carried out with current approaches given the paucity of experimental
1,4-dioxane data to inform individual steps.
An adequate range of animal toxicology data are available for the hazard assessment of
1,4-dioxane, as described throughout the previous section (Chapter 4). The database of oral
toxicity studies includes chronic drinking water studies in rats and mice, multiple subchronic
drinking water studies conducted in rats and mice, and a developmental study in rats. Toxicity
associated with oral exposure to 1,4-dioxane is observed predominately in the liver and kidney.
The database of inhalation toxicity studies in animals includes one subchronic bioassay in
rabbits, guinea pigs, and rats, and a chronic inhalation bioassay in rats. Although the subchronic
bioassay observed degenerative effects in the liver, kidney, and lungs of all species tested, the
99	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
information reported from the study was insufficient to determine an exposure level below which
these effects did not occur. The only available chronic inhalation bioassay did not indicate any
treatment related effects due to exposure to 1,4-dioxane. Thus, the inhalation database lacked
sufficient information to derive toxicity values relevant to this route of exposure for 1,4-dioxane.
In addition to oral and inhalation data, there are PBPK models and genotoxicity studies of
1,4-dioxane. Critical data gaps have been identified and uncertainties associated with data
deficiencies of 1,4-dioxane are more fully discussed below.
Consideration of the available dose-response data led to the selection of the two-year
drinking water bioassay in Sherman rats (Kociba et al., 1974) as the principal study and
increased liver and kidney degeneration as the critical effects for deriving the RfD for
1,4-dioxane. The dose-response relationship for oral exposure to 1,4-dioxane and cortical tubule
degeneration in Osborne-Mendel rats (NCI, 1978) was also suitable for deriving a RfD, but it is
associated with higher a POD and potential RfD compared to Kociba et al. (1974).
The RfD was derived by applying UFs to a NOAEL for degenerative liver and kidney
effects. The incidence data for the observed effects were not reported in the principal study
(Kociba et al., 1974), precluding modeling of the dose-response. However confidence in the
LOAEL can be derived from additional studies (JBRC, 1998a; NCI, 1978; Argus et al., 1973;
1965) that observed effects on the same organs at comparable dose levels and by the BMDL
generated by modeling of the kidney dose-response data from the chronic NCI (1978) study.
Extrapolating from animals to humans embodies further issues and uncertainties. The
effect and the magnitude associated with the dose at the POD in rodents are extrapolated to
human response. Pharmacokinetic models are useful to examine species differences in
pharmacokinetic processing; however, it was determined that dosimetric adjustment using
pharmacokinetic modeling was to reduce uncertainty following oral exposure to 1,4-dioxane was
not supported. Insufficient information was available to quantitatively assess toxicokinetic or
toxicodynamic differences between animals and humans, so a 10-fold UF was used to account
for uncertainty in extrapolating from laboratory animals to humans in the derivation of the RfD.
Heterogeneity among humans is another uncertainty associated with extrapolating doses
from animals to humans. Uncertainty related to human variation needs consideration. In the
absence of 1,4-dioxane-specific data on human variation, a factor of 10 was used to account for
uncertainty associated with human variation in the derivation of the RfD. Human variation may
be larger or smaller; however, 1,4-dioxane-specific data to examine the potential magnitude of
over- or under-estimation are unavailable.
Uncertainties in the assessment of the health hazards of ingested 1,4-dioxane are
associated with deficiencies in reproductive toxicity information. The oral database lacks a
multigeneration reproductive toxicity study. A single oral prenatal developmental toxicity study
in rats was available for 1,4-dioxane (Giavini et al., 1985). This developmental study indicates
that the developing fetus may be a target of toxicity. The database of inhalation studies is of
100
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
particular concern due to the lack of a basic toxicological studies, a multigenerational
reproductive study, and developmental toxicity studies.
5.4. CANCER ASSESSMENT
5.4.1. Choice of Study/Data - with Rationale and Justification
Three chronic drinking water bioassays provided incidence data for liver tumors in rats
and mice, and nasal cavity, peritoneal, and mammary gland tumors in rats only (Kano et al.,
2009; JBRC, 1998a; Yamazaki et al., 1994); NCI, 1978; Kociba et al., 1974). The dose-response
data from each of these studies are summarized in Table 5-5. With the exception of the NCI
(1978) study, the incidence of nasal cavity tumors was generally lower than the incidence of liver
tumors in exposed rats. The Kano et al. (2009) drinking water study was chosen as the principal
study for derivation of an oral cancer slope factor (CSF) for 1,4-dioxane. This study used three
dose groups in addition to controls and characterized the dose-response relationship at lower
exposure levels, as compared to the high doses employed in the NCI (1978) bioassay (see Table
5-5). The Kociba et al. (1974) study also used three dose groups and low exposures; however,
the study authors only reported the incidence of hepatocellular carcinoma, which may
underestimate the combined incidence of rats with adenoma or carcinoma. In addition to
increased incidence of liver tumors, chosen as the most sensitive target organ for tumor
formation, the Kano et al. (2009) study also noted increased incidence of peritoneal and
mammary gland tumors. Nasal cavity tumors were also seen in high-dose male and female rats;
however, the incidence of nasal tumors was much lower than the 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
Species/strain/gender
Animal dose
(mg/kg-day)
Tumor Incidence
Liver
Nasal
cavity
Peritoneal
Mammary
gland
Kociba et al., 1974
Sherman rats, male
and female
combined'
0
l/106h
0/106h
NA
NA
14
0/110
0/110
NA
NA
121
1/106
0/106
NA
NA
1,307
10/661
3/66
NA
NA
NCI, 1978
Male Osborne-Mendel
ratsb
0
NA
0/3 3h
NA
NA
240
NA
12/26
NA
NA
530
NA
16/3 31
NA
NA
Female Osborne-
Mendel ratsb,c
0
0/3 lh
0/3 4h
NA
NA
350
10/301
10/301
NA
NA
640
11/291
8/291
NA
NA
101	DRAFT - DO NOT CITE OR QUOTE

-------



Tumor Incidence
Study
Species/strain/gender
Animal dose
(mg/kg-day)
Liver
Nasal
cavity
Peritoneal
Mammary
gland

Male B6C3Fi miced
0
8/4 9h
NA
NA
NA


720
19/501
NA
NA
NA


830
28/471
NA
NA
NA

Female B6C3Fi miced
0
0/5 0h
NA
NA
NA


380
21/481
NA
NA
NA


860
35/371
NA
NA
NA
Kano et al, 2009
Male F344/DuCrj
ratsd'e'f,s
0
3/50
0/50
2/50
1/50

11
4/50
0/50
2/50
2/50


55
7/50
0/50
5/50
2/50


274
39/50>'k
7/50k
28/50>'k
6/5 0k

Female F344/DuCrj
ratsd'e'f'8
0
3/50
0/50
1/50
8/50

18
1/50
0/50
0/50
8/50


83
6/50
0/50
0/50
11/50


429
48/50J'k
8/50>'k
0/50
18/50''k

Male Crj:BDFi miced
0
23/50
0/50
NA
NA


49
31/50
0/50
NA
NA


191
37/501
0/50
NA
NA


677
40/50>'k
1/50
NA
NA

Female Crj:BDFi
miced
0
5/50
0/50
NA
NA

66
35/50"
0/50
NA
NA


278
41/50"
0/50
NA
NA


967
46/50''k
1/50
NA
NA
"Incidence of hepatocellular carcinoma,
incidence of nasal squamous cell carcinoma.
"Incidence of hepatocellular adenoma,
incidence of hepatocellular adenoma or carcinoma.
"Incidence (sum) of all nasal tumors including squamous cell carcinoma, sarcoma, rhabdomyosarcoma, and
esthesioneuroepithelioma.
incidence of peritoneal tumors (mesothelioma).
incidence of mammary gland tumors (fibroadenoma or adenoma)
hp < 0.05; positive dose-related trend (Cochran-Armitage or Peto's test).
'Significantly different from control atp < 0.05 by Fisher's Exact test.
J Significantly different from control atp < 0.01 by Fisher's Exact test.
kp < 0.01; positive dose-related trend (Peto's 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 Kano et al. (2009) 2-year drinking water study. There were statistically
3	significant increasing trends in tumorigenic response for males and females of both species. The
4	dose-response curve for female mice is steep, with 70% incidence of liver tumors occurring in
102
DRAFT - DO NOT CITE OR QUOTE

-------
1	the low-dose group (66 mg/kg-day). Exposure to 1,4-dioxane increased the incidence of these
2	tumors in a dose-related manner.
3	A significant increase in the incidence of peritoneal mesothelioma was observed in high-
4	dose male rats only (28/50 rats, see Table 5-5). The incidence of peritoneal mesothelioma was
5	lower than the observed incidence of hepatocellular adenoma or carcinoma in male rats (see
6	Table 5-6); therefore, hepatocellular adenoma or carcinoma data were used to derive an oral CSF
7	for 1,4-dioxane.
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
Animal dose
(mg/kg-day)
Incidence of liver tumors3
Male F344/DuCrj rats
0
3/50
11
4/50
55
7/50
274
39/50b'°
Female F344/DuCrj rats
0
3/50
18
1/50
83
6/50
429
48/50b'°
Male Cr^BDFj mice
0
23/50
49
31/50
191
37/50d
677
40/50b'°
Female Crj:BDFi mice
0
5/50
66
35/50°
278
41/50°
967
46/50b'°
"Incidence of either hepatocellular adenoma or carcinoma.
hp < 0.05; positive dose-related trend (Peto's test).
"Significantly different from control atp< 0.01 by Fisher's Exact test.
dSignificantly different from control atp < 0.01 by Fisher's Exact test.
Source: Kano et al. (2009).
5.4.3. Dose Adjustments and Extrapolation Method(s)
9
10
5.4.3.1. Dose Adjustments
Human equivalent doses (HEDs) were calculated from the administered animal doses
0 75
using a BW scaling factor (BW ). This was accomplished using the following equation:
HED = animal dose (mg/kg) x
animal BW (kg)
human BW (kg)
0.25
103
DRAFT - DO NOT CITE OR QUOTE

-------
1	HEDs for the principal study (Kano et al., 2009) are given in Table 5-7. HEDs were also
2	calculated for supporting studies (NCI, 1978; Kociba et al., 1974) and are also shown in Table 5-
3	7.
Table 5-7. Calculated HEDs for the tumor incidence data used for dose-
response modeling
Study
Species/strain/gender
Animal BW (g)
TWA
Animal dose
(mg/kg-day)
HED
(mg/kg-day)d
Kano et al., 2009
Male F344/DuCrj rats
432a
11
3.1
432a
81
23
432a
398
112
Female F344/DuCrj rats
267a
18
4.5
267a
83
21
267a
429
107
Male Crj iBDFj mice
47.9a
49
7.9
47.9a
191
31
47.9a
677
110
Female Crj :BDF i mice
35.9a
66
10
35.9a
278
42
35.9a
967
145
Kociba et al., 1974
Male and female (combined)
Sherman rats
325b
14
3.7
325b
121
32
285°
1,307
330
NCI, 1978
Male Osborne-Mendel rats
470b
240
69
470b
530
152
Female Osborne-Mendel rats
310b
350
90
310b
640
165
Male B6C3FJ mice
32b
720
105
32b
830
121
Female B6C3Fi mice
30b
380
55
30b
860
124
a TWA BWs were determined from BW growth curves provided for each species and gender.
bTWA BWs were determined from BW curve provided for control animals.
°BWs 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)025.
Sources: Kano et al. (2009); Kociba et al. (1974); and NCI (1978).
104
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
5.4.3.2. Extrapolation Metftodfs)
The U.S. EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a)
recommend that the method used to characterize and quantify cancer risk from a chemical is
determined by what is known about the mode of action of the carcinogen and the shape of the
cancer dose-response curve. The linear approach is recommended if the mode-of-action of
carcinogenicity is not understood (U.S. EPA, 2005a). In the case of 1,4-dioxane, the mode of
carcinogenic action for peritoneal, mammary, nasal, and liver tumors is unknown. Therefore, a
linear low-dose extrapolation approach was used to estimate human carcinogenic risk associated
with 1,4-dioxane exposure.
However, several of the external peer review panel members (see Appendix A: Summary
of External Peer Review and Public Comments and Disposition) recommended that the mode of
action data support the use of a non-linear extrapolation approach to estimate human
carcinogenic risk associated with exposure to 1,4-dioxane and that such an approach should be
presented in the Toxicological Review. As discussed in Section 4.7.3., numerous short-term in
vitro and a few in vivo tests were nonpositive for 1,4-dioxane-induced genotoxicity. Results
from two-stage mouse skin tumor bioassays demonstrated that 1,4-dioxane does not initiate
mouse skin tumors, but it is a promoter of skin tumors initiated by DMBA (King et al., 1973).
These data suggest that a potential mode of action for 1,4-dioxane-induced tumors may involve
proliferation of cells initiated spontaneously, or by some other agent, to become tumors
(Miyagawa et al., 1999; Uno et al., 1994; Goldsworthy et al., 1991; Lundberg et al., 1987; Bull
et al., 1986; Stott et al., 1981; King et al., 1973). However, key events related to the promotion
of tumor formation by 1,4-dioxane are unknown. Therefore, under the U.S. EPA Guidelines for
Carcinogen Risk Assessment (U.S. EPA, 2005a), EPA concluded that the available information
does not establish a plausible mode of action for 1,4-dioxane and data are insufficient to establish
significant biological support for a non-linear approach. EPA determined that there are no data
available to inform the low-dose region of the dose response, and thus, a non-linear approach
was not included.
Accordingly, the CSF for 1,4-dioxane was derived via a linear extrapolation from the
POD calculated by curve fitting the experimental dose-response data. The POD is the 95%
lower confidence limit on the dose associated with a benchmark response (BMR) near the lower
end of the observed data. The BMD modeling analysis used to estimate the POD is described in
detail in Appendix D and is summarized below in Section 5.4.4.
Model estimates were derived for all available bioassays and tumor endpoints (see
Appendix D); however, the POD used to derive the CSF is based on the most sensitive species
and target organ in the principal study (female mice; liver tumors; Kano et al., 2009).
The oral CSF was calculated using the following equation:
105
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
rep =	
BMDL10
5.4.4. Oral Slope Factor and Inhalation Unit Risk
The dichotomous models available in the Benchmark Dose Software (BMDS, version
2.1.1) were fit to the incidence data for "either hepatocellular carcinoma or adenoma" in rats and
mice, as well as mammary and peritoneal tumors in rats exposed to 1,4-dioxane in the drinking
water (Kano et al., 2009; NCI, 1978; Kociba et al., 1974) (Table 5-5). Animal doses are used for
BMD modeling and HED BMD and BMDL values are calculated using the animal TWAs (Table
5-7) and a human BW of 70kg. Doses associated with a BMR of 10% extra risk were calculated.
BMDs and BMDLs from all models are reported, and the output and plots corresponding to the
best-fitting model are shown (see Appendix D). When the best-fitting model is not a multistage
model, the multistage model output and plot are also provided (see Appendix D). A summary of
the BMDS model predictions for the Kano et al. (2009), NCI (1978), and Kociba et al. (1974)
studies is shown in Table 5-8.
Table 5-8. BMD hid and BMDLhed 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
Gender/strain/species
Tumor type
BMDhed3
(mg/kg-day)
BMDLhed3
(mg/kg-day)
Oral CSF
(mg/kg-day)1
Kano et al.,
2009
Male F344/DuCrj ratsb
Hepatocellular
adenoma or
carcinoma
17.43
14.33
7.0 x 10"3
Female F344/DuCrj rats0
19.84
14.43
6.9 x 10-3
Male CrjiBDFj miced
5.63
2.68
3.7 x 10-2
Female CrjiBDFj miced
0.83
0.55
0.18
Female Crj:BDFi miced'e
3.22e
2.12e
0.14
Female Crj:BDFi miced'f
7.51f
4.96f
0.10
Female F344/DuCrj rats8
Nasal
squamous cell
carcinoma
94.84
70.23
1.4 x 10-3
Male F344/DuCrj rats8
91.97
68.85
1.5 x 10-3
Male F344/DuCrj ratsb
Peritoneal
mesothelioma
26.09
21.39
4.7 x 10-3
Female F344/DuCrj ratsd
Mammary
gland adenoma
40.01
20.35
4.9 x 10-3
Kociba et al.,
1974
Male and female (combined)
Sherman rats8
Nasal
squamous cell
carcinomas
448.24
340.99
2.9 x 10"4
Male and female (combined)
Sherman ratsb
Hepatocellular
carcinoma
290.78
240.31
4.2 x 10-4
106
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
Study
Gender/strain/species
Tumor type
BMDhed3
(mg/kg-day)
BMDLhed3
(mg/kg-day)
Oral CSF
(mg/kg-day)1
NCI, 1978
Male Osborne Mendel ratsd
Nasal
squamouse cell
carcinomas
16.10
10.66
9.4 x 10"3
Female Osborne Mendel ratsd
40.07
25.82
3.9 x 10-3
Female Osborne Mendel ratsd
Hepatocellular
adenoma
28.75
18.68
5.4 x 10-3
Female B6C3FJ mice0
Hepatocellular
adenoma or
carcinoma
23.12
9.75
1.0 x 10"2
Male B6C3FJ miceh
87.98
35.67
2.8 x 10-3
aValues associated with a BMR of 10% unless otherwise noted.
bProbit model, slope parameter not restricted.
cMultistage model, degree of polynomial = 2.
dLog-logistic model, slope restricted > 1.
eValues associated with a BMR of 30%.
Values associated with a BMR of 50%.
8Multistage model, degree of polynomial =3.
hGamma model.
The multistage model did not provide an adequate fit (as determined by AIC, p-value <
0.1, and x P > |0.11) to the data for the incidence of hepatocellular adenoma or carcinoma in
female mice (see Appendix D). The high dose was dropped for the female mouse liver tumor
dataset in an attempt to achieve an adequate fit; however, an adequate fit was still not achieved.
Because the female mice were clearly the most sensitive group tested, other BMD models were
applied to the female mouse liver tumor dataset to achieve an adequate fit. The log-logistic
model was the only model that provided adequate fit for this data set due to the steep rise in the
dose-response curve (70% incidence at the low dose) followed by a plateau at near maximal
tumor incidence in the mid- and high-dose regions (82 and 92% incidence, respectively). The
predicted BMDio and BMDLio for the female mouse data are presented in Table 5-8, as well as
BMDhed and BMDLhed values associated with BMRs of 30 and 50% .
The multistage model also did not provide an adequate fit to mammary tumor incidence
data for the female rat or male rat peritoneal tumors. The predicted BMDio and BMDLio for
female rat mammary tumors and male peritoneal tumors obtained from the log-logistic and
probit models, respectively, are presented in Table 5-8.
A comparison of the model estimates derived for rats and mice from the Kano et al.
(2009), 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. The BMDL50 hed for the female mouse data was chosen as the POD and the CSF of
0.10 (mg/kg-day)"1 was calculated as follows:
CSF =		= 0.10 (mg/kg - day)"1
4.96 mg/kg - day (BMDL50HED for female mice)
Calculation of a CSF for 1,4-dioxane is based upon the dose-response data for the most
sensitive species and gender.
107
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
Inhalation studies for 1,4-dioxane evaluated in this assessment were not adequate for the
determination of an inhalation unit risk. 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). A route
extrapolation from oral bioassay data was not performed (see Section 5.2). In addition, available
kinetic models are not suitable for this purpose (see Appendix B).
During the development of this assessment, new data regarding the toxicity of 1,4-
dioxane through the inhalation route of exposure became available. The IRIS Program will
evaluate the more recently published 1,4-dioxane inhalation data for the potential to derive an
inhalation unit risk in a separate assessment.
5.4.5. Previous Cancer Assessment
A previous cancer assessment was posted for 1,4-dioxane on IRIS in 1988. 1,4-Dioxane
was classified as a Group B2 Carcinogen (probable human carcinogen; sufficient evidence from
animal studies and inadequate eveident or no data from human epidemiology studies [U.S. EPA,
1986c]) based on the induction of nasal cavity and liver carcinomas in multiple strains of rats,
liver carcinomas in mice, and gall bladder carcinomas in guinea pigs. An oral CSF of 0.011
(mg/kg-day)"1 was derived from the tumor incidence data for nasal squamous cell carcinoma in
male rats exposed to 1,4-dioxane in drinking water for 2 years (NCI, 1978). The linearized
multistage extra risk procedure was used for linear low dose extrapolation.
5.5. UNCERTAINTIES IN CANCER RISK VALUES
As in most risk assessments, extrapolation of study data to estimate potential risks to
human populations from exposure to 1,4-dioxane has engendered some uncertainty in the results.
Several types of uncertainty may be considered quantitatively, but other important uncertainties
cannot be considered quantitatively. Thus an overall integrated quantitative uncertainty analysis
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
The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible
MOA(s) for the observed tumors. The MOA is a key consideration in clarifying how risks
should be estimated for low-dose exposure. Exposure to 1,4-dioxane has been observed in
animal models to induce multiple tumor types, including liver adenomas and carcinomas, nasal
carcinomas, mammary adenomas and fibroadenomas, and mesothiolomas of the peritoneal cavity
(Kano et al., 2009; JBRC, 1998a; NCI, 1978; Kociba et al., 1974). MOA information that is
108
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
available for the carcinogenicity of 1,4-dioxane has largely focused on liver adenomas and
carcinomas, with little or no MOA information available for the remaining tumor types. In
Section 4.7.3, hypothesized MOAs were explored for 1,4-dioxane. Information that would
provide sufficient support for any MOA is not available. In the absence of a MOA(s) for the
observed tumor types, a linear low-dose extrapolation approach was used to estimate human
carcinogenic risk associated with 1,4-dioxane exposure.
It is not possible to predict how additional MOA information would impact the dose-
response assessment for 1,4-dioxane because of the variety of tumors observed and the lack of
data on how 1,4-dioxane or a metabolite thereof, interacts with cells starting the progression to
the observed tumors.
In general, the Agency has preferred to use the multistage model for analyses of tumor
incidence and related endpoints because they have a generic biological motivation based on
long-established mathematical models such as the Moolgavkar-Venzon-Knudsen (MVK) model.
The MVK model does not necessarily characterize all modes of tumor formation, but it is
a starting point for most investigations and, much more often than not, has provided at least an
adequate description of tumor incidence data.
In the studies evaluated (Kano et al., 2009; NCI, 1978; Kociba et al., 1974), the
multistage model provided good descriptions of the incidence of a few tumor types in male
(nasal cavity) and female (hepatocellular and nasal cavity) rats and in male mice (hepatocellular)
exposed to 1,4-dioxane (see Appendix D for details). However, the multistage model did not
provide an adequate fit for the female mouse liver tumor dataset based upon the following (U.S.
EPA, 2000b):
•	Goodness-of-fit p-value was not greater than 0.10;
•	Akaike's Information Criterion (AIC) was larger than other acceptable models;
•	Data deviated from the fitted model, as measured by their x residuals (values were
greater than an absolute value of one).
BMDS software typically implements the guidance in the external peer review draft
BMD technical guidance document (U.S.EPA, 2000b) by imposing constraints on the values of
certain parameters of the models. When these constraints were imposed, the multistage model
and most other models did not fit the incidence data for female mouse liver adenomas or
carcinomas.
The log-logistic model was selected because it provides an adequate fit for the female
mouse data (Kano et al., 2009). A BMR of 50% was used because it is proximate to the response
at the lowest dose tested and the BMDL50 hed was derived by applying appropriate parameter
constraints, consistent with recommended use of BMDS in the external peer review draft BMD
technical guidance document (U.S. EPA, 2000b).
109
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
The human equivalent oral CSFs estimated from tumor datasets with statistically
significant increases ranged from 4.2 x 10"4 to 0.18 per mg/kg-day (Table 5-8), a range of about
three orders of magnitude, with the extremes coming from the combined male and female rat
data for hepatocellular carcinomas (Kociba et al., 1974) and the female mouse combined liver
adenoma and carcinomas (Kano et al., 2009).
5.5.1.2.	Dose Metric
1,4-Dioxane is known to be metabolized in vivo. However, it is unknown whether a
metabolite or the parent compound, or some combination of parent compound and metabolites, is
responsible for the observed toxicity. If the actual carcinogenic moiety is proportional to
administered exposure, then use of administered exposure as the dose metric is the least biased
choice. On the other hand, if this is not the correct dose metric, then the impact on the CSF is
unknown.
5.5.1.3.	Cross-Species Scaling
0 75
An adjustment for cross-species scaling (BW ) was applied to address toxicological
equivalence of internal doses between each rodent species and humans, consistent with the 2005
Guidelines for Carcinogen Risk Assessment (US EPA, 2005a). It is assumed that equal risks
result from equivalent constant lifetime exposures.
5.5.1.4.	Statistical Uncertainty at the PO/)
Parameter uncertainty can be assessed through confidence intervals. Each description of
parameter uncertainty assumes that the underlying model and associated assumptions are valid.
For the log-logistic model applied to the female mouse data, there is a reasonably small degree of
uncertainty at the 10% excess incidence level (the POD for linear low-dose extrapolation).
5.5.1.5.	Bioassaf Selection
The study by Kano et al. (2009) was used for development of an oral CSF. This was a
well-designed study, conducted in both sexes in two species with a sufficient number of animals
per dose group. The number of test animals allocated among three dose levels and an untreated
control group was adequate, with examination of appropriate toxicological endpoints in both
sexes of rats and mice. Alternative bioassays (NCI, 1978; Kociba et al., 1974) are available and
were fully considered for the derivation of the oral CSF.
5.5.1.6.	Choice of Species/Gender
The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the
female mouse, which was thought to be more sensitive than male mice or either sex of rats to the
carcinogenicity of 1,4-dioxane. While all data, both species and sexes reported from the Kano et
al. (2009) study, were suitable for deriving an oral CSF, the female mouse data represented the
most sensitive indicator of carcinogenicity in the rodent model. The lowest exposure level
(66 mg/kg-day or 10 mg/kg-day [HED]) observed a considerable and significant increase in
110	DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
combined liver adenomas and carcinomas. Additional testing of doses within the range of
control and the lowest dose (66 mg/kg-day or 10 mg/kg-day [HED]) could refine and reduce
uncertainty for the oral CSF.
5.5.1.7.	Jie/evance to Humans
The derivation of the oral CSF is derived using the tumor incidence in the liver of female
mice. A thorough review of the available toxicological data available for 1,4-dioxane provides
no scientific justification to propose that the liver adenomas and carcinomas observed in animal
models due to exposure to 1,4-dioxane are not relevant to humans. As such, liver adenomas and
carcinomas were considered relevant to humans due to exposure to 1,4-dioxane.
5.5.1.8.	Human Population Variability
The extent of inter-individual variability in 1,4-dioxane metabolism has not been
characterized. A separate issue is that the human variability in response to 1,4-dioxane is also
unknown. Data exploring whether there is differential sensitivity to 1,4-dioxane carcinogenicity
across life stages are unavailable. This lack of understanding about potential differences in
metabolism and susceptibility across exposed human populations thus represents a source of
uncertainty. Also, the lack of information linking a MOA for 1,4-dioxane to the observed
carcinogenicity is a source of uncertainty.
Ill
DRAFT - DO NOT CITE OR QUOTE

-------
Table 5-9. Summary of uncertainty in the 1,4-dioxane cancer risk
assessment
Consideration/
approach
Impact on oral slope
factor
Decision
Justification
Low-dose
extrapolation
procedure
Departure from
EPA's Guidelines for
Carcinogen Risk
Assessment POD
paradigm, if justified,
could | or | unit risk
an unknown extent
Log-logistic model
to determine POD,
linear low-dose
extrapolation from
POD
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 MO A, EPA's 2005 Guidelines for
Carcinogen Risk Assessment recommend application
of a linear low-dose extrapolation approach.
Dose metric
Alternatives could t
or I CSF by an
unknown extent
Used administered
exposure
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.
Cross-species
scaling
Alternatives could j
or t CSF [e.g., 3.5-
fold i (scaling by
BW) or | twofold
(scaling by BW°67)]
BW°75 (default
approach)
There are no data to support alternatives. BW°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.
Bioassay
Alternatives could t
or I CSF by an
unknown extent
JBRC 1998a
Alternative bioassays were available and considered
for derivation of oral CSF.
Species /gender
combination
Human risk could j
or |, depending on
relative sensitivity
Female mouse
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.
Human
relevance of
mouse tumor
data
If rodent tumors
proved not to be
relevant to humans,
unit risk would not
apply i.e., could j
CSF
Liver adenomas and
carcinomas are
relevant to humans
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.
Human
population
variability in
metabolism and
response/
sensitive
subpopulations
Low-dose risk f or J,
to an unknown extent
Considered
qualitatively
No data to support range of human
variability/sensitivity, including whether children are
more sensitive.
112	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
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE
6.1. HUMAN HAZARD POTENTIAL
1,4-Dioxane is absorbed rapidly following oral and inhalation exposure, with much less
absorption occurring from the dermal route. 1,4-Dioxane is primarily metabolized to HEAA,
which is excreted in the urine. Liver and kidney toxicity are the primary noncancer health
effects associated with exposure to 1,4-dioxane in humans and laboratory animals. Several fatal
cases of hemorrhagic nephritis and centrilobular necrosis of the liver were related to
occupational exposure (i.e., inhalation and dermal contact) to 1,4-dioxane (Johnstone, 1959;
Barber, 1934). Neurological changes were also reported in one case, including headache,
elevation in blood pressure, agitation and restlessness, and coma (Johnstone, 1959). Perivascular
widening was observed in the brain of this worker, with small foci of demyelination in several
regions (e.g., cortex, basal nuclei). Severe liver and kidney degeneration and necrosis were
observed frequently in acute oral and inhalation studies (> 1,000 mg/kg-day oral, > 1,000 ppm
inhalation) (JBRC, 1998b; Drew et al., 1978; David, 1964; Kesten et al., 1939; Laug et al., 1939;
Schrenk and Yant, 1936; de Navasquez, 1935; Fairley et al., 1934).
Liver and kidney toxicity were the primary noncancer health effects of subchronic and
chronic oral exposure to 1,4-dioxane in animals. Hepatocellular degeneration and necrosis were
observed (Kociba et al., 1974) and preneoplastic changes were noted in the liver following
chronic administration of 1,4-dioxane in drinking water (Kano et al., 2009; JBRC, 1998a, Argus
et al., 1973). Liver and kidney toxicity appear to be related to saturation of clearance pathways
and an increase in the 1,4-dioxane concentration in the blood (Kociba et al., 1975). Kidney
damage was characterized by degeneration of the cortical tubule cells, necrosis with hemorrhage,
and glomerulonephritis (NCI, 1978; Kociba et al., 1974; Argus et al., 1973, 1965; Fairley et al.,
1934).
Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and
guinea pigs (Kano et al., 2009; JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Torkelson et al.,
1974; Argus et al., 1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al.,
1965). Liver tumors (hepatocellular adenomas and carcinomas) have been observed following
drinking water exposure in several species and strains of rats, mice, and guinea pigs. Nasal
(squamous cell carcinomas), peritoneal, and mammary tumors were also observed in rats, but
were not seen in mice. With the exception of the NCI (1978) study, the incidence of nasal cavity
tumors was generally lower than that of liver tumors in the same study population.
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), 1,4-dioxane
can be classified as "likely to be carcinogenic to humans," based on evidence of liver
113
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
carcinogenicity in several 2-year bioassays conducted in three strains of rats, two strains of mice,
and in guinea pigs (Kano et al., 2009; JBRC, 1998a; NCI, 1978; Kociba et al., 1974; Argus et al.,
1973; Hoch-Ligeti and Argus, 1970; Hoch-Ligeti et al., 1970; Argus et al., 1965). Studies in
humans found no conclusive evidence for a causal link between occupational exposure to
1,4-dioxane and increased risk for cancer; however, only two studies were available and these
were limited by small cohort size and a small number of reported cancer cases (Buffler et al.,
1978; Thiess etal., 1976).
The available evidence is inadequate to establish a MOA by which 1,4-dioxane induces
liver tumors in rats and mice. The genotoxicity data for 1,4-dioxane is generally characterized as
negative, although several studies may suggest the possibility of genotoxic effects (Roy et al.,
2005; Morita and Hayashi, 1998; Mirkova, 1994; Kitchin and Brown, 1990; Galloway et al.,
1987). A MOA hypothesis involving sustained proliferation of spontaneously transformed liver
cells has some support by evidence that suggests 1,4-dioxane is a tumor promoter in mouse skin
and rat liver bioassays (Lundberg et al., 1987; King et al., 1973). Some dose-response and
temporal evidence support the occurrence of cell proliferation and hyperplasia prior to the
development of liver tumors (JBRC, 1998a; Kociba et al., 1974). However, the dose-response
relationship for the induction of hepatic cell proliferation has not been characterized, and it is
unknown if it would reflect the dose-response relationship for liver tumors in the 2-year rat and
mouse studies. Conflicting data from rat and mouse bioassays (JBRC, 1998a; Kociba et al.,
1974) suggest that cytotoxicity is not a required precursor event for 1,4-dioxane-induced cell
proliferation. Liver tumors were observed in female rats and female mice in the absence of
lesions indicative of cytotoxicity (Kano et al., 2008; JBRC, 1998a; NCI, 1978). Data regarding a
plausible dose response and temporal progression from cytotoxicity to cell proliferation and
eventual liver tumor formation are not available.
6.2. DOSE RESPONSE
6.2.1. Noncancer/Oral
The RfD of 3 x 10" mg/kg-day was derived based on liver and kidney toxicity in rats
exposed to 1,4-dioxane in the drinking water for 2 years (Kociba et al., 1974). This study was
chosen as the principal study because it provides the most sensitive measure of adverse effects
by 1,4-dioxane. The incidence of liver and kidney lesions was not reported for each dose group.
Therefore, BMD modeling could not be used to derive a POD. Instead, the RfD is derived by
dividing the NOAEL of 9.6 mg/kg-day by a composite UF of 300 (factors of 10 for animal-to-
human extrapolation and interindividual variability, and an UF of 3 for database deficiencies).
Information was unavailable to quantitatively assess toxicokinetic or toxicodynamic differences
between animals and humans and the potential variability in human susceptibility; thus, the
interspecies and intraspecies uncertainty factors of 10 were applied. In addition, a threefold
114	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
database uncertainty factor was applied due to the lack of information addressing the potential
reproductive toxicity associated with 1,4-dioxane.
The overall confidence in the RfD is medium. Confidence in the principal study (Kociba
et al., 1974) is medium. Confidence in the database is medium due to the lack of a
multigeneration reproductive toxicity study. Reflecting medium confidence in the principal
study and medium confidence in the database, confidence in the RfD is medium.
6.2.2.	Noncancer/Inhalation
No inhalation RfC was derived for 1,4-dioxane. Inhalation data were inadequate and a
route extrapolation from oral toxicity data was not performed, due to direct effects of
1,4-dioxane on the respiratory tract (i.e., respiratory irritation in humans, pulmonary congestion
in animals) (Wirth and Klimmer, 1936; Fairley et al., 1934; Yant et al., 1930) and lack of a
suitable kinetic model (see Appendix B).
Note that during the development of this assessment, new data regarding the toxicity of
1,4-dioxane through the inhalation route of exposure became available and have not been
included in the current assessment. The IRIS Program will evaluate the more recently published
1,4-dioxane inhalation in a separate assessment.
6.2.3.	Cancer/Oral
An oral CSF for 1,4-dioxane of 0.10 (mg/kg-day)"1 was based on liver tumors in female
mice from a chronic study (Kano et al., 2009). The available data indicate that the MOA(s) by
which 1,4-dioxane induces peritoneal, mammary, or nasal tumors in rats and liver tumors in rats
and mice is unknown (see Section 4.7.3 for a more detailed discussion of 1,4-dioxane's
hypothesized MO As). Therefore, based on the U.S. EPA Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005a), a linear low dose extrapolation was used. The POD was
calculated by curve fitting the animal experimental dose-response data from the range of
observation and converting it to a HED (BMDL50 hed of 4.96 mg/kg-day).
The uncertainties associated with the quantitation of the oral CSF are discussed below.
6.2.3.1. Choice of Low-Dose Extrapolation Approach
The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible
MOA(s) for the observed tumors. The MOA is a key consideration in clarifying how risks
should be estimated for low-dose exposure. Exposure to 1,4-dioxane has been observed in
animal models to induce multiple tumor types, including liver adenomas and carcinomas, nasal
carcinomas, mammary adenomas and fibroadenomas, and mesothiolomas of the peritoneal cavity
(Kano et al., 2009). MOA information that is available for the carcinogenicity of 1,4-dioxane
has largely focused on liver adenomas and carcinomas, with little or no MOA information
115
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
available for the remaining tumor types. In Section 4.7.3, hypothesized MO As were explored
for 1,4-dioxane. Data are not available to support a carcinogenic MO A for 1,4-dioxane. In the
absence of a MOA(s) for the observed tumor types due to exposure to 1,4-dioxane, a linear low-
dose extrapolation approach was used to estimate human carcinogenic risk associated with
1,4-dioxane exposure.
In general, the Agency has preferred to use the multistage model for analyses of tumor
incidence and related endpoints because they have a generic biological motivation based on
long-established mathematical models such as the MVK model. The MVK model does not
necessarily characterize all modes of tumor formation, but it is a starting point for most
investigations and, much more often than not, has provided at least an adequate description of
tumor incidence data.
In the studies evaluated (Kano et al., 2009; NCI, 1978; Kociba et al., 1974) the multistage
model provided good descriptions of the incidence of a few tumor types in male (nasal cavity)
and female (hepatocellular and nasal cavity) rats and in male mice (hepatocellular) exposed to
1,4-dioxane (see Appendix D for details). However, the multistage model did not provide an
adequate fit for female mouse liver tumor dataset based upon the following (U.S. EPA, 2000b):
•	Goodness-of-fit /;-value was not greater than 0.10;
•	AIC was larger than other acceptable models;
•	Data deviated from the fitted model, as measured by their x residuals (values were
greater than an absolute value of one).
BMDS software typically implements the guidance in the external peer review draft
BMD technical guidance document (U.S.EPA, 2000b) by imposing constraints on the values of
certain parameters of the models. When these constraints were imposed, the multistage model
and most other models did not fit the incidence data for female mouse liver adenomas or
carcinomas.
The log-logistic model was selected because it provides an adequate fit for the female
mouse data (Kano et al., 2009). A BMR of 50% was used because it is proximate to the response
at the lowest dose tested and the BMDL50 was derived by applying appropriate parameter
constraints, consistent with recommended use of BMDS in the external peer review drfat BMD
technical guidance document (U.S. EPA, 2000b).
The human equivalent oral CSF estimated from liver tumor datasets with statistically
significant increases ranged from 4.2 x 10"4 to 0.18 per mg/kg-day, a range of about three orders
of magnitude, with the extremes coming from the combined male and female data for
hepatocellular carcinomas (Kociba et al., 1974) and the female mouse liver adenoma and
carcinoma dataset (Kano et al., 2009).
116
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
6.2.3.2.	Dose Metric
1,4-Dioxane is known to be metabolized in vivo. However, evidence does not exist to
determine whether the parent compound, metabolite(s), or a combination of the parent compound
and metabolites is responsible for the observed toxicity following exposure to 1,4-dioxane. If the
actual carcinogenic moiety is proportional to administered exposure, then use of administered
exposure as the dose metric is the least biased choice. On the other hand, if this is not the correct
dose metric, then the impact on the CSF is unknown.
6.2.3.3.	Cross-Species Scaling
0 75
An adjustment for cross-species scaling (BW ) was applied to address toxicological
equivalence of internal doses between each rodent species and humans, consistent with the 2005
Guidelines for Carcinogen Risk Assessment (US EPA, 2005a). It is assumed that equal risks
result from equivalent constant lifetime exposures.
6.2.3.4.	Statistical Uncertainty at the POD
Parameter uncertainty can be assessed through confidence intervals. Each description of
parameter uncertainty assumes that the underlying model and associated assumptions are valid.
For the log-logistic model applied to the female mouse data, there is a reasonably small degree of
uncertainty at the 10% excess incidence level (the POD for linear low-dose extrapolation).
6.2.3.5.	Bioassaf Selection
The study by Kano et al. (2009) was used for development of an oral CSF. This was a
well-designed study, conducted in both sexes in two species with a sufficient number of animals
per dose group. The number of test animals allocated among three dose levels and an untreated
control group was adequate, with examination of appropriate toxicological endpoints in both
sexes of rats and mice. Alternative bioassays (NCI, 1978; Kociba et al., 1974) are available and
were fully considered for the derivation of the oral CSF.
6.2.3.6.	Choice of Species/Gender
The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the
female mouse, which was thought to be more sensitive than male mice or either sex of rats to the
carcinogenicity of 1,4-dioxane. While all data, both species and sexes reported from the Kano et
al. (2009) study, were suitable for deriving an oral CSF, the female mouse data represented the
most sensitive indicator of carcinogenicity in the rodent model. The lowest exposure level
(66 mg/kg-day or 10 mg/kg-day [HED]) observed a considerable and significant increase in
combined liver adenomas and carcinomas. Additional testing of doses within the range of
control and the lowest dose (66 mg/kg-day or 10 mg/kg-day [HED]) could refine and reduce
uncertainty for the oral CSF.
117
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
6.2.3.7.	fle/evance to Humans
The oral CSF is derived using the tumor incidence in the liver of female mice. A
thorough review of the available toxicological data available for 1,4-dioxane provides no
scientific justification to propose the liver adenomas and carcinomas observed in animal models
due to exposure to 1,4-dioxane are not plausible in humans. Liver adenomas and carcinomas
were considered as a plausible outcome in humans due to exposure to 1,4-dioxane.
6.2.3.8.	Human Population Variability
The extent of inter-individual variability in 1,4-dioxane metabolism has not been
characterized. A separate issue is that the human variability in response to 1,4-dioxane is also
unknown. Data exploring whether there is differential sensitivity to 1,4-dioxane carcinogenicity
across life stages is unavailable. This lack of understanding about potential differences in
metabolism and susceptibility across exposed human populations thus represents a source of
uncertainty. Also, the lack of information linking a MO A for 1,4-dioxane to the observed
carcinogenicity is a source of uncertainty.
6.2.4. Cancer/Inhalation
Inhalation studies for 1,4-dioxane 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). Route extrapolation from oral
bioassay data was not performed because available kinetic models were not considered suitable
for this purpose.
Note that during the development of this assessment, new data regarding the toxicity of
1,4-dioxane through the inhalation route of exposure became available and have not been
included in the current assessment. The IRIS Program will evaluate the more recently published
1,4-dioxane inhalation data in a separate assessment.
118
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 ChemRefData. 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, RJ; 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.
119
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. J Hyg 35:540-548.
Derosa, CT; Wilbur, S; Holler, J; et al. (1996) Health evaluation of 1,4-dioxane. Toxicol Ind Health 12(1): 1-43.
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.
Ernstgard, 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
Assoc J58(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.
120
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 Ser 21 National
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 71:482^188.
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 in F344 rats and BDFi mice (drinking water studies). Kanagawa,
Japan: Japan Bioassay Research Center.
Johnstone, RT. (1959) Death due to dioxane? AMA Arch Ind Health 20:445-447.
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, H; Umeda, Y; Kasai, T; et al. (2009) Carcinogenicity studies of 1,4-dioxane administered in drinking-water to
rats and mice for 2 years. Food Chem Toxicol 47:2776-2784.
Kano, J; Umeda, Y; Saito, M; et al. (2008) Thirteen-week oral toxicity of 1,4-dioxane in rats and mice. J Toxicol
Sci 33(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. MutRes 627:106-116.
Katagiri, T; Nagano, K; Shigetoshi, A; et al. (1998) A pathological study on spontaneous hepatic neoplasms in
BDFI mice. J Toxicol Pathol 11:21-25.
Kesten, HD; Mulinos, MG; Pomerantz L. (1939) Pathologic effects of certain glycols and related compounds. Arch
Pathol 27:447-465.
Khudoley, VV; 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 57:453-462.
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.
121	DRAFT - DO NOT CITE OR QUOTE

-------
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.
Kociba, RJ; Torkelson, TR; Young, JD; et al. (1975) 1,4-dioxane: correlation of the results of chronic ingestion and
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; et al. (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. Environ Pollut 65:323-332.
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. MutatRes 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.
122
DRAFT - DO NOT CITE OR QUOTE

-------
Mirkova, ET. (1994) Activity of the rodent carcinogen 1,4-dioxane in the mouse bone marrow micronucleus assay.
Mutat Res 322:142-144.
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.
MMS Commun 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
in Drosophila melanogaster females. Mutat Res 517(1—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. J Ind Hyg 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 CD1 mice. Mutation Research 586:28-37.
123
DRAFT - DO NOT CITE OR QUOTE

-------
Schrenk, HH; Yant, WP. (1936) Toxicity of dioxane. J Ind Hyg Toxicol 18:448-460.
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/mutagenic potential. Mutat Res 113:357-3 91.
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(l):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
124
DRAFT - DO NOT CITE OR QUOTE

-------
U.S. EPA. (1988) Recommendations for and documentation of biological values for use in risk assessment. EPA
600/6-87/008. Available from National Technical Information Service, Springfield, VA; PB88-179874/AS.
U.S. EPA. (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.
125
DRAFT - DO NOT CITE OR QUOTE

-------
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.
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, 0. (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. Biochem Pharmacol 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 carciongenesis 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 and beta-
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.
126
DRAFT - DO NOT CITE OR QUOTE

-------
Young, JD; Braun, WH; Gehring, PJ. (1978a) The dose-dependent fate of 1,4-dioxane in rats. J Environ Pathol
Toxicol 2:263-282.
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.
127
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
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION
The Toxicological Review of 1,4-Dioxane has undergone formal external peer review
performed by scientists in accordance with EPA guidance on peer review (U.S. EPA, 2006a,
2000a). The external peer reviewers were tasked with providing written answers to general
questions on the overall assessment and on chemical-specific questions in areas of scientific
controversy or uncertainty. A summary of significant comments made by the external reviewers
and EPA's responses to these comments arranged by charge question follow. In many cases the
comments of the individual reviewers have been synthesized and paraphrased for development of
Appendix A. The majority of the specific observations (in addition to EPA's charge questions)
made by the peer reviewers were incorporated into the document and are not discussed further in
this Appendix. Public comments that were received are summarized and addressed following the
peer-reviewers' comments and disposition.
EXTERNAL PEER REVIEW PANEL COMMENTS
The reviewers made several editorial suggestions to clarify portions of the text. These
changes were incorporated in the document as appropriate and are not discussed further.
In addition, the external peer reviewers commented on decisions and analyses in the
Toxicological Review of 1,4-Dioxane under multiple charge questions, and these comments were
organized and summarized under the most appropriate charge question.
A. General Charge Questions
1. Is the Toxicological Review logical, clear and concise? Has EPA accurately, clearly and
objectively represented and synthesized the scientific evidence for noncancer and cancer
hazards?
Comment. All reviewers found the Toxicological Review to be logical, clear, and concise.
One reviewer remarked that it was an accurate, open-minded and balanced analysis of the
literature. Most reviewers found that the scientific evidence was presented objectively
and transparently; however, one reviewer suggested two things to improve the objectivity
and transparency (1) provide a clear description of the mode of action and how it feeds
into the choice of the extrapolation for the cancer endpoint and (2) provide a presentation
of the outcome if internal dose was used in the cancer and noncancer assessments.
One reviewer commented that conclusions could not be evaluated in a few places
where dose information was not provided (sections 3.2, 3.3 and 4.5.2.2). The same
A-l	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
reviewer found the MOA schematics, key event temporal sequence/dose-response table,
and the POD plots to be very helpful in following the logic employed in the assessment.
Response. The mode of action analysis and how conclusions from that analysis fed into
the choice of extrapolation method for the cancer assessment are discussed further under
charge questions C2 and C5. Because of the decision not to utilize the PBPK models,
internal doses were not calculated and thus were not included as alternatives to using the
external dose as the POD for the cancer and noncancer assessments.
In the sections noted by the reviewer (3.2, 3.3, and 4.5.2.2) dose information was
added as available. In section 3.2, Mikheev et al. (1990) did not report actual doses,
which is noted in this section. All other dose information in this section was found to be
present after further review by the Agency. In section 3.3, dose information for Woo et
al. (1978, 1977c) was added to the paragraph. In section 4.5.2.2, study details for
Nannelli et al. (2005) were provided earlier in section 3.3 and a statement referring the
reader to this section was added.
2. Please identify any additional studies that should be considered in the assessment of the
noncancer and cancer health effects of 1,4-dioxane.
Comment. Five reviewers stated they were unaware of any additional studies available to
add to the oral toxicity evaluation of 1,4-dioxane. These reviewers also acknowledged
the Kasai et al. (2009, 2008) publications that may be of use to derive toxicity values
following inhalation of 1,4-dioxane.
a.	Kasai T; Saito H; Senoh Y; et al. (2008) Thirteen-week inhalation toxicity of 1,4-
dioxane in rats. Inhal Toxicol 20: 961-971.
b.	Kasai T; Kano Y; Umeda T; et al. (2009) Two-year inhalation study of
carcinogenicity and chronic toxicity of 1,4-dioxane in male rats. Inhal Toxicol in
press.
Other references suggested by reviewers include:
c.	California Department of Health Services (1989) Risk Specific Intake Levels for
the Proposition 65 Carcinogen 1, 4-dioxane. Reproductive and Cancer Hazard
Assessment Section. Office of Environmental Health Hazard Assessment
d.	National Research Council (2009) Science and Decisions: Advancing Risk
Assessment. Committee on Improving Risk Analysis Approaches Used by the
U.S. EPA. Washington, D.C., National Academy Press.
e.	ATSDR (2007) Toxicological Profile for 1,4-dioxane. Agency for Toxic
Substances and Disease Registry. Atlanta, GA.
f.	Stickney JA; Sager SL; Clarkson JR; et al. (2003) An updated evaluation of the
carcinogenic potential of 1,4-dioxane. Regul Toxicol Pharmacol 38: 183-195.
A-2
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
g. Yamamoto S; Ohsawa M; Nishizawa T; et al. (2000) Long-term toxicology
study of 1,4-dioxane in R344 rats by multiple-route exposure (drinking water and
inhalation). J Toxicol Sci 25: 347.
Response: The references a-b above will be evaluated for derivation of an RfC and IUR,
which will follow as an update to this oral assessment. References c and e noted above
were considered during development of this assessment as to the value they added to the
cancer and noncancer analyses. Reference g listed above is an abstract from conference
proceedings from the 27th Annual Meeting of the Japanese Society of Toxicology;
abstracts are not generally considered in the development of an IRIS assessment.
Reference d reviews EPA's current risk assessment procedures and provides no specific
information regarding 1,4-dioxane. The Stickney et al. (2003) reference (letter f above)
was a review article and no new data were presented, thus it was not referenced in this
Toxicological Review but the data were considered during the development of this
assessment.
Following external peer review (as noted above) Kano et al. (2009) was added to
the assessment, which was an update and peer-reviewed published manuscript of the
JBRC (1998a) report.
3. Please discuss research that you think would be likely to increase confidence in the database
for future assessments of 1,4-dioxane.
Comment. All reviewers provided suggestions for additional research that would
strengthen the assessment and reduce uncertainty in several areas. The following is a
brief list of questions that were identified that could benefit from further research. What
are the mechanisms responsible for the acute and chronic nephrotoxicity? Is the acute
kidney injury (AKI) multifactorial? Are there both tubular and glomerular/vascular
toxicities that result in cortical tubule degeneration and evidence for glomerulonephrities?
What are the functional correlates of the histologic changes in terms of assessment of
renal function? What is the exposure in utero and risk to the fetus and newborn? What are
the concentrations in breast milk following maternal exposure to 1,4-dioxane? What is
the risk for use of contaminated drinking water to reconstitute infant formula? What are
the exposures during early human development? What is the pharmacokinetic and
metabolic profile of 1,4-dioxane during development? What are the susceptible
populations (e.g., individuals with decreased renal function or chronic renal disease,
obese individuals, gender, age)?
Additional suggestions for future research include: evaluation of potential
epigenetic mechanisms of carcinogenicity, additional information on sources of exposure
and biological concentrations as well as human toxicokinetic data for derivation of
A-3
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
parameter to refine PBPK model, studies to determine toxic moiety, focused studies to
inform mode of action, additional inhalation studies and a multigeneration reproductive
toxicity study.
One reviewer suggested additional analyses of the existing data including a
combined analysis of the multiple datasets and outcomes for cancer and non-cancer
endpoints, evaluation of the dose metrics relevant to the MOA to improve confidence in
extrapolation approach and uncertainty factors, and complete a Bayesian analysis of
human pharmacokinetic data to estimate human variability in key determinants of
toxicity (e.g., metabolic rates and partition coefficients).
Response. A number of research suggestions were provided for further research that may
enhance future health assessments of 1,4-dioxane. Regarding the suggested additional
analyses for the existing data, EPA did not identify a MOA in this assessment, thus
combined analysis of the cancer and non-cancer endpoints as well as application of
various dose metrics to a MOA is not applicable. Because the human PBPK model was
not implemented in this assessment for oral exposure to 1,4-dioxane a Bayesian analysis
was not completed. No additional changes to the Toxicological Review of 1,4-Dioxane
were made in response to these research recommendations.
4. Please comment on the identification and characterization of sources of uncertainty in
Sections 5 and 6 of the assessment document. Please comment on whether the key sources of
uncertainty have been adequately discussed. Have the choices and assumptions made in the
discussion of uncertainty been transparently and objectively described? Has the impact of the
uncertainty on the assessment been transparently and objectively described?
Comment. Six reviewers stated Sections 5 and 6 adequately discussed and characterized
uncertainty, in a succinct, and transparent manner. One reviewer suggested adding
additional discussion of uncertainty relating to the critical study used in the cancer
assessment and another reviewer suggested adding more discussion around the
uncertainty of the toxic moiety.
One reviewer made specific comments on uncertainty surrounding the Kociba et
al. (1974) study as used for derivation of the RfD, choice of the non-cancer dose metric,
and use of a 10%BMR as the basis for the CSF derivation. These comments and
responses are summarized below under their appropriate charge question.
Response. The majority of the reviewers thought the amount of uncertainty discussion
was appropriate. Since the external review, Kano et al. (2009) was published and this
assessment was updated accordingly (previously JBRC (1998a). It is assumed the
A-4
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
uncertainty referred to by the reviewer was addressed by the published Kano et al. (2009)
paper.
Clarification regarding the uncertainty surrounding the identification of the toxic
moiety was added to section 4.6.3 stating that the mechanism by which 1,4-dioxane
induces tissue damage is not known, nor is it known whether the toxic moiety is 1,4-
dioxane or a metabolite of 1,4-dioxane. Additional text was added to section 4.7.3
clarifying that available data also do not clearly identify whether 1,4-dioxane or one of its
metabolites is responsible for the observed effects. The impact of the lack of evidence to
clearly identify a toxic moiety related to 1,4-dioxane exposure was summarized in
sections 5.5.1.2 and 6.2.3.2.
B. Oral reference dose (RfD) for 1,4-dioxane
1. A chronic RfD for 1,4-dioxane has been derived from a 2-year drinking water study (Kociba
et al., 1974) in rats and mice. Please comment on whether the selection of this study as the
principal study has been scientifically justified. Has the selection of this study been
transparently and objectively described in the document? Are the criteria and rationale for
this selection transparently and objectively described in the document? Please identify and
provide the rationale for any other studies that should be selected as the principal study.
Comment. Seven of the reviewers agreed that the use of the Kociba et al. (1974) study
was the best choice for the principal study.
One reviewer stated that Kociba et al. (1974) was not the best choice because it
reported only NOAEL and LOAELs without providing incidence data for the endpoints.
This reviewer also stated that the study should not have been selected based on sensitivity
of the endpoints, but rather study design and adequacy of reporting of the study results.
Additionally, this reviewer suggested a better principal study would be either the NCI
(1978) or JBRC (1998a) study.
Response: The reviewer is correct that Kociba et al. (1974) did not provide incidence
data; however, Kociba et al. (1974) identified a NOAEL (9.6 mg/kg-day) and LOAEL
(94 mg/kg-day) within the text of the manuscript. Kociba et al. (1974) was a well
conducted chronic bioassay (four dose levels, including controls, with 60 rats/sex/group)
and seven of the peer reviewers found this study to be appropriate as the basis for the
RfD. Further support for the selection of the Kociba et al. (1974) as the principal study
comes from comparison of the liver and kidney toxicity data reported by JBRC (1998a)
and NCI (1978), which was presented in Section 5.1. The effects reported by JBRC
(1998a) and NCI (1978) were consistent with what was observed by Kociba et al. (1974)
A-5
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
and within a similar dose range. Derivation of an RfD from these datasets resulted in a
similar value (section 5.1.).
2.	Degenerative liver and kidney effects were selected as the critical effect. Please comment on
whether the rationale for the selection of this critical effect has been scientifically justified.
Are the criteria and rationale for this selection transparently and objectively described in the
document? Please provide a detailed explanation. Please comment on whether EPA's
rationale regarding adversity of the critical effect for the RfD has been adequately and
transparently described and is scientifically supported by the available data. Please identify
and provide the rationale for any other endpoints that should be considered in the selection of
the critical effect.
Comment. Five of the reviewers agreed with the selection of liver and kidney effects as
the critical effect. One of these reviewers suggested analyzing all datasets following dose
adjustment (e.g., body weight scaling or PBPK model based) to provide a better rationale
for selection of a critical effect.
One reviewer stated that 1,4-dioxane causing liver and kidney organ specific
effects is logical; however, with regards to nephrotoxicity, the models and limited human
data have not addressed the mechanisms of injury or the clinical correlates to the
histologic data. Also, advances in the field of biomarkers have not yet been used for the
study of 1,4-dioxane.
One reviewer found the selection of these endpoints to be 'without merit' because
of the lack of incidence data to justify the NOAEL and LOAEL values identified in the
study. This reviewer suggested selecting the most sensitive endpoint(s) from the NCI
(1978) or JBRC (1998) studies for the basis of the RfD, but did not provide a suggestion
as to what effect should be selected.
Response. The liver and kidney effects from Kociba et al. (1974) was supported as the
critical effect by most of the reviewers. PBPK model adjustment was not performed
because the PBPK model was found to be inadequate for use in the assessment. EPA
acknowledges that neither the mechanisms of injury nor the clinical correlates to
histologic data exist for 1,4-dioxane. This type of information could improve future
health assessments of 1,4-dioxane.
	As stated above, Kociba et al. (1974) identified a NOAEL (9.6 mg/kg-day) and
LOAEL (94 mg/kg-day) within the text of the manuscript and was a well conducted
chronic bioassay (four dose levels, including controls, with 60 rats/sex/group).
3.	Kociba et al. (1974) derived a NOAEL based upon the observation of degenerative liver and
kidney effects and these data were utilized to derive the point of departure (POD) for the
A-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
RfD. Please provide comments with regard to whether the NOAEL approach is the best
approach for determining the POD. Has the approach been appropriately conducted and
objectively and transparently described? Please identify and provide rationales for any
alternative approaches for the determination of the POD and discuss whether such
approaches are preferred to EPA's approach.
Comment. Seven reviewers agreed with the NOAEL approach described in the
document. One of these reviewers also questioned whether any attempt was made to
"semi-qualitatively represent the histopathological observations to facilitate a quantitative
analysis".
One reviewer stated that data were not used to derive the POD, but rather a claim
by the authors of Kociba et al. (1974) of the NOAEL and LOAEL for the endpoints. This
reviewer preferred the use of a BMD approach for which data include the reported
incidence rather than a study reported NOAEL or LOAEL.
Response. The suggestion to "semi-qualitatively represent the histopathological
observations to facilitate a quantitative analysis" was not incorporated into the document
because it is unclear how this would be conducted since Kociba et al. (1974) did not
provide incidence data and the reviewer did not illustrate their suggested approach. See
responses to B1 and B2 regarding the NOAEL and LOAEL approach. The Agency
agrees that a Benchmark Dose approach is preferred over the use of a NOAEL or
LOAEL for the POD if suitable data (e.g., reflecting the most sensitive sex, species, and
endpoint identified) are available for modeling and, if suitable data are not available, then
NOAEL and LOAEL values are utilized. In this case, the data were not suitable for
BMD modeling and the LOAEL or NOAEL approach was used.
4. EPA evaluated the PBPK and empirical models available to describe kinetics following
inhalation of 1,4-dioxane (Reitz et al., 1990; Young et al., 1978, 1977). EPA concluded that
the use of existing, revised, and recalibrated PBPK models for 1,4-dioxane were not superior
to default approaches for the dose-extrapolation between species. Please comment on
whether EPA's rationale regarding the decision to not utilize existing or revised PBPK
models has been adequately and transparently described and is supported by the available
data. Please identify and provide the rationale for any alternative approaches that should be
considered or preferred to the approach presented in the toxicological review.
Comment. Six reviewers found the decision not to utilize the available PBPK models to
be appropriate and supported by available data. One of these reviewers suggested
presenting as part of the uncertainty evaluation an adjustment of the experimental doses
based on metabolic saturation. Another reviewer stated Appendix B was hard to follow
A-7
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
and that the main document should include a more complete description of the model
refinement effort performed by Sweeney et al. (2008).
Two reviewers noted a complete evaluation of the models was evident; one of the
reviewers questioned the decision not to use the models on the basis that they were
unable to fit the human blood PK data for 1,4-dioxane. This reviewer suggested the rat
model might fit the human blood PK data, thus raising concern in the reliance on the
human blood PK data to evaluate the PBPK model for 1,4-dioxane. Instead, the reviewer
suggested the human urinary metabolite data may be sufficient to give confidence in the
model. One other reviewer also questioned the accuracy of the available human data.
One reviewer commented that the the rationale for not using the PBPK model to
extrapolate from high to low dose was questioned. In addition, the reviewer suggested
that two aspects of the model code for Reitz et al. (1990) need to be verified:
a.	In the document, KLC is defined as a first-order rate constant and is scaled by
0 7
BW . This is inconsistent when multiplied by concentration does not result
in units of mg/hr. However, if the parameter is actually considered a
clearance constant (zero-order rate constant) then the scaling rule used, as well
as the interpretations provided, would be acceptable.
b.	It is unclear as to why AM is calculated on the basis of RAM and not RMEX.
RMEX seems to represent the amount metabolized per unit time.
Response: The USEPA performed a rigorous evaluation of the PBPK models available
for 1,4-dioxane. This effort was extensively described in Section 3.5 and in Appendix B.
In short, several procedures were applied to the human PBPK model to determine if an
adequate fit of the model to the empirical model output or experimental observations
could be attained using biologically plausible values for the model parameters. The re-
calibrated model predictions for blood 1,4-dioxane levels did not come within 10-fold of
the experimental values using measured tissue:air partition coefficients of Gargas et al.
(1989) (Leung and Paustenbach, 1990) or Soelberg et al. (2006; Sweeney et al., 2007)
(Figures B-8 and B-9). The utilization of a slowly perfused tissue:air partition coefficient
10-fold lower than measured values produces exposure-phase predictions that are much
closer to observations, but does not replicate the elimination kinetics (Figure B-10). Re-
calibration of the model with upper bounds on the tissue:air partition coefficients results
in predictions that are still six- to sevenfold lower than empirical model prediction or
observations (Figures B-12 and B-13). Exploration of the model space using an
assumption of first-order metabolism (valid for the 50 ppm inhalation exposure) showed
that an adequate fit to the exposure and elimination data can be achieved only when
unrealistically low values are assumed for the slowly perfused tissue:air partition
coefficient (Figure B-16). Artificially low values for the other tissue:air partition
A-8
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
coefficients are not expected to improve the model fit, as these parameters are shown in
the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km.
In the absence of actual measurements for the human slowly perfused tissue:air partition
coefficient, high uncertainty exists for this model parameter value. Differences in the
ability of rat and human blood to bind 1,4-dioxane may contribute to the difference in Vd.
However, this is expected to be evident in very different values for rat and human
blood:air partition coefficients, which is not the case (Table B-l). Therefore, some other,
as yet unknown, modification to model structure may be necessary.
The results of USEPA's model evaluation were confirmed by other investigators
(Sweeney et al., 2008). Sweeney et al. (2008) concluded that the available PBPK model
with refinements resulted in an under-prediction of human blood levels for 1,4-dioxane
by six- to seven fold. It is anticipated that the high uncertainty in predictions of the
PBPK model for 1,4-dioxane would not result in a more accurate derivation of human
health toxicity values.
Because it is unknown whether the parent or the metabolite is the toxic moiety,
analyses were not conducted to adjust the experimental doses on the basis of metabolic
saturation.
The discussion of Sweeney et al. (2008) was expanded in the main document in
section 3.5.3. In the absence of evidence to the contrary, the Agency cannot discount the
human blood kinetic data published by Young et al. (1977). Even though the PBPK
model provided satisfactory fits to the rodent kinetic data, it was not used to extrapolate
from high dose to low dose in the animal because an internal dose metric was not
identified and external doses were utilized in derivation of the toxicity values.
KLC was implemented by USEPA during the evaluation of the model and should
have been described as a clearance constant (zero-order rate constant) with units of
L/hr/kg0'70. These corrections have been made in the document; however, this does not
impact the model predictions because it was in reference to the terminology used to
describe this constant.
The reviewer is correct that RMEX is the rate of metabolism of 1,4-dioxane per
unit time; however an amount of 1,4-dioxane metabolized was not calculated in the Reitz
et al. (1990) model code. Thus, AM is the amount of the metabolite (i.e., HEAA) in the
body rather than the amount metabolized of 1,4-dioxane. RAM was published by Reitz
et al. (1990) as equation 2 for the change in the amount of metabolite in the body per unit
time. AMEX is the amount of the metabolite excreted in the urine. While the variables
used are confusing, the code describes the metabolism of 1,4-dioxane as published in the
manuscripts. The comments in the model code were updated to make this description
more clear (see Appendix B).
A-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
5. Please comment on the selection of the uncertainty factors applied to the POD for the
derivation of the RfD. For instance, are they scientifically justified and transparently and
objectively described in the document? If changes to the selected uncertainty factors are
proposed, please identify and provide a rationale(s). Please comment specifically on the
following uncertainty factors:
•	An interspecies uncertainty factor of 10 was used to account for uncertainties in
extrapolating from laboratory animals to humans because a PBPK model to support
interspecies extrapolation was not suitable.
•	An intraspecies (human variability) uncertainty factor of 10 was applied in deriving the
RfD because the available information on the variability in human response to 1,4-
dioxane is considered insufficient to move away from the default uncertainty factor of
10.
•	A database uncertainty factor of 3 was used to account for lack of adequate
reproductive toxicity data for 1,4-dioxane, and in particular absence of a
multigeneration reproductive toxicity study. Has the rationale for the selection of these
uncertainty factors been transparently and objectively described in the document?
Please comment on whether the application of these uncertainty factors has been
scientifically justified.
Comment.
One reviewer noted the uncertainty factors appear to be the standard default choices and
had no alternatives to suggest,
o Five reviewers agreed that the use of an uncertainty factor of 10 for the interspecies
extrapolation is fully supportable. One reviewer suggested using BW3 4 scaling
rather than an uncertainty factor of 10 for animal to human extrapolation. Along
the same lines, one reviewer suggested a steady-state quantitative analysis to
determine the importance of pulmonary clearance and hepatic clearance and stated
that if hepatic clearance scales to body surface and pulmonary clearance is
negligible, then an adjusted uncertainty factor based on body surface scaling would
be more appropriate,
o Seven reviewers stated that the uncertainty factor of 10 for interindividual
variability (intraspecies) is fully supportable,
o Six reviewers commented that the uncertainty factor of 3 for database deficiencies
is fully justifiable. One reviewer suggested adding text to clearly articulate the
science policy for the use of a factor of 3 for database deficiencies.
A-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
Response. Body weight scaling based on body surface for noncancer endpoints is not
standard practice within the Agency and the default was implemented in this assessment.
The text states in section 5.1.3 that because of the absence of a multigenerational
reproductive study for 1,4-dioxane an uncertainty factor of 3 was used for database
deficiencies. No other changes regarding the use of the uncertainty factors were made to
the document.
C. Carcinogenicity of 1,4-dioxane
1.	Under the EPA's 2005 Guidelines for Carcinogen Risk Assessment
(www.epa.gov/iris/backgr-d.htm), the Agency concluded that 1,4-dioxane is likely to be
carcinogenic to humans. Please comment on the cancer weight of evidence characterization.
Has the scientific justification for the weight of evidence descriptor been sufficiently,
transparently and objectively described? Do the available data for both liver tumors in rats
and mice and nasal, mammary, and peritoneal tumors in rats support the conclusion that 1,4-
dioxane is a likely human carcinogen?
Comment. All reviewers agreed with the Agency's conclusion that 1,4-dioxane is likely
to be carcinogenic to humans. However, two reviewers also thought 1,4-dioxane could
be categorized as a potential human carcinogen, since low-dose environmental exposures
would be unlikely to result in cancer. One reviewer also suggested providing a brief
recapitulation of the guidance provided by the 2005 Guidelines for Carcinogen Risk
Assessment regarding classification of a compound as likely to be carcinogenic to
humans and how a chemical falls into this category.
Response. The document includes a weight-of-evidence approach to categorize the
carcinogenic potential of 1,4-dioxane. This was included in Section 4.7.1 based upon the
Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a). 1,4-Dioxane can be
described as likely to be carcinogenic to humans based on evidence of liver
carcinogenicity in several 2-year bioassays conducted in three strains of rats, two strains
of mice, and in guinea pigs. Additionally, tumors in other organs and tissues have been
observed in rats due to exposure to 1,4-dioxane.
2.	Evidence indicating the mode of action of carcinogenicity of 1,4-dioxane was considered.
Several hypothesized MO As were evaluated within the Toxicol ogical Review and EPA
reached the conclusion that a MOA(s) could not be supported for any tumor types observed
in animal models. Please comment on whether the weight of the scientific evidence supports
this conclusion. Please comment on whether the rationale for this conclusion has been
transparently and objectively described. Please comment on data available for 1,4-dioxane
that may provide significant biological support for a MOA beyond what has been described
A-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
in the Toxicological Review. Considerations should include the scientific support regarding
the plausibility for the hypothesized MOA(s), and the characterization of uncertainty
regarding the MOA(s).
Comment. Three reviewers commented that the weight of evidence clearly supported the
conclusion that a mode of action could not be identified for any of the tumor sites. One
reviewer commented that there is inadequate evidence to support a specific MOA with
any confidence and low-dose linear extrapolation is necessary. The reviewer also pointed
out that EPA should not rule out a metabolite as the toxic moiety.
One reviewer stated this was outside of their area of expertise but indicated that
the discussion was too superficial and suggested adding statements as to what the Agency
would consider essential information to make a determination about a MOA.
Two reviewers commented that even though the MOA for 1,4-dioxane is not clear
there is substantial evidence that the MOA is non-genotoxic, and one reviewer suggested
a non-linear cancer risk assessment model should be utilized.
One reviewer suggested adding more text to the summary statement to fully
reflect the MOA information available whichshould be tied to the conclusion and choice
of an extrapolation model.
Response. The Agency agrees with the reviewer not to rule out a toxic metabolite as the
toxic moiety. In Section 5.5.1.2 text is included relating that there is not enough
information to determine whether the parent compound, its metabolite, or a combination
is responsible for the observed toxicities following exposure to 1,4-dioxane.
It is not feasible to describe the exact data that would be necessary to conclude
that a particular MOA was operating to induce the tumors observed following 1,4-
dioxane exposure. In general, the data would fit the general criteria described in the 2005
Guidelines for Carcinogen Risk Assessment. For 1,4-dioxane, several MOA hypotheses
have been proposed and are explored for the observed liver tumors in Section 4.7.3. This
analysis represents the extent to which data could provide support for any particular
MOA.
One reviewer suggested that the evidence indicating that 1,4-dioxane is not
genotoxic supports a nonlinear approach to low-dose extrapolation. Following the 2005
Cancer Guidelines, the absence of evidence for genotoxicity does not invoke the use of
nonlinear low-dose extrapolation, nor does it define a MOA. A nonlinear low-dose
extrapolation can be utilized when a MOA supporting a nonlinear dose response is
identified. For 1,4-dioxane this is not the case; a cancer MOA for any of the tumor types
observed in animal models has not been elucidated. Therefore, as concluded in the
Toxicological Review, the application of a nonlinear low-dose extrapolation approach
was not supported.
A-12
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
Additional text has been added to Section 5.4.3.2 to relay the fact that several
reviewers recommended that the MOA data support the use of a non-linear extrapolation
approach to estimate human carcinogenic risk associated with exposure to 1,4-dioxane
and that such an approach should be presented in the Toxicological Review. Additional
text has also been added to the summary statement in section 6.2.3 stating that the weight
of evidence is inadequate to establish a MOA(s) by which 1,4-dioxane induces peritoneal,
mammary, or nasal tumors in rats and liver tumors in rats and mice (see Section 4.7.3 for
a more detailed discussion of 1,4-dioxane's hypothesized MOAs).
3. A two-year drinking water cancer bioassay (JBRC, 1998a) was selected as the principal study
for the development of an oral slope factor (OSF). Please comment on the appropriateness of
the selection of the principal study. Has the rationale for this choice been transparently and
objectively described?
Comment.
Seven reviewers agreed with the choice of the JBRC (1998a) study as the
principal study for the development of an OSF. However, two reviewers that agreed with
the choice of JBRC (1998a) also commented on the description and evaluation of the
study. One reviewer commented the evaluation of the study should be separated from the
evaluation/selection of endpoints within the study. The other reviewer suggested that
details on the following aspects should be added to improve transparency of the study: (1)
rationale for selection of doses; (2) temporal information on body weight for individual
treatment groups; (3) temporal information on mortality rates; and (4) dosing details.
One reviewer thought that the complete rationale for selection of the JBRC
(1998a) study was not provided because there was no indication of whether the study was
conducted under GLP conditions, and the study was not peer reviewed or published. This
reviewer noted the NCI (1978) study was not appropriate for use, but that the Kociba et
al. (1974) study may have resulted in a lower POD had they employed both sexes of mice
and combined benign and malignant tumors.
Response: Since the Toxicological Review of 1,4-Dioxane completed external peer
review, the cancer portion of the JBRC (1998a) study was published in the peer-reviewed
literature as Kano et al. (2009). This manuscript was reviewed by EPA and it was
determined that the data published by Kano et al. (2009) should be used in the assessment
of 1,4-dioxane for several reasons: (1) while the JBRC (1998a) was a detailed laboratory
report, it was not peer-reviewed; (2) the JBRC improved the diagnosis of pre- and
neoplastic lesions in the liver according to the current diagnostic criteria and submitted
the manuscript based on this updated data; (3) the Kano et al. (2009) peer-reviewed
A-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
manuscript included additional information such as body weight growth curves and
means and standard deviations of administered dose for both rats and mice of both sexes.
The Toxicological Review was updated to reflect the inclusion of the data from
Kano et al. (2009). Text was added to Section 4.2.1.2.6 regarding the choice of high dose
selection as included in the Kano et al. (2009) manuscript. Dose information was
updated throughout the assessment and are also provided in detail in Section 4.2.1.2.6,
along with temporal information on body weights and mortality. Documentation that the
study was conducted in accordance with Organization for Economic Co-operation and
Development (OECD) Principles of Good Laboratory Practice (GLP) is provided in the
manuscript and this was added to the text in Section 4.2.1.2.6.
4. Combined liver tumors (adenomas and carcinomas) in female Cjr:BDFi mice from the JBRC
(1998a) study were chosen as the most sensitive species and gender for the derivation of the
final OSF. Please comment on the appropriateness of the selections of species and gender.
Please comment on whether the rationale for these selections is scientifically justified. Has
the rationale for these choices been transparently and objectively described?
Comment. Six reviewers agreed the female Cjr:BDFi mice should be used for the
derivation of the OSF. Five of these reviewers agreed with the rationale for the selection
of the female Cjr:BDFi mouse as the most sensitive gender and species. However, one
reviewer suggested that the specific rationale (i.e., that the final OSF is determined by
selecting the gender/species that gives the greatest OSF value) be stated clearly in a
paragraph separate from the other considerations of study selection.
One reviewer was unsure of both the scientific justification for combining benign
and malignant liver tumors, as well as the background incidence of the observed liver
tumors in historical control Cjr:BDFi male and female mice.
One reviewer commented that the scientific basis for the selection of female
Cjr:BDFi mice was unclear. This reviewer thought that the rationale for the choice of
this strain/sex compared to all others was not clearly articulated.
Response: Using the approach described in the Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 2005a) studies were first evaluated based on their quality and
suitability for inclusion in the assessment. Once the studies were found to be of sufficient
quality for inclusion in the assessment, the dose-response analysis was performed with
the goal of determining the most appropriate endpoint and species for use in the
derivation of an OSF. These topics are discussed in detail in Section 4.7 and 5.4.
Benign and malignant tumors that arise from the same cell type (e.g.,
hepatocellular) may be combined to more clearly identify the weight of evidence for a
chemical. This is in accordance with the US EPA's 2005 Guidelines for Carcinogen Risk
A-14
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
Assessment as referenced in the Toxicological Review. In the absence of a MOA (MOA
analysis described in detail in Section 4.7.) for 1,4-dioxane carcinogenicity, it is not
possible to determine which species may more closely resemble humans. Text in Section
5.4.4 indicates that the calculation of an OSF for 1,4-dioxane is based upon the dose-
response data for the most sensitive species and gender.
5. Has the scientific justification for deriving a quantitative cancer assessment been
transparently and objectively described? Regarding liver cancer, a linear low-dose
extrapolation approach was utilized to derive the OSF. Please provide detailed comments on
whether this approach to dose-response assessment is scientifically sound, appropriately
conducted, and objectively and transparently described in the document. Please identify and
provide the rationale for any alternative approaches for the determination of the OSF and
discuss whether such approaches are preferred to EPA's approach.
Comment. Four reviewers agreed with the approach for the dose-response assessment.
One reviewer commented that even if a nongenotoxic MOA were identified for 1,4-
dioxane it may not be best evaluated by threshold modeling. One reviewer commented
the use of the female mouse data provided an appropriate health protective and
scientifically valid approach.
One reviewer commented that the basic adjustments and extrapolation method for
derivation of the OSF were clearly and adequately described, but disagreed with the
linear low-dose extrapolation. This reviewer suggested that the lack of certainty regarding
the MOA was not a sufficient cause to default to a linear extrapolation. Another reviewer
commented that the rationale for a linear low-dose extrapolation to derive the OSF was
not clear, but may be in accordance with current Agency policy in the absence of a
known MOA. This reviewer also commented that 1,4-dioxane appears to be non-
genotoxic and non-linear models should be tested on the available data to determine if
they provide a better fit and are more appropriate.
One reviewer thought that the justification for a linear extrapolation was not
clearly provided and that a disconnect between the MOA summary and the choice of a
linear extrapolation model existed. In addition, this reviewer commented that the
pharmacokinetic information did not support the use of a linear extrapolation approach,
but rather use of animal PBPK models to extrapolate from high to low dose that would
result in a mixture of linear and nonlinear extrapolation models was warranted.
One reviewer suggested consideration of an integrated assessment of the cancer
and noncancer endpoints; however, if linear low-dose extrapolation remains the approach
of choice by the Agency, then the effect of choosing BMRs other than 10% was
recommended to at least be included in the uncertainty discussion. Using BMRs lower
A-15
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
than 10% may allow for the identification of a risk level for which the low-dose slope is
'best' estimated.
Response: The EPA conducted a cancer MOA analysis evaluating all of the
available data for 1,4-dioxane. Application of the framework in the USEPA's Guidelines
for Carcinogen Risk Assessment (2005) demonstrates that the available evidence to
support any hypothesized MOA for 1,4-dioxane-induced tumors does not exist. In the
absence of a MOA, the USEPA's Guidelines for Carcinogen Risk Assessment (2005)
indicate that a low dose linear extrapolation should be utilized for dose response analysis
(see Section 5.4). Some of the potential uncertainty associated with this conclusion was
characterized in Section 5.5. Note that there is no scientific basis to indicate that in the
absence of evidence for genotoxicity a nonlinear low-dose extrapolation should be used.
As concluded in the Toxicological Review, the application of a nonlinear low-dose
extrapolation approach was not supported.
With regards to the PBPK model available for 1,4-dioxane, it is clear that there
currently exist deficiences within the model and as such, the model was not utilized for
interspecies extrapolation. Given the deficiencies and uncertainty in the 1,4-dioxane
model it also does not provide support for a MOA.
Lastly, in the absence of a MOA for 1,4-dioxane carcinogenicity it is not possible
to harmonize the cancer and noncancer effects to assess the risk of health effects due to
exposure. However, the choice of the BMDLu),which was more than 15-fold lower than
the response at the lowest dose (66 mg/kg-dav\ was reconsidered. BMDs and BMDLs
were calculated using a BMR of 30 and 50% extra risk (BMD^n. BMDUn, BMDso. and
BMDLsn). A BMR of 50% was used as it resulted in a BMDL closest to the response
level at the lowest dose tested in the bioassav.
PUBLIC COMMENTS
A. Carcinogenicity of 1,4-dioxane
Comment: Low-dose linear extrapolation for the OSF is not appropriate nor justified by
the data. WOE supports a threshold (non-linear) MOA when metabolic pathway is
saturated at high doses.
Response: The absence of evidence for genotoxicity/mutagenicity does not indicate the
use of nonlinear low-dose extrapolation. For 1,4-dioxane, a MOA to explain the
induction of tumors does not exist so the nature of the low-dose region of the dose-
response is unknown.
A-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
Comment: POD for BDFi female mouse is 15-fold lower than the lowest dose in the
bioassay, thus the POD is far below the lower limit of the data and does not follow the
2005 Cancer Guidelines.
Response: The comment is correct that the animal BMDLm was more than 15-fold
lower than the response at the lowest dose (66 mg/kg-dav) in the bioassay. BMDs and
BMDLs were calculated using a BMR of 30 and 50% extra risk (BMD30, BMDLjo,
BMDsn, and BMDLsn). A BMR of 50% was chosen as it resulted in a BMDL closest to
the response level at the lowest dose tested in the bioassay.
Comment: The OSF was based on the most sensitive group, BDFi mice; however BDFi
mice have a high background rate of liver tumors. Should consider incidence of liver
tumors in historical controls.
Response. Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas
and carcinomas in control male and female BDFi mice from ten 2-year bioassays at the
JBRC. For female mice, out of 499 control mice, the incidence rates were 4.4% for
hepatocellular adenomas and 2.0% for hepatocellular carcinomas. Kano et al. (2009)
reported a 10% incidence rate for hepatocellular adenomas and a 0% incidence rate for
hepatocellular carcinomas in control female BDFi. These incidence rates are well below
the historical control values and thus are appropriate for consideration in this assessment.
Additional text regarding these historical controls was added to the study description in
Section 4.2.1.2.6.
Comment. Should have used geometric mean of slope factors (as done with B[a]P &
DDT) instead of relying on the female BDF i mouse data, since MOA could not be
determined.
Response: In accordance with the external peer review draft Benchmark Dose Technical
Guidance (U.S. EPA, 2000b), averaging tumor incidence is not a standard or default
approach.
Comment. Critically reexamine the choice of JBRC (1998) as the principal study since it
has not been published or peer-reviewed. Provide transcript of e-mail correspondence.
Response. JBRC (1998a) was published as conference proceedings as Yamazaki et al.
(1994) and recently in the peer-reviewed literature as Kano et al. (2009). Additional study
A-17
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
information was also gathered from the authors. A transcript of the email
correspondence is also available via the IRIS Hotline.
Comment. WOE does not support likely to be carcinogenic to humans determination, but
rather suggestive human carcinogen at the high dose levels used in rodent studies for the
following reasons: 1) lack of conclusive human epidemiological data; 2) 1,4-dioxane is
not mutagenic; 3) evidence at high doses it would act via cell proliferation MOA.
Response: Classification of likely based on evidence of liver carcinogenicity in several
two-year bioassays conducted in three strains of rats, two strains of mice, and in guinea
pigs. Also, mesotheliomas of the peritoneum, mammary, and nasal tumors have been
observed in rats. The Agency agrees human epidemiological studies are inconclusive.
The evidence at any dose is insufficient to determine a MOA.
B.	PBPK Model.
Comment. Should have used and considered PBPK models to derive the oral toxicity
values (rat to human extrapolation) rather than relying on default. The draft did not
consider the Sweeney et al. (2008) model. The PBPK model should be used for both
noncancer and cancer dose extrapolation.
Response: The Agency evaluated the Sweeney et al. (2008) publication and this was
included in Appendix B of the document. Text was added to the main document in
Section 3.5.2.4 and 3.5.3 regarding the evaluation of Sweeney et al. (2008).
Comment: EPA should use the modified inhalation inputs used in the Reitz et al. (1990)
model and the updated input parameters provided in Sweeney et al. (2008) and add a
compartment for the kidney
Response: See response to previous comment regarding evaluation of Sweeney et al.
(2008). Modification of the model to add a kidney compartment is not within the scope
of this assessment.
C.	Other Comments
Comment: EPA should consider the Kasai et al. (2008, 2009) studies for inhalation and
MOA relevance.
A-18
DRAFT - DO NOT CITE OR QUOTE

-------
1	Response: Literature was reviewed through August 2008 and these studies were
2	published afterward. Kasia et al. (2008, 2009) will be reviewed for the derivation of
3	inhalation toxicity values in an update to this assessment.
4
5	Comment: 1,4-Dioxane is not intentionally added to cosmetics and personal care
6	products - correct sentence on page 4.
7
8	Response: This oversight has been corrected.
9
A-19
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
APPENDIX B. EVALUATION OF EXISTING PBPK MODELS FOR 1,4-DIOXANE
B.l. BACKGROUND
Several pharmacokinetic models have been developed to predict the absorption,
distribution, metabolism, and elimination of 1,4-dioxane in rats and humans. Single
compartment, empirical models for rats (Young et al., 1978a, b) and humans (Young et al., 1977)
were developed to predict blood levels of 1,4-dioxane and urine levels of the primary metabolite,
P-hydroxyethoxy acetic acid (HEAA). Physiologically based pharmacokinetic (PBPK) models
that describe the kinetics of 1,4-dioxane using biologically realistic flow rates, tissue volumes
and affinities, metabolic processes, and elimination behaviors, were also developed (Fisher et al.,
1997; Leung and Paustenbach, 1990; Reitz et al., 1990).
In developing updated toxicity values for 1,4-dioxane, the available PBPK models were
evaluated for their ability to predict observations made in experimental studies of rat and human
exposures to 1,4-dioxane. The model of Reitz et al. (1990) was identified for further
consideration to assist in the derivation of toxicity values. Issues related to the biological
plausibility of parameter values in the Reitz et al. (1990) human model were identified. The
model was able to predict the only available human inhalation data set (50 ppm 1,4-dioxane for 6
hours; Young et al., 1977) by increasing (i.e., doubling) parameter values for human alveolar
ventilation, cardiac output, and the blood:air partition coefficient above the measured values.
Furthermore, the measured value for the slowly perfused tissue:air partition coefficient (i.e.,
muscle) was replaced with the measured liver value to improve the fit. Analysis of the Young
et al. (1977) human data suggested that the apparent volume of distribution (Vd) for 1,4-dioxane
was approximately 10-fold higher in rats than humans, presumably due to species differences in
tissue partitioning or other process not represented in the model. Subsequent exercising of the
model demonstrated that selecting a human slowly perfused tissue:air partition coefficient much
lower than the measured rat value resulted in better agreement between model predictions of
1,4-dioxane in blood and experimental observations. Based upon these observations, several
model parameters (e.g., metabolism/elimination parameters) were re-calibrated using
biologically plausible values for flow rates and tissue:air partition coefficients.
This appendix describes activities conducted in the evaluation of the empirical models
(Young et al. 1978a, b, 1977), and re-calibration and exercising of the Reitz et al. (1990) PBPK
model, and evaluation of the Sweeney et al. (2008) model to determine the potential utility of the
PBPK models for 1,4-dioxane for interspecies and route-to-route extrapolation.
B-l
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
B.2. SCOPE
The scope of this effort consisted of implementation of the Young et al. (1978a, b, 1977)
empirical rat and human models using the acslXtreme simulation software, re-calibration of the
Reitz et al. (1990) human PBPK model, and evaluation of model parameters published by
Sweeney et al. (2008). Using the model descriptions and equations given in Young et al. (1978a,
b, 1977), model code was developed for the empirical models and executed, simulating the
reported experimental conditions. The model output was then compared with the model output
reported in Young et al. (1978a, b, 1977).
The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for
cardiac and alveolar flow rates and tissue:air partition coefficients. The predictions of blood and
urine levels of 1,4-dioxane and HEAA, respectively, from the re-calibrated model were
compared with the empirical model predictions of the same dosimeters to determine whether the
re-calibrated PBPK model could perform similarly to the empirical model. As part of the PBPK
model evaluation, EPA performed a sensitivity analysis to identify the model parameters having
the greatest influence on the primary dosimeter of interest, the blood level of 1,4-dioxane.
Variability data for the experimental measurements of the tissue:air partition coefficients were
incorporated to determine a range of model outputs bounded by biologically plausible values for
these parameters. Model parameters from Sweeney et al. (2008) were also tested to evaluate the
ability of the PBPK model to predict human data following exposure to 1,4-dioxane.
B.3. IMPLEMENTATION OF THE EMPIRICAL MODELS IN ACSLXTREME
The empirical models of Young et al. (1978a, b, 1977) for 1,4-dioxane in rats and
humans were reproduced using acslXtreme, version 2.3 (Aegis Technologies, Huntsville, AL).
Model code files were developed using the equations described in the published papers.
Additional files containing experiment-specific information (i.e., BWs, exposure levels, and
duration) were also generated.
B.3.1. Model Descriptions
The empirical model of Young et al. (1978a, b) for 1,4-dioxane in rats is shown in Figure
B-l. This is a single-compartment model that describes the absorption and metabolism kinetics
of 1,4-dioxane in blood and urine. No information is reported describing pulmonary absorption
or intravenous (i.v.) injection/infusion of 1,4-dioxane. The metabolism of 1,4-dioxane and
subsequent appearance of HEAA is described by Michaelis-Menten kinetics governed by a
maximum rate (Vmax, (J-g/mL-hour) and affinity constant (Km, (j,g/mL) . Both 1,4-dioxane and
HEAA are eliminated via the first-order elimination rate constants, ke and kme, respectively
(hour"1) by which 35% of 1,4-dioxane and 100% of HEAA appear in the urine, while 65% of
1,4-dioxane is exhaled. Blood concentration of 1,4-dioxane is determined by dividing the
instantaneous amount of 1,4-dioxane in blood by a Yd of 301 mL/kg BW.
B-2
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
Inhalation (k,,,,)
i.v. admin
dDioxbod^ Vmax xDioxbgd
—- =		 - k xDiox,

dt
Km + Dioxbody
body
Exhaled (65%)
k. xDiox f
s ~ Urine (35%)
dHEAAody VmzxXDiOXl
'body
dt
Km+Dioxbodf
~ k„ex HEAAboc
k xHEAA
-~ Urine
Source: Young et al. (1978a, b).
Figure B-l. Schematic representation of empirical model for 1,4-dioxane in rats.
Figure B-2 illustrates the empirical model for 1,4-dioxane in humans as described in
Young et al. (1977). Like the rat model, the human model predicts blood 1,4-dioxane and
urinary 1,4-dioxane and HEAA levels using a single-compartment structure. However, the
metabolism of 1,4-dioxane to HEAA in humans is modeled as a first-order process governed by
a rate constant, Km (hour"1). Urinary deposition of 1,4-dioxane and HEAA is described using the
first order rate constants, ke(diox) and kme(HEAA), respectively. Pulmonary absorption is described
by a fixed rate of 76.1 mg/hour (kiNH)- Blood concentrations of 1,4-dioxane and HEAA are
calculated as instantaneous amount (mg) divided by Vd(diox) or Vd(HEAA), respectively (104 and
480 mL/kg BW, respectively).
Inhalation (kr.:| ()

Dioxane
A
Diox
^d(Diax)
XC°nCD,oX


Km

1
r

HEAA
A
HEAA
= Vd(HEAA) one HEAA
e (diox)
Urine
Cumulative
Dioxane and
HEAA
km,
le (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
Several modifications were made to the empirical models. The need for the
modifications arose in some cases from incomplete reporting of the Young et al. (1978a, b, 1977)
B-3
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
studies and in other cases from the desire to add capabilities to the models to assist in the
derivation of toxicity values.
For the rat model, no information was given by Young et al. (1978a, b) regarding the
parameterization of pulmonary absorption (or exhalation) or i.v. administration of 1,4-dioxane.
Therefore, additional parameters were added to simulate these processes in the simplest form.
To replicate 1,4-dioxane inhalation, a first-order rate constant, kiNH (hour"1), was introduced.
kiNH was multiplied by the inhalation concentration and the respiratory minute volume of
0.238 L/minute (Young et al., 1978a, b). The value for k^H was estimated by optimization
against the blood time course data of Young et al. (1978a, b). Intravenous (i.v.) administration
was modeled as instantaneous appearance of the full dose at the start of the simulation. Rat
urinary HEAA data were reported by Young et al. (1978a, b) in units of concentration. To
simulate urinary HEAA concentration, an estimate of urine volume was required. Since
observed urinary volumes were not reported by Young et al. (1978a, b), a standard rat urine
production rate of 0.00145 L/hour was used.
For humans, Young et al. (1977) used a fixed 1,4-dioxane inhalation uptake rate of
76.1 mg/hour, which corresponded to observations during a 50 ppm exposure. In order to
facilitate user-specified inhalation concentrations, pulmonary absorption was modeled. The
modeling was performed identically to the rat model, but using a human minute volume of
7 L/minute. Urinary HEAA data were reported by Young et al. (1977) as a cumulative amount
(mg) of HEAA. Cumulative amount of HEAA in the urine is readily calculated from the rate of
transfer of HEAA from plasma to urine, so no modification was necessary to simulate this dose
metric for humans.
Neither empirical model of Young et al. (1978a, b;1977) described oral uptake of
1,4-dioxane. Adequate data to estimate oral absorption parameters are not available for either
rats or humans; therefore, neither empirical model was modified to include oral uptake.
B.3.3. Results
The acslXtreme implementation of the Young et al. (1978a, b) rat empirical model
simulates the 1,4-dioxane blood levels from the i.v. experiments identically to the model output
reported in the published paper (Figure B-3). However, the acslXtreme version predicts urinary
HEAA concentrations in rats that are approximately threefold lower and reach a maximum
sooner than the predicted levels reported in the paper (Figure B-4). These discrepancies may be
due, at least in part, to the reliance in the acslXtreme implementation on a constant, standard,
urine volume rather than experimental measurements, which may have been different from the
assumed value and may have varied over time. Unreported model parameters (e.g., lag times for
appearance of excreted HEAA in bladder urine) may also contribute to the discrepancy.
B-4
DRAFT - DO NOT CITE OR QUOTE

-------
acsl version - Young et al.
(1978a, b) empirical model
~ Young et al. (1978a, b)
observations
Observations and predictions of 1,4-dioxane in rat blood
following 3 to 1000 mg/kg IV injection
1000.0
100.0
30 40
Time (hrs)
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
10000
~ d?D
~ ~
~~
1000
_l
cn
E
~ ~
x
acsl version - Young et al.
(1978a, b) empirical model
~ Young et al. (1978a, b)
0
10
20
30
40
50
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.
B-5
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
The Young et al. (1978a, b) report did not provide model predictions for the 50-ppm
inhalation experiment. However, the acslXtreme implementation produces blood 1,4-dioxane
predictions that are quite similar to the reported observations (Figure B-5). As with the urine
data from the i.v. experiment, the acslXtreme-predicted urinary HEAA concentrations are
approximately threefold lower than the observations, presumably for the same reasons discussed
above for the i.v. predictions.
Observations and predictions of 1,4-dioxane in rat blood
following a 6 hour 50 ppm inhalation exposure
ui 1.0
acsl version - Young et al
(1978a, b) empirical model
Young et al. (1978a, b)
observations
Time (hrs)
Observations and predictions of HEAA in rat urine
following a 6 hour 50 ppm inhalation exposure
25.0 -
20.0 -
£.15.0 -
: 10.0 -
5.0 -
0.0 +-
acsl version - Young et
al. (1978a, b) empirical
model
Young et al. (1978a, b)
observations
20	30
Time (hrs)
40
50
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.
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
100.0
	acsl version - Young et
al (1977) empirical model
~ observed
10.0
Time (hrs)
o
Cft
E

c
o
1
Exposure Time, 6 Hours
Time, Hours
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.
B-7
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
Observations and predictions of HEAA in human urine
following a 6 hour 50 ppm inhalation exposure
700.0
~
~
600.0
~
~
>> 500.0
~
0.0
	acsl version - Young et al
(1977) empirical rrodel
~ observed
0
5
10
15
20
25
Time (hrs)
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
The empirical models described by Young et al. (1978a, b, 1977) for rats and humans
were implemented using acslXtreme. The models were modified to allow for user-defined
inhalation levels by addition of a first-order rate constant for pulmonary uptake of 1,4-dioxane,
fitted to the inhalation data. No modifications were made for oral absorption as adequate data
are not available for parameter estimation. The acslXtreme predictions of 1,4-dioxane in the
blood are identical to the published predictions for simulations of 6-hour, 50-ppm inhalation
exposures in rats and humans and 3 to 1,000 mg/kg i.v. doses in rats (Figures B-3, B-5, and
B-6). However, the acslXtreme version predicts lower urinary HEAA concentrations in rats
appearing earlier than either the Young et al. (1978a, b) model predictions or the experimental
observations. The lower predicted urinary HEAA levels in the acslXtreme implementation for
rats is likely due to use of default values for urine volume in the absence of measured volumes.
The reason for differences in time-to-peak levels is unknown, but may be the result of an
unreported adjustment by Young et al. (1978a, b) in model parameter values. For humans,
Young et al. (1977) did not report model predictions of urinary HEAA levels. The urinary
HEAA levels predicted by acslXtreme were low relative to the observations. However, unlike
the situation in rats, these data are not dependent on unreported urine volumes (observations
were reported as cumulative HEAA amount rather than HEAA concentration), but reflect the
model parameter values reported by Young et al. (1977). Presently, there is no explanation for
the lack of fit of the reported urinary HEAA elimination rate constant to the observations.
B-8
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
B.4. INITIAL RE-CALIBRATION OF THE PBPK MODEL
Concern regarding adjustments made to some of the parameter values in Reitz et al.
(1990) prompted a re-calibration of the Reitz et al. (1990) 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 liver:air 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
The cardiac output of 30 L/hour/kg0'74 (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/kg0 74 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
30
--
--
17.0
Alveolar ventilation (QPC)a
30
--
--
17.7
Partition Coefficients (PCs)
Blood:air (PB)
3,650
1,825 ±94
1,666 ±287
1,850
Fat:air (PFA)
851
851 ±118
--
851
Liver: air (PLA)
1,557
1,557 ± 114
1,862 ± 739b
1,557
Rapidly perfused tissue:air (PRA)
1,557
--
--
1,557
Slowly perfused tissue:air (PSA)
1,557
997 ± 254
1,348 ±290b
166
Metabolic Constants
Maximum rate for 1,4-dioxane
metabolism (VmaxC)d
6.35
—
—
5.49
B-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
Parameter
Reitz et al. (1990)
Leung and
Paustenbach (1990)
Sweeney et al.
(2008)
EPAC
Metabolic affinity constant (Km)e
3.00
--
--
9.8
HEAA urinary elimination rate
constant (kme)f
0.56
—
—
0.44
aL/hour/kg BW°74
bMeasurement for rat tissue
°Biologically plausible values utilized by EPA in this assessment
dmg/hour/kg BW°75
emg/L
fhour_1
Examination of the experimental data of Young et al. (1977) yields an estimated alveolar
ventilation to be 7 L/minute (or 16 L/hour/kg0'74) for volunteers having a mean BW of 84 kg.
This rate is based on the Young et al. (1977) estimate of 76.1 mg/hour for 1,4-dioxane uptake.
Based on these findings, the cardiac output and alveolar ventilation rates of 17.0 and 17.7
L/hour/kg0'74 were biologically plausible for the experimental subjects. These rate estimates are
based on calculations made using empirical data and are consistent with standard human values
and the experimental conditions (i.e., subject exertion level) reported by Young et al. (1977).
Therefore, these flow values were chosen for the model re-calibration.
B.4.2. Sources of Values for Partition Coefficients
Two data sources are available for the tissue:air equilibrium partition coefficients for
1,4-dioxane: Leung and Paustenbach (1990) and Sweeney et al. (2008). Both investigators
report mean values and standard deviations for human blood:air, rat liver:air, and rat muscle:air
(e.g., slowly perfused tissue:air), while Leung and Paustenbach et al. (1990) also reported values
for rat fat:air (Table B-l).
B.4.3. Calibration Method
The PBPK model was twice re-calibrated using the physiological flow values suggested
values (current EPA assessment, see Table B-l) and the partition coefficients of Leung and
Paustenbach (1990) and Sweeney et al. (2008) separately. For each calibration, the metabolic
parameters Vmaxc and Km, were simultaneously fit (using the parameter estimation tool provided
in the acslXtreme software) to the output of 1,4-dioxane blood concentrations generated by the
acslXtreme implementation of the Young et al. (1977) empirical human model for a 6 hour,
50 ppm inhalation exposure. Subsequently, the HEAA urinary elimination rate constant, kme,
was fitted to the urine HEAA predictions from the empirical model. The empirical model
predictions, rather than experimental observations, were used to provide a more robust data set
for model fitting, as the empirical model simulation provided 240 data points (one prediction
every 0.1 hour) compared with hourly experimental observations, and to avoid introducing error
by calibrating the model to data digitally captured from Young et al. (1977).
B-10
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
B.4.4. Results
Results of the model re-calibration are provided in Table B-2. The re-calibrated values
for Vmaxc and kme associated with the Leung and Paustenbach (1990) or Sweeney et al. (2008)
tissue:air partition coefficients are very similar. However, the fitted value for Km using the
Sweeney et al. (2008) partition coefficients is far lower (0.0001 mg/L) than that resulting from
use of the Leung and Paustenbach (1990) partition coefficients (2.5 mg/L). This appears to be
due to the higher slowly perfused tissue:air partition coefficient determined by Sweeney et al.
(2008) (1,348 vs. 997), resulting in a higher apparent Vd than if the Leung and Paustenbach
(1990) value is used. Thus, the optimization algorithm selects a low Km, artificially saturating
metabolism in an effort to drive predicted blood 1,4-dioxane levels closer to the empirical model
output. Saturation of metabolism during a 50 ppm inhalation exposure is inconsistent with the
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 tissue:air partition coefficients
Source of Partition Coefficients
Leung and Paustenbach (1990)
Sweeney et al. (2008)
Maximum rate for 1,4-dioxane metabolism (VmaxC)b
16.9
20.36
Metabolic affinity constant (Km)°
2.5
0.0001
HEAA urinary elimination rate constant (kme)d
0.18
0.17
aCardiac output = 17.0 L/hour/kg BW°74, alveolar ventilation = 17.7 L/hour/kg BW074
bmg/hour/kg BW°75
°mg/L
dhour_1
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.
B-ll
DRAFT - DO NOT CITE OR QUOTE

-------
ObQte^atiarw^plreal^iolnsnaftf^r^Dkarantti ihilnoiaiiabl• o)500.0
|£
5 1400.0
<0 o
¦« £
« < 300.0
^ <
E <
5 x 200.0
100.0
0.0
	PBPK predicted
~ observed
- - - empirical predicted
10	15
Time (hrs)
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.
4	Outputs of the blood 1,4-dioxane and urinary HEAA levels using the suggested (see
5	Table B-l) parameters are shown in Figure B-10. These outputs rely on a very low value for the
6	slowly perfused tissue:air partition coefficient (166) that is six- to eightfold lower than the
B-12
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
measured values reported in Leung and Paustenbach (1990) and Sweeney et al. (2008), and 10-
fold lower than the value used by Reitz et al. (1990). While the predicted maximum blood
1,4-dioxane levels are much closer to the observations, the elimination kinetics are markedly
different, producing higher predicted elimination rates compared to observations during the post-
exposure phase of the experiment.
EPA parameter estimates used
100.0
	predicted
~ observed
10.0
1.0
0.1
0
5
10
15
Time (hrs)
bservatio and Predict onsof HEAA in human urine
EPA parameter estimates used
700.0
600.0
^ "3500.0
d = 400.0
12 < 300.0
g ^ 200.0
	predicted
~ observed
100.0
0.0
Time (hrs)
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 PBPK Model Implementation
Re-calibration of the human PBPK model was performed using experiment-specific
values for cardiac output and alveolar ventilation (values derived from Young et al., 1977) and
measured mean tissue:air 1,4-dioxane partition coefficients reported by Leung and Paustenbach
(1990) or Sweeney et al. (2008). The resulting predictions of 1,4-dioxane in blood following a
6-hour, 50-ppm inhalation exposure were 10-fold (or more) lower than either the observations or
the empirical model predictions, while the predictions of urinary HEAA by the PBPK and
empirical models were similar to each other, but lower than observed values (Figures B-8 and
B-9). Output from the model using biologically plausible parameter values (see Table B-l),
Figure B-10 shows that application of a value for the slowly perfused tissue:air partition
coefficient, which is 10-fold lower than the measured value reported by Leung and Paustenbach
(1990), results in closer agreement of the predictions to observations during the exposure phase,
but not during the elimination phase. Thus, model re-calibration using experiment-specific flow
rates and mean measured partition coefficients does not result in an adequate fit of the PBPK
model to the available data.
B-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
B.4.6. SENSITIVITY ANALYSIS
A sensitivity analysis of the Reitz et al. (1990) model was performed to determine which
PBPK model parameters exert the greatest influence on the outcome of dosimeters of interest—
in this case, the concentration of 1,4-dioxane in blood. Knowledge of model sensitivity is useful
for guiding the choice of parameter values to minimize model uncertainty.
B.4.7. Method
A univariate sensitivity analysis was performed on all of the model parameters for two
endpoints: blood 1,4-dioxane concentrations after 1 and 4 hours of exposure. These time points
were chosen to assess sensitivity during periods of rapid uptake (1 hour) and as the model
approached steady state (4 hours) for blood 1,4-dioxane. Model parameters were perturbated 1%
above and below nominal values and sensitivity coefficients were calculated as follows:
y (T) - /(x + Ax)~/(x) *
Ax	f{x)
where x is the model parameter, f(x) is the output variable, Ax is the perturbation of the
parameter from the nominal value, and f'(x) is the sensitivity coefficient. The sensitivity
coefficients were scaled to the nominal value of x and f(x) to eliminate the potential effect of
units of expression. As a result, the sensitivity coefficient is a measure of the proportional
change in the blood 1,4-dioxane concentration produced by a proportional change in the
parameter value, with a maximum value of 1.
B.4.8. Results
The sensitivity coefficients for the seven most influential model parameters at 1 and
4 hours of exposure are shown in Figure B-l 1. The three parameters with the highest sensitivity
coefficients in descending order are alveolar ventilation (QPC) (1.0), the blood:air partition
coefficient (PB) (0.65), and the slowly perfused tissue:air partition coefficient (PSA) (0.51). Not
surprisingly, these were the parameters that were doubled or given surrogate values in the Reitz
et al. (1990) model in order to achieve an adequate fit to the data. Because of the large influence
of these parameters on the model, it is important to assign values to these parameters in which
high confidence is placed, in order to reduce model uncertainty.
B-14
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Sensitivity Coefficients: CV -1 hr
0.01	0.10	1.00
QPC
PB	~|
PSA	I
_(D		1
<1>
£	QSC I
2		
CO
QCC	I
^rnaxC	^
Sensitivity Coefficients: CV - 4 hr
0.01	0.10	1.00
QPC		
PB	~l
5 PSA	~
(D
E \/	I
CD w maxC	1
OJ
" K.	I
PRA		1
QSC	^
Figure B-ll. 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.5. PBPK MODEL EXERCISES USING BIOLOGICALLY PLAUSIBLE PARAMETER
BOUNDARIES
The PBPK model includes numerous physiological parameters whose values are typically
taken from experimental observations. In particular, values for the flow rates (cardiac output and
alveolar ventilation) and tissue:air partition coefficients (i.e., mean and standard deviations) are
available from multiple sources as means and variances. The PBPK model was exercised by
varying the partition coefficients over the range of biological plausibility (parameter mean ±
2 standard deviations), re-calibrating the metabolism and elimination parameters, and exploring
the resulting range of blood 1,4-dioxane concentration time course predictions. Cardiac output
and alveolar ventilation were not varied because the experiment-specific values used did not
include any measure of inter-individual variation.
B.5.1. Observations Regarding the Volume of Distribution
Young et al. (1978a, b) used experimental observations to estimate a Vd for 1,4-dioxane
in rats of 301 mL, or 1,204 mL/kg BW. For humans, the Vd was estimated to be 104 mL/kg BW
(Young et al., 1977). It is possible that a very large volume of the slowly perfused tissues in the
body of rats and humans may be a significant contributor to the estimated 10-fold difference in
distribution volumes for the two species. This raises doubt regarding the appropriateness of
using the measured rat slowly perfused tissue:air partition coefficient as a surrogate values for
humans in the PBPK model.
B-15
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
B.5.2. Defining Boundaries for Parameter Values
Given the possible 10-fold species differences in the apparent Vd for 1,4-dioxane in rats
and humans, boundary values for the partition coefficients were chosen to exercise the PBPK
model across its performance range to either minimize or maximize the simulated Vd. This was
accomplished by defining biologically plausible values for the partition coefficients as the
mean ± 2 standard deviations of the measured values. Thus, to minimize the simulated Vd for
1,4-dioxane, the selected blood:air partition coefficient was chosen to be the mean + 2 standard
deviations, while all of the other tissue:air partition coefficients were chosen to be the mean - 2
standard deviations. This created conditions that would sequester 1,4-dioxane in the blood, away
from other tissues. To maximize the simulated 1,4-dioxane Vd, the opposite selections were
made: blood and other tissue:air partition coefficients were chosen as the mean - 2 standard
deviations and mean + 2 standard deviations, respectively. Subsequently, Vmaxc, Km, and kme
were optimized to the empirical model output data as described in Section B.4.3. This procedure
was performed for both the Leung and Paustenbach (1990) and Sweeney et al. (2008) partition
coefficients (Table B-l). The two predicted time courses resulting from the re-calibrated model
with partition coefficients chosen to minimize or maximize the 1,4-dioxane Vd represent the
range of model performance as bounded by biologically plausible parameter values.
B.5.3. Results
The predicted time courses for a 6-hour, 50-ppm inhalation exposure for the re-calibrated
human PBPK model with mean (central tendency) and ± 2 standard deviations from the mean
values for partition coefficients are shown in Figure B-12 for the Leung and Paustenbach (1990)
values and Figure B-13 for the Sweeney et al. (2008) values. The resulting fitted values for
Vmaxc, Km, and kme, are given in Table B-3. By bounding the tissue:air partition coefficients with
upper and lower limits on biologically plausible values from Leung and Paustenbach (1990) or
Sweeney et al. (2008), the model predictions are still at least six- to sevenfold lower than either
the empirical model output or the experimental observations. The range of possible urinary
HEAA predictions brackets the prediction of the empirical model, but this agreement is not
surprising, as the cumulative rate of excretion depends only on the rate of metabolism of
1,4-dioxane, and not on the apparent Vd for 1,4-dioxane. These data show that the PBPK model
cannot adequately reproduce the predictions of blood 1,4-dioxane concentrations of the Young
et al. (1977) human empirical model or the experimental observations when constrained by
biologically plausible values for physiological flow rates and tissue:air partition coefficients.
B-16
DRAFT - DO NOT CITE OR QUOTE

-------
Cumulative HEAA in human urine from a 6-hour, 50 ppm
exposure
1,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure
700
100.0
600
10.0
« < 300
o I 200
100
oarc
Young etal. ^
observations
0
5
10
15
20
25
14
Time (hrs)
Time (hrs)
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.
Cumulative HEAA in human urine from a 6-hour, 50 ppm
exposure
1,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure
700
100.0
model
.... Sw^(^^0gLPC-UCL
	Swg^^f^8iiPr&-Central
600
10.0

to <300
O i 200
100
.CL
o\
Time (hrs)
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.
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 rates" and selected upper and lower boundary values for
tissuerair partition coefficients
Source of partition coefficients
Leung and Pausenbach (1990)
Sweeney et al. (2008)
For maximal Vd
For minimal Vd
For maximal Vd
For minimal Vd
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)b
14.95
18.24
17.37
21.75
Metabolic dissociation constant
(Km)°
5.97
0.0001
4.88
0.0001
HEAA urinary elimination rate
constant (kme)d
0.18
0.17
0.26
0.19
aCardiac output = 17.0 L/hour/kg BW0 74, alveolar ventilation = 17.7 L/hour/kg BW074
bmg/hour/kg BW°75
°mg/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 zero-order, non-saturable
6	metabolism rate constant, kit- 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.
B-18
DRAFT - DO NOT CITE OR QUOTE

-------
1,4-Dioxane in human blood from a 6-hourt 50 ppm
exposure: kLC (3.0) fitted to all observations
6.1
	Young et al. (1977) empirical
model
4.1
	K.c ~ fitted model
2.1
~ Young etal. (1977)
observation data
0.1

8.1
6.1
i ~
4.1
2.1
0.1
0
2
4
6
8
10
12
14
Time (hrs)
Figure B-14. Predictions of blood 1,4-dioxane concentration following calibration of
a zero-order metabolism rate constant, kLc, 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-15. 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-15).
B-19
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
1,4-Dioxane in human blood from a 6-hour,"50 ppm
exposure: kLC (0.1) fitted 1 to 6-hour observations
Young etal. (1977) empirical
model
k,_c - fitted model
Young etal. (1977)
observation data
Time (hrs)
Figure B-15. Predictions of blood 1,4-dioxane concentration following calibration of
a zero-order metabolism rate constant, kLc? to only the exposure phase of the
experimental data.
Finally, the model was re-calibrated by simultaneously fitting kLc and the slowly
perfused tissue:air partition coefficient to the experimental data with no bounds on possible
values (except that they be non-zero). The fitted slowly perfused tissue:air partition coefficient
was an extremely low (and biologically unlikely) value of 0.0001. The resulting model
predictions, however, were closer to the observations than even the empirical model predictions
(Figure B-16). These exercises show that better fits to the observed blood 1,4-dioxane kinetics
are achieved only when parameter values are adjusted in a way that corresponds to a substantial
decrease in apparent Vd of 1,4-dioxane in the human, relative to the rat (e.g., decreasing the
slowly perfused tissue:air partition coefficient to extremely low values, relative to observations).
Downward adjustment of the elimination parameters (e.g., decreasing k^) increases the
predicted blood concentrations of 1,4-dioxane, achieving better agreement with observations
during the exposure phase of the experiment; however, it results in unacceptably slow
elimination kinetics, relative to observations following cessation of exposure. These
observations suggest that some other process not captured in the present PBPK model structure is
responsible for the species differences in 1,4-dioxane Vd and the inability to reproduce the
human experimental inhalation data with biologically plausible parameter values.
B-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
1,4-Dioxane in human blood from a 6-hour,
50 ppm exposure
16.
c 14.
~
Young et al. (1977) empirical
model

~
Metabolism and slow PC-
fitted model
|-| Young et al. (1977)
observation data
0
2
4
6
8
10
12
14
Time (hrs)
Figure B-16. Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a zero-order metabolism rate constant, kLc, and slowly perfused
tissuerair partition coefficient to the experimental data.
B.6. CONCLUSIONS
The rat and human empirical models of Young et al. (1978a, b, 1977) were successfully
implemented in acslXtreme and perform identically to the models reported in the published
papers (Figures 3-3 through 3-6), with the exception of the lower predicted HEAA
concentrations and early appearance of the peak HEAA levels in rat urine. The early appearance
of peak HEAA levels cannot presently be explained, but may result from manipulations of kme or
other parameters by Young et al. (1978a, b) that were not reported. The lower predictions of
HEAA levels are likely due to reliance on a standard urine volume production rate in the absence
of measured (but unreported) urine volumes. While the human urinary HEAA predictions were
lower than observations, this is due to parameter fitting of Young et al. (1977). No model output
was published in Young et al. (1977) for comparison. The empirical models were modified to
allow for user-defined inhalation exposure levels. However, no modifications were made to
model oral exposures because adequate data to parameterize such modifications do not exist for
rats or humans.
Several procedures were applied to the human PBPK model to determine if an adequate
fit of the model to the empirical model output or experimental observations could be attained
using biologically plausible values for the model parameters. The re-calibrated model
predictions for blood 1,4-dioxane levels do not come within 10-fold of the experimental values
using measured tissue:air partition coefficients from Leung and Paustenbach (1990) or Sweeney
et al. (2008) (Figures B-8 and B-9). Use of a slowly perfused tissue:air partition coefficient 10-
B-21
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
fold lower than measured values produces exposure-phase predictions that are much closer to
observations, but does not replicate the elimination kinetics (Figure B-10). Re-calibration of the
model with upper bounds on the tissue:air partition coefficients results in predictions that are still
six- to sevenfold lower than empirical model prediction or observations (Figures B-12 and B-13).
Exploration of the model space using an assumption of first-order metabolism (valid for the 50-
ppm inhalation exposure) showed that an adequate fit to the exposure and elimination data can
be achieved only when unrealistically low values are assumed for the slowly perfused tissue:air
partition coefficient (Figure B-16). Artificially low values for the other tissue:air partition
coefficients are not expected to improve the model fit, because the sensitivity analysis to exert
less influence on blood 1,4-dioxane than Vmaxc and Km. This suggests that the model structure is
insufficient to capture the apparent 10-fold species difference in the blood 1,4-dioxane Vd
between rats and humans. In the absence of actual measurements for the human slowly perfused
tissue:air partition coefficient, high uncertainty exists for this model parameter value.
Differences in the ability of rat and human blood to bind 1,4-dioxane may contribute to the
difference in Vd. However, this is expected to be evident in very different values for rat and
human blood:air partition coefficients, which is not the case (Table B-l). Therefore, some other,
as yet unknown, modification to model structure may be necessary.
B.7. RECOMMENDATIONS FOR UTILIZING EXISTING PBPK MODELS
The use of empirical or PBPK models to reduce uncertainty in extrapolation of dose-
responses (in terms of internal dosimetry) requires accurate representation of exposure and
biological reality. In the case of the empirical models of Young et al. (1978a, b, 1977), the
acslXtreme implementations are adequate for predicting blood 1,4-dioxane levels for a variety of
inhalation exposure levels in rats and up to 50 ppm in humans. However, the absence of data
with which to evaluate simulated oral absorption in either species precludes the inclusion of this
route of exposure in the models. Therefore, the empirical models may be useful for assessment
of toxicity by inhalation exposure, but not by oral exposure, and not for route-to-route
extrapolation. For the PBPK model, an apparent gap in the model structure exists such that
experimental observations of blood 1,4-dioxane levels in humans during and following
inhalation exposures to 1,4-dioxane cannot be reproduced under the constraints of biologically
plausible parameter values for all parameters. Therefore, the use of the PBPK model (in its
present form) is not recommended for application to the derivation of toxicity values for
1,4-dioxane.
B-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
B.8. ACSLXTREME CODE FOR THE YOUNG ET AL. (1978A, B) EMPIRCAL MODEL
FOR 1,4-DIOXANE IN RATS
PROGRAM: Young 1978 rat.csl
Created by Michael Lumpkin, Syracuse Research Corporation, 08/06
This program implements the 1-compartment empirical model for 1,4-dioxane
in rats, developed by Young et al. 1978a, b. Program was modified to run
in ACSL Xtreme and to include user-defined i.v. and inhalation concentrations
(MLumpkin, 08/06)
INITIAL
!*****Timing and Integration Commands*****
ALGORITHM IALG=2 !Gear integration algorithm for stiff systems
IMERROR %%%%=0.01 IRelative error for lead in plasma
NSTEPS NSTP=1000 INumber of integration steps per communication interval
CINTERVAL CINT=0.1 !Communication interval
CONSTANT TSTART=0. ! Start of simulation (hr)
CONSTANT TSTOP=70. !End of simulation (hr)
! * * * * *MODEL PARAMETERS *****
CONSTANT BW=0.215 !Body weight (kg)
CONSTANT MINVOL=0.238 Irespiratory minute volume (L/min) estimated from Young et al.
(1978)
CONSTANT IVDOSE = 0. !IV dose (mg/kg)!
CONSTANT CONC = 0. !inhalation concentration (ppm)
CONSTANT MOLWT=88.105 !mol weight of 1,4-dioxane
CONSTANT TCHNG=6.0 lExposure pulse 1 width (hr)
CONSTANT TDUR=24.0 lExposure duration (hr)
CONSTANT TCHNG2=120.0 lExposure pulse 2 width (hr)
CONSTANT TDUR2= 168.0 lExposure duration 2 (hr)
CONSTANT Vmax=4.008 !(mcg/mL/hr)
CONSTANT Km=6.308 !(mcg/mL)
CONSTANT Kinh=0.43 Ipulmonary absorption constant (/hr)
CONSTANT Ke=0.0149 I (/hr)
CONSTANT Kme=0.2593 I (/hr)
CONSTANT Vd=0.3014 !(L)
IV = IVDOSE*BW
AmDIOXi=IV
END	! Of Initial Section
DYNAMIC
B-23
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
DERIVATIVE
!*** Dioxane inhalation concentration ***
CIZONE=PULSE(0.0, TDUR, TCHNG) * PULSE(0.0, TDUR2, TCHNG2)
IFirst pulse is hours/day, second pulse is hours/week
CI=CONC*CIZONE*MOLWT/24450. IConvert to mg/L
!*** Dioxane metabolism/1 st order elimination ***
d AmDIOX=(Kinh* CI* (MINV OL* 60))-(( V max* (AmDIOX))/(Km+(AmDIOX)))-
(Ke*(AmDIOX))
AmDIOX=INTEG(dAmDIOX,AmDIOXi)
ConcDIOX=AmDIOX/Vd Iplasma dioxane concentration (mcg/mL)
AUCDIOX=INTEG(ConcDIOX,0) Iplasma dioxane AUC
!*** HEAA production and 1st order metabolism ***
d AmHE A A=(( V max * (AmDIOX))/(Km+( AmDIOX)))-(Kme * ( AmHE A A))
AmHE A A=INTEG(d AmHE A A, 0.)
ConcHEAA=AmHEAA/Vd Iplasma HEAA concentration
!*** 1st order dioxane elimination to urine ***
d AmDIOXu=(Ke * (AmDIOX)) *0.35
AmDIOXu=INTEG(d AmDIOXu, 0.)
ConcDIOXu=Ke*AmDIOX*0.35/1.45e-3 lurine production approx 1.45e-3 L/hr in SD rats
!*** 1st order dioxane exhaled ***
d AmDIOXex=(Ke * (AmDIOX)) *0.65
AmDIOXex=INTEG(dAmDIOXex, 0.)
!*** 1st order HEAA elimination to urine ***
d AmHE A Au=(Kme * (AmHE A A))
AmHE A Au=INTEG(d AmHE A Au, 0.)
ConcHEAAu=Kme*AmHEAA/1.45e-3 lurine production approx 1.45e-3 L/hr in SD rats
END ! of Derivative Section
DISCRETE
END ! of Discrete Section
TERMT (T GT. TSTOP)
END ! of Dynamic Section
TERMINAL
END ! of Terminal Section
END ! of Program
B-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
B.9. ACSLXTREME CODE FOR THE YOUNG ET AL. (1977) EMPIRICAL MODEL
FOR 1,4-DIOXANE IN HUMANS
PROGRAM: Young 1977 human.csl
Created by Michael Lumpkin, Syracuse Research Corporation, 01/06
This program implements the 1-compartment model for 1,4-dioxane in humans,
developed by Young et al., 1977. Program was modified to run
in acslXtreme (MLumpkin, 08/06)
INITIAL
!*****Timing and Integration Commands*****
ALGORITHM IALG=2 !Gear integration algorithm for stiff systems
IMERROR %%%%=0.01 [Relative error for lead in plasma
NSTEPS NSTP=1000 INumber of integration steps per communication interval
CINTERVAL CINT=0.1 !Communication interval
CONSTANT TSTART=0. ! Start of simulation (hr)
CONSTANT TSTOP=120. !End of simulation (hr)
! * * * * *MODEL PARAMETERS *****
! CONSTANT DATA=1 ! Optimization dataset
CONSTANT MOLWT=88.105 !mol weight for 1,4-dioxane
CONSTANT DOSE=0.
CONSTANT CONC=0.
CONSTANT BW=84.1
CONSTANT MINVOL=7.0
CONSTANT F=1.0
CONSTANT kinh=1.06
CONSTANT ke=0.0033
CONSTANT km=0.7096
CONSTANT kme=0.2593
CONSTANT VdDkg=0.104
CONSTANT VdMkg=0.480
CONSTANT OStart=0.
CONSTANT OPeriod=120.
CONSTANT OWidth=l.
CONSTANT IStart=0.
CONSTANT IPeriod=120.
CONSTANT IWidth=6.
!Dose (mg/kg
!Inhalation concentration (ppm)
!Body weight (kg)
Ipulmonary minute volume (L/min)
[Fraction of dose absorbed
!Rate constant for inhalation (mg/hr); optimized by MHL
!Rate constant for dioxane elim to urine (hr-1)
!Rate constant for metab of dioxane to HEAA (hr-1)
!Rate constant for transfer from rapid to blood (hr-1)
! Volume of distribution for dioxane (L/kg BW)
! Volume of distribution for HEAA (L/kg BW)
!Time of first oral dose (hr)
!Oral Dose pulse period (hr)
! Width (gavage/drink time) of oral dose (hr)
!Time of inhalation onset (hr)
!Inhalation pulse period (hr)
! Width (duration) of inhalation exposure (hr)
END
! Of Initial Section
DYNAMIC
DERIVATIVE
B-25
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
I****VARIABLES and DEFINED VALUES*****
VdD=BW*VdDkg ! Volume of distribution for dioxane
VdM=BW*VdMkg ! Volume of distribution for HEAA
InhalePulse=PULSE(IStart,IPeriod,IWidth)
Inhale=CONC*InhalePulse*MOLWT/24450. IConvertto mg/L
l*****DIFFERENTIAL EQUATIONS for compartments****
!*** Dioxane in the body (plasma) ***
dAMTbD=(Kinh*Inhale*(MINVOL*60))-(AMTbD*km)-(AMTbD*ke)
AMTbD=INTEG(d AMTbD, 0.)
CbD=AMTbD/VdD
AU CbD=INTEG(CbD, 0)
! * * * HEAA in the body (plasma)* * *
dAMTbM=AMTbD * km-AMTbM * kme
AMTbM=INTEG(d AMTbM, 0.)
CbM=AMTbM/V dM
!*** Cumulative Dioxane in the urine ***
d AMTuD=( AMTbD * ke)
AMTuD=INTEG(dAMTuD,0.)
!*** Cumulative HEAA in the urine ***
d AMTuM=( AMTbM * kme)
AMT uM=INTEG(d AMTuM, 0.)
END	! Of Derivative Section
DISCRETE
END	! of Discrete Section
TERMT (T GT. TSTOP)
END	! Of Dynamic Section
TERMINAL
END	! of Terminal Section
END	! of Program
B-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
B.10. ACSLXTREME CODE FOR THE REITZ ET AL. (1990) PBPK MODEL FOR 1,4-
DIOXANE
PROGRAM: DIOXANE.CSL (Used in Risk Estimation Procedures)
! Added a venous blood compartment and 1st order elim of metab.'
!Mass Balance Checked OK for Inhal, IV, Oral, and Water RHR'
IDefined Dose Surrogates for Risk Assessment 01/04/89'
IModified the Inhal Route to use PULSE for exposure conditions'
Modifications by GLDiamond, Aug2004, marked as !**
I
IMetabolism of dioxane modified by MLumpkin, Oct2006, to include 1st order
lor saturable kinetics. For 1st order, set VmaxC=0; for M-Menten, set K1C=0.
I
INITIAL
INTEGER I
1=1
! ARRAY TDATA(20) ! CONSTANT TDATA=999, 19*1.OE-6 !**
CONSTANT BW = 0.40 !'Body weight (kg)'
CONSTANT QPC = 15. !'Alveolar ventilation rate (1/hr)'
CONSTANT QCC = 15. !'Cardiac output (1/hr)'
IFlows to Tissue Compartments'
CONSTANT QLC = 0.25 I'Fractional blood flow to liver'
CONSTANT QFC = 0.05 I'Fractional blood flow to fat'
CONSTANT QSC = 0.18 I'Fractional blood flow to slow'
QRC = 1.0 - (QFC + QSC + QLC)
CONSTANT SPDC = 1.0 I diffusion constant for slowly perfused tissues
I Volumes of Tissue/Blood Compartments'
CONSTANT VLC = 0.04 I'Fraction liver tissue'
CONSTANT VFC = 0.07 I'Fraction fat tissue'
CONSTANT VRC = 0.05 I'Fraction Rapidly Perf tissue'
CONSTANT VBC = 0.05 I'Fraction as Blood'
VSC = 0.91 - (VLC + VFC + VRC + VBC)
IPartition Coefficients'
CONSTANT PLA= 1557. I'Liver/air partition coefficient'
CONSTANT PFA = 851. I'Fat/air partition coefficient'
CONSTANT PSA = 2065. I'Muscle/air (Slow Perf) partition'
CONSTANT PRA= 1557. I'Richly perfused tissue/air partition'
CONSTANT PB = 1850. I'Blood/air partition coefficient'
I Other Compound Specific Parameters'
CONSTANT MW = 88.1 I'Molecular weight (g/mol)'
CONSTANT KLC = 12.0 I temp zero-order metab constant
CONSTANT VMAXC =13.8 I 'Maximum Velocity of Metabol.'
CONSTANT KM = 29.4 I'Michaelis Menten Constant'
CONSTANT ORAL = 0.0 I'Oral Bolus Dose (mg/kg)'
B-27
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
CONSTANT KA = 5.0 I'Oral uptake rate (/hr)'
CONSTANT WATER = 0.0 I'Conc in Water (mg/liter, ppm)'
CONSTANT WDOSE=0.0 IWater dose (mg/kg/day) **
CONSTANT IV = 0.0 ! 'IV dose (mg/kg)'
CONSTANT CONC = 0.0 !'Inhaled concentration (ppm)'
CONSTANT KME = 0.276 !'Urinary Elim constant for met (hr-1)'
! Timing commands'
CONSTANT TSTOP = 50 !'Length of experiment (hrs)'
CONSTANT TCHNG = 6 !'Length of inhalation exposure (hrs)'
CINTERVAL CINT=0.1
CONSTANT WIDD=24. !**
CONSTANT PERD=24. ! **
CONSTANT PERW= 168. !**
CONSTANT WIDW= 168. !**
CONSTANT DAT=0.017 !**
! Scaled parameters calculated in this section of Program'
QC=QCC*BW**0.74
QP=QPC*BW**0.74
QL=QLC*QC
QF=QFC*QC
QS=QSC*QC
QR=QRC*QC
VL=VLC*BW
VF=VFC*BW
VS=VSC*BW
VR=VRC*BW
VB=VBC*BW
PL=PLA/PB
PR=PRA/PB
PS=PSA/PB
PF=PFA/PB
KL = KLC*bw**0.7 ! Zero-order metab constant
VMAX = VMAXC*BW**0.7
DOSE = ORAL*BW	!'Initial Amount in Stomach'
ABO = IV*BW	!'Initial Amount in Blood'
IDRINK = 0.102*BW**0.7*WATER/24 !'Input from water (mg/hr)' !**
! DRINK A = 0.102*BW**0.7*WATER/DAT !'Input from water (mg/hr)' ! **
DRINKA=WDOSE*BW/DAT
CV = AB0/VB	! 'Initialize CV'
END !'End of INITIAL'
DYNAMIC
ALGORITHM IALG = 2 I'Gear method for stiff systems'
TERMT( T GE. TSTOP )
CR = AR/VR
B-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
CS = AS/VS
CF = AF/VF
BODY = AL + AR + AS + AF + AB + TUMMY
BURDEN = AM + BODY
TMASS = BURDEN + AX + AMEX
!Calculate the Interval Excretion Data here:'
! DAX = AMEX-AMEX2
! IF( DOSE .LE. 0.0 AND. IV .LE. 0.0 ) GO TO SKIP1
! PCTAX = 100*(AX - AX2)/(DOSE + IV*BW)
! PCTMX = 100*(AMEX - AMEX2)/(DOSE + IV*BW)
! SKIP 1. CONTINUE
! IF( T LT. TDATA(I) OR. I GE. 20 ) GO TO SKIP
! AX2=AX
! AMEX2=AMEX
! 1=1+1
! SKIP.. CONTINUE
IDISCRETE EXPOSE
! CIZONE =1.0 ! CALL LOGD(.TRUE.) Turns on inhalation exposure?
!END
IDISCRETE CLEAR
! CIZONE = 0.0 ! CALL LOGD(.TRUE )
!END
DERIVATIVE
!Use Zero-Crossing Form of DISCRETE Function Here'
! SCHEDULE command must be in DERIVATIVE section'
! DAILY = PULSE (0.0, PERI, TCHNG)
! WEEKLY = PULSE ( 0.0, PER2, LEN2 )
! SWITCHY = DAILY * WEEKLY
! SCHEDULE EXPOSE XP. SWITCHY - 0.995
!SCHEDULE CLEAR XN. SWITCHY - 0.005
DAILY=PULSE(0.0,PERD,WIDD)
WEEKL Y=PULSE(0.0,PERW, WID W)
SWITCHY = DAILY * WEEKLY
I*********************j—|qj-q	ong
CI = CONC * MW/24451.0 * SWITCHY!**
!CA = Concentration in arterial blood (mg/1)'
CA = (QC*CV+QP*CI)/(QC+(QP/PB))
CX = CA/PB
DRINK=DRINK A * SWITCHY ! * *
! TUMMY = Amount in stomach'
B-29
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
RTUMMY = -KA* TUMMY
TUMMY = INTEG(RTUMMY,DO SE)
!RAX = Rate of Elimination in Exhaled air'
RAX = QP*CX
AX = INTEG(RAX, 0.0)
! AS = Amount in slowly perfused tissues (mg)'
RAS = SPDC*(CA-CVS) !now governed by diffusion-limited constant, SPDC, instead of QS
AS = INTEG(RAS,0.)
CVS = AS/(VS*PS)
! AR = Amount in rapidly perfused tissues (mg)'
RAR = QR*(CA-CVR)
AR = INTEG(RAR,0.)
CVR = AR/(VR*PR)
! AF = Amount in fat tissue (mg)'
RAF = QF*(CA-CVF)
AF = INTEG(RAF,0.)
CVF = AF/(VF*PF)
! AL = Amount in liver tissue (mg)'
RAL = QL*(CA-CVL) - KL*CVL - VMAX*CVL/(KM+CVL) + KA*TUMMY + DRINK
AL = INTEG(RAL,0.)
CVL = AL/(VL*PL)
IMetabolism comments updated by EDM on 2/1/10
! AM = Amount metabolized (mg)'
RMEX = (KL*CVL)+(VMAX*CVL/(KM+CVL)) !Rate of 1,4-dioxane metabolism
RAM = (KL*CVL)+(VMAX*CVL)/(KM+CVL) - KME* AM !Rate of change of metabolite
! AB = Amount in Venous Blood'
RAB = QF*CVF + QL*CVL + QS*CVS + QR*CVR - QC*CV
AB = INTEG(RAB, ABO)
CV = AB/VB
AUCV = INTEG(CV, 0.0)
IPossible Dose Surrogates for Risk Assessment Defined Here'
CEX = 0.667*CX + 0.333 *CI I'Conc in Exhal Air'
AVECON = PLA * (CEX+CI)/2	!'Ave Cone in Nose Tissue'
AUCCON = INTEG(AVECON, 0.0) I'Area under Curve (Nose)'
AUCMET = INTEG(CAM, 0.0)	!'Area under Curve (Metab)'
in body
AM = INTEG(RAM, 0.0)
CAM = AM/BW
!'Amt Metabolite in body
I'Conc Metabolite in body'
)	! 'Amt Metabolite Excreted via urine'
AMEX = INTEG(KME * AM, 0.0)
B-30
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
CL = AL/VL
I'Conc Liver Tissue'
AUCL = INTEG(CL, 0.0)	I'Area under Curve (Liver)'
A AU CL=AU CL/TIME
! Dose Surrogates are Average Area under Time/Cone Curve per 24 hrs'
IF (T GT. 0) TIME=T
DAYS = TIME/24.0
NOSE = AUCCON/DAYS	!'Nasal Turbinates'
LIVER = AUCL/DAYS	! 'Liver Tissues'
METAB = AUCMET/DAYS	!'Stable Metabolite'
END !'End of dynamic'
END ! End of TERMINAL
END ! 'End of PROGRAM
B-31
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
APPENDIX C. DETAILS OF BMD ANALYSIS FOR ORAL RfD FOR 1,4-DIOXANE
C.l. CORTICAL TUBULE DEGENERATION
All available dichotomous models in the Benchmark Dose Software (version 2.1.1) 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 were 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)
Females (mg/kg-day)
0
240
530
0
350
640
0/3 r
20/3 lb
27/33b
0/3 r
0/34
10/3 2b

(65%)
(82%)


(31%)
aStatistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this
review.
incidence significantly elevated compared to control by Fisher's exact test (p < 0.05) performed for this review.
Source: NCI (1978).
As assessed by the x 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
(x P- o . 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 could be used to derive a POD of 38.5 mg/kd-ay for this endpoint. The
Weibull model provided the best fit to the data for female rats (Table C-2, Figure C-5) and could
be used to derive a POD of 452.4 mg/kg-day for this endpoint. 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. BMDS modeling results for all dichotomous
models are shown in Table C-2.
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 (NCI, 1978) exposed to 1,4-dioxane in drinking
water
1



Scaled





Residual of
BMD10
BMDL10
Model
AIC
/j-valuc"
Interest
(mg/kg-day)
(mg/kg-day)
Male
Gamma"
74.458
0.6514
0
28.80
22.27
Logistic
89.0147
0.0011
-1.902
88.48
65.84
Log-logistic0
75.6174
1
0
20.85
8.59
Log-probit°
74.168
0.7532
0
51.41
38.53
Multistage
(2 degree)d
74.458
0.6514
0
28.80
22.27
Probit
88.782
0.0011
-1.784
87.10
66.32
Weibulf
74.458
0.6514
0
28.80
22.27
Quantal-Linear
74.458
0.6514
0
28.80
22.27
Female
Gamma"
41.9712
0.945
0.064
524.73
437.08
Logistic
43.7495
0.9996
0
617.44
471.92
Log-logistic0
41.7501
0.9999
0
591.82
447.21
Log-probit°
43.7495
0.9997
0
584.22
436.19
Multistage
(2 degree)d
48.1969
0.1443
-1.693
399.29
297.86
Probit
43.7495
0.9997
0
596.02
456.42
Weibulf
41.75
0.9999
0
596.45
452.36
Quantal-Linear
52.3035
0.03
-2.086
306.21
189.49
a/>-Value from the x2 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.
dBetas restricted to >0.
Source: NCI (1978).
C-2
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
0.6
"O
0)
-i—¦
O
= 1
Total number of observations = 3
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
LogProbit Model with 0.95 Confidence Level
LogProbit
BMDL
C-3
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
User has chosen the log transformed model
Default Initial	(and Specified) Parameter Values
background =	0
intercept =	-5.14038
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe model parameter(s) -background -slope have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
intercept
intercept	1
Parameter Estimates
Variable
background
intercept
slope
Estimate
-5.22131
1
Std. Err.
NA
0 . 172682
NA
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-5 . 55976
-4 .88286
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
Log(1i ke1i hood)
-35 . 8087
-36.084
-65 .8437
# Param's
3
1
1
Deviance Test d.f.
0 .550629
60.07
P-value
0 . 7593
<.0001
AIC:
74 .168
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0.0000
0.0000
0 . 000
0 . 000
31
0 .000
240.0000
0.6023
18.672
20.000
31
0.487
530.0000
0 . 8535
28.166
27.000
33
-0.574
Chi *2 = 0.57	d.f. = 2	P-value = 0.7532
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	51.4062
BMDL =	38.5284
C-4
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
Weibull Model with 0.95 Confidence Level
Weibull
0.5
0.4
0.3
0.2
0
BMDL
BMD
0	100	200	300	400	500	600
dose
14:20 12/04 2009
Source: NCI (1978).
Figure C-2. 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.
Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: Z:\14Dioxane\BMDS\wei_nci_frat_cortdeg_Wei-BMR10-Restrict.(d)
Gnuplot Plotting File: Z:\14Dioxane\BMDS\wei_nci_frat_cortdeg_Wei-BMR10-Restrict.plt
Fri Dec 04 14:20:41 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-slope*dose*power)]
Dependent variable = Effect
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 3
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
C-5
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
Default Initial (and Specified) Parameter Values
Background =	0.015625
Slope = 1.55776e-010
Power =	3.33993
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe model parameter(s) -Background -Power have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
Slope
Slope
-1.$
Variable
Background
Slope
Power
Estimate
1.15454e-051
18
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit
NA
1.#QNAN	1.#QNAN	1.#QNAN
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
Log(likelihood) # Param's Deviance Test d.f. P-value
-19.8748
-19.875
-32.1871
0.000487728
24.6247
0.9998
<.0001
AIC:
41. 75
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.0000 0.000 0.000	31	0.000
350.0000 0.0000 0.000 0.000	34	-0.016
640.0000 0.3125 9.999 10.000	32	0.000
Chi*2 = 0.00	d.f. = 2	P-value = 0.9999
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	596.445
BMDL =	4 52.3 59
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
C.2. LIVER HYPERPLASIA
All available dichotomous models in the Benchmark Dose Software (version 2.1.1) were
fit to the incidence data shown in Table C-3, for liver hyperplasia in male and female
F344/DuCij rats exposed to 1,4-dioxane in the drinking water (Kano et al., 2009; JBRC, 1998a).
Benchmark doses associated with a BMR of a 10% extra risk were calculated.
Table C-3. Incidence of liver hyperplasia in F344/DuCrj rats exposed to
1,4-dioxane in drinking water
Males (mg/kg-day)
Females (mg/kg-day)
0
11
55
274
0
18
83
429
3/40
2/45
9/35a
12/22b
0/3 8a
0/37
1/38
14/24b
Statistically significant compared to controls by the Dunnett's test (p < 0.05).
incidence significantly elevated compared to control by x2 test (p < 0.01).
Sources: Kano et al. (2009); JBRC (1998a).
For incidence of liver hyperplasia in F344 male rats, the logistic, probit, and
dichotomous-Hill models all exhibited a statistically significant lack of fit (i.e., % />value <0.1;
see Table C-4), 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 gamma, multistage, quantal-
linear, and Weibull models were lower than the AIC values for the log-logistic and log-probit
models. Finally, the AIC values for gamma, multistage, quantal-linear, and Weibull models in
Table C-4 are equivalent and, in this case, essentially represent the same model. Therefore,
consistent with the external review draft Benchmark Dose Technical Guidance (EPA, 2000b),
any of them with equal AIC values (gamma, multistage, quantal-linear, or Weibull) could be
used to identify a POD for this endpoint of 23.8 mg/kg-day.
For liver hyperplasias in F344 female rats exposed to 1,4-dioxane, none of the models
exhibited a statistically significant lack of fit (i.e., % p-value < 0.1; See Table C-5). The log-
probit model had the lowest AIC value and was selected as the best-fitting model. Therefore,
consistent with the external review draft Benchmark Dose Technical Guidance (EPA, 2000b),
the BMDL from the log-probit model was selected to yield a POD for this endpoint of 88.9
mg/kg-day.
C-7
DRAFT - DO NOT CITE OR QUOTE

-------
Table C-4. Benchmark dose modeling results based on the incidence of liver
hyperplasias in male and female F344 rats exposed to 1,4-dioxane in drinking
water for 2 years
Model
AIC
/j-valuc"
Scaled
Residual of
Interest
BMD10
(mg/kg-day)
BMDL10
(mg/kg-day)
Male
Gammab
114.172
0.3421
0.886
35.90
23.81
Logistic
117.047
0.0706
1.869
83.56
63.29
Log-logistic0
115.772
0.1848
0.681
33.39
16.96
Log-probit°
115.57
0.1431
1.472
54.91
37.05
Multistage"1
(2 degree)
114.172
0.3421
0.886
35.90
23.81
Probit
116.668
0.0859
1.804
76.69
58.57
Weibullb
114.172
0.3421
0.886
35.90
23.81
Quantal-Linear
114.172
0.3421
0.886
35.90
23.81
Dichotomous-Hill
117.185
NCe
-0.2398
32.01
14.84
Female
Gammab
45.8849
0.9908
0.042
150.69
94.38
Logistic
46.9807
0.6605
0.659
241.49
182.17
Log-logistic0
45.8983
0.9874
0.046
151.25
92.66
Log-probit°
45.8529
0.9992
0.005
137.25
88.87
Multistage11
(2 degree)
44.0038
0.9923
-0.187
150.32
101.88
Probit
46.6775
0.7459
0.54
212.66
160.89
Weibullb
45.9215
0.9811
0.067
161.35
96.21
Quantal-Linear
51.1591
0.1478
-1.637
75.67
50.55
Dichotomous-Hill
47.8499
0.9997
-1.51 xlO"9
95.95
83.42
a/?-Value from the % 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.
dBetas restricted to >0.
eNC=Not calculated.
Sources: Kano et al. (2009); JBRC (1998a).
C-8
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
Gamma Multi-Hit Model with 0.95 Confidence Level
0.6
0.7
0.6
0.5
"O
CD
-I—1
o
=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
Gamma Multi-Hit
BMDL
BMD
C-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
Parameter Convergence has been set to: le-008
Default Initial
Background =
Slope =
Power =
(and Specified) Parameter Values
0.0853659
0.00479329
1.3
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe model parameter(s) -Power have been estimated at a boundary point,
been specified by the user, and do not appear in the correlation matrix )
or have
Background
Slope
Background
1
-0.36
Slope
-0.36
1
Variable
Background
Slope
Power
Parameter Estimates
95.0% Wald Confidence Interval
Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
0.0569658	0.0278487	0.00238329	0.111548
0.00293446	0.000814441	0.00133818	0.00453073
1	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
Log(likelihood) # Param's Deviance Test d.f. P-value
-53.9471
-55.0858
-67 .6005
2 .27725
27 .3066
0 .3203
< .0001
AIC:
114.172
Dose
Est. Prob.
Goodness of Fit
Expected
Observed
Size
Scaled
Residual
0 .0000
11 .0000
55.0000
274.0000
0570
0869
1975
5780
2	.279
3	.911
6 .913
12.715
3 . 000
2 . 000
9 . 000
12.000
40
45
35
22
0 .492
-1. 011
0.886
-0.309
Chi 2 =2.15
d.f. = 2
P-value = 0.3421
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0 .1
Extra risk
0 . 95
35.9046
23.8065
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
Multistage Model with 0.95 Confidence Level
0.8
0.7
0.6
0.5
"O
CD
c
o
-I—1
o
2	0.3
LL
0.2
0.1
0
0	50	100	150	200	250
dose
14:35 12/04 2009
Figure C-4. BMD multistage (2 degree) 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.
Multistage Model. (Version: 3.0; Date: 05/16/2008)
Input Data File: Z:\14Dioxane\BMDS\mst_jbrcl998_mrat_liver_hyper_Mst-BMR10-
restrict.(d)
Gnuplot Plotting File: Z:\14Dioxane\BMDS\mst_jbrcl998_mrat_liver_hyper_Mst-BMR10-
Restrict.pit
Fri Dec 04 14:35:06 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-betal*dose^l-beta2*dose^2) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
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
Multistage
BMDL
BMD
C-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
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 = 0.0750872
Beta (1) = 0.00263797
Beta (2) =	0
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe model parameter(s) -Beta(2) have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
Background	Beta(l)
Background	1	-0.49
Beta(1)	-0.49	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf.	Limit Upper Conf. Limit
Background 0.0569658 * *	*
Beta(1) 0.00293446 * *	*
Beta(2) 0 * *	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-53.9471
-55.0858
-67 .6005
# Param's
4
2
1
Deviance Test d.f.
2 .27725
27 .3066
P-value
0 .3203
< .0001
AIC:
114.172
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0.0000
0 .0570
2 .279
3 . 000
40
0 .492
11.0000
0 .0869
3 . 911
2 . 000
45
-1.011
55.0000
0.1975
6 . 913
9 . 000
35
0.886
274.0000
0 .5780
12.715
12.000
22
-0 .309
Chi *2 = 2.15	d.f. = 2	P-value = 0.3421
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	3 5.9046
BMDL =	23 . 8 06 5
BMDU =	82 .1206
Taken together, (23.8065, 82.1206) is a 90% two-sided confidence interval for the BMD
C-12
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
Weibull Model with 0.95 Confidence Level
0.8
0.7
0.6
0.5
"O
CD
c
o
-I—1
o
,2	0.3
0.2
0.1
0
0	50	100	150	200	250
dose
14:35 12/04 2009
Figure C-5. 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.
Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: Z:\14Dioxane\BMDS\wei_jbrcl998_mrat_liver_hyper_Wei-BMR10-
Restrict.(d)
Gnuplot Plotting File: Z:\14Dioxane\BMDS\wei_jbrcl998_mrat_liver_hyper_Wei-BMR10-
Restrict.pit
Fri Dec 04 14:35:08 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-slope*dose*power)]
Dependent variable = Effect
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
Weibull
BMDL
BMD
C-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
46
47
48
49
50
51
52
53
54
Default Initial	(and Specified) Parameter Values
Background =	0.0853659
Slope =	0.00253609
Power =	1
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
Variable
Background
Slope
Power
Estimate
0.0569661
0.00293445
1
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit
0.0278498	0.00238155	0.111551
0.000814445	0.00133816	0.00453073
NA
NA - Indicates that this parameter has hit
constraint and thus has no standard error.
a bound implied by some inequality
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-53.9471
-55.0858
-67.6005
# Param's
4
2
1
Deviance Test d.f.
2 .27725
27.3066
P-value
0 .3203
<.0001
AIC:
114.172
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.0570 2.279 3.000	40	0.492
11.0000 0.0869 3.911 2.000	45	-1.011
55.0000 0.1975 6.913 9.000	35	0.886
274.0000 0.5780 12.715 12.000	22	-0.309
Chi*2 = 2.15	d.f. = 2	P-value = 0.3421
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	3 5.9047
BMDL =	23.8065
C-14
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
Quantal Linear Model with 0.95 Confidence Level
0.8
0.7
0.6
0.5
"O
CD
c
o
-I—1
o
2	0.3
LL
0.2
0.1
0
0	50	100	150	200	250
dose
14:35 12/04 2009
Figure C-6. 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.
Quantal Linear Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
Input Data File: Z:\14Dioxane\BMDS\qln_jbrcl998_mrat_liver_hyper_Qln-BMR10.(d)
Gnuplot Plotting File: Z:\14Dioxane\BMDS\qln_jbrcl998_mrat_liver_hyper_Qln-BMR10.plt
Fri Dec 04 14:35:09 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-slope*dose)]
Dependent variable = Effect
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 =	0.0853659
Slope =	0.00253609
Power =	1 Specified
Quantal Linear
BMDL
BMD
C-15
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
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe 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
Variable
Background
Slope
Estimate
0.0569665
0.00293447
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err.
0.02785
0.000814452
Lower Conf. Limit
0 .00238157
0.00133818
Upper Conf. Limit
0 .111551
0 .00453077
Model
Full model
Fitted model
Reduced model
Analysis of Deviance Table
#
Log(likelihood)
-53.9471
-55.0858
-67.6005
Param's
4
2
1
Deviance Test d.f.
2 .27725
27.3066
P-value
0 .3203
<.0001
AIC:
114.172
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.0570 2.279 3.000	40	0.492
11.0000 0.0869 3.911 2.000	45	-1.011
55.0000 0.1975 6.913 9.000	35	0.886
274.0000 0.5780 12.716 12.000	22	-0.309
Chi*2 = 2.15	d.f. = 2	P-value = 0.3421
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	35.9044
BMDL =	23.8065
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
LogProbit
BMDL
BMD
LogProbit Model with 0.95 Confidence Level
0	50 100 150 200 250 300 350 400 450
dose
15:00 02/01 2010
Source: JBRC (1998a).
Figure C-7. 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: 3.1; Date: 05/16/2008)
Input Data File: C:\14DBMDS\lnp_jbrcl998_frat_liver_hyper_Lnp-BMR10-restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnp_jbrcl99 8_frat_liver_hyper_Lnp-BMR10-
restrict.pit
Mon Feb 01 15:00:38 2010
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 = Effect
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
C-17
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
User has chosen the log transformed model
Default Initial	(and Specified) Parameter Values
background =	0
intercept =	-6.0748
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe 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	-0.99
slope	-0.99	1
Parameter Estimates
Variable
background
intercept
slope
Estimate
0
-7.72641
1.30946
Std. Err.
NA
1 .704
0 .300762
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-11. 0662
0.719976
-4 .38663
1 . 89894
NA - Indicates that this parameter has hit
constraint and thus has no standard error.
a bound implied by some inequality
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-20 . 9249
-20 . 9265
-47.3261
# Param's
4
2
1
Deviance Test d.f.
0 . 0030237
52 .8022
P-value
0.9985
<.0001
AIC :
45 . 8529
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0.0000
0.0000
0 . 000
0 . 000
38
0 .000
18.0000
0.0000
0 . 001
0 . 000
37
-0 . 039
83.0000
0.0262
0 . 995
1. 000
38
0 .005
429.0000
0.5835
14.004
14.000
24
-0.002
Chi *2 = 0.00	d.f. = 2	P-value = 0.9992
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
0 . 1
Extra risk
0 .95
137.246
88 .8743
C-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
APPENDIX D. DETAILS OF BMD ANALYSIS FOR ORAL CSF FOR 1,4-DIOXANE
Dichotomous models available in the Benchmark Dose Software (BMDS) (version 2.1.1)
were fit to the incidence data for hepatocellular carcinoma and/or adenoma for mice and rats, as
well as nasal cavity tumors, peritoneal mesotheliomas, and mammary gland adenomas in rats
exposed to 1,4-dioxane in the drinking water. Doses associated with a benchmark response
(BMR) of a 10% extra risk were calculated. BMDio and BMDLio values from the best fitting
model, determined by adequate global- fit (x p > 0.1) and AIC values, are reported for each
endpoint (U.S. EPA, 2000b). If the multistage cancer model is not the best fitting model for a
particular endpoint, the best-fitting multistage cancer model for that endpoint is also presented as
a point of comparison.
A summary of the model predictions for the Kano et al. (2009) study are shown in Table
D-l. The data and BMD modeling results are presented separately for each dataset as follows:
•	Hepatic adenomas and carcinomas in female F344 rats (Tables D-2 and D-3; Figure D-l)
•	Hepatic adenomas and carcinomas in male F344 rats (Tables D-4 and D-5; Figures D-2
and D-3)
•	Significant tumor incidence data at sites other than the liver (i.e., nasal cavity, mammary
gland, and peritoneal) in male and female F344 rats (Table D-6)
o Nasal cavity tumors in female F344 rats (Table D-7; Figure D-4)
o Nasal cavity tumors in male F344 rats (Table D-8; Figure D-5)
o Mammary gland adenomas in female F344 rats (Table D-9; Figures D-6 and D-7)
o Peritoneal mesotheliomas in male F344 rats (Table D-10; Figures D-8 and D-9)
•	Hepatic adenomas and carcinomas in female BDFi mice (Tables D-l 1, D-l2, and D-l3;
Figures D-10, D-l 1, D-12, and D-13)
•	Hepatic adenomas and carcinomas in male BDFi mice (Tables D-14 and D-l5; Figures
D-14 and D-l5)
Data and BMD modeling results from the additional chronic bioassays (NCI, 1978; Kociba et al.,
1974) were evaluated for comparison with the data from Kano et al. (2009). These results are
presented as follows:
•	Summary of BMDS dose-response modeling estimates associated with liver and nasal
tumor incidence data resulting from chronic oral exposure to 1,4-dioxane in rats and mice
(Table D-l 6)
•	Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in male and
female Sherman rats (combined) (Kociba et al., 1974) treated with 1,4-dioxane in the
drinking water for 2 years (Table D-l7)
D-l	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
o BMDS dose-response modeling results for incidence of hepatocellular carcinoma in
male and female Sherman rats (combined) (Kociba et al., 1974) exposed to
1,4-dioxane in drinking water for 2 years (Table D-18; Figures D-16 and D-17)
o BMDS dose-response modeling results for incidence of nasal squamous cell
carcinoma in male and female Sherman rats (combined) (Kociba et al., 1974) exposed
to 1,4-dioxane in the drinking water for 2 years (Table D-19; Figure D-18)
•	Incidence of nasal cavity squamous cell carcinoma and hepatocellular adenoma in
Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the drinking water (Table D-
20)
o BMDS dose-response modeling results for incidence of hepatocellular adenoma in
female Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the drinking
water for 2 years (Table D-21; Figures D-19 and D-20)
o BMDS dose-response modeling results for incidence of nasal cavity squamous cell
carcinoma in female Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the
drinking water for 2 years (Table D-22; Figures D-21 and D-22)
o BMDS dose-response modeling results for incidence of nasal cavity squamous cell
carcinoma in male Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the
drinking water for 2 years (Table D-23; Figures D-23 and D-24)
•	Incidence of hepatocellular adenoma or carcinoma in male and female B6C3F i mice
(NCI, 1978) exposed to 1,4-dioxane in drinking water (Table D-24)
o BMDS dose-response modeling results for the combined incidence of hepatocellular
adenoma or carcinoma in female B6C3Fi mice (NCI, 1978) exposed to 1,4-dioxane in
the drinking water for 2 years (Table D-25; Figure D-25)
o BMDS dose-response modeling results for incidence of combined hepatocellular
adenoma or carcinoma in male B6C3Fi mice (NCI, 1978) exposed to 1,4-dioxane in
the drinking water for 2 years (Table D-26; Figures D-26 and D-27).
D.l. GENERAL ISSUES AND APPROACHES TO BMDS MODELING
D.l.l. Combining Data on Adenomas and Carcinomas
The incidence of adenomas and the incidence of carcinomas within a dose group at a site
or tissue in rodents are sometimes combined. This practice is based upon the hypothesis that
adenomas are a severe endpoint by themselves and most would have developed into carcinomas
if exposure at the same dose was continued (U.S. EPA, 2005a). The incidence at high doses of
both tumors in rat and mouse liver is high in the key study (Kano et al., 2009). The incidence of
hepatic adenomas and carcinomas was summed without double-counting them so as to calculate
the combined incidence of either a hepatic carcinoma or a hepatic adenoma in rodents.
D-2
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
The variable N is used to denote the total number of animals tested in the dose group.
The variable Y is used here to denote the number of rodents within a dose group that have
characteristic X, and the notation Y(X) is used to identify the number with a specific
characteristic X. Modeling was performed on the adenomas and carcinomas separately and the
following combinations of tumor types:
•	Y(adenomas) = number of animals with adenomas, whether or not carcinomas are
present;
•	Y(carcinomas) = number of animals with carcinomas, whether or not adenomas are also
present;
•	Y(either adenomas or carcinomas) = number of animals with adenomas or carcinomas,
not both = Y(adenomas) + Y(carcinomas) - Y(both adenomas and carcinomas);
•	Y(neither adenomas nor carcinomas) = number of animals with no adenomas and no
carcinomas = N - Y(either adenomas or carcinomas).
D.1.2. Model Selection Criteria
Multiple models were fit to each dataset. The model selection criteria used in the
external review draft Benchmark Dose Technical Guidance Document (U.S. EPA, 2000b) were
applied as follows:
•	p-v alue for goodness-of-fit >0.10
•	AIC smaller than other acceptable models
•	x residuals as small as possible
•	No systematic patterns of deviation of model from data
Additional criteria were applied to eliminate implausible dose-response functions:
•	Monotonic dose-response functions, e.g. no negative coefficients of polynomials in MS
models
•	No infinitely steep dose-response functions near 0 (control dose), achieved by requiring
the estimated parameters "power" in the Weibull and Gamma models and "slope" in the
log-logistic model to have values >1.
Because no single set of criteria covers all contingencies, an extended list of preferred models are
presented below in Table D-l.
D.1.3. Summary
The BMDS models recommended to calculate rodent BMD and BMDL values and
corresponding human BMDhed and BMDLhed values are summarized in Table D-l.
D-3
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-l. Recommended models for rodents exposed to 1,4-dioxane in
drinking water (Kano et al., 2009)

Model



BMDa
BMDLa
BMD||||)a


selection
Model

P-
mg/kg-
mg/kg-
mg/kg-
BMDL|ni)a
Endpoint
criterion
Type
AIC
value
day
day
day
mg/kg-day
Female F344 Rat

Hepatic
Lowest AIC
Multistage
91.5898
0.4516
79.83
58.09
19.84
14.43

Tumors

(2 degree)







Mammary
Lowest AIC
LogLogistic
194.151
0.8874
161.01
81.91
40.01
20.35

Gland









Tumors









Nasal
Lowest AIC
Multistage
42.6063
0.9966
381.65
282.61
94.84
70.23

Cavity
Tumors

(3 degree)






Male R344 Rat

Hepatic
Lowest AIC
Probit
147.787
0.9867
62.20
51.12
17.43
14.33

Tumors









Peritoneal
Lowest AIC
Probit
138.869
0.9148
93.06
76.32
26.09
21.39

Meso-









thelioma









Nasal
Lowest AIC
Multistage
24.747
0.9989
328.11
245.63
91.97
68.85

Cavity
Tumors

(3 degree)






Female BDFiMouse

Hepatic
Lowest AIC
LogLogistic
176.225
0.1411
5.54
3.66
0.83
0.55

Tumors
BMR 50%
LogLogistic
176.225
0.1411
49.90b
32.94b
7.51b
4.96b
Male BDFj Mouse

Hepatic
Lowest AIC
Log-
248.839
0.3461
34.78
16.60
5.63
2.68

Tumors

Logistic






1	aValues for BMR 10% unless otherwise noted.
2	bBMR 50%.
D.2. FEMALE F344 RATS: HEPATIC CARCINOMAS AND ADENOMAS
3	The incidence data for hepatic carcinomas and adenomas in female F344 rats (Kano et
4	al., 2009) are shown in Table D-2.
Table D-2. Data for hepatic adenomas and carcinomas in female F344 rats
(Kano et al., 2009)
Tumor type
Dose (mg/kg-day)
0
18
83
429
Hepatocellular adenomas
3
1
6
48
Hepatocellular carcinomas
0
0
0
10
Either adenomas or carcinomas
3
1
6
48
Neither adenomas nor carcinomas
47
49
44
2
Total number per group
50
50
50
50
Source: Kano et al. (2009).
5	Note that the incidence of rats with adenomas, with carcinomas, and with either
6	adenomas or carcinomas are monotone non-decreasing functions of dose except for 3 female rats
D-4
DRAFT - DO NOT CITE OR QUOTE

-------
1	in the control group. These data therefore appear to be appropriate for dose-response modeling
2	using BMDS.
3	The results of the BMDS modeling for the entire suite of models are presented in Table
4	D-3.
Table D-3. BMDS dose-response modeling results for the combined
incidence of hepatic adenomas and carcinomas in female F344 rats (Kano et
al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
93.1067
0.3024
89.46
62.09
0.027
22.23
15.43
Logistic
91.7017
0.4459
93.02
71.60
0.077
23.12
17.79
LogLogistic
93.102
0.3028
88.34
65.52
0.016
21.95
16.28
LogProbitb
93.0762
0.3074
87.57
66.19
0.001
21.76
16.45
Multistage-Cancer
(1 degree)
114.094
0.0001
25.58
19.92
-1.827
6.36
4.95
Multistage-Cancer
(2 degree)0
91.5898
0.4516
79.83
58.09
-0.408
19.84
14.43
Multistage-Cancer
(3 degree)
93.2682
0.2747
92.81
59.31
0.077
23.06
14.74
Probit
91.8786
0.3839
85.46
67.84
-0.116
21.24
16.86
Weibull
93.2255
0.2825
92.67
59.89
0.088
23.03
14.88
Quantal-Linear
114.094
0.0001
25.58
19.92
-1.827
6.36
4.95
Dichotomous-Hill
4458.37
NCd
NCd
NCd
0
0
0
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bSlope restricted > 1.
°Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
D-5
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
Multistage Cancer Model with 0.95 Confidence Level
i
0.6
0.4
0.2
07:20 10/26 2009
50	100	150 200 250 300 350 400 450
dose
Source: Kano et al. (2009).
Figure D-l. Multistage BMD model (2 degree) for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_hepato_adcar_Msc-
BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_frat_hepato_adcar_Msc-BMR10-2poly.pit
Mon Oct 26 08:20:52 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l-beta2*dose*2)]
The parameter betas are restricted to be positive
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0281572
I	I	I	I	I	I	I
Multistage Cancer 	
Linear extrapolation 	
BMDL BMD
D-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
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
Beta (1) =	0
Beta (2) = 1.73306e-005
Asymptotic Correlation Matrix of Parameter Estimates ( *** The model parameter(s)
Beta(1)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
-0.2
Beta(2)
-0.2
1
Variable
Background
Beta (1)
Beta(2)
Estimate
0.0362773
0
1.65328e-005
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err. Lower Conf. Limit Upper Conf. Limit
- Indicates that this value is not calculated.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Log(likelihood) # Param's Deviance Test d.f. P-value
-42 .9938
-43 .7949
-120.43
91.5898
1.60218
154.873
0.4488
<.0001
Dose
Est. Prob.
Goodness of Fit
Expected Observed	Size
0.0000
18 .0000
83.0000
429 . 0000
0.0363
0.0414
0.1400
0 .9540
1	.814
2	. 071
7 . 001
47.701
3 . 000
1. 000
6 . 000
48 . 000
Chi *2 = 1.59	d.f. = 2
Benchmark Dose Computation
P-value = 0.4516
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0 .1
Extra risk
0 . 95
79 .8299
58.085
94.0205
50
50
50
50
Scaled
Residual
0 . 897
-0.760
-0.408
0 .202
Taken together, (58.085 , 94.0205) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00172161
D-7
DRAFT - DO NOT CITE OR QUOTE

-------
D.3. MALE F344 RATS: HEPATIC CARCINOMAS AND ADENOMAS
1	The data for hepatic adenomas and carcinomas in male F344 rats (Kano et al., 2009) are
2	shown in Table D-4.
Table D-4. Data for hepatic adenomas and carcinomas in male F344 rats
(Kano et al., 2009)
Tumor type
Dose (mg/kg-day)
0
11
55
274
Hepatocellular adenomas
3
4
7
32
Hepatocellular carcinomas
0
0
0
14
Either adenomas or carcinomas
3
4
7
39
Neither adenomas nor carcinomas
47
46
43
11
Total number per group
50
50
50
50
Source: Kano et al. (2009).
3	Note that the incidence of rats with hepatic adenomas, carcinomas, and with either
4	adenomas or carcinomas are monotone non-decreasing functions of dose. These data therefore
5	appear to be appropriate for dose-response modeling using BMDS.
6	The results of the BMDS modeling for the entire suite of models tested using the data for
7	hepatic adenomas and carcinomas for male F344 rats are presented in Table D-5.
D-8
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-5. BMDS dose-response modeling results for the combined
incidence of adenomas and carcinomas in livers of male F344 rats (Kano et
al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10 hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
149.884
0.7257
62.41
30.79
-0.03
17.49
8.63
Logistic
147.813
0.9749
68.74
55.39
0.097
19.27
15.53
LogLogistic
149.886
0.7235
62.10
34.61
-0.021
17.41
9.70
LogProbitb
149.913
0.6972
61.70
37.49
-0.003
17.29
10.51
Multistage-Cancer
(1 degree)
152.836
0.0978
23.82
18.34
-0.186
6.68
5.14
Multistage-Cancer
(2 degree)
149.814
0.8161
61.68
28.26
-0.063
17.29
7.92
Multistage-Cancer
(3 degree)
149.772
0.9171
63.62
27.49
-0.024
17.83
7.71
Probit0
147.787
0.9867
62.20
51.12
-0.05
17.43
14.33
Weibull
149.856
0.7576
62.63
30.11
-0.039
17.56
8.44
Quantal-Linear
152.836
0.0978
23.82
18.34
-0.186
6.68
5.14
Dichotomous-Hill
4441.71
NCd
NCd
NCd
0
0
0
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bSlope restricted > 1.
°Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
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
Probit Model with 0.95 Confidence Level
Source: Kano et al. (2009).
Figure D-2. Probit BMD model for the combined incidence of hepatic adenomas
and carcinomas in male F344 rats.
50	100	150	200	250
dose
07:32 10/26 2009
BMDL BMD
—I	i	I	lL_J	L±J	I	1_
Probit
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-
BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2 00 9_mrat_hepato_adcar_Prb-BMR10.pit
Mon Oct 26 08:32:08 2009
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 = Effect
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 = 2 50
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 =	-1.51718
slope =	0.00831843
D-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
( *** xhe 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
-0.69
slope
-0.69
1
Variable
intercept
slope
Estimate
1 . 53138
0 . 00840347
Parameter Estimates
95 . 0?
Wald Confidence Interval
Std. Err.
0 .160195
0.000976752
Lower Conf. Limit
-1.84535
0 .00648907
Upper Conf. Limit
-1 .2174
0 . 0103179
Analysis of Deviance Table
Model	Log(likelihood)	# Param's	Deviance	Test d.f.	P-value
Full model	-71.8804	4
Fitted model	-71.8937	2	0.0265818	2	0.9868
Reduced model	-115.644	1	87.528	3	<.0001
AIC:	147.787
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0.0000	0.0628	3.142	3.000	50	-0.083
11.0000	0.0751	3.754	4.000	50	0.132
55.0000	0.1425	7.125	7.000	50	-0.050
274.0000	0.7797	38.985 39.000	50	0.005
Chi*2 = 0.03	d.f. = 2	P-value = 0.9867
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	62 .1952
BMDL =	51.1158
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.8
0
BMDI
BMD
0
50
100
150
200
250
dose
07:32 10/26 2009
Source: Kano et al. (2009).
Figure D-3. Multistage BMD model (3 degree) for the combined incidence of
hepatic adenomas and carcinomas in male F344 rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_hepato_adcar_Msc-
BMR10-3poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_mrat_hepato_adcar_Msc-BMR10-3poly.pit
Mon Oct 26 08:32:08 2009
BMDS Model Run
The form of the probability function is: P [response] = background + (1-background)*[1-
EXP ( - betal*dose^l-beta2*dose^2-beta3 *dose*3) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
D-12
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
Background = 0.0623822
Beta (1) = 0.00142752
Beta (2) =	0
Beta (3) = 5.14597e-008
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -Beta(2)have been estimated at a boundary point, or have
been specified by the user,and do not appear in the correlation matrix )
Background	Beta(l)	Beta (3)
Background 1	-0.67	0.58
Beta (1) -0.67	1	-0.95
Beta (3) 0.58	-0.95	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
Background	0.0619918	*	*	*
Beta (1)	0.001449	*	*	*
Beta (2)	0	*	*	*
Beta (3)	5.11829e-008	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-71 .8804
-71.8858
-115 . 644
# Param's
4
3
1
Deviance Test d.f.
0 . 0107754
87 . 528
P-value
0 . 9173
< .0001
AIC :
149.772
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 . 0000
0 .0620
3 . 100
3 . 000
50
-0.058
11.0000
0 .0769
3 . 844
4 . 000
50
0 .083
55.0000
0.1412
7 . 059
7 . 000
50
-0.024
274.0000
0.7799
38 . 997
39.000
50
0 .001
Chi *2 = 0.01	d.f. = 1	P-value = 0.9171
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	63 .6179
BMDL =	27.4913
BMDU =	123.443
Taken together, (27.4913, 123.443) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00363752
D-13
DRAFT - DO NOT CITE OR QUOTE

-------
D.4. F344 RATS: TUMORS AT OTHER SITES
1	The data for tumors at sites other than the liver in male and female F344 rats (Kano et al.,
2	2009) are shown in Table D-6. Note that the incidence of rats with these endpoints are monotone
3	non-decreasing functions (except female peritoneal mesotheliomas). These data therefore appear
4	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 (Kano et al., 2009)
Tumor site and type
Dose (mg/kg-day)
Female
Male
0
18
83
429
0
11
55
274
Nasal cavity squamous cell carcinoma
0
0
0
7
0
0
0
3
Peritoneal mesothelioma
1
0
0
0
2
2
5
28
Mammary gland adenoma
6
7
10
16
0
1
2
2
Total number per group
50
50
50
50
50
50
50
50
Source: Kano et al., (2009).
5	The results of the BMDS modeling for the entire suite of models are presented in Tables
6	D-7 through Table D-10 for tumors in the nasal cavity, mammary gland, and peritoneal cavity.
D-14
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-7. BMDS dose-response modeling results for the incidence of nasal
cavity tumors in female F344 ratsa (Kano et al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x2"
BMDio hed
mg/kg-day
BMDLio hed
mg/kg-day
Gamma
44.4964
1
403.82
269.03
0
100.35
66.85
Logistic
44.4963
1
421.54
351.74
0
104.75
87.41
LogLogistic
44.4963
1
413.69
268.85
0
102.80
66.81
LogProbit0
44.4963
1
400.06
260.38
0
99.42
64.71
Multistage-Cancer
(1 degree)
45.6604
0.6184
375.81
213.84
0.595
93.39
53.14
Multistage-Cancer
(2 degree)
43.0753
0.9607
366.07
274.63
0.109
90.97
68.24
Multistage-Cancer
(3 degree)d
42.6063
0.9966
381.65
282.61
0.021
94.84
70.23
Probit
44.4963
1
414.11
333.31
0
102.91
82.83
Weibull
44.4963
1
414.86
273.73
0
103.09
68.02
Quantal-Linear
45.6604
0.6184
375.81
213.84
0.595
93.39
53.14
Dichotomous-Hill
46.4963
0.9997
413.96
372.57
1.64xl0-8
102.87
92.58
aNasal cavity tumors in female F344 rats include squamous cell carcinoma and esthesioneuro-epithelioma.
bMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
°Slope restricted > 1.
dBest-fitting model.
D-15
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
Multistage Cancer Model with 0.95 Confidence Level
0.3
Multistage Cancer
Linear extrapolation
0.25
0.2
0.15
0.1
0.05
0
BMDL
BMD
0
50
100
150
200
250
300
350
400
450
dose
07:28 10/26 2009
Source: Kano et al. (2009).
Figure D-4. Multistage BMD model (3 degree) for nasal cavity tumors in female
F344 rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-
BMR10-3poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_frat_nasal_car_Msc-BMR10-3poly.pit
Mon Oct 26 08:28:58 2009
BMDS Model Run
The form of the probability function is: P[response] = background + (1-
background)* [1-EXP(-betal*dose*l-beta2*dose*2-beta3*dose*3) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background =	0
Beta (1) =	0
Beta (2) =	0
Beta (3) = 1.91485e-009
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
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -Background -Beta(l) -Beta(2)
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Beta(3)
Beta (3)
Parameter Estimates
Variable
Background
Beta (1)
Beta (2)
Beta(3)
Estimate
0
0
0
1.89531e-009
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
Log(1i ke1i hood)
-20.2482
-20 .3031
-30 .3429
# Param's
4
1
1
Deviance Test d.f.
0 .109908
20.1894
P-value
0.9906
0.0001551
AIC :
42 .6063
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0.0000
0.0000
0 . 000
0 . 000
50
0 .000
18.0000
0.0000
0 . 001
0 . 000
50
-0.024
83.0000
0.0011
0 . 054
0 . 000
50
-0.233
429.0000
0.1390
6 . 949
7 . 000
50
0 . 021
Chi *2 = 0.06	d.f. = 3	P-value = 0.9966
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	3 81.651
BMDL =	2 8 2.609
BMDU =	500.178
Taken together, (282.609, 500.178) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000353846
D-17
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-8. BMDS dose-response modeling results for the incidence of nasal
cavity tumors in male F344 rats" (Kano et al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x2b
BMDio hed
mg/kg-day
BMDLio hed
mg/kg-day
Gamma
26.6968
1
299.29
244.10
0
83.89
68.42
Logistic
26.6968
1
281.06
261.29
0
78.78
73.24
LogLogistic
26.6968
1
288.31
245.29
0
80.81
68.75
LogProbit0
26.6968
1
303.06
238.86
0
84.94
66.95
Multistage-Cancer
(1 degree)
26.0279
0.8621
582.49
256.43
0.384
163.28
71.88
Multistage-Cancer
(2 degree)
24.9506
0.988
365.19
242.30
0.073
102.37
67.92
Multistage-Cancer
(3 degree)d
24.747
0.9989
328.11
245.63
0.015
91.97
68.85
Probit
26.6968
1
287.96
257.01
0
80.72
72.04
Weibull
26.6968
1
288.00
246.36
0
80.73
69.06
Quantal-Linear
26.0279
0.8621
582.49
256.43
0.384
163.28
71.88
Dichotomous-Hill
28.6968
0.9994
290.52
261.47
6.25x 10"5
81.44
73.29
aNasal cavity tumors in male F344 rats include squamous cell carcinoma, Sarcoma: NOS, rhabdomyosarcoma, and
esthesioneuro-epithelioma.
bMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
°Slope restricted > 1.
dBest-fitting model.
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.15
0.05
0
BMDL
BMC)
0
50
100
150
200
250
300
dose
07:34 10/26 2009
Source: Kano et al. (2009).
Figure D-5. Multistage BMD model (3 degree) for nasal cavity tumors in male F344
rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_nasal_car_Msc-
BMR10-3poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_mrat_nasal_car_Msc-BMR10-3poly.pit
Mon Oct 26 08:34:20 2009
BMDS Model Run
The form of the probability function is: P [response] = background + (1-background)* [1-
EXP(-betal*dose^l-beta2*dose^2-beta3 *dose*3) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background =	0
D-19
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
Beta (1) =	0
Beta (2) =	0
Beta(3) = 3.01594e-009
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -Background -Beta(l) -Beta(2)
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Beta (3)
Beta(3)	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf.	Limit Upper Conf. Limit
Background 0 * *	*
Beta(1) o * *	*
Beta(2) 0 * *	*
Beta(3) 2.98283e-009 * *	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model	Log(likelihood)	# Param's	Deviance Test d.f.	P-value
Full model	-11.3484	4
Fitted model	-11.3735	1	0.0502337 3	0.9971
Reduced model	-15.5765	1	8.45625 3	0.03747
AIC:	24.747
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 . 0000
0.0000
0 . 000
0 . 000
50
0 .000
11.0000
0.0000
0 . 000
0 . 000
50
-0.014
55.0000
0.0005
0 . 025
0 . 000
50
-0 .158
274.0000
0.0595
2 . 976
3 . 000
50
0 .015
Chi *2 = 0.03	d.f. = 3	P-value = 0.9989
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	328.108
BMDL =	245.634
BMDU =	1268 .48
Taken together, (245.634, 1268.48) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00040711
D-20
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-9. BMDS dose-response modeling results for the incidence of
mammary gland adenomas in female F344 rats (Kano et al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
r'
BMDiohed
mg/kg-day
BMDLio hed
mg/kg-day
Gamma
194.222
0.8559
176.66
99.13
0.465
43.90
24.63
Logistic
194.475
0.7526
230.35
159.73
0.612
57.24
39.69
LogLogisticb
194.151
0.8874
161.01
81.91
0.406
40.01
20.35
LogProbit0
195.028
0.5659
270.74
174.66
-0.075
67.28
43.41
Multistage-Cancer
(1 degree)
194.222
0.8559
176.66
99.13
0.465
43.90
24.63
Multistage-Cancer
(2 degree)
194.222
0.8559
176.66
99.13
0.465
43.90
24.63
Multistage-Cancer
(3 degree)
194.222
0.8559
176.66
99.13
0.465
43.90
24.63
Probit
194.441
0.7656
223.04
151.60
0.596
55.43
37.67
Weibull
194.222
0.8559
176.65
99.13
0.465
43.90
24.63
Quantal-Linear
194.222
0.8559
176.65
99.13
0.465
43.90
24.63
Dichotomous-Hill
197.916
NCd
94.06
14.02
3.49x 10"5
23.37
3.48
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
hBest-fitting model.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
D-21
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
Log-Logistic Model with 0.95 Confidence Level
0.5
Log-Logistic
0.4
0.3
0.2
1
BMDL
BMD
0	50	100	150	200	250	300	350	400	450
dose
11:31^02/01 201{L	,
Source: Kano et al. (2009).
Figure D-6. LogLogistic BMD model for mammary gland adenomas in female F344
rats.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnl_kano2 00 9_frat_mamm_ad_Lnl-BMR10-Restrict.plt
Mon Feb 01 11:31:31 2010
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/[1 + EXP(-intercept-slope*Log(dose)) ]
Dependent variable = Effect
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 =	0.12
intercept =	-7.06982
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
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
( *** xhe 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	-0.53
intercept	-0.53	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
background	0.130936	*	*	*
intercept	-7.2787	*	*	*
slope	1	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(likelihood)
-94.958
-95.0757
-98 . 6785
# Param's
4
2
1
Deviance Test d.f.
0 .235347
7.4409
P-value
0.889
0 .0591
AIC:
194.151
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.1309 6.547 6.000	50	-0.229
18.0000 0.1416 7.080 7.000	50	-0.032
83.0000 0.1780 8.901 10.000	50	0.406
429.0000 0.3294 16.472 16.000	50	-0.142
Chi*2 = 0.24	d.f. = 2	P-value = 0.8874
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	161.012
BMDL =	81.9107
D-23
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
Multistage Cancer Model with 0.95 Confidence Level
0.5
Multistage Cancer
Linear extrapolation
0.4
0.3
0.2
0.1
BMDL
BMD
0
50
100
150
200
250
300
350
400
450
dose
07:27 10/26 2009
Source: Kano et al. (2009).
Figure D-7. Multistage BMD model (1 degree) for mammary gland adenomas in
female F344 rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_mamm_ad_Msc-BMR10-
lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_frat_mamm_ad_Msc-BMR10-lpoly.pit
Mon Oct 26 08:27:02 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background =	0.136033
Beta (1) = 0.000570906
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
Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(l)
Background	1	-0.58
Beta (1)	-0.58	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
Background	.133161	*	*	*
Beta (1)	0.000596394	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-94.958
-95.Ill
-98.6785
# Param's
4
2
1
Deviance Test d.f.
0 .305898
7.4409
P-value
0 .8582
0.0591
AIC :
194.222
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 . 0000
0 .1332
6 . 658
6 . 000
50
-0.274
18 . 0000
0 .1424
7 . 121
7 . 000
50
-0 . 049
83 . 0000
0.1750
8 . 751
10 . 000
50
0.465
429.0000
0.3288
16 . 442
16.000
50
-0.133
Chi *2 = 0.31	d.f. = 2	P-value = 0.8559
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	176.663
BMDL =	99.1337
BMDU =	501.523
Taken together, (99.1337, 501.523) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00100874
D-25
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-10. BMDS dose-response modeling results for the incidence of
peritoneal mesotheliomas in male F344 rats (Kano et al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x2a
BMDio hed
mg/kg-day
BMDLio hed
mg/kg-day
Gamma
140.701
0.9189
73.52
35.62
0.018
20.61
9.98
Logistic
139.016
0.8484
103.52
84.35
0.446
29.02
23.65
LogLogistic
140.699
0.9242
72.56
36.37
0.014
20.34
10.19
LogProbitb
140.69
0.9852
70.29
52.59
0.001
19.70
14.74
Multistage-Cancer
(1 degree)
140.826
0.3617
41.04
30.51
-1.066
11.50
8.55
Multistage-Cancer
(2 degree)
140.747
0.8135
77.73
35.43
0.067
21.79
9.93
Multistage-Cancer
(3 degree)
140.747
0.8135
77.73
35.43
0.067
21.79
9.93
Probit0
138.869
0.9148
93.06
76.32
0.315
26.09
21.39
Weibull
140.709
0.8915
74.77
35.59
0.027
20.96
9.97
Quantal-Linear
140.826
0.3617
41.04
30.51
-1.066
11.50
8.55
Dichotomous-Hill
2992
NCd
NCd
NCd
0
0
0
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bSlope restricted > 1.
°Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
D-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
Probit Model with 0.95 Confidence Level
Probit
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
BMDI
BMD
0
50
100
150
200
250
dose
07:41 10/26 2009
Source: Kano et al. (2009).
Figure D-8. Probit BMD model for peritoneal mesotheliomas in male F344 rats.
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_peri_meso_Prb-
BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2 00 9_mrat_peri_meso_Prb-BMR10.pit
Mon Oct 26 08:41:29 2009
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 = Effect
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 = 2 50
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 =	-1.73485
slope = 0.00692801
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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 )
D-27
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
intercept
slope
intercept
1
-0 . 75
slope
-0 . 75
1
Variable
intercept
slope
Estimate
-1 . 73734
0 .00691646
Parameter Estimates
95.0% Wald Confidence Interval
Std. Err.	Lower Conf. Limit Upper Conf. Limit
0.18348	-2.09695	-1.37772
0.000974372	0.00500672	0.00882619
Analysis of Deviance Table
Model	Log(likelihood) # Param's Deviance Test d.f. P-value
Full model	-67.3451	4
Fitted model	-67.4344	2	0.178619	2	0.9146
Reduced model	-95.7782	1	56.8663	3	<.0001
AIC:	138.869
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 .0000
0.0412
2 .058
2 . 000
50
-0.041
11 .0000
0 . 0483
2 .417
2 . 000
50
-0.275
55 .0000
0.0874
4 .370
5 . 000
50
0 .315
274 .0000
0 .5627
28 . 134
28 . 000
50
-0.038
Chi*2 = 0.18 d.f.	= 2 P-value = 0.9148
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	93 . 0615
BMDL =	76 .3242
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
Multistage Cancer Model with 0.95 Confidence Level
0.7
Multistage Cancer
Linear extrapolation
0.6
0
BMDL
BMD
0
50
100
150
200
250
dose
07:41 10/26 2009
Source: Kano et al. (2009).
Figure D-9. Multistage BMD (2 degree) model for peritoneal mesotheliomas in male
F344 rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_peri_meso_Msc-
BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_mrat_peri_meso_Msc-BMR10-2poly.pit
Mon Oct 26 08:41:28 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l-beta2*dose*2)]
The parameter betas are restricted to be positive
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-29
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
Default Initial Parameter Values
Background = 0.0358706
Beta (1) = 0.000816174
Beta (2) = 7.47062e-006
Asymptotic Correlation Matrix of Parameter Estimates
Background
Beta (1)
Beta (2)
Background
1
-0.67
0 . 59
Beta(1)
-0.67
1
-0.98
Beta (2)
0	.59
-0 . 98
1
Variable
Background
Beta (1)
Beta (2)
Estimate
0.0366063
0.000757836
7.68 93e-006
Parameter Estimates
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
Log(likelihood)	# Param's	Deviance Test d.f.
-67.3451	4
-67.3733	3	0.056567 1
-95.7782	1	56.8663 3
P-value
0 . 812
< .0001
AIC :
140.747
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 . 0000
0 . 0366
1. 830
2 . 000
50
0 .128
11.0000
0.0455
2 .275
2 . 000
50
-0.186
55.0000
0.0972
4 . 859
5 . 000
50
0 .067
274.0000
0.5605
28.027
28 . 000
50
-0.008
Chi *2 = 0.06	d.f. = 1	P-value = 0.8135
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	77.7277
BMDL =	3 5.4296
BMDU =	118.349
Taken together, (35.4296, 118.349) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor =	0.0028225
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
D.5. FEMALE BDFi MICE: HEPATIC CARCINOMAS AND ADENOMAS
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 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 66 mg/kg-day and then decreases sharply with increasing dose. This cannot be
modeled by a multistage model using only non-negative coefficients. To some extent the
incidence of "either adenomas or carcinomas" 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 278 mg/kg-day), thus is not well characterized by any multistage model.
Table D-ll. Data for hepatic adenomas and carcinomas in female BDFi mice
(Kano et al., 2009)

Dose (mg/kg-day)
Tumor type
0
66
278
967
Hepatocellular adenomas
5
31
20
3
Hepatocellular carcinomas
0
6
30
45
Either adenomas or carcinomas
5
35
41
46
Neither adenomas nor carcinomas
45
15
9
4
Total number per group
50
50
50
50
Source: Kano et al. (2009).
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-12. The multistage models did
not provide reasonable fits to the incidence data for hepatocellular adenoma or carcinoma in
female BDFi mice. The log-logistic model provided the best-fit to the data as indicated by the
AIC andp-value as was chosen as the best-fitting model to carry forward in the analysis;
however, this model resulted in a BMDLio much lower than the response level at the lowest dose
in the study (Kano et al., 2009). Thus, the log-logistic model was run for BMRs of 30 and 50%.
The output from these models are shown in Figures D-l 1 and D-12. A summary of the BMD
results for BMRs of 10, 30, and 50% are shown in Table D-13. Using a higher BMR resulted in
BMDLs closer to the lowest observed response data, and a BMR of 50% was chosen to carry
forward in the analysis.
The graphical output from fitting these models suggested that a simpler model obtained
by dropping the data point for the highest dose (967 mg/kg-day) might also be adequate. This
was tested and the results did not affect the choice of the model, nor significantly affect the
resulting BMDs and BMDLs.
D-31
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-12. BMDS dose-response modeling results for the combined
incidence of hepatic adenomas and carcinomas in female BDFi mice (Kano et
al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
2a
X
BMDiohed
mg/kg-day
BMDLio hed
mg/kg-day
Gamma
203.409
0
26.50
19.55
-2.661
3.99
2.94
Logistic
215.019
0
58.21
44.54
3.198
8.76
6.70
LogLogisticb
176.225
0.1411
5.54
3.66
-0.122
0.83
0.55
LogProbit0
198.414
0
26.39
19.58
-1.168
3.97
2.94
Multistage-Cancer
(1 degree)
203.409
0
26.50
19.55
-2.661
3.99
2.94
Multistage-Cancer
(2 degree)
203.409
0
26.50
19.55
-2.661
3.99
2.94
Multistage-Cancer
(3 degree)
203.409
0
26.50
19.55
-2.661
3.99
2.94
Probit
217.735
0
70.11
56.38
3.111
10.55
8.48
Weibull
203.409
0
26.50
19.55
-2.661
3.99
2.94
Quantal-Linear
203.409
0
26.50
19.55
-2.661
3.99
2.94
Dichotomous-Hill
7300.47
NCd
NCd
NCd
0
0
0
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
hBest-fitting model, lowest AIC value.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
Table D-13. BMDS LogLogistic dose-response modeling results using BMRs
of 10, 30, and 50% for the combined incidence of hepatic adenomas and
carcinomas in female BDFi mice (Kano et al., 2009).
BMR
AIC
/j-value
BMD
mg/kg-day
BMDL
mg/kg-day
x23
BMDini)
mg/kg-day
BMDLhed
mg/kg-day
10%
176.225
0.1411
5.54
3.66
-0.122
0.83
0.55
30%
176.225
0.1411
21.39
14.12
-0.122
3.22
2.12
50%
176.225
0.1411
49.90
32.94
1.238
7.51
4.96
"Maximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
D-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
Log-Logistic Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
MDLBMD
Log-Logistic
200
400
600
800
1000
07:12 10/26 2009
dose
Source: Kano et al. (2009).
Figure D-10. LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDFi mice with a BMR of 10%.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2 00 9_fmouse_hepato_adcar_Lnl-BMR10-
Restrict.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2 00 9_fmouse_hepato_adcar_Lnl-BMR10-
Restrict.pit
Mon Oct 26 08:12:42 2009
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/ [1+EXP(- intercept-slope*Log(dose))]
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
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
Default Initial Parameter Values
background =	0.1
intercept =	-4.33842
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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	-0.32
intercept	-0.32	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
background	0.105274	*	*	*
intercept	-3.91	*	*	*
slope	1	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(likelihood)
-84.3055
-86 .1125
-131.248
# Param's
4
2
1
Deviance Test d.f.
3.61404
93 .8853
P-value
0 .1641
< .0001
AIC:
176.225
Dose
Est. Prob.
Goodness of Fit
Expected
Observed
Size
Scaled
Residual
0 .0000
66 .0000
278.0000
967 .0000
1053
6148
8638
9561
5 .264
30.739
43.192
47.805
5 .000
35.000
41.000
46 . 000
50
50
50
50
-0.122
1. 238
-0.904
-1.246
Chi*2 =3.92
d.f. = 2
P-value = 0.1411
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	5.54431
BMDL =	3.65971
D-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
Log-Logistic Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
Log-Logistic
I3MDL BMD
200
400
600
800
1000
dose
09:51^2/01 20m	, , ,
Source: Kano et al. (2009).
Figure D-ll. LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDFi mice with a BMR of 30%.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\14DBMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR30-Restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnl_kano2 00 9_fmouse_hepato_adcar_Lnl-BMR3 0-
Restrict.pit
Mon Feb 01 09:51:15 2010
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/[1 + EXP(-intercept-slope*Log(dose))]
Dependent variable = Effect
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 =	0.1
intercept =	-4.33842
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
D-35
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
( *** xhe 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	-0.32
intercept	-0.32	1
Parameter Estimates
Variable
background
intercept
slope
Estimate
0 . 105274
-3 . 91
1
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
Log (likelihood) # Param's Deviance Test d.f.
-84.3055
-86 .1125
-131.248
3.61404
93.8853
P-value
0 .1641
< .0001
AIC:
176.225
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.1053 5.264 5.000	50	-0.122
66.0000 0.6148 30.739 35.000	50	1.238
278.0000 0.8638 43.192 41.000	50	-0.904
967.0000 0.9561 47.805 46.000	50	-1.246
Chi *2 = 3.92	d.f. = 2	P-value = 0.1411
Benchmark Dose Computation
Specified effect =	0.3
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	21.3852
BMDL =	14.116
D-36
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
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
0.8
0.6
0.4
0.2
0
BMDL
BMD
0
200
400
600
800
1000
dose
09:51 02/01 2010
Source: Kano et al. (2009).
Figure D-12. LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDFi mice with a BMR of 50%.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\14DBMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR50-Restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnl_kano2 00 9_fmouse_hepato_adcar_Lnl-BMR5 0-
Restrict.pit
Mon Feb 01 09:51:15 2010
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/[1 + EXP(-intercept-slope*Log(dose))]
Dependent variable = Effect
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
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
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 Parameter Values
background =	0.1
intercept =	-4.33842
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
(*** xhe 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
1	-0.32
-0.32	1
background
intercept
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf.	Limit Upper Conf. Limit
background 0.105274 * *	*
intercept -3.91 * *	*
slope 1 * *	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model	Log(likelihood)	# Param's	Deviance Test d.f.	P-value
Full model	-84.3055	4
Fitted model	-86.1125	2	3.61404 2	0.1641
Reduced model	-131.248	1	93.8853 3	<.0001
AIC:	176.225
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.1053 5.264 5.000	50	-0.122
66.0000 0.6148 30.739 35.000	50	1.238
278.0000 0.8638 43.192 41.000	50	-0.904
967.0000 0.9561 47.805 46.000	50	-1.246
Chi*2 = 3.92	d.f. = 2	P-value = 0.1411
Benchmark Dosi
Specified effect
Risk Type
Confidence level
BMD
BMDL
Computation
0 . 5
Extra risk
0 . 95
49.8988
32 . 9374
D-38
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
"O
©
<
c
o
0.4
0.8
0.6
/
0.2
0
BMDLBMD
0
200
400
600
800
1000
dose
07:12 10/26 2009
Source: Kano et al. (2009).
Figure D-13. Multistage BMD model (1 degree) for the combined incidence of
hepatic adenomas and carcinomas in female BDFi mice.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_fmouse_hepato_adcar_Msc-BMR10-lpoly. (d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_fmouse_hepato_adcar_Msc-BMR10-lpoly.pit
Mon Oct 26 08:12:43 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-39
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
Default Initial Parameter Values
Background =	0.51756
Beta (1) = 0.00200935
Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(l)
Background	1	-0.65
Beta (1)	-0.65	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
Background	0.266368	*	*	*
Beta (1)	0.0039752	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(likelihood)	# Param's	Deviance Test d.f. P-value
-84.3055	4
-99.7043	2	30.7975 2
-131.248	1	93.8853 3
2.0530531e-007
<.0001
AIC:	203.409
Goodness of Fit
Dose	Est._Prob. Expected Observed	Size
Scaled
Residual
0 . 0000
0.2664
13.318
5 . 000
50
-2.661
66.0000
0.4357
21.783
35.000
50
3 .770
278.0000
0.7570
37.852
41.000
50
1 . 038
967.0000
0 .9843
49.215
46 . 000
50
-3 . 657
Chi *2 = 35.74	d.f. = 2	P-value = 0.0000
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	26.504 5
BMDL =	19.5505
BMDU =	3 7.6816
Taken together, (19.5505, 37.6816) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00511497
D-40
DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
8
9
10
11
12
D.6. MALE BDFi MICE: HEPATIC CARCINOMAS AND ADENOMAS
Data for hepatic carcinomas and adenomas in male BDFi mice (Kano et al., 2009) are
shown in Table D-14. Note that the incidence of carcinomas and the incidence of either
adenomas or carcinomas 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 191 mg/kg-day and then decreases sharply with
increasing dose. This cannot be modeled by a multistage 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 (49 and 191 mg/kg-day), thus is not well characterized
by any multistage model.
Table D-14. Data for hepatic adenomas and carcinomas in male BDFi mice
(Kano et al., 2009)
Tumor type
Dose (mg/kg-day)
0
49
191
677
Hepatocellular adenomas
9
17
23
11
Hepatocellular carcinomas
15
20
23
36
Either adenomas or carcinomas
23
31
37
40
Neither adenomas nor carcinomas
27
19
13
10
Total number per group
50
50
50
50
Source: Kano et al. (2009).
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-15.
D-41
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-15. BMDS dose-response modeling results for the combined
incidence of hepatic adenomas and carcinomas in male BDFi mice (Kano et
al., 2009)
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10 hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
250.551
0.1527
70.99
44.00
0.605
11.48
7.12
Logistic
251.187
0.112
91.89
61.98
0.529
14.86
10.02
LogLogisticb
248.839
0.3461
34.78
16.60
0.656
5.63
2.68
LogProbit0
252.244
0.0655
133.53
78.18
0.016
21.60
12.64
Multistage-Cancer
(1 degree)
250.551
0.1527
70.99
44.00
0.605
11.48
7.12
Multistage-Cancer
(2 degree)
250.551
0.1527
70.99
44.00
0.605
11.48
7.12
Multistage-Cancer
(3 degree)
250.551
0.1527
70.99
44.00
0.605
11.48
7.12
Probit
251.326
0.1048
97.01
67.36
0.518
15.69
10.90
Weibull
250.551
0.1527
70.99
44.00
0.605
11.48
7.12
Quantal-Linear
250.551
0.1527
70.99
44.00
0.605
11.48
7.12
Dichotomous-Hill
250.747
NCd
11.60
1.63
-1.25xl0-5
1.88
0.26
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bBest-fitting model.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
D-42
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
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
0.9
0.8
0.7
0.6
0.5
0.4
0.3
BMDL BMD
0
100
200
300
400
500
600
700
dose
07:30 10/26 2009
Source: Kano et al. (2009).
Figure D-14. LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in male BDFi mice.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2 00 9_mmouse_hepato_adcar_Lnl-BMR10-
Restrict.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2 00 9_mmouse_hepato_adcar_Lnl-BMR10-
Restrict.pit
Thu Nov 12 09:09:36 2009
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/[1+EXP(- intercept-slope*Log(dose))]
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
D-43
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
Default Initial Parameter Values
background =	0.46
intercept =	-5.58909
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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
Variable
background
intercept
slope
Estimate
0 .507468
-5 . 74623
1
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	Log(likelihood)	# Param's	Deviance Test d.f. P-value
Full model	-121.373	4
Fitted model	-122.419	2	2.09225 2	0.3513
Reduced model	-128.859	1	14.9718 3	0.001841
AIC:	248.839
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.5075 25.373 23.000	50	-0.671
49.0000 0.5741 28.707 31.000	50	0.656
191.0000 0.6941 34.706 37.000	50	0.704
677.0000 0.8443 42.214 40.000	50	-0.863
Chi*2 = 2.12	d.f. = 2	P-value = 0.3461
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	34 .7787
BMDL =	16.5976
D-44
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.9
0.8
0.7
0.6
0.5
0.4
0.3
BMD
BMDL
0
100
200
300
400
500
600
700
dose
07:30 10/26 2009
Source: Kano et al. (2009).
Figure D-15. Multistage BMD model (1 degree) for the combined incidence of
hepatic adenomas and carcinomas in male BDFi mice.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_mmouse_hepato_adcar_Msc-BMR10-lpoly. (d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2 00 9_mmouse_hepato_adcar_Msc-BMR10-lpoly.pit
Mon Oct 26 08:30:50 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
D-45
DRAFT - DO NOT CITE OR QUOTE

-------
Background =	0.5 73 75 6
Beta (1) = 0.00123152
D-46	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
Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(l)
Background	1	-0.58
Beta (1)	-0.58	1
Parameter Estimates
Variable
Background
Beta (1)
Estimate
0 . 545889
0.00148414
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
Log(1i ke1i hood)
-121.373
-123.275
-128.859
# Param's
4
2
1
Deviance Test d.f.
3.80413
14 .9718
P-value
0 .1493
0 . 001841
AIC :
250.551
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 . 0000
0 . 5459
27.294
23.000
50
-1.220
49 . 0000
0 .5777
28.887
31.000
50
0 . 605
191.0000
0 . 6580
32.899
37 . 000
50
1 .223
677.0000
0 .8337
41.687
40 . 000
50
-0.641
Chi *2 = 3.76	d.f. = 2	P-value = 0.1527
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	70.9911
BMDL =	44.0047
BMDU =	150.117
Taken together, (44.0047, 150.117) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00227248
D-47
DRAFT - DO NOT CITE OR QUOTE

-------
D.7. BMD MODELING RESULTS FROM ADDITIONAL CHRONIC BIOASSAYS (NCI,
1978; KOCIBA ET AL., 1974)
1	Data and BMDS modeling results for the additional chronic bioassays (NCI, 1978;
2	Kociba et al., 1974) were evaluated for comparison with the Kano et al. (2009) study. These
3	results are presented in the following sections.
4	The BMDS dose-response modeling estimates and HEDs that resulted are presented in
5	detail in the following sections and a summary is provided in Table D-16.
Table D-16. Summary of BMDS dose-response modeling estimates
associated with liver and nasal tumor incidence data resulting from chronic
oral exposure to 1,4-dioxane in rats and mice
Endpoint
Model
selection
criterion
Model
Type
AIC
P-
value
BMD10
mg/kg-
day
BMDL10
mg/kg-
day
BMD io hed
mg/kg-
day
BMDLio hed
mg/kg-day
Kociba et al., 1974
Male and Female (combined) Sherman Rats

Hepatic
Tumors3
Lowest AIC
Probit
84.3126
0.606
1113.94
920.62
290.78
240.31
Nasal
Cavity
Tumorsb
Lowest AIC
Multistage
(3 degree)
26.4156
0.9999
1717.16
1306.29
448.24
340.99
NCI, 1978
Female Osborne-Mendel Rats

Hepatic
Tumors0
Lowest AIC
LogLogistic
84.2821
0.7333
111.46
72.41
28.75
18.68
Nasal
Cavity
Tumorsb
Lowest AIC
LogLogistic
84.2235
0.2486
155.32
100.08
40.07
25.82
NCI, 1978
Male Osborne-Mendel Rats

Nasal
Cavity
Tumorsb
Lowest AIC
LogLogistic
92.7669
0.7809
56.26
37.26
16.10
10.66
NCI, 1978
Female B6C3Fi Mice

Hepatic
Tumorsd
Lowest
AIC,
Multistage
model
Multistage
(2 degree)
85.3511
1
160.68
67.76
23.12
9.75
NCI, 1978
Male B6C3Fi Mice

Hepatic
Tumorsd
Lowest AIC
Gamma
177.539
0.7571
601.69
243.92
87.98
35.67
incidence of hepatocellular carcinoma,
incidence of nasal squamous cell carcinoma.
Incidence of hepatocellular adenoma,
incidence of hepatocellular adenoma or carcinoma.
D-48
DRAFT - DO NOT CITE OR QUOTE

-------
D.7.1. 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-17. The predicted BMDio hed and BMDLio hed values are also presented in
3	Tables D-18 and D-19 for hepatocellular carcinomas and nasal squamous cell carcinomas,
4	respectively.
Table D-17. Incidence of hepatocellular carcinoma and nasal squamous cell
carcinoma in male and female Sherman rats (combined) (Kociba et al., 1974)
treated with 1,4-dioxane in the drinking water for 2 years
Animal Dose (mg/kg-day)
(average of male and female dose)
Incidence of hepatocellular
carcinoma3
Incidence of nasal
squamous cell carcinoma3
0
1/106b
0/106°
14
0/110
0/110
121
1/106
0/106
1307
10/66d
3/66d
aRats surviving until 12 months on study.
hp < 0.001; positive dose-related trend (Cochran-Armitage test).
cp < 0.01; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: Kociba et al. (1974).
D-49
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-18. BMDS dose-response modeling results for the incidence of
hepatocellular carcinoma in male and female Sherman rats (combined)
(Kociba et al., 1974) exposed to 1,4-dioxane in the drinking water for 2 years
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
86.2403
0.3105
985.13
628.48
-0.005
257.15
164.05
Logistic
84.3292
0.6086
1148.65
980.95
-0.004
299.84
256.06
LogLogistic
86.2422
0.3103
985.62
611.14
-0.005
257.28
159.53
LogProbitb
84.4246
0.5977
1036.97
760.29
-0.011
270.68
198.46
Multistage-Cancer
(1 degree)
85.1187
0.3838
940.12
583.58
0.279
245.40
152.33
Multistage-Cancer
(2 degree)
86.2868
0.3109
1041.72
628.56
-0.006
271.92
164.07
Multistage-Cancer
(3 degree)
86.2868
0.3109
1041.72
628.56
-0.006
271.92
164.08
Probit0
84.3126
0.606
1113.94
920.62
-0.005
290.78
240.31
Weibull
86.2443
0.3104
998.33
629.93
-0.005
260.60
164.43
Quantal-Linear
85.1187
0.3838
940.12
583.58
0.279
245.40
152.33
Dichotomous-Hill
1503.63
NCd
NCd
NCd
0
0
0
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bSlope restricted > 1.
°Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS..
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
Probit Model with 0.95 Confidence Level
Probit
0.25
0.2
0.15
0.05
BMDI
BMD
0
200
400
600
800
1000
1200
dose
11:54 10/27 2009
Source: Kociba et al. (1974).
Figure D-16. Probit BMD model for the incidence of hepatocellular carcinoma in
male and female Sherman rats exposed to 1,4-dioxane in drinking water.
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-
BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.pit
Tue Oct 27 12:54:14 2009
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 = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Initial (and Specified) Parameter Values
background =	0 Specified
intercept =	-2.62034
slope = 0.0012323
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
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
intercept	-2.55961	0.261184	-3.07152	-2.0477
slope	0.00117105	0.000249508	0.000682022	0.00166008
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-39.3891
-40 .1563
-53 . 5257
# Param's
4
2
1
Deviance Test d.f.
1. 53445
28 .2732
P-value
0.4643
<.0001
AIC :
84 .3126
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.0052 0.555 1.000	106	0.598
14.0000 0.0055 0.604 0.000	110	-0.779
121.0000 0.0078 0.827 1.000	106	0.191
1307.0000 0.1517 10.014 10.000	66	-0.005
Chi*2 = 1.00	d.f. = 2	P-value = 0.6060
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	1113.94
BMDL =	92 0.616
D-52
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.25
0.2
0.15
1
0.05
0
BMDL
BMD
0
200
400
600
800
1000
1200
dose
11:54 10/27 2009
Source: Kociba et al. (1974).
Figure D-17. Multistage BMD model (1 degree) for the incidence of hepatocellular
carcinoma in male and female Sherman rats exposed to 1,4-dioxane in drinking
water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_hepato_car_Msc-
BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_hepato_car_Msc-BMR10-lpoly.pit
Tue Oct 27 12:54:10 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
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 = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-53
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
Default Initial Parameter Values
Background = 0.000925988
Beta (1) = 0.000124518
Asymptotic Correlation Matrix of Parameter Estimates
Background	Beta(l)
Background	1	-0.44
Beta (1)	-0.44	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
Background	0.0038683	*	*	*
Beta (1)	0.000112071	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(1i ke1i hood)
-39.3891
-40.5594
-53.5257
# Param's
4
2
1
Deviance Test d.f.
2 . 34056
28 .2732
P-value
0 .3103
< .0001
AIC :
35.1187
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.0039 0.410 1.000	106	0.923
14.0000 0.0054 0.597 0.000	110	-0.775
121.0000 0.0173 1.832 1.000	106	-0.620
1307.0000 0.1396 9.213 10.000	66	0.279
Chi*2 = 1.92	d.f. = 2	P-value = 0.3838
Benchmark Dose Computation
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
BMDU =
0 .1
Extra risk
0 . 95
940.124
583.576
1685 . 88
Taken together, (583.576, 1685.88) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000171357
D-54
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-19. BMDS dose-response modeling results for the incidence of nasal
squamous cell carcinoma in male and female Sherman rats (combined) (Kociba et
al., 1974) exposed to 1,4-dioxane in the drinking water for 2 years
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
7.2"
BMD10 hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
28.4078
1
1572.09
1305.86
0
410.37
340.87
Logistic
28.4078
1
1363.46
1306.67
0
355.91
341.09
LogLogistic
28.4078
1
1464.77
1306.06
0
382.35
340.93
LogProbitb
28.4078
1
1644.38
1305.49
0
429.24
340.78
Multistage-Cancer
(1 degree)
27.3521
0.9163
3464.76
1525.36
0.272
904.42
398.17
Multistage-Cancer
(2 degree)
26.4929
0.9977
1980.96
1314.37
0.025
517.10
343.10
Multistage-Cancer
(3 degree)0
26.4156
0.9999
1717.16
1306.29
0.002
448.24
340.99
Probit
28.4078
1
1419.14
1306.44
0
370.44
341.03
Weibull
28.4078
1
1461.48
1306.11
0
381.50
340.94
Quantal-Linear
27.3521
0.9163
3464.76
1525.35
0.272
904.42
398.17
Dichotomous-Hill
30.4078
0.9997
1465.77
1319.19
5.53x 10"7
382.62
344.35
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bSlope restricted > 1.
°Best-fitting model.
D-55
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.14
0.12
0.08
0.06
0.04
0.02
0
BMDL
BME)
0
200
400
600
800
1000
1200
1400
1600
1800
dose
06:25 10/27 2009
Figure D-18. Multistage BMD model (3 degree) for the incidence of nasal squamous
cell carcinoma in male and female Sherman rats exposed to 1,4-dioxane in drinking
water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_nasal_car_Msc-
BMR10-3poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_nasal_car_Msc-BMR10-3poly.pit
Tue Oct 27 07:25:02 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l-beta2*dose*2-
beta3*dose*3)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0D
egree of polynomial = 3
D-56
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
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background =	0
Beta (1) =	0
Beta (2) =	0
Beta(3) = 2.08414e-011
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -Background -Beta(l) -Beta (2)
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Beta (3)
Beta(3)	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf.	Limit Upper Conf. Limit
Background 0 * *	*
Beta(1) o * *	*
Beta(2) 0 * *	*
Beta(3) 2.08088e-011 * *	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model	Log(likelihood)	# Param's Deviance Test d.f.	P-value
Full model	-12.2039	4
Fitted model	-12.2078	1 0.00783284 3	0.9998
Reduced model	-17.5756	1 10.7433 3	0.0132
AIC:	26.4156
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed	Size	Residual
0.0000 0.0000 0.000 0.000	106	0.000
14.0000 0.0000 0.000 0.000	110	-0.003
121.0000 0.0000 0.004 0.000	106	-0.063
1307.0000 0.0454 2.996 3.000	66	0.002
Chi*2 = 0.00	d.f. = 3	P-value = 0.9999
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	1717.16
BMDL =	1306.29
BMDU =	8354.46
Taken together, (1306.29, 8354.46) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 7.65529e-005
D-57
DRAFT - DO NOT CITE OR QUOTE

-------
D.7.2. 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-20. The log-logistic model
3	adequately fit both the male and female rat nasal squamous cell carcinoma data, as well as
4	female hepatocellular adenoma incidence data. For all endpoints and genders evaluated in this
5	section, compared to the multistage models, the log-logistic model had a higherp-value, as well
6	as both a lower AIC and lower BMDL. The results of the BMDS modeling for the entire suite of
7	models are presented in Tables D-21 through D-23.
Table D-20. Incidence of nasal cavity squamous cell carcinoma and
hepatocellular adenoma in Osborne-Mendel rats (NCI, 1978) exposed to
1,4-dioxane in the drinking water
Male rat Animal Dose (mg/kg-day)a

0
240b
530
Nasal cavity squamous cell carcinoma
0/33°
12/26d
16/3 3 d
Female rat Animal Dose (mg/kg-day)a

0
350
640
Nasal cavity squamous cell carcinoma
0/34°
10/30d
8/29d
Hepatocellular adenoma
0/31°
10/30d
1 l/29d
aTumor 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.
cp < 0.001; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: NCI (1978).
D-58
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-21. BMDS dose-response modeling results for the incidence of
hepatocellular adenoma in female Osborne-Mendel rats (NCI, 1978) exposed
to 1,4-dioxane in the drinking water for 2 years
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
84.6972
0.5908
132.36
94.06
0
34.144
24.26
Logistic
92.477
0.02
284.09
220.46
1.727
73.29
56.87
LogLogisticb
84.2821
0.7333
111.46
72.41
0
28.75
18.68
LogProbit
85.957
0.3076
209.47
160.66
1.133
54.04
41.45
Multistage-Cancer
(1 degree)
84.6972
0.5908
132.36
94.06
0
34.14
24.26
Multistage-Cancer
(2 degree)
84.6972
0.5908
132.36
94.06
0
34.14
24.26
Probit
91.7318
0.0251
267.02
207.18
1.7
68.88
53.44
Weibull
84.6972
0.5908
132.36
94.06
0
34.14
24.26
Quantal-Linear
84.6972
0.5908
132.36
94.06
0
34.14
24.26
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
hBest-fitting model.
D-59	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
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
0.5
0.4
0.3
0.2
0.1
0
BMDL
BMD
0
100
200
300
400
500
600
dose
06:32 10/27 2009
Source: NCI (1978).
Figure D-19. LogLogistic BMD model for the incidence of hepatocellular adenoma
in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_hepato_ad_Lnl-BMR10-
Restrict.(d)
Gnuplot Plotting File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_hepato_ad_Lnl-
BMRIO-Restrict.pit
Tue Oct 27 07:32:13 2009
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/[1+EXP(- intercept-slope*Log(dose)) ]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background =	0
intercept =	-6.62889
slope =	1
D-60
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
( *** xhe model parameter(s) -background -slope have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
intercept
intercept
1
Parameter Estimates
Variable
background
intercept
slope
Estimate
0
-6.91086
1
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
Log(1i ke1i hood)
-40 .8343
-41.141
-50.4308
# Param's	Deviance Test d.f.
3
1	0.613564 2
1	19.1932 2
P-value
0.7358
< .0001
AIC :
84 .2821
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0.0000
0.0000
0 . 000
0 . 000
31
0 .000
350.0000
0 .2587
8 . 536
10 . 000
33
0 .582
640.0000
0.3895
12.464
11.000
32
-0 . 531
Chi *2 = 0.62	d.f. = 2	P-value = 0.7333
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	111.457
BMDL =	72 .4092
D-61
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.5
0.4
0
BMDL
BMD
0
100
200
300
400
500
600
dose
06:32 10/27 2009
Source: NCI (1978).
Figure D-20. Multistage BMD model (1 degree) for the incidence of hepatocellular
adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_hepato_ad_Msc-BMR10-
lpoly.(d)
Gnuplot Plotting File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_hepato_ad_Msc-
BMR10-lpoly.pit
Tue Oct 27 07:32:16 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-62
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
Default Initial Parameter Values
Background = 0.0385912
Beta (1) = 0.000670869
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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 (1)
Beta(1)	1
Parameter Estimates
Variable
Background
Beta(1)
Estimate
0
0 .00079602
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	Log(likelihood)	# Param's	Deviance Test d.f. P-value
Full model	-40.8343	3
Fitted model	-41.3486	1	1.02868 2	0.5979
Reduced model	-50.4308	1	19.1932 2	<.0001
AIC:	84.6972
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 .0000
0.0000
0 . 000
0 . 000
31
0 . 000
350 . 0000
0 .2432
8 . 024
10 . 000
33
0 . 802
640 . 0000
0 .3992
12.774
11.000
32
-0.640
Chi*2 = 1.05	d.f. = 2	P-value = 0.5908
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	132.359
BMDL =	94 .0591
BMDU =	194.33
Taken together, (94.0591, 194.33 ) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00106316
D-63
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-22. BMDS dose-response modeling results for the incidence of nasal cavity
squamous cell carcinoma in female Osborne-Mendel rats (NCI, 1978) exposed to
1,4-dioxane in the drinking water for 2 years
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
84.7996
0.1795
176.28
122.27
1.466
45.47
31.54
Logistic
92.569
0.0056
351.51
268.75
2.148
90.68
69.33
LogLogisticb
84.2235
0.2486
155.32
100.08
0
40.07
25.82
LogProbit0
87.3162
0.0473
254.73
195.76
1.871
65.71
50.50
Multistage-Cancer
(1 degree)
84.7996
0.1795
176.28
122.27
1.466
45.47
31.54
Multistage-Cancer
(2 degree)
84.7996
0.1795
176.28
122.27
1.466
45.47
31.54
Probit
91.9909
0.0064
328.46
251.31
2.136
84.73
64.83
Weibull
84.7996
0.1795
176.28
122.27
1.466
45.47
31.54
Quantal-Linear
84.7996
0.1795
176.28
122.27
1.466
45.47
31.54
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
hBest-fitting model.
°Slope restricted > 1.
D-64
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
Log-Logistic Model with 0.95 Confidence Level
0.5
Log-Logistic
0.4
~T
BMDI
BMD
0
100
200
300
400
500
600
dose
06:30 10/27 2009
Source: NCI (1978).
Figure D-21. LogLogistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_nasal_car_Lnl-BMR10-
Restrict.(d)
Gnuplot Plotting File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_nasal_car_Lnl-
BMRIO-Restrict.pit
Tue Oct 27 07:30:09 2009
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/ [1+EXP(- intercept-slope*Log(dose))]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-65
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
User has chosen the log transformed model
Default Initial Parameter Values
background =	0
intercept =	-6.64005
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -background -slope have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
intercept
intercept	1
Parameter Estimates
Variable
background
intercept
slope
Estimate
0
-7 .24274
1
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
Log(likelihood)	# Param's Deviance Test d.f.
-39.7535	3
-41.1117	1	2.71651	2
-47.9161	1	16.3252	2
P-value
0 .2571
0.0002851
AIC :
84 .2235


Goodness of Fit







Scaled
Dose
Est. Prob.
Expected
Observed
Size
Residual
0.0000
0.0000
0 . 000
0 . 000
34
0 .000
350 . 0000
0 .2002
7 . 008
10 . 000
35
1 .264
640.0000
0 .3140
10.992
8 . 000
35
-1.090
Chi 2 = 2.7E
d.f. = 2
P-value = 0.2486
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	155.324
BMDL =	100.081
D-66
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
Multistage Cancer Model with 0.95 Confidence Level
0.5
Multistage Cancer
Linear extrapolation
0.4
0.3
0.2
0.1
0
BMDL
BMD
0
100
200
300
400
500
600
dose
06:30 10/27 2009
Source: NCI (1978).
Figure D-22. Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_nasal_car_Msc-BMR10-
lpoly.(d)
Gnuplot Plotting File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_nasal_car_Msc-
BMR10-lpoly.pit
Tue Oct 27 07:30:12 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-67
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
Default Initial Parameter Values
Background = 0.0569154
Beta (1) =	0.00042443
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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(1)
Beta (1)
Parameter Estimates
Variable
Background
Beta(1)
Estimate
0
0 .000597685
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
Log(likelihood)	# Param's Deviance Test d.f.
-39.7535	3
-41.3998	1	3.29259 2
-47.9161	1	16.3252 2
P-value
0 .1928
0.0002851
AIC:	84.7996
Goodness of Fit
Dose	Est._Prob. Expected Observed	Size
0 .0000
350 . 0000
640 . 0000
Chi*2 = 3.44
0.0000
0 .1888
0.3179
d.f. = 2
0.000	0.000	34
6.607 10.000	35
11.125	8.000	35
P-value = 0.1795
Scaled
Residual
0 . 000
1.466
-1.134
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	176.281
BMDL =	122.274
BMDU =	271.474
Taken together, (122.274, 271.474) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000817837
D-68
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-23. BMDS dose-response modeling results for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel rats (NCI, 1978) exposed to
1,4-dioxane in the drinking water for 2 years
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10 hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
93.6005
0.5063
73.94
54.724
0
21.17
15.66
Logistic
103.928
0.0061
179.05
139.26
2.024
51.25
39.86
LogLogisticb
92.7669
0.7809
56.26
37.26
0
16.10
10.66
LogProbit0
95.0436
0.2373
123.87
95.82
1.246
35.46
27.43
Multistage-Cancer
(1 degree)
93.6005
0.5063
73.94
54.72
0
21.16
15.66
Multistage-Cancer
(2 degree)
93.6005
0.5063
73.94
54.72
0
21.16
15.66
Probit
103.061
0.0078
168.03
131.61
2.024
48.10
37.67
Weibull
93.6005
0.5063
73.94
54.72
0
21.17
15.66
Quantal-Linear
93.6005
0.5063
73.94
54.72
0
21.17
15.66
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bBest-fitting model.
°Slope restricted > 1.
D-69
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
Log-Logistic Model with 0.95 Confidence Level
~i—¦—¦—¦—¦—¦—¦—¦—¦—¦—i—¦—¦—¦—¦—¦—¦—¦—¦—¦—i—¦—¦—¦—¦—¦—>—>—>—>—r
Log-Logistic 	
100	200	300	400	500
dose
06:27 10/27 2009
Source: NCI (1978).
Figure D-23. LogLogistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water.
BMDL BMD
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_mrat_nasal_car_Lnl-BMR10-
Restrict.(d)
Gnuplot Plotting File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_mrat_nasal_car_Lnl-
BMRIO-Restrict.pit
Tue Oct 27 07:27:57 2009
BMDS Model Run
The form of the probability function is:
P [response] = background+(1-background)/[1+EXP(- intercept-slope*Log(dose))]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
D-70
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
Default Initial Parameter Values
background =	0
intercept =	-6.08408
slope =	1
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -background -slope have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
intercept
intercept	1
Parameter Estimates
95.0% Wald Confidence Interval
Variable	Estimate	Std. Err.	Lower Conf. Limit Upper Conf. Limit
background	0	*	*	*
intercept	-6.2272	*	*	*
slope	1	*	*	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(likelihood)	# Param's	Deviance	Test d.f.	P-value
-45.139	3
-45.3835	1	0.488858	2	0.7832
-59.2953	1	28.3126	2	<.0001
AIC :
92 .7669
Dose
Est. Prob.
Goodness of Fit
Expected
Observed
Size
Scaled
Residual
0.0000
0.0000
0 . 000
0 . 000
33
0 .000
240.0000
0.3216
10.612
12.000
33
0 .517
530.0000
0.5114
17.388
16.000
34
-0.476
Chi *2 = 0.49	d.f. = 2	P-value = 0.7809
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	56 .2596
BMDL =	3 7.256
D-71
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
Multistage Cancer Model with 0.95 Confidence Level
-i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—r-
Multistage Cancer 	
Linear extrapolation 	
100	200	300	400	500
dose
06:28 10/27 2009
Source: NCI (1978).
Figure D-24. Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
BMDL BMD
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mrat_nasal_car_Msc-BMR10-
lpoly.(d)
Gnuplot Plotting File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mrat_nasal_car_Msc-
BMR10-lpoly.pit
Tue Oct 27 07:28:00 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)*[1-EXP(-betal*dose*l)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0578996
D-72
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
Beta (1) =
0.00118058
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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 (1)
Beta(1)
Variable
Background
Beta (1)
Parameter Estimates
Estimate
0
0 . 00142499
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
Log(likelihood)	# Param's	Deviance Test d.f.
-45.139	3
-45.8002	1	1.32238 2
-59.2953	1	28.3126 2
P-value
0.5162
<.0001
AIC:	93.6005
Goodness of Fit
Dose	Est._Prob. Expected Observed	Size
Scaled
Residual
0.0000
240 . 0000
530.0000
Chi*2 = 1.3 6
0.0000
0.2896
0.5301
d.f. = 2
0.000	0.000	33
9.558 12.000	33
18.024 16.000	34
P-value = 0.5063
-0.000
0 . 937
-0.695
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	73 . 9379
BMDL =	54 . 7238
BMDU =	103 . 07
Taken together, (54.7238, 103.07 ) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00182736
D-73
DRAFT - DO NOT CITE OR QUOTE

-------
D.7.3. 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. The 2-degree polynomial model (betas restricted > 0)
3	was the lowest degree polynomial that provided an adequate fit to the female mouse data
4	(Figure D-25), while the gamma model provided the best fit to the male mouse data
5	(Figure D-26). The results of the BMDS modeling for the entire suite of models are
6	presented in Tables D-25 and D-26 for the female and male data, respectively.
Table D-24. Incidence of hepatocellular adenoma or carcinoma in male and
female B6C3Fi mice (NCI, 1978) exposed to 1,4-dioxane in drinking water
Male mouse Animal Dose (mg/kg-day)a
Female mouse Animal Dose (mg/kg-day)a
0
720
830
0
380
860
8/4 9b
19/50d
28/47°
0/5 0b
21/48°
35/37°
aTumor incidence values were not adjusted for mortality.
hp < 0.001, positive dose-related trend (Cochran-Armitage test).
cp < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (1978).
D-74
DRAFT - DO NOT CITE OR QUOTE

-------
Table D-25. BMDS dose-response modeling results for the combined
incidence of hepatocellular adenoma or carcinoma in female B6C3Fi mice
(NCI, 1978) exposed to 1,4-dioxane in the drinking water for 2 years
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gamma
85.3511
1
195.69
105.54
0
28.16
15.19
Logistic
89.1965
0.0935
199.63
151.35
0.675
28.72
21.78
LogLogistic
85.3511
1
228.08
151.16
0
32.82
21.75
LogProbitb
85.3511
1
225.8
150.91
0
32.49
21.71
Multistage-Cancer
(1 degree)
89.986
0.0548
49.10
38.80
0
7.06
5.58
Multistage-Cancer
(2 degree)0
85.3511
1
160.68
67.76
0
23.12
9.75
Probit
88.718
0.1165
188.24
141.49
-1.031
27.08
20.36
Weibull
85.3511
1
161.77
89.27
0
23.28
12.84
Quantal-Linear
89.986
0.0548
49.10
38.80
0
7.065
5.58
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bSlope restricted > 1.
"Best-fitting model.
D-75
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
Multistage Cancer Model with 0.95 Confidence Level
1
Multistage Cancer
Linear extrapolation
0
BMDL
BMD
0
100
200
300
400
500
600
700
800
900
dose
06:36 10/27 2009
Source: NCI (1978).
Figure D-25. Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in female B6C3Fi mice exposed to 1,4-dioxane in drinking
water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_fmouse_hepato_adcar_Msc-
BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_fmouse_hepato_adcar_Msc-BMR10-2poly.pit
Tue Oct 27 07:36:26 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background)* [1-EXP(-betal*dose*l-beta2*dose*2)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
D-76
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
Parameter Convergence has been set to: le-008
Default Initial
Background =
Beta (1) =
Beta (2) =
Parameter Values
0
2 . 68591e-005
3.91383e-006
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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(1)
Beta(2)
Beta (1)
1
-0 . 92
Beta(2)
-0 . 92
1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf.	Limit Upper Conf. Limit
Background 0 * *	*
Beta(1) 2.686e-005 * *	*
Beta (2) 3.91382e-006 * *	*
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(likelihood)	# Param's Deviance Test d.f. P-value
-40.6756	3
-40.6756	2 3.20014e-010	1	]
-91.606	1 101.861	2	<.0001
AIC:
35.3511


Goodness of Fit







Scaled
Dose
Est. Prob.
Expected
Observed
Size
Residual
0 .0000
0.0000
0 . 000
0 . 000
50
0 . 000
380 . 0000
0.4375
21.000
21.000
48
0 . 000
860 . 0000
0.9459
35.000
35.000
37
0 . 000
Chi 2 = 0.00
d.f. = 1
P-value = 1.0000
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	160.678
BMDL =	67.7635
BMDU =	18 6.58 7
Taken together, (67.7635, 186.587) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00147572
D-77	DRAFT - DO NOT CITE OR QUOTE

-------
Table D-26. BMDS dose-response modeling results for the combined
incidence of hepatocellular adenoma or carcinoma in male B6C3Fi mice
(NCI, 1978) exposed to 1,4-dioxane in drinking water
Model
AIC
/j-value
BMD10
mg/kg-day
BMDL10
mg/kg-day
x23
BMD10 hed
mg/kg-day
BMDL10 hed
mg/kg-day
Gammab
177.539
0.7571
601.69
243.92
-0.233
87.98
35.67
Logistic
179.9
0.1189
252.66
207.15
0.214
36.94
30.29
LogLogistic
179.443
NC°
622.39
283.04
0
91.01
41.39
LogProbitd
179.443
NC°
631.51
305.44
0
92.34
44.66
Multistage-Cancer
(1 degree)
180.618
0.0762
164.29
117.37
0.079
24.02
17.16
Multistage-Cancer
(2 degree)
179.483
0.1554
354.41
126.24
0.124
51.82
18.46
Probit
179.984
0.1128
239.93
196.90
0.191
35.08
28.79
Weibull
179.443
NC°
608.81
249.71
0
89.02
36.51
Quantal-Linear
180.618
0.0762
164.29
117.37
0.079
24.02
17.16
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are
undesirable.
bBest-fitting model.
°Value unable to be calculated (NC: not calculated) by BMDS.
dSlope restricted > 1.
D-78
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
Gamma Multi-Hit Model with 0.95 Confidence Level
Gamma Multi-Hit
0.7
0.6
0.5
0.4
0.3
0.2
0.1
BMDI
BMD
0
100
200
300
400
500
600
700
800
dose
06:34 10/27 2009
Source: NCI (1978).
Figure D-26. Gamma BMD model for the incidence of hepatocellular adenoma or
carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking water.
Gamma Model. (Version: 2.13; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-
BMRIO-Restrict.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-BMR10-Restrict.pit
Tue Oct 27 07:34:35 2009
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 = Effect
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial	(and Specified) Parameter Values
Background = 0.17
Slope =	0.000671886
Power = 1.3
D-79
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
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe model parameter(s) -Power have been estimated at a boundary point,
been specified by the user, and do not appear in the correlation matrix)
or have
Background
Slope
Background
1
-0 . 52
Slope
-0 . 52
1
Parameter Estimates
Variable
Background
Slope
Power
Estimate
0.160326
0.0213093
18
Std. Err.
0.0510618
0.000971596
NA
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
0.060247	0.260405
0.019405	0.0232136
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
Log(likelihood)	# Param's	Deviance Test d.f.	P-value
-86.7213	3
-86.7693	2	0.096042 1	0.7566
-96.715	1	19.9875 2	<.0001
AIC :
177.539


Goodness of Fit







Scaled
Dose
Est. Prob.
Expected
Observed
Size
Residual
0.0000
0.1603
7 . 856
8 . 000
49
0 .056
720 . 0000
0.3961
19.806
19 . 000
50
-0 .233
830.0000
0.5817
27.339
28 . 000
47
0 .196
Chi *2 = 0.10 d.f.	= 1
Benchmark Dose Computation
Specified effect =	0.1
Risk Type =	Extra risk
Confidence level =	0.95
BMD =	601.692
BMDL =	243.917
P-value = 0.7571
D-80
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
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
0.7
0.6
,,
BMDL
BMD
0
100
200
300
400
500
600
700
800
dose
06:34 10/27 2009
Source: NCI (1978).
Figure D-27. Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking
water
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File: L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-
BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-BMR10-2poly.pit
Tue Oct 27 07:34:42 2009
BMDS Model Run
The form of the probability function is: P [response] = background + (1-background)* [1-
EXP(-betal*dose^l-beta2*dose^2)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 2 50
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
D-81
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
Background =	0.131156
Beta (1) =	0
Beta (2) =	9.44437e-007
Asymptotic Correlation Matrix of Parameter Estimates
( *** xhe 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)
1	-0.72
-0.72	1
Background
Beta(2)
Parameter Estimates
Variable
Background
Beta (1)
Beta (2)
Estimate
0 .1568
0
8.38821e-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
Log(1i ke1i hood)
-86 . 7213
-87.7413
-96.715
# Param's
3
2
1
Deviance Test d.f.
2 . 04001
19.9875
P-value
0.1532
<.0001
AIC :
179 .483
Goodness of Fit
Scaled
Dose	Est._Prob. Expected Observed	Size	Residual
0 . 0000
0 .1568
7 . 683
8 . 000
49
0 .124
720.0000
0 .4541
22 . 707
19 . 000
50
-1.053
830.0000
0 . 5269
24 . 764
28.000
47
0 . 946
Chi *2 = 2.02	d.f. = 1	P-value = 0.1554
Benchmark Dose Computation
Specified effect =	0.1
Risk Type	=	Extra risk
Confidence level =	0.95
BMD =	354.409
BMDL =	12 6.241
BMDU =	447.476
Taken together, (126.241, 447.476) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000792138
D-82
DRAFT - DO NOT CITE OR QUOTE

-------