EPA/635/R-11/003F
www.epa.gov/iris
United
Environmental Protection
TOXICOLOGICAL REVIEW
OF
1,4-DIOXANE
(WITH INHALATION UPDATE)
(CAS No. 123-91-1)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
September 2013
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
-------
CONTENTS: TOXICOLOGICAL REVIEW OF
1,4-DIOXANE (CAS NO. 123-91-1)
LIST OF ABBREVIATIONS AND ACRONYMS
FOREWORD
XIII
xv
AUTHORS, CONTRIBUTORS, AND REVIEWERS
1. INTRODUCTION
XVI
CHEMICAL AND PHYSICAL INFORMATION
Figure 2-1.
Table 2-1
TOXICOKINETICS
1,4-Dioxane chemical structure.
Physical properties and chemical identity of 1,4-dioxane_
HAZARD IDENTIFICATION
4.1. Studies in Humans- Epidemiology, Case Reports, Clinical Controls
4.1.1. Thiess et al.
4.1.2. Buffleretal.
Table 4-1
Table 4-2
Table 4-3
Table 4-4
Table 4-5
Table 4-6
Table 4-7
Table 4-8
Table 4-9
Table 4-10
Table 4-11
Incidence of histopathological lesions in Crj:BDF1 mice exposed to
1,4-dioxane in drinking water for 13 weeks
Incidence of nasal cavity sguamous cell carcinoma and liver
hepatocellular adenoma in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water
Incidence of hepatocellular adenoma or carcinoma in B6C3F-: mice
exposed to 1,4-dioxane in drinking water
Incidence of histopathological lesions in male F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years
Incidence of histopathological lesions in female F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years
3.1. Absorption
3.2. Distribution
3.3. Metabolism
Figure 3-1. Suggested metabolic pathways of 1,4-dioxane in the rat.
Figure 3-2. Plasma 1,4-dioxane levels in rats following i.v. doses of
3-5,600 mg/kg
3.4. Elimination
3.5. Physiologically Based Pharmacokinetic Models
Figure 3-3. General PBPK model structure.
3.5.1. Available Pharmacokinetic Data
3.5.2. Published PBPK Models for 1,4-Dioxane
3.5.3. Implementation of Published PBPK Models for 1 ,4-Dioxane
3.6. Rat Nasal Exposure via Drinking Water
7
8
9
10
11
12
13
14
14
16
20
23
4.2. Subchronic and Chronic Studies and Cancer Bioassays in Animals - Oral and Inhalation
4.2.1. Oral Toxicity
Incidence of histopathological lesions in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 13 weeks
.24
24
25
26
"27
.28
.31
33
Number of incipient liver tumors and hepatomas in male
Sprague-Dawley rats exposed to 1,4-dioxane in drinking water for
13 months 36
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 38
Incidence of nonneoplastic lesions in Osborne-Mendel rats exposed
to 1,4-dioxane in drinking water 40
41
42
46
47
Incidence of nasal cavity, peritoneum, and mammary gland tumors in
F344/DuCrj rats exposed to 1,4-dioxane in drinking water for 2 years 49
Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane
in drinking water for 2 years 49
111
-------
Table 4-12 Incidence of histopathological lesions in male Crj:BDF1 mice exposed
to 1,4-dioxane in drinking water for 2 years 51
Table 4-13 Incidence of histopathological lesions in female Crj:BDF1 mice
exposed to 1,4-dioxane in drinking water for 2 years 51
Table 4-14 Incidence of tumors in Crj:BDF1 mice exposed to 1,4-dioxane in
drinking water for 2 years 53
4.2.2. Inhalation Toxicity 53
Table 4-15 Terminal body weights and relative organ weights of F344/DuCrj rats
exposed to 1,4-dioxane vapor by whole-body inhalation for 13 weeks 56
Table 4-16 Hematology and clinical chemistry of F344/DuCrj rats exposed to
1,4-dioxane vapor by whole-body inhalation for 13 weeks 57
Table 4-17 Incidence data of histopathological lesions in F344/DuCrj rats
exposed to 1,4-dioxane vapor by whole-body inhalation for 13 weeks 58
Table 4-18 Terminal body and relative organ weights of F344/DuCrj male rats
exposed to 1,4-dioxane vapor by whole-body inhalation for 2 years 62
Table 4-19 Hematology and clinical chemistry of F344/DuCrj male rats exposed
to 1,4-dioxane vapor by whole-body inhalation for 2 years 62
Table 4-20 Incidence of pre-and nonneoplastic lesions in male F344/DuCrj rats
exposed to 1,4-dioxane vapor by whole-body inhalation for 2 years 63
Table 4-21 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 2 years 64
4.2.3. Initiation/Promotion Studies 64
4.3. Reproductive/Developmental Studies—Oral and Inhalation 66
4.3.1. Giavini etal. 66
4.4. Other Duration or Endpoint Specific Studies 67
4.4.1. Acute and Short-term Toxicity 67
Table 4-22 Acute and short-term toxicity studies of 1,4-dioxane 68
4.4.2. Neurotoxicity 70
4.5. Mechanistic Data and Other Studies in Support of the Mode of Action 72
4.5.1. Genotoxicity 72
Table 4-23 Genotoxicity studies of 1,4-dioxane; in vitro 75
Table 4-24 Genotoxicity studies of 1,4-dioxane; mammalian in vivo 78
4.5.2. Mechanistic Studies 80
4.6. Synthesis of Major Noncancer Effects 82
4.6.1. Oral 83
Table 4-25 Oral toxicity studies (noncancer effects) for 1,4-dioxane 84
4.6.2. Inhalation 86
Table 4-26 Inhalation toxicity studies (noncancer effects) for 1,4-dioxane 88
4.7. Evaluation of Carcinogenicity 90
4.7.1. Summary of Overall Weight of Evidence 90
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence 91
4.7.3. Mode of Action Information 93
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. 94
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. 96
Table 4-27 Temporal sequence and dose-response relationship for possible key
events and liver tumors in rats and mice 98
Table 4-28 Temporal sequence and dose-response relationship for possible key
events and nasal tumors in rats and mice 100
4.8. Susceptible Populations and Life Stages 105
5. DOSE-RESPONSE ASSESSMENTS 107
5.1. Oral Reference Dose (RfD)
5.1.1. Choice of Principal Studies and Critical Effect with Rationale and Justification
5.1.2. Methods of Analysis— Including Models (PBPK, BMD, etc.)
107
107
108
Table 5-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years 109
Table 5-2 BMD and BMDL values derived from BMD modeling of the incidence
of cortical tubule degeneration in male and female Osborne-Mendel
rats exposed to 1,4-dioxane in drinking water for 2 years 109
5.1.3. RfD Derivation - Including Application of Uncertainty Factors (UFs) 110
5.1.4. RfD Comparison Information 111
-------
Figure 5-1. Potential points of departure (POD) based on liver toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane. 112
Figure 5-2. Potential points of departure (POD) based on kidney toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane. 113
Figure 5-3. Potential points of departure (POD) based on nasal inflammation with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane. 114
Figure 5-4. Potential points of departure (POD) based on organ-specific toxicity
endpoints with corresponding applied uncertainty factors and derived
candidate RfDs following chronic oral exposure to 1,4-dioxane. 115
5.1.5. Previous RfD Assessment 115
5.2. Inhalation Reference Concentration (RfC) 115
5.2.1. Choice of Principal Study and Candidate Critical Effect(s) with Rationale and
Justification 115
Table 5-3 Incidences of nonneoplastic lesions resulting from chronic exposure
(ppm) to 1,4-dioxane considered for identification of a critical effect. 118
5.2.2. Methods of Analysis 119
5.2.3. Exposure Duration and Dosimetric Adjustments 119
Table 5-4 Duration adjusted POD estimates for BMDLs (from best fitting BMDS
models) or NOAELs/LOAELs from chronic exposure to 1,4-dioxane 120
5.2.4. RfC Derivation- Including Application of Uncertainty Factors (UFs) 123
5.2.5. RfC Comparison Information 124
Figure 5-5. Potential points of departure (POD) for candidate endpoints with
corresponding applied uncertainty factors and derived candidate RfCs
following chronic inhalation exposure of F344 male rats to
1,4-dioxane. 124
5.2.6. Previous RfC Assessment 124
5.3. Uncertainties in the Oral Reference Dose and Inhalation Reference Concentration 125
5.4. Cancer Assessment 126
5.4.1. Choice of Study/Data -with Rationale and Justification 126
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) 127
Table 5-6 Incidence of liver, nasal cavity, kidney, peritoneal, and mammary
gland, Zymbal gland, and subcutis tumors in rats exposed to
1,4-dioxane via inhalation for 2 years. 129
5.4.2. Dose-Response Data 130
Table 5-7 Incidence of hepatocellular adenomas or carcinomas in rats and mice
exposed to 1,4-dioxane in drinking water for 2 years 130
Table 5-8 Incidence of tumors in F344 male rats exposed to 1,4-dioxane via
inhalation for 104 weeks (6 hours/day, 5 days/week) 131
5.4.3. Dose Adjustments and Extrapolation Method(s) 132
Table 5-9 Calculated HEDs for the tumor incidence data used for
dose-response modeling 133
5.4.4. Oral Slope Factor and Inhalation Unit Risk 136
Table 5-10 BMD HED and BMDI_HED values from best-fit 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 137
Table 5-11 Dose-response modeling summary results for male rat tumors
associated with inhalation exposure to 1,4-dioxane for 2 years 139
5.4.5. Previous Cancer Assessment 140
5.5. Uncertainties in Cancer Risk Values 140
5.5.1. Sources of Uncertainty 141
Table 5-12 Summary of uncertainty in the 1,4-dioxane cancer risk estimation 145
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE 146
6.1. Human Hazard Potential 146
6.2. Dose Response 147
6.2.1. Noncancer/Oral 147
6.2.2. Noncancer/lnhalation 148
6.2.3. Cancer 148
-------
REFERENCES 153
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC COMMENTS AND
DISPOSITION A-1
A. 1. External Peer Review Panel Comments — Oral Assessment A-1
A. 1.1. General Charge Questions A-2
A. 1.2. Oral reference dose (RfD) for 1,4-dioxane A-5
A. 1.3. Carcinogenicity of 1,4-dioxane and derivation of an oral slope factor A-11
A.2. Public Comments - Oral Assessment A-16
A.2.1. Oral reference dose (RfD) for 1,4-dioxane A-17
A.2.2. Carcinogenicity of 1,4-dioxane A-17
A.2.3. PBPK Modeling A-19
A.2.4. Other Comments A-20
A.3. External Peer Review Panel Comments — Inhalation Update A-20
A.3.1. General Charge Questions A-20
A.3.2. Inhalation reference concentration (RfC) for 1,4-dioxane A-23
A.3.3. Carcinogenicity of 1,4-dioxane and derivation of an inhalation unit risk A-28
A.4. Public Comments - Inhalation Update A-34
A.4.1. Inhalation reference concentration (RfC) for 1,4-dioxane A-34
A.4.2. Carcinogenicity of 1,4-dioxane A-34
A.4.3. PBPK modeling A-38
A.4.4. Other comments A-40
APPENDIX B. EVALUATION OF EXISTING PHARMACOKINETIC MODELS FOR 1,4-DIOXANE B-1
B.1. Background B-1
B.2. Implementation of the Empirical Models in acsIX B-2
B.2.1. Model Descriptions B-2
Figure B-1. Schematic representation of empirical model for 1,4-dioxane in rats. B-2
Figure B-2. Schematic representation of empirical model for 1,4-dioxane in
humans. B-3
B.2.2. Modifications to the Empirical Models B-3
B.2.3. Results B-4
Figure B-3. Output of 1,4-dioxane blood level data from the acsIX implementation
(left) and published (right) empirical rat model simulations of i.v.
administration experiments. B-4
Figure B-4. Output of HEAA urine level data from acslXtreme implementation of
the empirical rat model (left) and published (right) data following i.v.
administration experiments. B-5
Figure B-5. acsIX empirical rat model predictions of blood 1,4-dioxane
concentration and total amount of HEAA levels in the urine for a
6-hour, 50-ppm 1,4-dioxane inhalation exposure. B-6
Figure B-6. acsIX predictions of blood 1,4-dioxane levels using the Young et al.
(1978a, b) model compared with data from Kasai et al. (2008). B-7
Figure B-7. Output of 1,4-dioxane and HEAA blood concentrations from the acsIX
implementation of the empirical human model (left) and published
(right) data of a 6-hour, 50-ppm inhalation exposure. B-8
Figure B-8. Observations and acsIX predictions of the cumulative amount of
HEAA in human urine following a 6-hour, 50-ppm inhalation exposure. B-8
B.2.4. Conclusions for Empirical Model Implementation B-9
B.3. Initial Evaluation of the PBPK Models B-9
B.3.1. Initial Recalibration of the Reitz et al. PBPK Model B-10
B.3.2. Flow Rates B-10
Table B-1 Human PBPK model parameter values published in literature and
values used by EPA in this assessment for 1,4-dioxane B-11
B.3.3. Partition Coefficients B-12
B.3.4. Calibration Method B-13
B.3.5. Results B-13
Table B-2 PBPK metabolic and elimination parameter values resulting from
recalibration of the human model using alternative values for
physiological flow rates3 and tissue:air partition coefficients B-14
Figure B-9. Human predicted and observed blood 1,4-dioxane concentrations
(left) and urinary HEAA levels (right) following a 6-hour, 50 ppm
1,4-dioxane exposure and recalibration of the PBPK model with
VI
-------
tissue:air partition coefficient values from Leung and Paustenbach
(1990). B-14
Figure B-10. Human predicted and observed blood 1,4-dioxane concentrations
(left) and urinary HEAA levels (right) following a 6-hour, 50 ppm
1,4-dioxane exposure and recalibration of the PBPK model with
tissue:air partition coefficient values from Sweeney et al. (2008) B-15
Figure B-11. Human predicted and observed blood 1,4-dioxane concentrations
(left) and urinary HEAA levels (right) following a 6-hour, 50 ppm
1,4-dioxane exposure, using EPA biologically plausible parameters. B-16
B.3.6. Conclusions for PBPK Model Implementation B-16
B.3.7. Sensitivity Analysis B-17
B.3.8. Method B-17
B.3.9. Results B-18
Figure B-12. 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-18
B.4. PBPK Model Exercises Using Biologically Plausible Parameter Boundaries B-18
B.4.1. Observations Regarding the Volume of Distribution B-19
B.4.2. Defining Boundaries for Parameter Values B-19
B.4.3. Results B-19
Figure B-13. Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients from Leung & Paustenbach
(1990) with empirical model predictions and experimental
observations for human blood 1,4-dioxane concentrations (left) and
amount of HEAA in human urine (right) from a 6-hour, 50-ppm
inhalation exposure. B-20
Figure B-14. Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients from Sweeney et al (2008)
with empirical model predictions and experimental observations for
human blood 1,4-dioxane concentrations (left) and amount of HEAA
in human urine (right) from a 6-hour, 50-ppm inhalation exposure. B-21
Table B-3 PBPK metabolic and elimination parameter values resulting from
recalibration of the human model using biologically plausible values
for physiological flow rates3 and selected upper and lower boundary
values fortissue:air partition coefficients B-21
B.4.4. Alternative Model Parameterization B-22
Figure B-15. Predictions of human blood 1,4-dioxane concentration following
calibration of a first-order metabolism rate constant, kLc (1.2 hr"1), to
the experimental data. B-22
Figure B-16. Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, ki_c(0.1 hr"1), to only the
exposure phase of the experimental data. B-23
Figure B-17. Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a first-order metabolism rate constant ( ki_c = 0.28 hr"1)
and slowly perfused tissue:air partition coefficient (PSA = 10) to the
experimental data. B-24
B.5. Conclusions B-24
B.6. acsIX Model Code B-25
APPENDIX C. DETAILS OF BMD ANALYSIS FOR ORAL RFD FOR 1,4-DIOXANE C-1
C.1. Cortical Tubule Degeneration C-1
Table C-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years C-1
Table C-2 Goodness-of-fit statistics and BMD-io and BMDL-io 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
Figure C-1. BMD Log-probit model of cortical tubule degeneration incidence data
for male rats exposed to 1,4-dioxane in drinking water for 2 years. 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. C-5
APPENDIX D. DETAILS OF BMD ANALYSIS FOR ORAL CSF FOR 1,4-DIOXANE D-1
D. 1. General Issues and Approaches to BMDS Modeling D-2
VII
-------
D.1.1. Combining Data on Adenomas and Carcinomas D-2
D.1.2. Model Selection Criteria D-3
D.1.3. Summary D-4
Table D-1 Recommended models for rodents exposed to 1,4-dioxane in drinking
water (Kano et al., 2009). D-4
D.2. Female F344 Rats: Hepatic Carcinomas and Adenomas_
Table D-2
Table D-3
Data for hepatic adenomas and carcinomas in female F344 rats
(Kano et al., 2009).
BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats (Kano et al.,
2009).
Figure D-1. Multistage BMD model (2 degree) for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats.
D.3. Male F344 Rats: Hepatic Carcinomas and Adenomas
Table D-4
Table D-5
Figure D-2.
Figure D-3.
Data for hepatic adenomas and carcinomas in male F344 rats (Kano
et al., 2009).
BMDS dose-response modeling results for the combined incidence of
adenomas and carcinomas in livers of male F344 rats (Kano et al.,
2009).
Probit BMD model for the combined incidence of hepatic adenomas
and carcinomas in male F344 rats.
Multistage BMD model (3 degree) for the combined incidence of
hepatic adenomas and carcinomas in male F344 rats.
D.4. F344 Rats: Tumors at Other Sites
Table D-6
Table D-7
Figure D-4.
Table D-8
Figure D-5.
Table D-9
Figure D-6.
Figure D-7.
Table D-10
Figure D-8.
Figure D-9.
Data for significant tumors at other sites in male and female F344 rats
(Kano et al., 2009).
BMDS dose-response modeling results for the incidence of nasal
cavity tumors in female F344 rats3 (Kano et al., 2009).
Multistage BMD model (3 degree) for nasal cavity tumors in female
F344 rats.
BMDS dose-response modeling results for the incidence of nasal
cavity tumors in male F344 rats3 (Kano et al., 2009).
Multistage BMD model (3 degree) for nasal cavity tumors in male
F344 rats.
BMDS dose-response modeling results for the incidence of mammary
gland adenomas in female F344 rats (Kano et al., 2009).
Log-Logistic BMD model for mammary gland adenomas in female
F344 rats.
Multistage BMD model (1 degree) for mammary gland adenomas in
female F344 rats.
BMDS dose-response modeling results for the incidence of peritoneal
mesotheliomas in male F344 rats (Kano et al., 2009).
Probit BMD model for peritoneal mesotheliomas in male F344 rats.
Multistage BMD (2 degree) model for peritoneal mesotheliomas in
male F344 rats.
D.5. Female BDF1 Mice: Hepatic Carcinomas and Adenomas
Table D-11 Data for hepatic adenomas and carcinomas in female BDF1 mice
(Kano et al., 2009).
Table D-12
Table D-13
BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in female BDF1 mice (Kano et al.,
2009).
BMDS Log-Logistic dose-response modeling results using BMRs of
10, 30, and 50% for the combined incidence of hepatic adenomas
and carcinomas in female BDF1 mice (Kano et al., 2009).
Figure D-10. Log-Logistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 10%. _
Figure D-11. Log-Logistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 30%. _
Figure D-12. Log-Logistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 50%. _
Figure D-13. Multistage BMD model (1 degree) for the combined incidence of
hepatic adenomas and carcinomas in female BDF1 mice.
D.6. Male BDF1 Mice: Hepatic Carcinomas and Adenomas
Table D-14 Data for hepatic adenomas and carcinomas in male BDF1 mice
(Kano et al., 2009).
D-5
.D-6
D-7
.D-9
D-9
.D-10
.D-11
D-13
^D-15
.D-15
.D-16
.D-17
.D-19
. D-20
. D-22
. D-23
. D-25
D-27
^D-28
D-30
.D-32
D-32
D-33
. D-33
. D-34
. D-36
. D-38
D-40
^D-42
D-42
Vlll
-------
Table D-15 BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.,
2009).
Figure D-14. Log-Logistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice.
Figure D-15. Multistage BMD model (1 degree) for the combined incidence of
hepatic adenomas and carcinomas in male BDF1 mice.
D.7. BMD Modeling Results from Additional Chronic Bioassays
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.7.1.
Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma (Kociba et al.,
1974)
Table D-17
Table D-18
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.
Table D-19
D.7.2.
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.
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.
Nasal Cavity Squamous Cell Carcinoma and Liver Hepatocellular Adenoma in
Osborne-Mendel Rats (NCI, 1978)
Table D-20 Incidence of nasal cavity squamous cell carcinoma and hepatocellular
Table D-21
adenoma in Osborne-Mendel rats (NCI, 1978) exposed to
1,4-dioxane in drinking water._
BMDS dose-response modeling results for the incidence of
hepatocellular adenoma in female Osborne-Mendel rats (NCI, 1978)
exposed to 1,4-dioxane in drinking water for 2 years.
Table D-22
Figure D-21. Log-Logistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in female Osborne-Mendel rats exposed to
1,4-dioxane in drinking water.
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._
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.
Figure D-23. Log-Logistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane
in drinking water.
Figure D-24. Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel ratsexposed to
1,4-dioxane in drinking water.
D.7.3. Hepatocellular Adenoma or Carcinoma in B6C3F-I Mice (NCI, 1978)
D-43
D-44
D-46
'D-48
D-48
D-49
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
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-49
D-50
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-52
D-54
D-55
D-57
D-57
D-58
Figure D-19. Log-Logistic BMD model for the incidence of hepatocellular adenoma
in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water. D-59
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-61
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-63
D-64
D-66
D-68
D-69
D-71
D-73
IX
-------
Table D-24
Table D-25
Incidence of hepatocellular adenoma or carcinoma in male and
female B6C3F-: mice (NCI, 1978) exposed to 1,4-dioxane in drinking
water. D-73
BMDS dose-response modeling results for the combined incidence of
hepatocellular adenoma or carcinoma in female B6C3F-: mice (NCI,
1978) exposed to 1,4-dioxane in the drinking water for 2 years. D-73
Figure D-25. Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in female B6C3F-: mice exposed to
1,4-dioxane in drinking water._
Table D-26
Figure D-26.
Figure D-27.
BMDS dose-response modeling results for the combined incidence of
hepatocellular adenoma or carcinoma in male B6C3F-: mice (NCI,
1978) exposed to 1,4-dioxane in drinking water.
Table F-2
fit to incidence data for centrilobular necrosis of the liver in male
F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai et al., 2009) _
Figure F-1. BMD Dichotomous Hill model of centrilobular necrosis incidence data
for male rats exposed to 1,4-dioxane vapors for 2 years.
F.2. Squamous Cell Metaplasia
Table F-3 Incidence of squamous cell metaplasia of the respiratory epithelium in
male F344/DuCrj rats exposed to 1,4-dioxane via inhalation for
2 years
Table F-4 Goodness-of-fit statistics and BMD-io and BMDL-io values from models
fit to incidence data for squamous cell metaplasia of the respiratory
epithelium in male F344/DuCrj rats exposed to 1,4-dioxane vapors
(Kasai et al., 2009)
Figure F-2. BMD Log-probit model of squamous cell metaplasia of the respiratory
epithelium incidence data for male rats exposed to 1,4-dioxane
vapors for 2 years.
F.3. Squamous Cell Hyperplasia
D-74
D-76
Gamma BMD model for the incidence of hepatocellular adenoma or
carcinoma in male B6C3F-: mice exposed to 1,4-dioxane in drinking
water. D-77
Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in male B6C3F-: mice exposed to 1,4-dioxane
in drinking water. D-79
APPENDIX E. COMPARISON OF SEVERAL DATA REPORTS FOR THE JBRC 2-YEAR
1,4-DIOXANE DRINKING WATER STUDY
Table E-1
Table E-2
Table E-3
Table E-4
Table E-5
Table E-6
Table E-7
Table E-8
APPENDIX F. DETAILS OF
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinkinq water study in male F344 rats
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinkinq water study in female F344 rats
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinkinq water study in male F344 rats
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinkinq water study in female F344 rats
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinkinq water study in male Cri:BDF1 mice
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinkinq water study in female Cri:BDF1 mice
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinkinq water study in male Cri:BDF1 mice
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinkinq water study in female Cri:BDF1 mice
BMD ANALYSIS FOR INHALATION RFC FOR 1,4-DIOXANE
F.1. Centrilobular Necrosis of the Liver
Table F-1
Incidence of centrilobular necrosis of the liver in male F344/DuCrj rats
exposed to 1 ,4-dioxane via inhalation for 2 years
E-1
E-2
E-5
E-8
E-11
E-1 4
E-1 6
E-1 8
E-20
F-1
F-1
F-1
Table F-5 Incidence of squamous cell hyperplasia of the respiratory epithelium
in male F344/DuCrj rats exposed to 1,4-dioxane via inhalation for
2 years
Table F-6 Goodness-of-fit statistics and BMD-io and BMDL-io values from models
fit to incidence data for squamous cell hyperplasia of the respiratory
F-2
F-3
F-5
F-5
F-6
F-7
F-9
F-9
-------
epithelium in male F344/DuCrj rats exposed to 1,4-dioxane vapors
(Kasaietal.,2009).
Figure F-3. BMD Log-probit model of squamous cell hyperplasia of the respiratory
epithelium incidence data for male rats exposed to 1,4-dioxane
vapors for 2 years.
F.4. Respiratory Metaplasia
Table F-7
Table F-8
Table F-9
Figure F-4.
F.5. Atrophy
Table F-10
Table F-11
Figure F-5.
F.6. Hydropic Change_
Table F-12 "
Table F-13
Figure F-6.
F.7. Sclerosis
Incidence of respiratory metaplasia of the olfactory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
Goodness-of-fit statistics and BMD-io and BMDL-io values from models
fit to incidence data for respiratory metaplasia of olfactory epithelium
in male F344/DuCrj rats (Kasai et al., 2009) exposed to 1,4-dioxane
vapors
Goodness-of-fit statistics and BMD-io and BMDL-io values from models
fit to incidence data for respiratory metaplasia of olfactory epithelium
with high dose group dropped in male F344/DuCrj rats (Kasai et al.,
2009) exposed to 1,4-dioxane vapors
BMD Gamma model of respiratory metaplasia of olfactory epithelium
incidence data for male rats exposed to 1,4-dioxane vapors for 2
years.
Incidence of atrophy of the olfactory epithelium in male F344/DuCrj
rats exposed to 1,4-dioxane via inhalation for 2 years
Goodness-of-fit statistics and BMDio and BMDLio values from models
fit to incidence data for atrophy of olfactory epithelium in male
F344/DuCrj rats (Kasai et al., 2009) exposed to 1,4-dioxane vapors
BMD Log-Logistic model of atrophy of olfactory epithelium incidence
data for male rats exposed to 1,4-dioxane vapors for 2 years.
Incidence of hydropic change of the lamina propria in the nasal cavity
of F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
Goodness-of-fit statistics and BMD-|0 and BMDLio values from models
fit to incidence data for hydropic change of the lamina propria in the
nasal cavity of male F344/DuCrj rats exposed to 1,4-dioxane vapors
(Kasaietal.,2009).
BMD Log-logistic model of hydropic change of lamina propria (nasal
cavity) incidence data for male rats exposed to 1,4-dioxane vapors for
2 years.
Table F-14 Incidence of sclerosis of the lamina propria in the nasal cavity of
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
Table F-15 Goodness-of-fit statistics and BMDio and BMDLio values from models
fit to incidence data for sclerosis of the lamina propria in the nasal
cavity of male F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai
et al., 2009)
Figure F-7. BMD Log-logistic model of sclerosis of lamina propria (nasal cavity)
incidence data for male rats exposed to 1,4-dioxane vapors for 2
years.
APPENDIX G. DETAILS OF BMD ANALYSIS FOR INHALATION UNIT RISK FOR 1,4-DIOXANE _
G.1. General Issues and Approaches to BMDS and Multitumor Modeling
G.1.1. Combining Data tumor types
G.1.2. Summary
Table G-1 Summary of BMC-io and BMCLio model results for individual tumor
types and combined tumor analysis for male rats exposed to
1,4-dioxane vapors (Kasai et al., 2009)
G.2. BMDS Model Output for Multistage Cancer Models for Individual Tumor Types
Table G-2 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 2 years
Nasal Squamous Cell Carcinoma
G.2.1.
Table G-3
Figure G-1.
BMDS Multistage cancer dose-response modeling results for the
incidence of nasal squamous cell carcinomas in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai et al., 2009)
Multistage model (First (1°)-degree) for male rat nasal squamous cell
carcinomas.
F-10
F-11
F-13
F-13
F-14
F-15
F-16
F-18
F-18
F-19
F-20
F-22
F-22
F-23
F-24
F-26
F-26
F-27
F-28
G-1
G-1
"G-1
"G-1
G-2
'
G-3
"G-3
G.2.2. Hepatocellular Adenoma and Carcinoma
.G-3
G-4
"G-6
XI
-------
Table G-4 BMDS Multistage cancer dose-response modeling results for the
incidence of either hepatocellular adenoma or carcinoma in male rats
exposed to 1,4-dioxane vapors for 2-years (Kasai etal., 2009) G-6
Figure G-2. Multistage model (First-degree (1°)) for male rat hepatocellular
adenomas and carcinomas. G-7
G.2.3. Renal Cell Carcinoma and Zymbal Gland Adenoma G-9
Table G-5 BMDS Multistage cancer dose-response modeling results for the
incidence of renal cell carcinomas and Zymbal gland adenomas in
male rats exposed to 1,4-dioxane vapors for 2-years (Kasai et al.,
2009) G-9
Figure G-3. Multistage model (Second-degree (2°)) for male rat renal cell
carcinomas and Zymbal gland adenomas. G-10
Figure G-4. Multistage model (Third-degree (3°)) for male rat renal cell
carcinomas. G-12
G.2.4. Peritoneal Mesothelioma G-14
Table G-6 BMDS Multistage cancer dose-response modeling results for the
incidence of peritoneal mesothelioma in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai et al., 2009) G-14
Figure G-5. Multistage model (First-degree (1°)) for male rat peritoneal
mesotheliomas. G-15
G.2.5. Mammary Gland Fibroadenoma G-16
Table G-7 BMDS Multistage cancer dose-response modeling results for the
incidence of mammary gland fibroadenoma in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai et al., 2009) G-17
Figure G-6. Multistage model (First-degree (1°)) for male rat mammary gland
fibroadenoma. G-17
G.2.6. Subcutis Fibroma G-20
Table G-8 BMDS Multistage cancer dose-response modeling results for the
incidence of subcutis fibromas in male rats exposed to 1,4-dioxane
vapors for 2-years (Kasai et al., 2009) G-20
Figure G-7. Multistage model (First-degree (1°)) for male rat subcutis fibroma
(high dose dropped). G-21
G.3. Multitumor Analysis Using BMDS MS_Combo G-23
G.4. Multitumor analysis using Bayesian Methods G-23
APPENDIX H. ASSESSMENTS BY OTHER NATIONAL AND INTERNATIONAL HEALTH
AGENCIES H-1
Table H-1 Health assessments, guideline levels, and regulatory limits by other
national and international agencies H-1
APPENDIX I. DOCUMENTATION OF IMPLEMENTATION OF THE 2011 NATIONAL RESEARCH
COUNCIL RECOMMENDATIONS 1-1
Table 1-1. National Research Council recommendations that EPA is
implementing in the short-term I-2
Table I-2. National Research Council recommendations that the EPA is
generally implementing in the long-term I-6
XII
-------
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
BMC benchmark concentration
BMCL benchmark concentration, lower
95% confidence limit
BMCL10 benchmark concentration, lower
95% confidence limit at 10% extra
risk
BMD benchmark dose
BMD10 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-bromo-2'-deoxyuridine
BUN blood urea nitrogen
BW(s) body weight(s)
CASE computer automated structure
evaluator
CASRN Chemical Abstracts Service
Registry Number
CFD computational fluid dynamic
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
GOT y-glutamyl transpeptidase
GST-P glutathione S-transferase, placental
form
HEAA (3-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
IDLH immediately dangerous to life and
health
i.p. intraperitoneal
i.v. intravenous
IRIS Integrated Risk Information System
JBRC Japan Bioassay Research Center
ke 1 st order elimination rate of
1,4-dioxane
kiKH 1 st order 1,4-dioxane inhalation
rate constant
kLC 1st order, non-saturable metabolism
rate constant for 1,4-dioxane in the
liver
Km Michaelis constant for metabolism
of 1,4-dioxane in the liver
kme 1 st order elimination rate of HEAA
(1,4-dioxane metabolite)
k0c soil organic carbon-water
portioning coefficient
LAP leucine aminopeptidase
LD50 median lethal dose
LDH lactate dehydrogenase
LOAEL lowest-observed-adverse
effect-level
MCH mean corpuscular hemoglobin
MCV mean corpuscular volume
Xlll
-------
MOA mode of action
MRL minimum risk level
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
OSF oral cancer slope factor
PB blood:air partition coefficient
PBPK physiologically based
pharmacokinetic
PC partition coefficient
PCB polychlorinated biphenyl
PCE polychromatic erythrocyte
PEL permissible exposure limit
PFA fat:air partition coefficient
PL A liver: air partition coefficient
POD point of departure
ppm parts per million
PRA rapidly perfused tissue:air partition
coefficient
PSA slowly 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
REL reference exposure level
SCE sister chromatid exchange
SDH sorbitol dehydrogenase
SMR standardized mortality ratio
SRC Syracuse Research Corporation
TLV threshold limit value
TPA 12-O-tetradecanoylphorbol-
1-3-acetate
TWA time-weighted average
UF uncertainty factor
UNEP United Nations Environment
Programme
U.S. United States of America
U.S. EPA U.S. Environmental Protection
Agency
V volts
VAS visual analogue scale
Vd volume of distribution
Vmax maximal rate of metabolism
VmaxC normalized maximal rate of
metabolism of 1,4-dioxane in liver
VOC(s) volatile organic compound(s)
WBC white blood cell
X2 Chi-squared
xiv
-------
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 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).
NOTE: New studies (Kasai et al., 2009; Kasai et al., 2008) regarding the toxicity of 1,4-dioxane
through the inhalation route of exposure became available during the finalization of the 1,4-dioxane oral
assessment that was posted on the IRIS database in 2010 (U.S. EPA. 2010). In this version of the
toxicological review, these studies have been incorporated into the previously posted assessment (U.S.
EPA. 2010). Although the focus of the most recent peer review was on the inhalation toxicity following
exposure to 1,4-dioxane, a few comments were received on the oral assessment and were addressed to
ensure scientific consistency between both routes of exposure. These comments did not impact the final
conclusions of the oral assessment. Also, to minimize changes to the oral portion of the assessment, the
NRC recommendations were not fully implemented (see Appendix I).
xv
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
Assessment Team
Patricia Gillespie, Ph.D. (Chemical Manager, Inhalation)
Eva D. McLanahan, Ph.D. (Chemical Manager, Oral/Inhalation)
Reeder Sams, Ph.D. (Chemical Manager, Oral)
John Stanek, Ph.D.
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
Scientific Support Team
Lyle Burgoon, Ph.D.
J.Allen Davis, MSPH
Jeff S. Gift, Ph.D.
Nagu Keshava, Ph.D.
Allan Marcus, Ph.D.
Connie Meacham, M.S.
Andrew Rooney, Ph.D.
Paul Schlosser, Ph.D.
John Vandenberg, Ph.D.
Jason Lambert, Ph.D.
Karen Hogan, M.S.
Leonid Kopylev, Ph.D.
Susan Rieth, M.S.
Anthony DeAngelo, Ph.D.
Hisham El-Masri, Ph.D.
William Lefew, Ph.D.
Douglas Wolf, Ph.D.
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
U.S. EPA/ORD/NCEA
Cincinnati, OH
U.S. EPA/ORD/NCEA
Washington, DC
U.S. EPA/ORD/NHEERL
Research Triangle Park, NC
Production Team
EllenLorang, M.S.
Deborah Wales
J. Sawyer Lucy
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
U.S. EPA/ORD/NCEA
Student Services Contractor
Research Triangle Park, NC
Contractor Support
Fernando Llados, Ph.D.
Michael Lumpkin, Ph.D.
Mark Odin, Ph.D.
Julie Stickney, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
xvi
-------
Executive Direction
Kenneth Olden, Ph.D., Sc.D., L.H.D. U.S. EPA/ORD/NCEA
Lynn Flowers, Ph.D., DABT Washington, DC
Vincent Cogliano, Ph.D.
Samantha Jones, Ph.D.
Lyle Burgoon, Ph.D. U.S. EPA/ORD/NCEA
Reeder Sams, Ph.D. Research Triangle Park, NC
John Vandenberg, Ph.D
Debra Walsh, M.S.
Oral Assessment Reviewers
The oral assessment was provided for review to scientists in EPA's Program and Region Offices.
Comments were submitted by:
Office of Air Quality and Planning Standards, Research Triangle Park, NC
Office of Pesticide Programs, Washington, DC
Office of Policy, Economics, and Innovation, Washington, DC
Office of Water, Washington, DC
Region 2, New York City, NY
Region 3, Philadelphia, PA
Region 6, Dallas, TX
Region 8, Denver, CO
The oral assessment was provided for review to other federal agencies and Executive Office of the
President. Comments were submitted by:
Department of Defense
National Aeronautics and Space Administration
Office of Management and Budget
The oral assessment was released for public comment in May 2009. A summary and EPA's disposition of
the comments from the public is included in Appendix A. Comments were received from the following:
The Alliance for Environmental Responsibility and
Openness (AERO)
Betty Locey, Ph.D., DABT
Ted Simon, Ph.D., DABT ARCADIS
Lu Yu, Ph.D. Novi, MI
P. Stephen Finn
„ T „ Golder Associates, Inc.
Gregory J. Garvey
TU T, cu T->AD^ Mt. Laurel, NJ
Theresa Repaso-Subang, DABT
Gradient Corporation
Lorenz R. Rhomberg, Ph.D. Cambridge, MA
John E. Bailey, Ph.D. Personal Care Products Council
xvii
-------
The oral assessment was peer reviewed by independent expert scientists external to EPA and a peer-
review meeting was held on August 17, 2009. The external peer-review comments are available on the
IRIS Web site. A summary and EPA's disposition of the comments received from the independent
external peer reviewers is included in Appendix A and is also available on the IRIS Web site
Oittp: //www. epa. gov/iris/).
George V. Alexeeff, Ph.D., DABT California Environmental Protection Agency
Sacramento, CA
Bruce C. Allen, M.S. Bruce Allen Consulting
Chapel Hill, NC
James V. Bruckner, Ph.D. University of Georgia
Athens, GA
Harvey J. Clewell III. Ph.D., DABT The Hamner Institutes for Health Sciences
Research Triangle Park, NC
Lena Ernstgard, Ph.D. Karolinska Institute!
Frederick J. Kaskel, M.D., Ph.D. Children's Hospital at Montefiore
Albert Einstein College of Medicine of Yeshiva University
Kantian Krishnan, PhD., DABT Universite de Montreal
Montreal, Canada
Raghubir P. Sharma, DVM, Ph.D. University of Georgia (retired)
Athens, GA
Inhalation Assessment Reviewers
The assessment with the inhalation update was provided for review to scientists in EPA's Program and
Region Offices. Comments were submitted by:
Office of Policy, Washington, DC
Office of Solid Waste and Emergency Response, Washington, DC
Office of Water, Washington, DC
Region 2, New York City, NY
Region 8, Denver, CO
The assessment with the inhalation update was provided for review to other federal agencies and
Executive Office of the President. Comments were submitted by:
Agency for Toxic Substances Disease Registry, Centers for Disease Control and Prevention,
Department of Health & Human Services
Council on Environmental Quality
Department of Defense
National Aeronautics and Space Administration
National Institute for Occupational Safety and Health
National Toxicology Program, National Institutes for Environmental Health Sciences, National
Institutes of Health, Department of Health & Human Services
Office of Management and Budget
Office of Science and Technology Policy
XVlll
-------
The assessment with the inhalation update was released for public comment on September 9,2011 and
comments were due on November 15, 2011. A summary and EPA's disposition of the comments from the
public is included in Appendix A. Comments were received from the following:
Michael Dourson, Ph.D., DABT, Fellow ATS
Patricia Nance, M.Ed., M.A.
John Reichard, Ph.D.
Mahta Mahdavi, J.D.
Lisa Goldberg
Toxicology Excellence for Risk Assessment
Cincinnati, OH
National Association of Manufacturers
Washington, DC
Aerospace Industries Association
Arlington, VA
The assessment with the inhalation update was peer reviewed by independent expert scientists external to
EPA and a peer-review meeting was held on March 19, 2012. The external peer-review comments are
available on the IRIS Web site. A summary and EPA's disposition of the comments received from the
independent external peer reviewers is included in Appendix A and is also available on the IRIS Web site
(http: //www. epa. gov/iris/).
James V. Bruckner, Ph.D.
Harvey J. Clewell III. Ph.D., DABT
David C. Dorman, DVM, Ph.D. DABVT,
DABT
Ronald L. Melnick, Ph.D.
Frederick J. Miller, Ph.D., Fellow ATS
RaghubirP. Sharma, DVM, Ph.D.
University of Georgia
Athens, GA
The Hamner Institutes for Health Sciences
Research Triangle Park, NC
NCSU-College of Veterinary Medicine
Raleigh, NC
Ron Melnick Consulting, LLC
Chapel Hill, NC
Fred J. Miller & Associates, LLC
Cary, NC
University of Georgia (retired)
Athens, GA
xix
-------
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 other exposure durations, and a carcinogenicity assessment.
The RfD and RfC, if derived, provide quantitative information for use in risk assessments for
health effects known or assumed to be produced through a nonlinear (presumed threshold) mode of
action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with uncertainty spanning
perhaps an order of magnitude) of a daily exposure to the human population (including sensitive
subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. The
inhalation RfC (expressed in units of mg/m3) is analogous to the oral RfD, but provides a continuous
inhalation exposure estimate. The inhalation RfC considers toxic effects for both the respiratory system
(portal-of-entry) and for effects peripheral to the respiratory system (extrarespiratory or systemic effects).
Reference values are generally derived for chronic exposures (up to a lifetime), but may also be derived
for acute (< 24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
lifetime) exposure durations, all of which are derived based on an assumption of continuous exposure
throughout the duration specified. Unless specified otherwise, the RfD and RfC are derived for chronic
exposure duration.
The carcinogenicity assessment provides information on the carcinogenic hazard potential of the
substance in question and quantitative estimates of risk from oral and inhalation exposure may be derived.
The information includes a weight-of-evidence judgment of the likelihood that the agent is a human
carcinogen and the conditions under which the carcinogenic effects may be expressed. Quantitative risk
estimates may be derived from the application of a low-dose extrapolation procedure. If derived, the oral
slope factor is a plausible upper bound on the estimate of risk per mg/kg-day of oral exposure. Similarly,
an inhalation unit risk is a plausible upper bound on the estimate of risk per ug/m3 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). U.S. Environmental Protection Agency (U.S. EPA) Guidelines and Risk Assessment Forum
technical panel reports that may have been used in the development of this assessment include the
following Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S. EPA. 1986c).
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 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. 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. 2006b). A Framework for Assessing Health Risks of Environmental Exposures to
Children (U.S. EPA. 2006a). and Benchmark Dose Technical Guidance Document (U.S. EPA. 2012b).
In 2010, an updated health assessment for oral exposures to 1,4-dioxane was released (U.S. EPA.
2010). During the development of the 2010 health assessment, new studies (Kasai et al., 2009; Kasai et
al.. 2008) regarding the toxicity of 1,4-dioxane through the inhalation route of exposure became available
during the finalization of the 1,4-dioxane assessment that was posted on the IRIS database in 2010 (U.S.
EPA. 2010). These new inhalation studies have been incorporated into the previously posted assessment
and are presented in this version of the toxicological review.
The literature search strategy employed for 1,4-dioxane was initially based on the chemical name,
Chemical Abstracts Service Registry Number (CASRN), and multiple common synonyms. A subsequent
search was completed which focused on the toxicology and toxicokinetics of 1,4-dioxane, particularly as
they pertain to target tissues, effects at low doses, mode of action (noncancer and cancer), and sensitive
populations. Following peer review of the assessment, a more targeted search was carried out based on
comments received from expert peer reviewers. Additionally, any pertinent scientific information
submitted by the public to the IRIS Submission Desk and by external peer reviewers during the
Independent Expert Peer Review meetings was also considered in the development of this document.
Selection of studies for inclusion in the Toxicological Review was based on consideration of the
extent to which the study was informative and relevant to the assessment, and general study
considerations as outlined in EPA guidance documents (A Review of the Reference Dose and Reference
Concentration Processes (U.S. EPA. 2002a) and Methods for Derivation of Inhalation Reference
Concentrations and Application of Inhaled Dosimetry (U.S. EPA. 1994b)).
Primary, peer-reviewed literature was reviewed through September 2009 for the oral assessment
and through May 2013 for the inhalation assessment and was included where the literature was
determined to be critical to the assessment. The relevant literature included publications on 1,4-dioxane
which were identified through Toxicology Literature Online (TOXLINE), PubMed, the Toxic Substance
Control Act Test Submission Database (TSCATS), the Registry of Toxic Effects of Chemical Substances
(RTECS), the Chemical Carcinogenesis Research Information System (CCRIS), the Developmental and
Reproductive Toxicology/Environmental Teratology Information Center (DART/ETIC), the
Environmental Mutagens Information Center (EMIC) and Environmental Mutagen Information Center
Backfile (EMICBACK) databases, the Hazardous Substances Data Bank (HSDB), the Genetic
Toxicology Data Bank (GENE-TOX), Chemical abstracts, and Current Contents. Other peer-reviewed
information, including health assessments developed by other organizations, review articles, and
independent analyses of the health effects data were retrieved and may be included in the assessment
where appropriate.
-------
The references considered and cited in this document, including bibliographic information and
abstracts, can be found on the Health and Environmental Research Online (HERO) website 1
(http://hero.epa.gov). 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.
Assessments by Other National and International Health Agencies
Toxicity information on 1,4-dioxane has been evaluated by several national and international
organizations. The results of these assessments are presented in Appendix H. It is important to recognize
that these assessments were prepared at different times, for different purposes, using different guidelines
and methods, and that newer studies have been included in the IRIS assessment.
:HERO is a database of scientific studies and other references used to develop EPA's risk assessments aimed at
understanding the health and environmental effects of pollutants and chemicals. It is developed and managed in
EPA's Office of Research and Development (ORD) by the National Center for Environmental Assessment (NCEA).
The database includes more than 700,000 scientific articles from the peer-reviewed literature. New studies are added
continuously to HERO.
-------
2.CHEMICAL AND PHYSICAL INFORMATION
1,4-Dioxane, a semi-volatile compound, is a colorless liquid with a pleasant odor (Hawlev and
Lewis. 2001; Lewis. 2000). Synonyms include diethylene ether, 1,4-diethylene dioxide, diethylene oxide,
dioxyethylene ether, and dioxane (Hawlev and Lewis. 2001). The chemical structure of 1,4-dioxane is
shown in Figure 2-1. Selected chemical and physical properties of this substance are in Table 2-1.
o
o
Figure 2-1. 1,4-Dioxane chemical structure.
Table 2-1 Physical properties and chemical identity of 1,4-dioxane
Property Value
CASRN:
123-91-1 (CRC Handbook (Lide. 2000))
Molecular weight:
88.10 (Merck Index (2001))
Chemical formula:
C4H8O2 (Merck Index (2001))
Boiling point:
101.1°C (Merck Index (2001))
Melting point:
11.8°C (CRC Handbook (Lide, 2000))
Vapor pressure:
40 mmHg at 25°C (Lewis, 2000)
Density:
1.0337 g/mL at 20°C (CRC Handbook (Lide, 2000))
Vapor density:
3.03 (air =1) (Lewis. 2000)
Water solubility:
Miscible with water (Hawlev and Lewis. 2001)
Other solubilities:
Miscible with ethanol, ether, acetone (CRC Handbook (Lide. 2000))
Log Kow:
-0.27 (Hanschetal.. 1995)
Henry's Law constant:
4.80 x 10"b atm-nY/molecule at 25°C (Parketal., 1987)
OH reaction rate constant:
1.09 x 10"11 cm3/molecule secat25°C (Atkinson, 1989)
17 (estimated using log Kow) (ACS Handbook (Lyman et al., 1990))
Bioconcentration factor:
0.4 (estimated using log Kow) (Mevlan et al.. 1999)
Conversion factors (in air):
1 ppm = 3.6 mg/m3; 1 mg/m3 = 0.278 ppm
(25 C and 1 atm) (HSDB. 2007)
-------
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-trichloroethane
(Surprenant. 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. 2012; UNEP. 2000; 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 (Hawley and Lewis. 2001; O'Neil et al.. 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 (Surprenant 2002). Reaction products of 1,4-dioxane are used in the manufacture of
insecticides, herbicides, plasticizers, and monomers (Surprenant. 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; Surprenant. 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, 2012; 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 (Mevlan et al.. 1999; Franke et al.. 1994). 1,4-Dioxane is not expected to
undergo hydrolysis or to biodegrade readily in the environment (ATSDR, 2012; HSDB. 2007). 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.
Recent environmental monitoring data for 1,4-dioxane in ambient air, drinking water, and food
samples are not available. Levels of 1,4-dioxane in ambient air ranged from 0.01-1.03 ppb in the mid
1980s (ATSDR, 2012; Spicer et al., 2002); however, concentrations in indoor may be greater.
1,4-Dioxane was found in groundwater samples in the United States at concentrations ranging from 1 ppb
to 109 ppb (ATSDR. 2005). Data indicate that 1,4-dioxane may leach from hazardous waste sites into
drinking water sources located nearby (Yasuhara et al.. 2003; Yasuhara et al.. 1997; Lesage etal.. 1990).
-------
1,4-Dioxane has been detected in contaminated surface and groundwater samples collected near
hazardous waste sites and industrial facilities (Derosa et al.. 1996). Total annual environmental releases of
1,4-dioxane reported from 1988 to 2011 by EPA's Toxics Release Inventory (TRI) ranged from 0.3
million to 1.3 million pounds, with approximately 0.9 million pounds released in 2011 (U.S. EPA. 2013b:
NTP. 2011). Dermal exposure to 1,4-dioxane may occur through contact with residues in contaminated
consumer products. The Environmental Working Group analyzed the ingredients of 15,000 personal care
products and reported that 22% of these products may contain 1,4-dioxane (EWG. 2012). The
concentrations of 1,4-dioxane in cosmetic products are declining over the past decade (ATSDR. 2012).
Additionally, occupational exposure to 1,4-dioxane may occur during its production and use as a solvent
(IARC. 1999).
-------
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. The metabolite most often measured and reported is (3-hydroxyethoxy acetic acid (HEAA),
which is predominantly excreted in the urine; however, other metabolites have also been identified.
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 ug/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 (Young etal.. 1978a. b).
Cumulative recovery of radiolabel in the feces was
-------
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 et al.. 1981). In this study, Rhesus
monkeys were exposed to [14C]-1,4-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]-1,4-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 of the median lethal 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.
Woo et al. (1977a) administered i.p. doses of [3H]-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 polychlorinated 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 would be unknown had saline perfusion been performed.
Covalent binding determined by gas chromatography 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 (U.S. Army Public Health
Command. 2010). although there is one report that identified l,4-dioxane-2-one as a major metabolite
(Woo et al.. 1977a). 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. Q978a, b;
1977)) employed an analytical technique that converted HEAA to the more volatile l,4-dioxane-2-one
prior to gas chromatography (GC); however, it is still unclear as to whether HEAA or l,4-dioxane-2-one
is the major metabolite of 1,4-dioxane. More recently, Koissi et al. (2012) found that l,4-dioxane-2-one is
rapidly degraded in rats (tl/2 is approximately 2 hours) at physiological conditions (pH=7.0 and 25 °C).
A proposed metabolic scheme for 1,4-dioxane metabolism (Woo et al., 1977a) 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 (Wooetal.. 1978. 1977b). 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 etal.. 1977c).
-------
C> OH ^OH 0
HOHC CHOH HOH2C
O'
[I] "---...._ ["] [III]
(b)"
+ hi,O
-hLO
Legend: I = 1,4-dioxane; II = diethylene glycol; III = p-hydroxyethoxy acetic acid (HEAA); IV = 1,4-dioxane-2-one;
V = 1,4-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.
Source: Adapted with permission of Elsevier Ltd., Woo et al. (1977b: 1977a).
Figure 3-1. Suggested metabolic pathways of 1,4-dioxane in the rat.
Metabolism of 1,4-dioxane in humans is extensive. In a survey of five 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]-1,4-dioxane (Young et al.. 1978a. b). Plasma 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 ug/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
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.
Induction of 1,4-dioxane metabolism was evaluated in a 13 week inhalation study by Kasai et al.
(2008). In this study, male and female F344 rats were exposed daily to concentrations of 0 (control), 100,
200, 400, 1,600, and 3,200 ppm. Plasma levels of 1,4-dioxane linearly increased with increasing
inhalation concentration, suggesting that metabolic saturation was not achieved during the course of the
experiments for plasma levels up to 730 and 1,054 ug/mL in male and female rats, respectively, at the
highest exposure concentration (3,200 ppm). In contrast, Young et al. (1978a) estimated from
experimentally determined Km values that metabolic saturation occurred near plasma levels of 100
ug/mL. Kociba et al. (1975) also estimated metabolic saturation near plasma levels of 100 ug/mL in rats
following a single i.v. dose. The lack of the metabolic saturation of 1,4-dioxane found in the Kasai et al.
(2008) study is likely attributed to enhanced metabolism by the induction of P450 enzymes, including
CYP2E1, by 13 weeks of repeated inhalation exposure to 1,4-dioxane at concentrations up to 3,200 ppm
(Kasai et al.. 2008).
10.000
55 60 65 7C
Note: y-axis is plasma concentration of 1,4-dioxane (|jg/mL) and x-axis is time (hr)
Source: Reprinted with permission of Taylor and Francis, Young et al. (|978a).
Figure 3-2. Plasma 1,4-dioxane levels in rats following i.v. doses of 3-5,600 mg/kg
11
-------
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. Takano et al. (2010) recently tested liver microsome contents from
male Sprague-Dawley rats treated with 500 mg 1,4-dioxane/kg BW intraperitoneally (i.p.) for 3 days for
CYP450 activities. CYP2B and CYP2E activities were significantly increased (p <0.05) compared to
control activity levels, while CYP2C activity was significantly decreased to approximately 50% of control
values. This is in contrast to Nannelli et al. (2005) where CYP2C values increased.
The increase in CYP2C or specifically, CYP2C11 activity reported by Nanelli et al. (2005) 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 etal.. 1977). There are no data for
1,4-dioxane elimination in humans from oral exposures.
12
-------
Elimination of 1,4-dioxane in rats (Young et al.. 1978a. b) was primarily via urine. As
comparably assessed in 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
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 14CO2 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 14CO2 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 14CO2 were measured in the
1,000 mg/kg group.
3.5. Physiologically Based Pharmacokinetic Models
Physiologically based pharmacokinetic models (PBPK) models have been developed for
1,4-dioxane in rats (Sweeney et al.. 2008; Leung and Paustenbach. 1990; Reitz et al.. 1990). mice (Reitz
etal.. 1990). humans (Sweeney et al.. 2008; Leung and Paustenbach. 1990; Reitz etal. 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 nonneoplastic endpoints (also see Appendix B).
13
-------
IV
infusion inhalation
J
O
O
DQ
en
Z3
O
£=
QJ
Lung blood
Rapidly
perfused
Slowly
perfused
"D
O
m
CD
t!
absorption
metabolism
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.
Figure 3-3. General PBPK model structure.
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: 1977c). 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 14CO2 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 etal.. 1978a. b). Although it may be concluded
that the rate of oral absorption was high enough to ensure nearly complete absorption by 72 hours, a more
14
-------
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 etal. 1978a. b), and metabolic induction was observed following exposure to high oral daily
doses (1,000 mg/kg-day) of 1,4-dioxane. 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 14CO2 (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 14CO2. 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. (I978a, b) corroborate previous studies describing the saturable nature of 1,4-dioxane metabolism in
rats (Woo et al.. 1977a: Woo et al.. 1977c) and are useful for optimizing metabolic parameters (Vmax and
Km)inaPBPKmodel.
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 could 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 etal. 1978a. b). The observed linear elimination of 1,4-dioxane after inhalation exposure
suggests that, via this route, metabolism follows a first-order process 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 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]-1,4-dioxane to white rats (strain not reported) and reported time-to-peak
blood, liver, kidney, and testes concentrations. They also reported ratios of tissue to blood concentrations
15
-------
at various time points after dosing. Woo et al. (1977a; 1977c) 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 and Paustenbach
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 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 etal. 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
16
-------
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. Reitz et al.
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 may not have reached equilibrium
in the highly gelatinous tissue homogenate (Reitz etal.. 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 etal.. 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 etal.. 1977). As noted previously, because the
human exposures were below the level needed to exhibit nonlinear kinetics, uncertainty exists in the
17
-------
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).
The Leung and Paustenbach (1990) model and 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. Fisher et al.
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.
3.5.2.4. Sweeney et al.
The Sweeney et al. (2008) model consisted of fat, liver, slowly perfused, and other well perfused
tissue compartments. Lung and stomach compartments were used to describe the route of exposure, and
an overall volume of distribution compartment was used for calculation of urinary excretion levels of
1,4-dioxane and HEAA. Blood, saline, and tissue to air partition coefficient values for 1,4-dioxane were
experimentally determined for rats and mice. Average values of the rat and mouse partition coefficients
were used for humans. Metabolic constants (VmaxC and Km) for the rat were derived by optimization of
data from an i.v. exposure of 1,000 mg/kg (Young et al.. 1978a) for inducible metabolism. For uninduced
VmaxC estimation, data generated by i.v. exposures to 3, 10, 30, and 100 mg/kg were used (Young et al..
1978a). Sweeney et al. (2008) determined best fit values for VmaxC by fitting to blood data in Young et al.
(1978a). The best fit VmaxC values were 7.5, 10.8, and 12.7 mg/hr-kg075 for i.v. doses of 3 to 100, 300, and
1,000 mg/kg, suggesting a gradual dose dependent increase in metabolic rate over i.v. doses ranging from
3 to 1,000 mg/kg. Although the Sweeney et al. (2008) model utilized two values for VmaxC (induced and
uninduced), the PBPK model does not include a dose-dependent function description of the change of
Vmax for i.v. doses between metabolic induced and uninduced exposures. Mouse VmaxC and absorption
constants were derived by optimizing fits to the blood 1,4-dioxane concentrations in mice administered
nominal doses of 200 and 2,000 mg/kg 1,4-dioxane via gavage in a water vehicle (Young et al.. 1978a).
The in vitro Vmax values for rats and mice determined by Sweeney et al. (2008) were scaled to estimate
18
-------
in vivo rates. The scaled and optimized rat VmaxC values were similar. The discrepancy between the scaled
and optimized mouse values was larger, which was attributed to possible induction in mice at the lowest
dose tested (200 mg/kg). The ratio of optimized/scaled values for the rat was used to adjust the scaled
human VmaxC and Km values to projected in vivo values.
The Sweeney et al. (2008) model outputs were compared, by visual inspection, with data not used
in fitting model parameters. The model predictions gave adequate match to the 1,4-dioxane exhalation
data in rats after a 1,000 mg/kg i.v. dose. 1,4-Dioxane exhalation was overpredicted by a factor of about
3, after a 10 mg/kg i.v. dose. Similarly, the simulations of exhaled 1,4-dioxane after oral dosing were
adequate at 1,000 mg/kg and 100 mg/kg (within 50%), but poor at 10 mg/kg (model over predicted by a
factor of 5). The model did not adequately fit the human data (Young etal.. 1977). Using physiological
parameters of Brown et al. (1997) and measured partitioning parameters (Sweeney et al.. 2008; Leung and
Paustenbach. 1990) with no metabolism, measured blood 1,4-dioxane concentrations reported by Young
et al. (1977) could not be achieved unless the estimated exposure concentration was increased by 2-fold.
As expected, inclusion of any metabolism resulted in a decrease in predicted blood concentrations. If
estimated metabolism rates were used with the reported exposure concentration, urinary metabolite
excretion was also underpredicted (Sweeney et al.. 2008).
3.5.2.5. Takano et al.
More recently, Takano et al. (2010) reported the development of a simplified rat and human
pharmacokinetic model. The purpose of this model was to provide a platform for a forward dosimetry
calculation using in vivo animal data and in vitro human and animal microsome data to predict the
1,4-dioxane concentrations in humans. The model had three nonphysiological compartments: absorption
compartment, metabolizing compartment, and a central compartment. Human metabolic parameters were
determined from in vitro data using liver microsomes, coefficients (octanol-water partition coefficient,
plasma unbound fraction) derived in silico, and physiological parameters (e.g., hepatic volume and blood
flow rate) obtained from the literature. Clearance was described as a first order rate of metabolism from
both the metabolizing compartment (e.g., hepatic metabolism) and the central compartment (e.g., renal
clearance). This is in contrast to the saturable metabolism used in previous models (Sweeney et al.. 2008;
ReitzetaL 1990).
The rat model outputs of Takano et al. (2010) were compared with 1,4-dioxane blood data at the
end of exposure in rats treated for 14 days with an oral dose of 500 mg/kg. The model adequately
predicted these rat data and showed a minimal amount of 1,4-dioxane remained in the blood 24 hrs after
the last exposure. The authors performed an in vitro to in vivo extrapolation to estimate human hepatic
intrinsic clearance for the human pharmacokinetic model. The ratio of rat in vivo/in vitro measurements
(0.0244/0.313) was multiplied by the human in vitro determination (22.9 L/hr) to yield 1.76 L/hr used in
the human pharmacokinetic model. The model was then used to simulate hypothetical human exposures;
however, no data were compared with model outputs. Thus, the ability of this model to adequately
simulate the available human data is unknown.
19
-------
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 etal.. 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 biologically realistic flow rates, tissue
volumes, enzyme affinities, metabolic processes, and elimination behaviors were also developed
(Sweeney et al.. 2008; Fisher etal.. 1997; Leung and Paustenbach. 1990; Reitz etal.. 1990). Most
recently, Takano et al. (2010) published a pharmacokinetic model utilizing hepatic volume, blood flow,
and an in vitro to in vivo extrapolation method for human intrinsic hepatic clearance.
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 Andersen. 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) and Takano et al. (2010) models were 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/hr/kg°74), cardiac output
(30 L/hr/kg°74), and the blood:air partition coefficient (3,650) above the measured values of
13 L/min/kg074 (Brown et al.. 1997). 14 L/hr/kg074 (Brown etal.. 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 recalibrated 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 recalibration 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, recalibration of the Reitz et al. (1990) human
PBPK model, and evaluation of the model parameters published by Sweeney et al. (2008). Using the
20
-------
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. Q978a, b; 1977).
The PBPK model of Reitz et al. (1990) was recalibrated 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 recalibrated model were compared with the empirical
model predictions of the same dosimeters to determine whether the recalibrated 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 acslX and perform identically to the models reported in the published papers (Figure B-3.
Figure B-4. Figure B-5. Figure B-7. and Figure B-8). 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 closer to the observed data 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, they were not modified to
describe oral exposures due to a lack of adequate human or animal data for parameterization.
Additionally, the inhalation Young et al. (1977) model did not provide adequate fits to the subchronic
exposure plasma levels of 1,4-dioxane in rats using the data from the Kasai et al. (2008) study, which is
likely due to the absence of a model description for metabolic induction.
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 recalibrated model predictions for blood 1,4-dioxane did
not adequately fit the experimental values using measured tissue:air partition coefficients from Leung and
Paustenbach (1990) or Sweeney et al. (2008) (Figure B-9 and Figure B-10). Use of a slowly perfused
tissue:air partition coefficient 4- to 7-fold lower than measured values produces exposure-phase
predictions that are much closer to observations, but does not replicate the elimination kinetics
(Figure B-16). Recalibration of the model with upper bounds on the tissue:air partition coefficients results
in predictions that are still 2- to 4-fold lower than empirical model prediction or observations
(Figure B-13 and Figure B-14). 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
21
-------
perfused tissue:air partition coefficient (Figure B-17). Artificially low values for the other tissue:air
partition coefficients are not expected to improve the model fit, because blood 1,4-dioxane is less
sensitive to these parameters than it is to VmaxC and Km. This suggests that the model structure is
insufficient to capture the apparent species difference in the blood 1,4-dioxane Vd between rats and
humans. 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). Additionally, the models do not account
for induction in metabolism, which may be present in animals repeatedly exposed to 1,4-dioxane.
Therefore, some other modification(s) to the Reitz et al. (1990) model structure may be necessary.
Similarly, Sweeney et al. (2008) also evaluated the available PBPK models (Leung and
Paustenbach. 1990; Reitz etal.. 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
Paustenbach. 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). The Takano (2010) model was only
tested by the authors using a single dose and route of exposure in rats, so the ability of the model to
predict over a range of exposures or exposure routes is unknown. Additionally, the human model (Takano
et al.. 2010) was not compared to the available published data (Young et al.. 1978a, b; Young et al.. 1977;
Young etal.. 1976)..
22
-------
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.
23
-------
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 (Yantetal. 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 (Fairlev
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 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;
Fairlev etal.. 1934; Yantetal.. 1930) did not provide information on the human subjects research ethics
24
-------
procedures undertaken in these studies; 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 etal.. 1976). These studies did not provide evidence
of effects in humans; however, the cohort size and number of reported cases were small.
4.1.1. Thiess et al.
A cross-sectional survey was conducted by Thiess et al. (1976) 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.
25
-------
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 grams 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 myelofibrosis 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 differ from the control population (SMR for
overall population = 0.83; SMR for 65-75-year-old men = 1.61; confidence intervals (CIs) were not
provided).
4.1.2. Buffleretal.
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
26
-------
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 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 and chronic studies conducted for 1,4-dioxane were drinking
water studies. To date, there are only two subchronic inhalation studies (Kasai et al.. 2008; Fairley et al.
1934) and two chronic inhalation studies (Kasai et al.. 2009; Torkelson et al.. 1974). The effects
following oral and inhalation exposures are described in detail below.
27
-------
4.2.1. Oral Toxicity
4.2.1.1. Subchronic Oral Toxicity
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 killed 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.Stoneretal.
1,4-Dioxane was evaluated by Stoner et al. (1986) 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 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. Stottetal.
In the Stott et al. (1981) study, 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 removed, weighed, and prepared for histopathology
evaluation. [3H]-Thymidine incorporation was measured by liquid scintillation spectroscopy.
28
-------
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 reported). Hepatic DNA synthesis,
measured by [3H]-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. Kanoetal.
In the Kano et al. (2008) study, groups of 6-week-old F344/DuCrj rats (10/sex/group) and
Crj:BDFl 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.
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 %2 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
29
-------
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 1,554 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 the olfactory epithelium,
nuclear enlargement of the tracheal epithelium, hepatocyte swelling of the centrilobular area of the liver,
vacuolar changes in the liver, granular changes in the liver, single cell necrosis in the liver, nuclear
enlargement of the proximal tubule of the kidneys, hydropic changes in the proximal tubule of the
kidneys, and vacuolar changes in the brain. The incidence data for histopathological lesions in rats are
presented in Table 4-1. The effects that occurred at the lowest doses were nuclear enlargement of the
respiratory epithelium in the nasal cavity and hepatocyte swelling in the central area of the liver in male
rats. Based on these histopathological findings the study authors identified the LOAEL as 126 mg/kg-day
and the NOAEL as 52 mg/kg-day.
Nuclear enlargement may be found in any cell type responding to microenvironmental stress or
undergoing proliferation. It may also be an indicator of exposure to a xenobiotic in that the cells are
responding by transcribing mRNA. Several studies indicate that it may also be identified as an early
change in response to exposure to a carcinogenic agent (Wiemann et al.. 1999; Enzmann et al.. 1995;
Clawson et al.. 1992; Ingram and Grasso. 1987. 1985); however, its relationship to the typical
pathological progression from initiated cell to tumor is unclear. Therefore, nuclear enlargement as a
specific morphologic diagnosis is not considered an adverse effect of exposure to 1,4-dioxane.
30
-------
Table 4-1 Incidence of histopathological lesions in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 13 weeks
Effect
Incidence
Male dose (mg/kg-day)a
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; renal proximal tubule
Hydropic change; renal proximal tubule
Vacuolic change; brain
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
52
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
126
9/1 Ob
0/10
0/10
9/1 Ob
1/10
0/10
0/10
0/10
0/10
0/10
Female dose
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; proximal tubule
Hydropic change; proximal tubule
Vacuolic change; brain
aData are presented for sacrificed animals.
bp < 0.01 by x2 test.
°p < 0.05.
Source: Reprinted with permission of the Japanese Society
0
0/10
0/10
0/10
0/10
0/10
2/10
2/10
0/10
0/10
0/10
of Toxicology;
83
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
Kano et al.
185
5/1 Oc
0/10
0/10
1/10
0/10
1/10
1/10
0/10
0/10
0/10
(2008)
274
10/10b
10/10b
10/10b
10/10b
0/10
5/1 Oc
5/1 Oc
1/10
0/10
0/10
657
9/1 Ob
9/1 Ob
10/10b
10/10b
10/10b
2/10
2/10
5/1 Oc
0/10
0/10
1,554
10/10b
10/10b
10/10b
10/10b
10/10b
10/10b
10/10b
9/1 Ob
7/1 Ob
10/10b
(mg/kg-day)a
427
10/10b
9/1 Ob
9/1 Ob
0/10
0/10
5/1 Oc
5/10
0/10
0/10
0/10
756
10/10b
10/10b
10/10b
9/1 Ob
0/10
5/1 Oc
5/10
8/1 Ob
0/10
0/10
1,614
8/9b
8/9b
9/9b
9/9b
9/9b
8/9b
8/9b
9/9b
5/9c
9/9b
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 (1,570 mg/kg-day). Hematological changes were within 2-15% of control values. Serum
31
-------
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). However, as noted above, EPA does not
consider nuclear enlargement an adverse effect of exposure to 1,4-dioxane.
32
-------
Table 4-2 Incidence of histopathological lesions in Crj:
in drinking water for 13 weeks
Effect
BDF1 mice exposed to 1,4-dioxane
Incidence
Male dose (mg/kg-day)a
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
86
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
170
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
1/10
0/10
231
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
Female dose
387
0/10
1/10
0/10
0/10
0/10
2/10
0/10
10/10C
0/10
1/10
0/10
585
2/10
0/10
9/1 Oc
0/10
0/10
7/1 Oc
0/10
9/1 Oc
0/10
10/10C
5/1 Ob
882
5/1 Ob
0/10
10/10C
0/10
0/10
9/1 Oc
0/10
9/1 Oc
0/10
10/10C
10/10C
1,570
0/9
5/9b
9/9c
6/9c
9/9c
9/9c
6/9c
9/9c
8/9c
9/9c
9/9c
(mg/kg-day)a
898
3/10
1/10
6/1 Ob
1/1 Oc
0/10
9/1 Oc
0/10
10/10C
0/10
10/10C
7/1 Oc
1,620
3/10
5/1 Ob
10/10C
6/1 Ob
2/10
10/10C
10/10C
10/10C
7/1 Oc
10/10C
10/10C
2,669
7/1 Oc
9/1 Oc
10/10C
6/1 Ob
8/1 Oc
10/10C
10/10C
10/10C
10/10C
9/1 Ob
9/1 Oc
"Data are presented for sacrificed animals.
bp < 0.01 by x2 test.
°p < 0.05.
Source: Kano et al (2008).
33
-------
4.2.1.1.4. Yamamoto et al.
Studies (Yamamoto etal. 1998a; Yamamoto et al., 1998b) in rasH2 transgenic mice carrying the
human prototype c-Ha-ras gene have been investigated as a bioassay model for rapid carcinogenicity
testing. As part of validation studies of this model, 1,4-dioxane was one of many chemicals that were
evaluated. RasH2 transgenic mice were Fl offspring of transgenic male C57BLr6J and normal female
BALB/cByJ mice. CB6Fi mice were used as a nontransgenic control. Seven- to nine-week-old mice (10-
15/group) were exposed to 0, 0.5, or 1% 1,4-dioxane in drinking water for 26 weeks. An increase in lung
adenomas was observed in treated transgenic mice, as 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 Oral Toxicity and Carcinogenicity
4.2.1.2.1. Argus etal.
Twenty-six adult male Wistar rats (Argus etal.. 1965) 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
34
-------
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 al.; Hoch-Ligeti et al.
Five groups (28-32/dose group) of male Sprague Dawley rats (2-3 months of age) weighing
110-230 g at the beginning of the experiment 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% (Argus et al.. 1973;
Hoch-Ligeti et al.. 1970). 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 necropsies 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 sacrificed 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 authors reported qualitatively that 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 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 study authors provided quantitation for the numbers of
incipient liver tumors and hepatomas in rats from this study (treated for 13 months and observed at
13-16 months) as 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)3 Incipient tumors Hepatomas Total
430 4 0 4
574 909
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. Hoch-Ligeti and Argus.
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
36
-------
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
between 944 and 1,019 mg 1,4-dioxane/kg-day. A group often 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. Kocibaetal.
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 (Kociba
et al.. 1974). 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 %2 Contingency Tables and Fisher's Exact test. Student's 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) calculated mean daily doses of 9.6, 94, and
37
-------
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 not different. 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.
Table 4-4 Incidence of liver and nasal tumors in male and female Sherman rats
(combined) treated with 1,4-dioxane in the drinking water for 2 years
Dose in mg/kg-day
(average of male and
female dose)
0
14
121
1,307
Effective
number of
animals3
106
110
106
66
Number of
tumor-bearing
animals
31
34
28
21
Number of animals
Hepatic tumors
(all types)
2
0
1
12b
Hepatocellular
carcinomas
1
0
1
10C
Nasal
carcinomas
0
0
0
3d
aRats surviving until 12 months on study.
bp = 0.00022 by one-tailed Fisher's Exact test.
°p = 0.00033 by one-tailed Fisher's Exact test.
dp = 0.05491 by one-tailed Fisher's Exact test.
Source: Reprinted with permission of Elsevier, Ltd., Kociba et al. (1974),
38
-------
The high-dose level was the only dose that increased the formation of liver tumors over control
(males 1,015 mg/kg-day; females 1,599 mg/kg-day) and also caused significant liver and kidney toxicity
in these animals. The mid-dose group (males 94 mg/kg-day; females 148 mg/kg-day) experienced hepatic
and renal degeneration and necrosis, as well as regenerative proliferation 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 either sex in the low-dose (males 9.6 mg/kg-day; females 19 mg/kg-day) group
of rats.
4.2.1.2.5. National Cancer Institute (NCI).
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% (NCI. 1978). 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.
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
39
-------
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)
Cortical tubule degeneration
Hepatocytomegaly
Gastric ulcer
Pneumonia
0
0/3 1a
5/31
(16%)
0/30a
8/30
(27%)
240
20/3 1b
(65%)
3/32
(9%)
5/28b
(18%)
15/31
(48%)
530
27/33b
(82%)
11/33
(33%)
5/30b
(17%)
14/33
(42%)
Females (mg/kg-day)
0
0/3 1a
7/31 a
(23%)
0/31
6/30a
(20%)
350
0/34
11/33
(33%)
1/33
(3%)
5/34
(15%)
640
10/32b
(31%)
17/32b
(53%)
1/30
(3%)
25/32b
(78%)
Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this review.
""Incidence significantly elevated compared to control by Fisher's Exact test (p < 0.05) performed for this review.
Source: NCI (1978).
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)).
40
-------
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
Effect
Males (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
Females (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/33 (0%)
2/31 (6%)
0
0/34 (0%)d
0/31 (0%)f
Incidence
240b
12/33(36%)
2/32 (6%)
350
10/35(29%)c
10/33(30%)e
530
16/34(47%)e
1/33 (3%)
640
8/35 (23%)c
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.
°p & 0.003 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.008 by Cochran-Armitage test.
ep < 0.001 by Fisher's Exact test pair-wise comparison with controls.
fp = 0.001 by Cochrai
Source: NCI (1978).
fp = 0.001 by Cochran-Armitage test.
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 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
41
-------
that hepatocytomegaly was observed in dosed and control mice but did not comment on the significance
of the effect. 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; aNOAEL 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.
Table 4-7 Incidence of hepatocellular adenoma or carcinoma in B6C3Fi mice exposed to
1,4-dioxane in drinking water
Effect
Males (mg/kg-day)a 0
Hepatocellular carcinoma 2/49 (4%)b
Hepatocellular adenoma or carcinoma 8/49 (16%)b
Females (mg/kg-day)a 0
Hepatocellular carcinoma 0/50 (0%)b
Hepatocellular adenoma or carcinoma 0/50 (0%)b
Incidence
720
18/50(36%)c
19/50(38%)d
380
12/48 (25%)c
21/48 (44%)c
830
24/47 (51 %)c
28/47 (60%)c
860
29/37 (78%)c
35/37 (95%)c
aTumor incidence values were not adjusted for mortality.
bp < 0.001 , positive dose-related trend (Cochran-Armitage test).
0 p £ 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 B6C3FJ 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.
42
-------
4.2.1.2.6. Kano et al.; Japan Bioassay Research Center; Yamazaki et al.
The Japan Bioassay Research Center (JBRC) conducted a 2-year drinking water study
determining the effects of 1,4-dioxane on both sexes of rats and mice. The study results have been
reported several times: once as conference proceedings (Yamazaki et al., 1994). once as a laboratory
report (JBRC. 1998). and most recently as a peer-reviewed manuscript (Kano et al.. 2009). Dr. Yamazaki
also provided some detailed information (Yamazaki. 2006). Variations in the data between these three
reports were noted and included: (1) the level of detail on dose information reported; (2) categories for
incidence data reported (e.g., all animals or sacrificed animals); and (3) analysis of non- and neoplastic
lesions.
The 1,4-dioxane dose information provided in the reports varied. Specifically, Yamazaki et al.
(1994) only included drinking water concentrations for each dose group. In contrast, JBRC (1998)
included drinking water concentrations (ppm), in addition using body weights and water consumption
measurements to calculate daily chemical intake (mg/kg-day). JBRC (1998) reported daily chemical
intake for each dose group as a range. Thus, for the External Peer Review draft of this Toxicological
Review of 1,4-Dioxane (U.S. EPA. 2009b). the midpoint of the range was used. Kano et al. (2009) also
reported a calculation of daily chemical intake based on body weight and water consumption
measurements; however, for each dose group they reported a mean and standard deviation estimate.
Therefore, because the mean more accurately represents the delivered dose than the midpoint of a range,
the Kano et al. (2009) calculated mean chemical intake (mg/kg-day) is used for quantitative analysis of
this data.
The categories for which incidence rates were described also varied among the reports. Yamazaki
et al. (1994) and Kano et al. (2009) reported histopathological results for all animals, including dead and
moribund animals; however, the detailed JBRC (1998) laboratory findings included separate incidence
reports for dead and moribund animals, sacrificed animals, and all animals.
Finally, the criteria used to evaluate some of the data were updated when JBRC published the
most recent manuscript by Kano et al. (2009). The manuscript by Kano et al. (2009) stated that the lesions
diagnosed in the earlier reports (JBRC. 1998; Yamazaki et al.. 1994) were re-examined and recategorized
as appropriate according to current pathological diagnostic criteria (see references in Kano et al. (2009)).
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:BDFl mice
(50/sex/dose level) were similarly exposed in the drinking water to 0, 500, 2,000, or 8,000 ppm of
1,4-dioxane. The high doses were 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) 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 an estimate of the daily intake of 1,4-dioxane (mg/kg-day) by male and
female rats and mice. Kano et al. (2009) reported a calculated mean ± standard deviation for the daily
43
-------
doses of 1,4-dioxane for the duration of the study. Male rats received doses of approximately 0, 11 ± 1,
55 ± 3, or 274 ± 18 mg/kg-day and female rats received 0, 18 ± 3, 83 ± 14, or 429 ± 69 mg/kg-day. Male
mice received doses of 0, 49 ± 5, 191 ± 21, or 677 ± 74 mg/kg-day and female mice received 0, 66 ± 10,
278 ± 40, or 964 ± 88 mg/kg-day. For the remainder of this document, including the dose-response
analysis, the mean calculated intake values are used to identify dose groups. 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).
No information was provided as to when urine samples were collected. Blood samples were
collected only at the end of the 2-year study (Yamazaki. 2006). 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 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 %2 test were used to
assess the statistical significance of changes in continuous and discrete variables, respectively.
For rats, 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) exhibited
slower growth rates and terminal body weights that were significantly different (p < 0.05) compared to
controls. 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). 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) .
All control and exposed rats lived at least 12 months following study initiation (Yamazaki. 2006);
however, survival at the end of the 2-year study in the high dose group of male and female rats (274 and
429 mg/kg-day, respectively) was approximately 50%, which was significantly different compared to
controls. The investigators 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. (Yamazaki. 2006).
Several hematological changes were noted in the JBRC (1998) report: Decreases in RBC (male
rats only), hemoglobin, hematocrit, and MCV; and increases in platelets in high-dose groups were
observed (JBRC, 1998). 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,
44
-------
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. 1998). 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. 1998). Also, blood in the urine was seen in female rats at
83 and 429 mg/kg-day (JBRC. 1998). 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 (Table 4-8 and Table 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. 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, and clear and mixed cell foci) and at 429 mg/kg-day in female rats (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, 1998). As noted previously in Section 4.2.1.1.3.
nuclear enlargement as a specific morphologic diagnosis is not considered an adverse effect of exposure
to 1,4-dioxane.
45
-------
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'b
Effect
Nuclear enlargement; nasal respiratory epithelium0
Squamous cell metaplasia; nasal respiratory epithelium0
Squamous cell hyperplasia; nasal respiratory epithelium0
Nuclear enlargement; nasal olfactory epithelium0
Respiratory metaplasia; nasal olfactory epitheliumd
Atrophy; nasal olfactory epitheliumd
Hydropic change; lamina propriad
Sclerosis; lamina propriad
Adhesion; nasal cavityd
Inflammation; nasal cavityd
Spongiosis hepatis; liverd
Clear cell foci; liver0'9
Acidophilic cell foci; liver0'9
Basophilic cell foci; liver0'9
Mixed-cell foci; liver0'9
Nuclear enlargement; kidney proximal tubuled
0
0/50
0/50
0/50
0/50
12/50
0/50
0/50
0/50
0/50
0/50
12/50
3/50
12/50
7/50
2/50
0/50
11
0/50
0/50
0/50
0/50
11/50
0/50
0/50
0/50
0/50
0/50
20/50
3/50
8/50
11/50
8/50
0/50
55
0/50
0/50
0/50
5/50f
20/50
0/50
0/50
1/50
0/50
0/50
25/50f
9/50
7/50
8/50
14/508
0/50
274
26/508
31/508
2/50
38/508
43/50
36/50
46/50
44/50
48/50
13/50
40/50
8/50
5/50
16/50f
13/508
50/50
"Data presented for all animals, including animals that became moribund or died before the end of the study.
bDose levels from Kano et al. (2009).
°Data from Kano et al. (2009).
dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
fp < 0.05 by x2 test.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-
examined according to updated criteria for liver lesions and were afterwards classified as either hepatocellular adenoma or
altered hepatocellular foci in Kano et al. (2009).
Sources: Kano et al. (2009) and JBRC (1998).
46
-------
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'b
Effect
Nuclear enlargement; nasal respiratory epithelium0
Squamous cell metaplasia; nasal respiratory epithelium0
Squamous cell hyperplasia; nasal cavity0
Nuclear enlargement; nasal olfactory epithelium0
Respiratory metaplasia; nasal olfactory epitheliumd
Atrophy; nasal olfactory epitheliumd
Hydropic change; lamina propriad
Sclerosis; lamina propriad
Adhesion; nasal cavityd
Inflammation; nasal cavityd
Proliferation; nasal glandd
Spongiosis hepatis; liverd
Cyst formation; liverd
Acidophilic cell foci; liver0'9
Basophilic cell foci; liver0'9
Clear cell foci; liver0'9
Mixed-cell foci; liver0'9
Nuclear enlargement; kidney proximal tubuled
0
0/50
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
23/50
1/50
1/50
0/50
18
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
1/50
27/50
1/50
1/50
0/50
83
0/50
0/50
0/50
28/508
2/50
1/50
0/50
0/50
0/50
1/50
0/50
1/50
1/50
1/50
31/50
5/50
3/50
6/50
429
13/508
35/508
5/50
39/50
42/50
40/50
46/50
48/50
46/50
15/50
11/50
20/50
8/50
1/50
8/50e
4/50
11/50f
39/50
"Data presented for all animals, including animals that became moribund or died before the end of the study.
"Dose levels from Kano et al. (2009).
°Data from Kano et al. (2009).
dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
fp < 0.05 by x2 test.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-
examined according to updated criteria for liver lesions and were afterwards classified as either hepatocellular adenoma or
altered hepatocellular foci in Kano et al. (2009).
Sources: Kano et al. (2009) and JBRC (1998).
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, as
noted previously, EPA does not consider it an adverse toxicological effect. 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
47
-------
lesions were also seen at a dose of 55 mg/kg-day in male rats; these changes are likely to 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 perisinusoidal (Ito) cells (PSC) 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.
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 (1998) 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. 1998) in the
high dose groups, 274 and 429 mg/kg-day, respectively.
Significantly increased incidences of liver tumors (adenomas and carcinomas) and tumors of the
nasal cavity occurred in high-dose male and female rats (Table 4-10 and Table 4-11) treated with
1,4-dioxane for 2 years (Kano et al.. 2009). The first liver tumor was seen at 85 weeks in high-dose male
rats and 73 weeks in high-dose female rats (versus 101-104 weeks in lower dose groups and controls)
(Yamazaki. 2006). 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 pretermination deaths
(Yamazaki. 2006). 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 Table 4-10 and Table 4-11 were not adjusted for survival.
48
-------
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
Effect Males
Dose (mg/kg-day) 0 11
55
274
0
Females
18
83
429
Nasal cavity
Squamous cell carcinoma 0/50 0/50
Sarcoma 0/50 0/50
Rhabdomyosarcoma 0/50 0/50
Esthesioneuroepithelioma 0/50 0/50
0/50
0/50
0/50
0/50
3/50a
2/50
1/50
1/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
7/50a'b
0/50
0/50
1/50
Peritoneum
Mesothelioma 2/50 2/50
5/50
28/50a'b
1/50
0/50
0/50
0/50
Mammary gland
Fibroadenoma 1/50 1/50
Adenoma 0/50 1/50
Either adenoma or fibroadenoma 1/50 2/50
0/50
2/50
2/50
4/50a
2/50
6/50a
3/50
6/50
8/50
2/50
7/50
8/50
1/50
10/50
11/50
3/50
16/50a'c
18/50a'c
Statistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
Significantly different from control by Fisher's exact test (p < 0.01).
Significantly different from control by Fisher's exact test (p < 0.05).
Source: Reprinted with permission of Elsevier, Ltd., 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
Effect
Dose (mg/kg-day)
Hepatocellular adenoma
Hepatocellular carcinoma
Either adenoma or carcinoma
Males
0
3/50
0/50
3/50
11
4/50
0/50
4/50
55
7/50
0/50
7/50
274
32/50a'b
14/50a'b
39/50a'b
0
3/50
0/50
3/50
Females
18
1/50
0/50
1/50
83
6/50
0/50
6/50
429
48/50a'b
10/50a'b
48/50a'b
Significantly different from control by Fisher's exact test (p < 0.01).
Statistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
For mice, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
study. Similar to rats, the growth rates of male and female mice 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 964 mg/kg-day, respectively). 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 (964 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 Kano et al. (2009) reported tumor incidence for all animals in the study
(N=50), including animals that became moribund or died before the end of the study. Additional data on
49
-------
survival rates of mice were provided in an email from Dr. Yamazaki (JBRC) to Dr. Stickney (SRC) on
12/18/2006 (2006). who reported that the survival of mice was low in all male groups (31/50, 33/50,
25/50 and 26/50 in control, low-, mid-, and high-dose groups, respectively) and particularly low in
high-dose females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-, and high-dose groups,
respectively). These deaths occurred primarily during the second year of the study. Survival at 12 months
in male mice was 50/50, 48/50, 50/50, and 48/50 in control, low-, mid-, and high-dose groups,
respectively. Female mouse survival at 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-,
mid-, and high-dose groups, respectively (Yamazaki. 2006). Furthermore, these deaths were primarily
tumor related. Liver tumors were listed as the cause of death for 31 of the 45 pretermination deaths in
high-dose female Crj:BDFl mice (Yamazaki. 2006). For mice, growth and mortality rates were reported
in Kano et al. (2009) for the duration of the study. Similar to rats, the growth rates of male and female
mice 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 964 mg/kg-day, respectively).
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. 1998). 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 (Table 4-12 and Table 4-13). Commonly seen alterations included nuclear
enlargement, atrophy, and inflammation of the epithelium. Other changes observed included nuclear
enlargement of the proximal tubule of the kidney and angiectasis in the liver in high-dose male mice.
50
-------
Table 4-12 Incidence of histopathological
1,4-dioxane in drinking water
Effect
Nuclear enlargement; nasal respiratory epithelium0
Nuclear enlargement; nasal olfactory epithelium0
Atrophy; nasal olfactory epitheliumd
Inflammation; nasal cavityd
Atrophy; tracheal epitheliumd
Nuclear enlargement; tracheal epitheliumd
Nuclear enlargement; bronchial epitheliumd
Atrophy; lung/bronchial epitheliumd
Accumulation of foamy cells; lungd
Angiectasis; liverd
Nuclear enlargement; kidney proximal tubuled
lesions in male Crj
for 2 years
0
0/50
0/50
0/50
1/50
0/50
0/50
0/50
0/50
1/50
2/50
0/50
:BDF1 mice exposed to
Dose
49
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
0/50
3/50
0/50
(mg/kg-day)a'b
191
0/50
9/50e
1/50
1/50
0/50
0/50
0/50
0/50
0/50
4/50
0/50
677
31/508
49/508
48/50
25/50
42/50
17/50
41/50
43/50
27/50
16/50
39/50
aData presented for all animals, including animals that became moribund or died before the end of the study.
"Dose levels from Kano et al. (2009).
°Data from Kano et al. (2009).
dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
Sources: Kano et al. (2009) and JBRC (1998).
Table 4-13 Incidence of histopathological
1,4-dioxane in drinking water
Effect
Nuclear enlargement; nasal respiratory epithelium0
Nuclear enlargement; nasal olfactory epithelium0
Atrophy; nasal olfactory epitheliumd
Inflammation; nasal cavityd
Atrophy; tracheal epitheliumd
Nuclear enlargement; bronchial epitheliumd
Atrophy; lung/bronchial epitheliumd
Accumulation of foamy cells; lungd
lesions in female CrjrBDFl mice exposed
for 2 years
0
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
Dose
66
0/50
0/50
0/50
0/50
0/50
1/50
0/50
1/50
(mg/kg-day)a'b
278
0/50
41/508
1/50
7/50
2/50
22/50
7/50
4/50
to
964
41/508
33/508
42/50
42/50
49/50
48/50
50/50
45/50
"Data presented for all animals, including animals that became moribund or died before the end of the study.
bDose levels from Kano et al. (2009).
°Data from Kano et al. (2009).
dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
Sources: Kano et al. (2009) and JBRC (1998).
51
-------
NOAEL and LOAEL values for mice in this study were identified by EPA as 66 and
278 mg/kg-day, respectively, based on nasal inflammation observed in female mice. Nuclear enlargement
of the nasal olfactory epithelium and bronchial epithelium was also observed at a dose of 278 mg/kg-day
in female mice; however, as described previously nuclear enlargement as a specific morphologic
diagnosis is not considered an adverse effect of exposure to 1,4-dioxane. 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 statistically
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 (Yamazaki. 2006). 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 (Yamazaki, 2006). 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 BDF1 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 BDF1. The background incidence rates for
male BDF1 mice were 15% and 22.8% for hepatocellular adenomas and carcinomas, respectively, out of
500 control mice in ten 2-year bioassays (Katagiri et al., 1998). Background rates for B6C3F] mice
evaluated by the National Toxicology Program are similar (10.3% and 21.3% for hepatocellular
adenomas and carcinomas in male mice, respectively; 4.0% and 4.1% for hepatocellular adenomas and
carcinomas in female mice, respectively) to the BDF1 mice background rates observed by JBRC
(Haseman et al.. 1984). Thus, the BDF1 mouse is not particularly sensitive compared to the commonly
used B6C3Fi strain and indicates that the results obtained by JBRC are reasonable.
52
-------
Table 4-14 Incidence of tumors in CrjrBDFl mice exposed to 1,4-dioxane in drinking
water for 2 years
Effect
Dose (mg/kg-day)
Males
0
49
191
677
0
Females
66
278
964
Nasal Cavity
Adenocarcinoma
Esthesioneuroepithelioma
0/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
0/50
Liver
Hepatocellular adenoma
Hepatocellular carcinoma
Either hepatocellular
adenoma or carcinoma
9/50
15/50
23/50
17/50
20/50
31/50
23/50a
23/50
37/50c
11/50
36/50a'b
40/50a'b
5/50
0/50
5/50
31/50a
6/50c
35/50a
20/50a
30/50a
41/50a
3/50
45/50a'b
46/50a'b
aSignificantly different from control by Fisher's exact test (p < 0.01).
bStatistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
""Significantly different from control by Fisher's exact test (p < 0.05).
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
A weight of evidence evaluation of the carcinogenicity studies presented in Section 4.2.1.2 is
located in Section 4.7 and Table 4-19.
4.2.2. Inhalation Toxicity
4.2.2.1. Subchronic Inhalation Toxicity
4.2.2.1.1. Fairley etal.
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) (Fairley etal.. 1934). 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
53
-------
exposed to this concentration. Animals exposed to 1,000 ppm were sacrificed 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.
4.2.2.1.2. Kasaietal.
Male and female 6-week-old F344/DuCrj rats (10/sex/group) were exposed to nominal
concentrations of 0 (clean air), 100, 200, 400, 800, 1,600, 3,200, or 6,400 ppm (0, 360, 720, 1,400, 2,900,
5,800, 12,000, and 23,000 mg/m3, respectively) of vaporized 1,4-dioxane (>99% pure) for 6 hours/day, 5
days/week, for 13 weeks in whole body inhalation chambers (Kasai et al., 2008). Each inhalation chamber
housed 20 individual cages for 10 males and 10 females. During exposure, the concentration of
1,4-dioxane vapor was determined every 15 minutes by gas chromatography. In addition, during
exposure, animals received food and water ad libitum and the following data were collected: 1) clinical
signs and mortality (daily); 2) BW and food intake (weekly); 3) urinary parameters using Ames reagent
strips (measured during week 13 of the exposure); and 4) 1,4-dioxane content in plasma from three rats of
both sexes (measured on the third day of exposure during weeks 12 and 13 at 1 hour after termination). At
the end of the 13-week exposure period or at the time of an animal's death during exposure, all organs
were collected, weighed, and evaluated for macroscopic lesions. Histopathological evaluations of organs
and tissues were conducted in accordance with the OECD test guidelines, including all tissues of the
respiratory tract. Liver sections from male and female rats exposed to 800, 1,600 and 3,200 ppm of
1,4-dioxane were also analyzed for foci (in the absence of tumor formation) by immunohistochemical
expression of glutathione S-transferase placental form (GST-P). Hematological and clinical chemistry
parameters were measured using blood collected from the abdominal aorta of rats following an overnight
fasting at the end of the 13-week exposure period. The measured hematological and clinical chemistry
parameters included: red blood cell count, hemoglobin, hematocrit, MCV, AST, ALT, glucose, and
triglyceride. Statistically significant differences (p-value of 0.05) between 1,4-dioxane and clean air
exposed groups were determined by study authors using Dunnett's test or %2test.
All rats exposed to 6,400 ppm of 1,4-dioxane died by the end of the first week of exposure; the
determined cause of death was renal failure and diagnosed as necrosis of the renal tubules. At
concentrations lower than 6,400 ppm, mortality was not observed and all exposed rats were absent of
clinical signs. Exposure-related effects on final BWs, organ weights, and hematological and clinical
chemistry parameters were reported as compared to controls and these changes are outlined in Table 4-15
and Table 4-16. Briefly, terminal BWs were significantly decreased in both sexes at 200 ppm; and
additionally in females at 800 and 1,600 ppm. Statistically significant increases in several organ weights
were observed, including lung (> 1,600 ppm, males; > 200 ppm, females); liver (> 800 ppm, both sexes),
and kidneys (3,200 ppm, males; > 800 ppm, females). Statistically significant changes in hematological
parameters and clinical chemistry were observed in both sexes at 3,200 ppm including increased levels of
hemoglobin ALT, RBC, AST ,and MCV. In females only, at 3,200 ppm, increased levels of hematocrit
was noted; and in males at this exposure concentration decreased levels of glucose and triglyceride were
54
-------
observed, in addition to slightly decreased urinary protein. However, the urinary protein data were not
shown in this study. At 200 ppm, an increased AST level in females was noted. Blood plasma levels of
1,4-dioxane were also evaluated and in both sexes, a linear increase in 1,4-dioxane levels was detected at
exposure concentrations of 400 ppm and above. The highest blood levels of 1,4-dioxane were detected in
females.
Exposure and/or sex-related histopathology findings also reported by the study authors included
nuclear enlargement of the nasal respiratory, nasal olfactory, tracheal, and bronchial epithelium; vacuolic
change in the olfactory and bronchial epithelium; atrophy of the nasal epithelium; hydropic change in the
proximal tubules of the kidney; and single-cell necrosis and centrilobular swelling in the liver. Table 4-17
presents a summary of these histopathological lesions, including incidence and severity data. Further
microscopic evaluation of liver tissue revealed GST-P positive liver foci in both sexes at 3,200 ppm (3/10
males, 2/10 females) and in females at 1,600 ppm (4/10).
The study authors determined nuclear enlargement in the respiratory epithelium as the most
sensitive lesion and a LOAEL value of 100 ppm was identified by the study authors based on the
incidence data of this lesion in both male and female rats. However, as noted for the oral studies, nuclear
enlargement may be found in any cell type responding to microenvironmental stress or undergoing
proliferation. It may also be an indicator of exposure to a xenobiotic in that the cells are responding by
transcribing mRNA. Several studies indicate that it may also be identified as an early change in response
to exposure to a carcinogenic agent (Wiemann et al.. 1999; Enzmann et al.. 1995; Clawson et al.. 1992;
Ingram and Grasso. 1987. 1985); however, its relationship to the typical pathological progression from
initiated cell to tumor is unclear. Therefore, as described in Section 4.2.1.1.3. nuclear enlargement as a
specific morphologic diagnosis is not considered an adverse effect of exposure to 1,4-dioxane.
55
-------
Table 4-15 Terminal body weights and relative organ weights
to 1,4-dioxane vapor by whole-body inhalation for
Males3
Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
Females3
Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
0 (clean air)
323 ± 14
0.310
±0.011
2.610
± 0.069
0.589
±0.016
0 (clean air)
187 ±5
0.402
±0.013
2.353 ±0.081
0.647
±0.014
1
100
323 ± 14
0.312
± 0.007
2.697
± 0.092
0.596
± 0.021
1
100
195 ±8
0.402
±0.015
2.338
± 0.092
0.631
±0.019
of F344/DuCrj rats exposed
13 weeks
,4-dioxane vapor concentration (ppm)
200
304 ± 11C
0.325
± 0.008C
2.613
± 0.084
0.612
±0.013
400
311 ± 19
0.320
± 0.009
2.666
± 0.080
0.601
± 0.020
800
317± 12
0.321
±0.011
2.726
± 0.082C
0.610
±0.015
1,600
312± 14
0.333
± 0.009b
2.737
± 0.077b
0.606
± 0.021
3,200
301 ± 11b
0.346
±0.017b
2.939
±0.101b
0.647
± 0.026b
,4-dioxane vapor concentration (ppm)
200
174 ± 10b
0.435
±0.018b
2.395
± 0.092
0.668
±0.012
400
180 ±5
0.429
± 0.029C
2.408
± 0.066
0.662
± 0.024
800
175±6b
0.430
±0.013b
2.513
± 0.076b
0.679
±0.018b
1,600
173±8b
0.454
±0.018b
2.630
±0.139b
0.705
± 0.028b
3,200
168 ±4b
0.457
±0.016b
2.828
±0.144b
0.749
± 0.024b
aData are presented for 10 sacrificed animals.
bp < 0.01 by Dunnett's test.
°p < 0.05 by Dunnett's test.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2008)
56
-------
Table 4-16 Hematology and clinical chemistry of F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 13 weeks
Males3
Red blood cell
(106/|jL)
Hemoglobin (g/dL)
Hematocrit (%)
MCV (fl_)
AST (IU/L)
ALT (IU/L)
Glucose (mg/dL)
Triglyceride (mg/dL)
Females3
Red blood cell
(106/|jL)
Hemoglobin (g/dL)d
Hematocrit (%)d
MCV (fL)d
AST (IU/L)d
ALT (IU/L)d
Glucose (mg/dL)d
Triglyceride (mg/dL)
0 (clean air)
9.55 ±0.17
16.0 ±0.2
46.2 ± 1.2
48.4 ±0.7
73 ±8
27 ±3
197 ± 17
125 ± 17
0 (clean air)
8.77 ±0.23
16.2 ±0.3
46.0 ± 1.5
52.5 ±0.7
64 ±6
23 ±3
143 ± 18
45 ±5
1
100
9.53 ±0.24
16.1 ±0.4
46.3 ± 1.3
48.6 ±0.7
75 ± 14
27 ±4
206 ± 13
148 ±37
1
100
8.69 ±0.21
16.0 ±0.3
45.5 ± 1.2
52.3 ±0.7
65 ±3
21 ±2
144 ± 18
48 ±6
,4-dioxane vapor concentration (ppm)
200
9.54 ±0.1 8
15.9 ±0.2
46.3 ±0.9
48.6 ±0.4
73 ± 10
27 ±4
192 ±9
118±33
400
9.59 ±0.26
16.1 ±0.3
46.3 ± 1.4
48.3 ±0.4
72 ±5
28 ± 1
190± 12
131 ±30
800
9.55 ±0.18
16.0 ±0.3
46.3 ± 1.1
48.5 ±0.6
72 ±3
27 ±2
187 ± 15
113±27
1,600
9.58 ±0.14
16.2 ±0.3
46.8 ±0.9
48.9 ±0.6
70 ±4
27 ±2
184± 12
106 ±24
3,200
9.57 ±0.37
16.4±0.4C
47.3 ± 1.7
49.4±0.5b
73 ±4
30 ±2
170± 11b
87 ± 22C
,4-dioxane vapor concentration (ppm)
200
8.73 ±0.25
16.3 ±0.4
45.8 ± 1.7
52.4 ±0.7
74 ± 14C
26 ± 10
137 ±9
42 ±4
400
8.88 ±0.21
16.2 ±0.4
46.5 ± 1.5
52.4 ±0.8
69 ±5
25 ±3
140± 15
47 ±8
800
8.68 ±0.69
16.2 ±0.6
45.4 ± 3.6
52.3 ± 0.6
68 ±6
24 ±4
141 ± 15
42 ±6
1,600
8.86 ±0.16
16.3 ±0.2
46.2 ±0.7
52.1 ±0.5
70 ±5
25 ±3
139± 11
39 ±7
3,200
9.15±0.12b
16.6±0.2C
47.5±0.6C
52.0 ±0.7
76±5b
30±3b
139± 18
42 ±7
"Data are presented for 10 sacrificed animals.
bp < 0.01 by Dunnett's test.
°p < 0.05 by Dunnett's test.
dData were reported for 9/10 female rats.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2008).
57
-------
Table 4-17 Incidence data of histopathological lesions in F344/DuCrj rats exposed
1,4-dioxane vapor by whole-body inhalation for 13 weeks
Males3
Effect13
Nuclear enlargement;
nasal respiratory epithelium
Nuclear enlargement;
nasal olfactory epithelium
Nuclear enlargement;
tracheal epithelium
Nuclear enlargement;
bronchial epithelium
Vacuolic change;
olfactory epithelium
Vacuolic change;
bronchial epithelium
to
1,4-dioxane vapor concentration (ppm)
0 (clean air)
0/10
0/10
0/10
0/10
0/10
0/10
100
7/1 Oc
(7, 1+)
0/10
0/10
0/10
1/10
(1,1+)
0/10
200
9/1 Oc
(9, 1 + )
5/1 Od
(5, 1 + )
0/10
0/10
3/10
(3, 1 + )
0/10
400
7/1 Oc
(7, 1+)
10/10C
(10,1 + )
0/10
0/10
6/1 Od
(6, 1+)
0/10
800
10/10C
(10,1+)
10/10C
(10,1+)
1/10
(1,1+)
0/10
10/10C
(10,1+)
4/10
(4, 1+)
1,600
10/10C
(10,2+)
10/10C
(10,2+)
10/10C
(10,1+)
9/1 Oc
(9, 1 + )
10/10C
(10, 1+)
6/1 Od
(6, 1 + )
3,200
10/10C
(10,2+)
10/10C
(10,2+)
10/10C
(10,1+)
10/10C
(10,1+)
9/1 Oc
(10,1+)
6/1 Od
(6, 1+)
Atrophy; olfactory epithelium8 - ------
Hepatocyte centrilobular
swelling
Hepatocyte single-cell necrosis
0/10
0/10
0/10
0/10
0/10
0/1
0/10
0/10
0/10
0/10
1/10
(1,1 + )
1/10
(1,1 + )
10/10C
(10,1+)
8/1 Oc
(8, 1+)
Hydropic change;
renal proximal tubule8
Females3
Effect"
Nuclear enlargement;
nasal respiratory epithelium
Nuclear enlargement;
nasal olfactory epithelium
Nuclear enlargement;
tracheal epithelium
Nuclear enlargement;
bronchial epithelium
Vacuolic change;
olfactory epithelium
Vacuolic change;
bronchial epithelium
Atrophy;
olfactory epithelium
Hepatocyte centrilobular
swelling
Hepatocyte single-cell necrosis
Hydropic change;
renal proximal tubule
1,4-dioxane vapor concentration (ppm)
0 (clean air)
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
100
5/1 Od
(5, 1+)
2/10
(2, 1+)
0/10
0/10
1/10
(1,1+)
0/10
0/10
0/10
0/10
0/10
200
9/1 Oc
(9, 1 + )
6/1 Od
(6, 1 + )
0/10
0/10
2/10
(2, 1 + )
0/10
2/10
(2, 1 + )
0/10
0/10
0/10
400
10/10C
(10, 1 + )
10/10C
(9, 1+;
1,2+)
0/10
0/10
3/10
(3, 1+)
1/10
(1,1+)
3/10
(3, 1+)
0/10
0/10
0/10
800
10/10C
(10,1+)
10/10C
(10,1+)
2/10
(2, 1+)
0/10
7/1 Oc
(7, 1+)
1/10
(1,1+)
5/1 Od
(5, 1+)
0/10
0/10
0/10
1,600
10/10C
(10,2+)
10/10C
(7, 1+;
3,2+)
7/1 Oc
(7, 1 + )
0/10
9/1 Oc
(9, 1 + )
3/10
(3, 1 + )
5/1 Od
(5, 1 + )
1/10
(1,1 + )
0/10
0/10
3,200
10/10C
(10,2+)
10/10C
(10,2+)
10/10C
(10,1+)
10/10C
(10,1+)
10/10C
(10,1+)
4/10
(4, 1+)
4/10
(4, 1+)
8/1 Oc
(8, 1+)
3/10
(3, 1+)
6/1 Od
(6, 1+)
"Data are presented for sacrificed animals.
bValues listed are the number of animals with the indicated lesion. Values in parentheses are the number of lesion-bearing animals
fora given grade of lesion severity. Severity key: 1+ = slight; and, 2+ = moderate.
°p<0.01 by x test.
dp < 0.05 by x2 test.
eData were not reported for male rats.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2008
58
-------
4.2.2.2. Chronic Inhalation Toxicity and Carcinogenicity
4.2.2.2.1. Torkelson et al.
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
(Torkelson et al.. 1974). 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 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 %2 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 on the
lack of significant effects on several endpoints, EPA identified the exposure concentration of 0.4 mg/L
(111 ppm) as a free standing NOAEL.
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 tumors were seen in
any rat and the presence or absence of nasal cavity tumors was not evaluated.
59
-------
4.2.2.2.2. Kasai et al.
Groups of male 6-week-old F344/DuCrj rats (50/group) weighing 120 ± 5g (mean ± SD) at the
beginning of the study were exposed via inhalation to nominal concentrations of 0 (clean air), 50, 250,
and 1,250 ppm (0, 180, 900, and 4,500 mg/m3, respectively) of vaporized 1,4-dioxane (>99%pure) for 6
hours/day, 5 days/week, for 104 weeks (2 years) in whole body inhalation chambers (Kasai et al.. 2009).
Each inhalation chamber housed male rats individually in stainless-steel wire hanging cages. The authors
stated female counterparts were not exposed given data illustrating the absence of induced mesotheliomas
following exposure to 1,4-dioxane in drinking water (Yamazaki et al.. 1994). During exposure, the
concentration of 1,4-dioxane vapor was determined every 15 minutes by gas chromatography and animals
received food and water ad libitum. In addition, during the 2-year exposure period, clinical signs and
mortality were recorded daily. BW and food intake were measured once weekly for the first 14 weeks of
exposure, and thereafter, every 4 weeks. At the end of the 2-year exposure period or at the time of an
animal's death during exposure, all organs were collected, weighed, and evaluated for macroscopic
lesions. Additional examinations were completed on rats sacrificed at the end of the 2-year exposure
period. Endpoints examined included: 1) measurement of hematological and clinical chemistry
parameters using blood collected from the abdominal aorta of rats following an overnight fasting at the
end of the 2-year exposure period; 2) measurement of urinary parameters using Ames reagent strips
during the last week of the exposure period; and 3) histopathological evaluations of organs and tissues
outlined in the OECD test guideline which included all tissues of the respiratory tract. For measured
hematological and clinical chemistry parameters, analyses included: red blood cell count, hemoglobin,
hematocrit, MCV, mean corpuscular hemoglobin (MCH), AST, ALT, ALP, and y-GTP. Organs and
tissues collected for histopathological examination were fixed in 10% neutral buffered formalin with the
exception of nasal cavity samples. Nasal tissue was trimmed transversely at three levels after
decalcification and fixation in a formic acid-formalin solution. The levels were demarcated at the
following points: at the posterior edge of the upper incisor teeth (level 1), at the incisive papilla (level 2),
and at the anterior edge of the upper molar teeth (level 3). All tissue samples were embedded in paraffin,
and then sectioned (at 5 (im thickness) and stained with hematoxylin and eosin (H&E). Dunnett's test, %2
test, and Fisher's exact test were used by study authors to determine statistical differences (p-value of
0.05) between 1,4-dioxane exposed and clean air exposed group data.
Deformity in the nose was the only clinical sign reported in this study. This deformity was seen at
exposure weeks 74 and 79 in one rat each, exposed to 250 ppm and 1,250 ppm of 1,4-dioxane,
respectively. Both of these rats did not survive the 2-year exposure with deaths caused by malignant nasal
tumors.
Growth rates and survival rates were analyzed. Growth rates were not significantly affected by
1,4-dioxane exposures, but a decreasing trend in growth was observed during the latter half of the 2-year
exposure period for all exposure doses (i.e., 50, 250, and 1,250 ppm). Survival rates were significantly
decreased following 91 weeks of exposure to 1,250 ppm of 1,4-dioxane. The authors attributed these
deaths to increased incidences of peritoneal mesotheliomas, but also noted that nasal tumors could have
been a contributing factor. Terminal survival rates were 37/50, 37/50, 29/50, and 25/50 for 0, 50, 250, and
1,250 ppm exposed groups, respectively.
60
-------
Exposure-related effects on final BWs, organ weights, and hematological and clinical chemistry
parameters were reported. Changes in these effects, as compared to control are outlined in Table 4-18 and
Table 4-19. Briefly, at 1,250 ppm terminal BWs were significantly decreased and relative liver and lung
weights were significantly increased. It is of note that the observed change in terminal body weight was
not an effect of food consumption, which was determined by the study authors to be unaltered. Altered
hematological and clinical chemistry parameters were also observed with significant changes at
1,250 ppm. Altered endpoints included decreased hemoglobin, MCV, and MCH, and increased AST,
ALT, ALP, and y-GTP (p < 0.01) levels. In addition, urine pH was significantly decreased in 1,250 ppm
exposed rats.
Histopathology findings of pre- and nonneoplastic lesions associated with 1,4-dioxane treatment
were seen in the nasal cavity, liver, and kidneys (Table 4-20). At the highest concentration of 1,250 ppm,
all pre- and nonneoplastic lesions were significantly increased, as compared to controls, with the
exception of clear and mixed cell foci in the liver. At the lowest concentration of 50 ppm, nuclear
enlargement of the respiratory epithelium was the most sensitive lesion observed in the nasal cavity.
Based on this finding, the study authors identified a LOAEL of 50 ppm in male rats. As noted earlier in
Section 4.2.1.1.3. nuclear enlargement as a specific morphologic diagnosis is not considered by EPA to be
an adverse effect of exposure to 1,4-dioxane.
Tumor development was observed in the nasal cavity (squamous cell carcinoma), liver
(hepatocellular adenoma and carcinoma), peritoneum (peritoneal mesothelioma), kidney (renal cell
carcinoma), mammary gland (fibroadenoma and adenoma), Zymbal gland (adenoma), and subcutaneous
tissue (subcutis fibroma). Tumor incidences with a dose-dependent, statistically significant positive trend
(Peto's test) included nasal squamous cell carcinoma, hepatocellular adenoma, peritoneal mesothelioma,
mammary gland fibroadenoma, and Zymbal gland adenoma. Renal cell carcinoma was also identified as
statistically significant with a positive dose-dependent trend; however, no tumor incidences were reported
at 50 and 250 ppm. At 1,250 ppm, significant increases in nasal squamous cell carcinoma, hepatocellular
adenoma, and peritoneal mesothelioma were observed. At 250 ppm, significant increases in peritoneum
mesothelioma and subcutis fibroma were observed. Table 4-21 presents a summary of tumor incidences
found in this study. Further characterizations of neoplasms revealed nasal squamous cell carcinoma
occurred at the dorsal area of the nose (levels 1-3) marked by keratinization and the progression of growth
into surrounding tissue. Peritoneal mesotheliomas were characterized by complex branching structures
originating from the mesothelium of the scrotal sac. Invasive growth into surrounding tissues was
occasionally observed for peritoneal mesotheliomas.
61
-------
Table 4-18 Terminal body and relative organ weights of F344/DuCrj male rats exposed to
1,4-dioxane vapor by whole-body inhalation for 2 years
Males
1,4-dioxane vapor concentration (ppm)
Number of animals examined
Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
0 (clean air)
37
383 ± 50
0.45 ±0.25
3.57 ±0.66
0.87 ±0.21
50
37
383 ± 53
0.49 ± 0.27
3.86 ± 1.05
0.93 ±0.32
250
29
376 ± 38
0.45 ±0.18
3.58 ± 0.52
0.81 ±0.13
1,250
25
359 ± 29b
0.46 ± 0.07a
4.53±0.71b
0.86 ±0.12
Bp < 0.01 by Dunnett's test.
bp < 0.05 by Dunnett's test.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2009).
Table 4-19 Hematology and clinical chemistry of F344/DuCrj male rats exposed to
1,4-dioxane vapor by whole-body inhalation for 2 years
Males
1,4-dioxane vapor concentration (ppm)
Number of animals examined
Red blood cell (106/|jL)
Hemoglobin (g/dL)
Hematocrit (%)
MCV (fL)
MCH (pg)
AST (IU/L)
ALT (IU/L)
ALP (IU/L)
y-GTP (IU/L)
Urinary pH
ap < 0.01 by Dunnett's test.
bp < 0.05 by Dunnett's test.
Source: Reprinted with permission of Informa
0 (clean air)
35
7.4 ± 1.8
12.5 ±3.5
38.6 ±8.7
52.4 ±5.7
16.9 ±2.2
67 ±31
37 ± 12
185 ±288
6±3
7.1 ±0.6
Healthcare; Kasai et al.
50
35
6.8± 1.8
12.0 ±3.1
36.9 ±7.9
55.6 ±8.7
17.8 ±2.4
95 ±99
42 ±21
166 ±85
8±5
7.1 ±0.6
(2009).
250
28
7.9± 1.0
13.4± 1.9
40.7 ±5.1
51.8±2.3
17.1 ± 1.2
95± 116
49 ±30
145 ±71
10±8
7.1 ±0.6
1,250
25
7.0 ± 1.8
10.9±2.8b
34.3 ± 7.6
49.4 ± 4.0b
15.5 ± 1.3a
98 ± 52a
72 ± 36a
212 ± 109a
40 ± 26a
6.6±0.4b
62
-------
Table 4-20 Incidence of pre-and nonneoplastic lesions in male F344/DuCrj rats exposed
to 1,4-dioxane vapor by whole-body inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
Effect
0 (clean air)
50
250
1,250
Nuclear enlargement; nasal respiratory epithelium
0/50
50/50a
48/50a
38/50a
Squamous cell metaplasia; nasal respiratory epithelium
0/50
0/50
7/50"
44/50a
Squamous cell hyperplasia; nasal respiratory epithelium
0/50
0/50
1/50
10/50a
Inflammation; nasal respiratory epithelium
13/50
9/50
7/50
39/50a
Nuclear enlargement; nasal olfactory epithelium
0/50
48/50a
48/50a
45/50a
Respiratory metaplasia; nasal olfactory epithelium
11/50
34/50a
49/50a
48/50a
Atrophy; nasal olfactory epithelium
0/50
40/50a
47/50a
48/50a
Inflammation; nasal olfactory epithelium
0/50
2/50
32/50a
34/50a
Hydropic change; lamina propria
0/50
2/50
36/50a
49/50a
Sclerosis; lamina propria
0/50
0/50
22/50a
40/50a
Proliferation; nasal gland
0/50
1/50
0/50
6/50b
Nuclear enlargement; liver centrilobular
0/50
0/50
1/50
30/50a
Necrosis; liver centrilobular
1/50
3/50
6/50
12/50a
Spongiosis hepatis; liver
7/50
6/50
13/50
19/50a
Clear cell foci; liver
15/50
17/50
20/50
23/50
Basophilic cell foci; liver
17/50
20/50
15/50
44/50a
Acidophilic cell foci; liver
5/50
10/50
12/50
25/50a
Mixed-cell foci; liver
5/50
3/50
4/50
14/50
Nuclear enlargement; kidney proximal tubule
0/50
1/50
20/50a
47/50a
Hydropic change; kidney proximal tubule
0/50
0/50
5/50
6/50a
ap < 0.01 by x2 test.
bp < 0.05 by x2 test.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. i
63
-------
Table 4-21 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane vapor by
whole-body inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
Effect
Nasal squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Mammary gland adenoma
Zymbal gland adenoma
Subcutis fibroma
0 (clean air)
0/50
1/50
0/50
0/50
2/50
1/50
0/50
0/50
1/50
50
0/50
2/50
0/50
0/50
4/50
2/50
0/50
0/50
4/50
250
1/50
3/50
1/50
0/50
14/50a
3/50
0/50
0/50
9/50a
1,250
6/50b'c
21/50a'c
2/50
4/50c
41/50a'c
5/50d
1/50
4/50c
5/50
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
°p < 0.01 by Peto's test for dose-related trend.
dp < 0.05 by Peto's test for dose-related trend.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2009)
4.2.3. Initiation/Promotion Studies
Bronaugh et al. (1982) reported more 1,4-dioxane absorption from occluded than unoccluded
surfaces. Due to the volatility of 1,4-dioxane, the unoccluded skin paint studies are unreliable; however,
all of the available skin paint initiation/promotion studies are summarized below.
4.2.3.1. Bulletal.
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 SENCARmice (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-O-tetradecanoylphorbol-13-acetate (TPA) (1.0 (ig in 0.2 mL of acetone) was applied to
the shaved back of mice 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.
64
-------
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 al.
1,4-Dioxane was evaluated for complete carcinogenicity and tumor promotion activity in mouse
skin (King etal.. 1973). 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 ug 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 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
65
-------
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.
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
GOT. 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 etal.
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 doses 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) (Giavini et al.,
1985). 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 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
66
-------
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 4-22. 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 (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 etal.. 1939; Schrenk and Yant. 1936; deNavasquez. 1935). Severe liver and
kidney degeneration and necrosis were often seen in acute studies (JBRC. 1998; David. 1964; Kesten et
al.. 1939; Laug etal.. 1939; Schrenk and Yant. 1936; deNavasauez. 1935). JBRC (1998) 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. Inhalation Toxicity
Acute and short-term toxicity studies (all routes) are summarized in Table 4-22. Mortality
occurred in many high-concentration studies (Pozzani etal.. 1959; Nelson. 1951; Wirth and Klimmer.
1936). Inhalation of 1,4-dioxane caused eye and nasal irritation, altered respiration, and pulmonary edema
and congestion (Yant etal.. 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;
Fairlev et al.. 1934).
67
-------
Table 4-22 Acute and short-term toxicity studies of 1,4-dioxane
Animal
Exposure
route
Test conditions
Results
Dose3
Reference
Oral studies
Rat (inbred strain
and gender
unspecified)
Rat (strain and
gender
unspecified)
F344/DuCrj rat
Female
Sprague Dawley
rat
Female Carworth
Farms-Nelson rat
Male Wistar rat,
guinea pig
Rat, mouse,
guinea pig
Rabbit
Rat, rabbit
Oral via
drinking
water
Oral via
drinking
water
Oral via
drinking
water
Gavage
Gavage
Gavage
Gavage
Gavage
Gavage
1-10 days of exposure
5-12 days of exposure
14-day exposure
0, 168, 840, 2550, or
4,200 mg/kg by
gavage,
21 and 4 hours prior to
sacrifice
Determination of a
single dose LD50
Single dose,
LD50 determination
Single dose; several
dose groups
Single gavage dose of
0,207, 1,034, or
2,068 mg/kg-day
Single dose; mortality
after 2 weeks
Ultrastructural changes
in the kidney,
degenerative nephrosis,
hyaline droplet
accumulation, crystal
formation in
mitochondria
Extensive degeneration
of the kidney, liver
damage, mortality in
8/10 animals by
12 days
Mortality, decreased
BWs, histopathological
lesions in the nasal
cavity, liver, kidney, and
brain
Increased ODC activity,
hepatic
CYP450 content, and
DMA single-strand
breaks
Lethality
Lethality
Clinical signs of CMS
depression, stomach
hemorrhage, kidney
enlargement, and liver
and kidney
degeneration
Clinical signs of CMS
depression, mortality at
2,068 mg/kg, renal
toxicity (polyuria
followed by anuria),
histopathological
changes in liver and
kidneys
Mortality and narcosis
11,000 mg/kg-day
(5%)
11,000 mg/kg-day
(5%)
2,500 mg/kg-day
(nuclear enlargement
of olfactory epithelial
cells),
>7,500 mg/kg-day for
all other effects
840 mg/kg (ODC
activity only)
LD50 = 6,400 mg/kg
(14,200 ppm)
LD50 (mg/kg):
rat - 7 190
guinea pig = 3,150
LD50 (mg/kg):
rat = 5,400
guinea pig = 4,030
1,034 mg/kg-day
3,160 mg/kg
David
(1964)
Kesten et
al. (1939)
JBRC
(1998)
Kitchin and
Brown
(1990)
Pozzani et
al. (1959)
Smyth et al.
(1941)
Laug et al.
(1939)
de
Navasquez
(1935)
Nelson
(1951)
68
-------
Table 4-22 (Continued): Acute and short-term toxicity studies of 1,4-dioxane
Animal
Crj:BDF1 mouse
Dog
Exposure
route
Oral via
drinking
water
Drinking
water
ingestion
Test conditions
14-day exposure
3-10 days of exposure
Results
Mortality, decreased
BWs, histopathological
lesions in the nasal
cavity, liver, kidney, and
brain
Clinical signs of CMS
depression, and liver
and kidney
degeneration
Dose3
10,800 mg/kg-day;
hepatocellular swelling
11,000 mg/kg-day
(5%)
Reference
JBRC
(1998)
Schrenk
and Yant
(1936)
Inhalation studies
Male CD1 rat
Rat
Female Carworth
Farms-Nelson rat
Mouse, cat
Guinea pig
Rabbit, guinea
pig, rat, mouse
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Vapor
inhalation
Serum enzymes
measured before and
after a single 4 hour
exposure
5 hours of exposure
Determination of a
4-hour inhalation LCso
8 hours/day for
17 days
8-Hour exposure to
0.1-3% by volume
3 hours exposure, for
5 days; 1.5 hour
exposure for 1 day
Increase in ALT, AST,
and OCT; no change in
G-6-Pase
Mortality and narcosis
Lethality
Paralysis and death
Eye and nasal irritation,
retching movements,
altered respiration,
narcosis, pulmonary
edema and congestion,
hyperemia of the brain
Degeneration and
necrosis in the kidney
and liver, vascular
congestion in the lungs
1,000 ppm
6,000 ppm
LC5o = 51.3 mg/L
8,400 ppm
0.5% by volume
10,000 ppm
Drew et al.
(1978)
Nelson
(1951)
Pozzani et
al. (1959)
Wirth and
Klimmer
(1936)
Yant et al.
(1930)
Fairley et
al.(1934)
Other routes
Male
COBS/Wistar rat
Rabbit, cat
Female
Sprague Dawley
rat
CBA/J mouse
Dermal
i.v.
injection
i.p.
injection
i.p.
injection
Nonoccluded
technique using
shaved areas of the
back and flank; single
application, 14-day
observation
Single injection of 0,
207, 1,034,
1,600 mg/kg-day
Single dose;
LDso values
determined 24 hours
and 14 days after
injection
Daily injection for
7 days, 0, 0.1, 1, 5,
and 10%
Negative; no effects
noted
Clinical signs of CMS
depression, narcosis at
1,034 mg/kg, mortality
at 1 ,600 mg/kg
Increased serum SDH
activity at 1/1 6th of the
LDso dose; no change
at higher or lower doses
Slightly lower
lymphocyte response to
mitogens
8,300 mg/kg
1,034 mg/kg-day
LDso (mg/kg):
94 hnnrc; - 4 848
14 days = 799
2,000 mg/kg-day
(10%)
Clark et al.
(1984)
de
Navasquez
(1935)
Lundberg et
al. (1986)
Thurman et
al. (1978)
aLowest effective dose for positive results/ highest dose tested for negative results.
ND = no data; OCT = ornithine carbamyl transferase; ODC = ornithine decarboxylase; SDH = sorbitol dehydrogenase
69
-------
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; Laugetal.. 1939; Schrenk and Yant. 1936; deNavasquez. 1935; Yantetal.. 1930).
The neurotoxicity of 1,4-dioxane was further investigated in several studies described below (Frantik et
al.. 1994; Kanadaetal. 1994; Goldberg et al.. 1964; Knoefel. 1935).
4.4.2.1. Frantik et al.
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) (Frantik etal.. 1994). 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 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 al.
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
70
-------
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. Kanada et al.
Kanada et al. evaluated the effect of oral exposure to 1,4-dioxane on the regional neurochemistry
of the rat brain (Kanada etal.. 1994). 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.
4.4.2.4. Knoefel
The narcotic potency of 1,4-dioxane was evaluated following i.p. injection in rats and gavage
administration in rabbits (Knoefel. 1935). 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.
71
-------
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 4-23 and Table 4-24 for in vitro and
in vivo tests, respectively. 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 half of the 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. It is important to note that three of the negative studies reported using closed systems to
control for evaporation of the test substance (McGregor et al.. 1991; Zimmermann et al.. 1985; Nestmann
etal.. 1984).
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 Havashi. 1998; Hellmer and Bolcsfoldi. 1992; Kwanetal. 1990; Khudolev et al.. 1987;
Nestmann et al.. 1984; Haworth et al.. 1983; Stottetal. 1981) (Table 4-23). 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 etal.. 1985; Zimmermann et al.. 1985). In the presence of toxicity, positive results were reported
for meiotic nondisjunction in Drosophila (Tvlunoz 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 formation in Chinese hamster ovary (CHO)
cells (Morita and Havashi. 1998; Galloway et al.. 1987). Results were negative in one assay for sister
chromatid exchange (SCE) in CHO (TVIorita and Havashi. 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 (Sinaet al..
1983) and produced a positive genotoxic response in a cell transformation assay with BALB/3T3 cells
also in the presence of toxicity (Sheuetal.. 1988).
1,4-Dioxane was not genotoxic in the majority of available in vivo mammalian assays
(Table 4-24). 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,
72
-------
CBA, and C57BL6 mice (McFee et al.. 1994; Mirkova. 1994; Tinwell and Ashbv. 1994) and
micronucleus formation in peripheral blood of CD1 mice (Morita and Havashi. 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., non-significant 1.6-fold increase over control) with the dose-related
positive findings reported by Mirkov (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 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 erythrocytes
(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.
73
-------
1,4-Dioxane did not affect in vitro or in vivo DNA repair in hepatocytes or in vivo DNA repair in
the nasal cavity (Golds-worthy et al.. 1991; Stottet al.. 1981). but increased hepatocyte DNA synthesis
indicative of cell proliferation in several in vivo studies (Mivagawa 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.. 1977c).
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 etal.. 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.
Gene expression profiling in cultured human hepatoma HepG2 cells was performed using DNA
microarrays to discriminate between genotoxic and other carcinogens (van Delft et al.. 2004). Van Delft
et al. (2004) examined this method using a training set of 16 treatments (nine genotoxins and seven
nongenotoxins) and a validation set (three and three), with discrimination models based on Pearson
correlation analyses for the 20 most discriminating genes. As reported by the authors (van Delft et al..
2004). the gene expression profile for 1,4-dioxane indicated a classification of this chemical as a
"nongenotoxic" carcinogen, and thus, 1,4-dioxane was included in the training set as a "nongenotoxic"
carcinogen. The accuracy for carcinogen classification using this method ranged from 33 to 100%,
depending on which chemical data sets and gene expression signals were included in the analysis.
74
-------
Table 4-23 Genotoxicity studies of 1,4-dioxane; in vitro
Test system Endpoint
Results3
Without With
Test conditions activation activation
Doseb
Source
Prokaryotic organisms in vitro
S. typhimurium Reverse
strains TA98, mutation
TA100, TA1535,
TA1537
S. typhimurium Reverse
strains TA98, mutation
TA100, TA1530,
TA1535,
TA1537
S. typhimurium Reverse
strains TA98, mutation
TA100, TA1535,
TA1537
S. typhimurium Reverse
strains TA1 00, mutation
TA1535
S. typhimurium Reverse
strains TA98, mutation
TA100, TA1535,
TA1537,
TA1538
£. co//K-12 DMA repair
uvrB/recA
£. co// Reverse
WP2/WP2uvrA mutation
P. phosphoreum Mutagenicity,
M169 DMA damage
Plate incorporation assay - -
Plate incorporation assay - -
Plate incorporation and
preincubation assays
Preincubation assay - -
Plate incorporation assay
Host mediated assay - -
Plate incorporation and
preincubation assays
Mutatox assay - ND
10,000
ug/plate
ND
5,000
ug/plate
103 mg
103 mg
1,150
mmol/L
5,000
ug/plate
NDS
Haworth et
al. (1983)
Khudoley et
al. (1987)
Morita and
Hayashi
(1998)
Nestmann et
al. (1984)
Stott et al.
(1981)
Hellmer and
Bolcsfoldi
(1992)
Morita and
Hayashi
(1998)
Kwan et al.
(1990)
Nonmammalian eukaryotic organisms in vitro
S. cerevisiae Aneuploidy
D61.M
D. melanogaster Meiotic
nondisjunction
D. melanogaster Sex-linked
recessive lethal
test
Standard 16-hour -T ND
incubation or
cold-interruption regimen
Oocytes were obtained for +TC NDd
evaluation 24 and
48 hours after mating
Exposure by feeding and - NDd
injection
4.75%
2% in sucrose
media
35,000 ppm in
feed,
7 days or
50,000 ppm
(5% in water)
by injection
Zimmerman
et al. (1985)
Munoz and
Barnett
(2002)
Yoon et al.
(1985)
75
-------
Table 4-23 (Continued): Genotoxicity studies of 1,4-dioxane; in vitro
Results3
Without With
Test system
Mammalian cells
Rat hepatocytes
Primary
hepatocyte
culture from
male F344 rats
L5178Y mouse
lymphoma cells
L5178Y mouse
lymphoma cells
BALB/3T3 cells
Endpoint
in vitro
DMA damage;
single-strand
breaks
measured by
alkaline elution
DMA repair
Forward
mutation assay
Forward
mutation assay
Cell
transformation
Test conditions
3-Hour exposure to
isolated primary
hepatocytes
Autoradiography
Thymidine kinase
mutagenicity assay
(trifluorothymidine
resistance)
Thymidine kinase
mutagenicity assay
(trifluorothymidine
resistance)
48-Hour exposure
followed by 4 weeks
incubation; 13 day
exposure followed by
2.5 weeks incubation
activation activation Dose
+Te NDd 0.3 mM
NDd 1 mM
5,000
ug/mL
-T 5,000
ug/mL
+Tf NDd 0.5
mg/mL
Source
Sina et al.
(1983)
Goldsworthy
et al. (1991)
McGregor et
al. (1991)
Morita and
Hayashi
(1998)
Sheu et al.
(1988)
CHO cells
CHO cells
CHO cells
CHO cells
SCE BrdU was added 2 hours ±g
after 1,4-dioxane addition;
chemical treatment was
2 hours with S9 and
25 hours without S9
Chromosomal Cells were harvested 8- -
aberration 12 hours or 18-26 hours
after treatment (time of
first mitosis)
SCE 3 hour pulse treatment;
followed by continuous
treatment of BrdU for
23 or 26 hours
Chromosomal 5 hour pulse treatment,
aberration 20 hour pulse and
continuous treatments, or
44 hour continuous
treatment; cells were
harvested 20 or 44 hours
following exposure
10,520
ug/mL
10,520
ug/mL
5,000
ug/mL
5,000
ug/mL
Galloway et
al. (1987)
Galloway et
al. (1987)
Morita and
Hayashi
(1998)
Morita and
Hayashi
(1998)
76
-------
Table 4-23 (Continued): Genotoxicity studies of 1,4-dioxane; in vitro
Test system Endpoint
Test conditions
Results3
Without With
activation activation
Dose0
Source
CHO cells
Calf thymus
DMA
Micronucleus
formation
Covalent
binding to DMA
5 hour pulse treatment or
44 hour continuous
treatment; cells were
harvested 42 hours
following exposure
Incubation with
microsomes from
3-methylcholanthrene
treated rats
5,000
ug/mL
0.04 pmol/mg
DMA (bound)
Morita and
Hayashi
(1998)
Woo et al.
(1977c)
a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity, ND = no data. Endogenous
metabolic activation is not applicable for in vivo studies.
Yowest effective dose for positive results/highest dose tested for negative results; ND = no data.
CA 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.
dExogenous 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. DMA
damage was observed at 0.3, 3, 10, and 30 mM; however, no dose-response trend was observed for the
extent of DMA damage (severity score related to the elution rate).
Vorthe 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).
gThe highest concentration tested (10,520 ug/L) produced a 27% increase in the number of SCE/cell in the
absence of S9 mix. No effect was seen at lower doses (1,050 and 3,500 ug/L) in the absence of S9 mix
or at any concentration level (1,050, 3,500, 10,500 ug/L) tested in the presence of S9.
77
-------
Table 4-24 Genotoxicity studies of 1,4-dioxane; mammalian in vivo
Test system Endpoint
Female DMA damage;
Sprague Dawley single-strand breaks
Rat measured by
alkaline elution
Male DMA alkylation in
Sprague Dawley hepatocytes
Rat
Male Micronucleus
B6C3F-: formation in bone
Mouse marrow
Male and Micronucleus
female formation in bone
C57BL6 marrow
Mouse;
Male BALB/c
Mouse
Male Micronucleus
CD1 formation in
Mouse peripheral blood
Male Micronucleus
CD1 formation in
Mouse hepatocytes
Male Micronucleus
CD1 formation in
Mouse peripheral blood
Male Micronucleus
CBA and formation in bone
C57BL6 marrow
Mouse
Male Micronuclei
CD1 formation in bone
Mouse marrow
Male Micronuclei
CD1 formation in
Mouse hepatocytes
Male DMA repair in
Sprague Dawley hepatocytes
Rat
Male DMA repair in
F344 hepatocytes
Rat (autoradiography)
Test Conditions
Two gavage doses given
21 and 4 hours prior to
sacrifice
Gavage; DMA isolation and
HPLC analysis 4 hours
after dosing
i.p. injection; analysis of
polychromatic erythrocytes
24 or 48 hours after dosing
Gavage; analysis of
polychromatic erythrocytes
24 or 48 hours after dosing
Two i.p. injections (1/day);
micronucleated
reticulocytes measured
24, 48, and 72 hours after
the 2nd dose
Gavage, partial
hepatectomy 24 hours after
dosing, hepatocytes
analyzed 5 days after
hepatectomy
Gavage, partial
hepatectomy 24 hours after
dosing, peripheral blood
obtained from tail vein
24 hours after hepatectomy
Gavage; analysis of
polychromatic erythrocytes
from specimens prepared
24 hours after dosing
Gavage; analysis for
micronucleated
erythrocytes 24 hours after
dosing
Gavage; analysis for
micronuclei 24 hours after
dosing
Drinking water; thymidine
incorporation with
hydroxyurea to repress
normal DMA synthesis
Gavage and drinking water
exposure; thymidine
incorporation
Results3 Doseb Source
+c 2,550 Kitchin and
mg/kg Brown (1990)
1,000 Stottetal.
mg/kg (1981)
- Single dose of McFee et al.
4,000 mg/kg; (1994)
3 daily doses of
2,000
(C57BL6)d 90° mg/kg
^ ; (C57BL6); Mirkova (1994)
(BALB/c) 5'000 mg/kg
(bALb/c) (BALB/C)
3,200 Morita (1994)
mg/kg
+e 2,000 Morita and
mg/kg Hayashi (1998)
3,000 Morita and
mg/kg Hayashi (1998)
3,600 Tinwell and
mg/kg Ash by (1994)
+f 1,500 mg/kg-day Roy et al.
for 5 days (2005)
+g 2,500 mg/kg-day Roy etal. (2005)
for 5 days
1,000 mg/kg-day Stottetal.
for 1 1 weeks (1981)
1,000 mg/kg for Goldsworthy et
2 or 12 hours; al. (1991)
1,500 mg/kg-day
for 2 weeks or
3,000 mg/kg-day
for 1 week
78
-------
Table 4-24 (Continued): Genotoxicity studies of 1,4-dioxane; mammalian in vivo
Test system
Male
F344
Rat
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Male
F344
Rat
Male
F344
Rat
Male
Endpoint
DMA repair in nasal
epithelial cells from
the nasoturbinate or
maxilloturbinate
Replicative DMA
synthesis (i.e., cell
proliferation) in
hepatocytes
Replicative DMA
synthesis (i.e., cell
proliferation) in nasal
epithelial cells
RNA synthesis;
inhibition of RNA
polymerase A and B
DMA synthesis in
hepatocytes
DMA synthesis in
hepatocytes
DMA synthesis in
Test Conditions
Gavage and drinking water
exposure; thymidine
incorporation
Gavage and drinking water
exposure; thymidine
incorporation
Drinking water exposure;
thymidine incorporation
i.v. injection; activity
measured in isolated
hepatocytes
Gavage; thymidine and
BrdU incorporation
Thymidine incorporation
Drinking water; thymidine
Results3 Doseb
1,500 mg/kg-day
for 8 days +
1,000 mg/kg
gavage dose
12 hours prior to
sacrifice
+h 1,000 mg/kg for
(1_2-week 24 or 48 hours;
exposure) 1,500 mg/kg-day
for 1 or 2 weeks
1,500 mg/kg-day
for 2 weeks
+' 10
mg/rat
+J 1,000
mg/kg
±k 2,000
mg/kg
+' 1,000 mg/kg-day
Source
Goldsworthy et
al. (1991)
Goldsworthy et
al. (1991)
Goldsworthy et
al. (1991)
Kurl et al.
(1981)
Miyagawa et al.
(1999)
Uno et al.
(1994)
Stott et al.
Sprague Dawley hepatocytes incorporation
Rat
for 11 weeks (1981)
a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity, ND = no data. Endogenous metabolic activation
is not applicable for in vivo studies.
bLowest effective dose for positive results/highest dose tested for negative results; ND = no data.
°Rats 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 DMA eluted was observed for doses of 2,550 and 4,200 mg/kg, respectively. Alkaline elution of DMA
was not significantly different from control in the two lowest dose groups (168 and 840 mg/kg).
dA 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.
eA 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.
'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).
9A 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.
hNo 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.
'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).
kReplicative 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%.
'Replicative 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.
79
-------
4.5.2. Mechanistic Studies
4.5.2.1. Free Radical Generation
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/m3 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 peroxidase activity, and protein
peroxidation. Inhalation of 100 mg/m3 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 of Metabolism
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.
80
-------
4.5.2.3. Mechanisms of Tumor Induction
Several studies have been performed to evaluate potential mechanisms for the carcinogenicity of
1,4-dioxane (Goldsworfhy et al.. 1991: Kitchin and Brown. 1990: Stottetal.. 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 (Stottetal.. 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 by acute gavage
exposure or ingestion of 1,4-dioxane in the drinking water for a 1- or 2-week period. Hepatocyte cell
81
-------
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, kidney, and nasal toxicity were the primary noncancer health effects associated with
exposure to 1,4-dioxane. In humans, 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). In laboratory animals, following oral and inhalation exposure
82
-------
to 1,4-dioxane, liver and kidney degeneration and necrosis were observed (JBRC. 1998; Drew et al..
1978: David. 1964: Kestenetal.. 1939: Laugetal.. 1939: Schrenk and Yant. 1936: deNavasauez. 1935:
Fairley etal.. 1934). in addition to changes in the nasal epithelium (Kano et al.. 2009: Kasai et al.. 2009:
Kano et al.. 2008: Kasai et al.. 2008: JBRC. 1998). The results of subchronic and chronic studies are
discussed below.
4.6.1. Oral
Table 4-25 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: Kocibaetal.. 1974: Argus etal.. 1965: Fairlev et al.. 1934). Renal cell
degeneration generally began with cloudy swelling of cells in the cortex (Fairlev et al.. 1934). Nuclear
enlargement of proximal tubule cells was observed at doses below those producing renal necrosis (Kano
et al., 2008: JBRC. 1998): however, its relationship to the typical pathological progression from initiated
cell to tumor is unclear. 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..
1973: Argus etal.. 1965).
83
-------
Table 4-25 Oral toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration
NOAEL LOAEL
(mg/kg-day) (mg/kg-day) Effect
Reference
Subchronic studies
Rat and Mouse Rats NA
(6/species); 0 or 1,900 mg/kg-day;
unknown strain Mice
0 or 3, 300 mg/kg-day
for 67 days
Male Rats 10
Sprague Dawley 0, 10, or 1,000 mg/kg-day
Rat for 1 1 weeks
(4-6/group)
F344/DuCrj Rat Rats 52
(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
Crj:BDF1 Mouse Mice 170
(10/sex/group) Males 0, 86, 231, 585,
1,900 rats Renal cortical degeneration
3,300 mice and necrosis, hemorrhage;
hepatocellular degeneration
1,000 Minimal centrilobular
hepatocyte swelling;
increased DMA synthesis
126 Nuclear enlargement of
nasal respiratory epithelium;
hepatocyte swelling
387 Nuclear enlargement of
bronchial epithelium
Fairley et al.
(1934)
Stott et al.
(1981)
Kano et al.
(2008)
Kano et al.
(2008)
882, or 1,570 mg/kg-day;
Females 0, 170, 387, 898,
1,620, or
2,669 mg/kg-day
for 13 weeks
Chronic studies
Male Rats NA
Wistar 0 or 640 mg/kg-day
Rat (26 treated, for 63 weeks
9 controls)
Male Rats NA
Sprague Dawley 0, 430, 574, 803, or
Rat (30/group) 1,032 mg/kg-day
for 13 months
Sherman Rat Rats 9.6
(60/sex/dose Males 0, 9.6, 94, or
group) 1,015 mg/kg-day;
Females 0, 19, 148, or
1,599 mg/kg-day
for 2 years
Osborne-Mendel Rats NA
Rat (35/sex/dose Males 0, 240, or
level) 530 mg/kg-day;
Females 0, 350, or
640 mg/kg-day
for 110 weeks
B6C3F-I Mouse Mice NA
(50/sex/dose Males 0, 720, or
level) 830 mg/kg-day;
Females 0, 380, or
860 mg/kg-day
for 90 weeks
640 Hepatocytes with enlarged
hyperchromic nuclei;
glomerulonephritis
430 Hepatocytomegaly;
glomerulonephritis
94 Degeneration and necrosis
of renal tubular cells and
hepatocytes
240 Pneumonia, gastric ulcers,
and cortical tubular
degeneration in the kidney
380 Pneumonia and rhinitis
Argus et al.
(1965)
Argus et al.
(1973)
Kociba et al.
(1974)
NCI (1978)
NCI (1978)
84
-------
Table 4-25 (Continued): Oral toxicity studies (noncancer effects) for 1,4-dioxane
Species Dose/duration
F344/DuCrj Rat Rats
(50/sex/dose Males 0, 11, 55, or
level) 274 mg/kg-day;
Females 0, 18, 83, or
429 mg/kg-day for 2 years
F344/DuCrj Rat Rats
(50/sex/dose Males 0, 11, 55, or
level) 274 mg/kg-day;
Females 0, 18, 83, or
429 mg/kg-day for 2 years
F344/DuCrj Rat Rats
(50/sex/dose Males 0, 11, 55, or
level) 274 mg/kg-day;
Females 0, 18, 83, or
429 mg/kg-day for 2 years
Crj:BDF1 Mouse Mice
(50/sex/dose Males 0, 49, 191 or
level) 677 mg/kg-day;
Females 0, 66, 278, or
964 mg/kg-day for 2 years
Crj:BDF1 Mouse Mice
(50/sex/dose Males 0, 49, 191 or
level) 677 mg/kg-day;
Females 0, 66, 278, or
964 mg/kg-day for 2 years
NOAEL LOAEL
(mg/kg-day) (mg/kg-day) Effect
55 274 Atrophy of nasal olfactory
epithelium; nasal adhesion
and inflammation
11 55 Mixed cell liver foci
55 274 Increases in serum liver
enzymes (GOT, GPT,
LDH, and ALP)
66 278 Nasal inflammation
49 191 Increases in serum liver
enzymes (GOT, GPT,
LDH, and ALP)
Reference
JBRC (1998):
Kano et al.
(2009)
JBRC (1998);
Kano et al.
(2009)
JBRC (1998):
Kano et al.
(2009)
JBRC (1998):
Kano et al.
(2009)
JBRC (1998):
Kano et al.
(2009)
Developmental studies
Sprague Dawley Rats
Rat Pregnant dams 0, 250,
(18-20/group) 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)
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: Kocibaetal. 1974: Argus etal.. 1973: Argus etal.. 1965: Fairlev et al.. 1934V Hepatocellular
degeneration and necrosis were seen at high doses in a subchronic study (1,900 mg/kg-day in rats)
(Fairlev etal.. 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 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. 1998). 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 (Bannasch. 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
85
-------
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 (Bannasch. 2003; Stroebel etal. 1995). In the case of the JBRC (1998) 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. 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 (1998) 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.
Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice). The
JBRC (1998) 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. 1998; Kocibaetal.. 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. 1998); 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, EPA does not consider it to be 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. 1998).
4.6.2. Inhalation
Two subchronic (Kasai et al., 2008; Fairley et al.. 1934) and two chronic inhalation studies (Kasai
et al.. 2009; Torkelson et al.. 1974) were identified. Nasal, liver, and kidney toxicity were the primary
noncancer health effects of inhalation exposure to 1,4-dioxane in rodents. Table 4-26 presents a summary
86
-------
of the noncancer results for the subchronic and chronic inhalation studies of 1,4-dioxane toxicity in
laboratory animals.
Of the inhalation studies, nasal tissue was only evaluated in rat studies conducted by Kasai et al.
(2009; 2008). Adverse effects in nasal tissue were observed frequently in these studies, and statistically
significant changes were noted at vapor concentrations as low as 50 ppm. Nasal effects included
deformity of the nose and histopathological changes characterized by enlarged epithelial nuclei
(respiratory epithelium, olfactory epithelium, trachea, and bronchus), atrophy (olfactory epithelium),
vacuolic change (olfactory epithelium and bronchial epithelium), squamous cell metaplasia and
hyperplasia (respiratory epithelium), respiratory metaplasia (olfactory epithelium), inflammation
(respiratory and olfactory epithelium), hydropic change (lamina propria), and sclerosis (lamina propria).
In both studies, a concentration-dependent, statistically significant incidence of enlarged nuclei of the
respiratory epithelium were reported by the study authors; however, nuclear enlargement as a specific
morphologic diagnosis is not considered by EPA to be an adverse effect of exposure to 1,4-dioxane.
At high doses, liver damage was characterized by hepatocellular degeneration which varied from
swelling (Kasai et al.. 2008; Fairlev et al.. 1934) to necrosis (Kasai et al.. 2009; Kasai et al.. 2008; Fairlev
et al.. 1934). spongiosis hepatis (Kasai et al.. 2009). nuclear enlargement of centrilobular cells (Kasai et
al.. 2009) and basophilic and acidophilic cell foci (Kasai et al.. 2009). At concentrations ranging from
200 to 3,200 ppm, altered liver enzymes (i.e., AST, ALT, ALP, and y-GTP), increased liver weights, and
induction of GST-P were also observed (Kasai et al.. 2009; Kasai et al.. 2008). Changes in the activity of
serum enzymes were mostly observed in exposed rat groups at high 1,4-dioxane concentrations (Kasai et
al.. 2009; Kasai et al.. 2008). Induction of GST-P positive hepatocytes was observed in female rats at
1,600 ppm and male and female rats at 3,200 ppm following 13 weeks of exposure (Kasai et al.. 2008).
GST-P is considered a good enzymatic marker for early detection of chemical hepatocarcinogenesis
(Sato. 1989). GST-P positive altered cell foci are commonly considered preneoplastic changes and would
not be considered evidence of noncancer toxicity when observed in conjunction with tumor formation
(Bannasch et al.. 1982). Although, GST-P positive liver foci were not observed in the 2-year bioassay
(Kasai et al.. 2009). the focally and proliferating GST-P positive hepatocytes noted in the 13- week study
suggest eventual progression to hepatocellular tumors after 2 years of exposure and therefore would not
be considered a potential noncancer effect.
The lowest vapor concentration reported to produce liver lesions after 2 years of exposure was
1,250 ppm. The lesions were characterized by necrosis of centrilobular cells, spongiosis hepatis, and
nuclear enlargement in the Kasai et al. (2009) study. However, as previously stated, it was not considered
to be an adverse effect.
Kidney effects were reported less frequently than other effects in these inhalation studies and
were generally observed at higher exposure concentrations than nasal and liver effects. Kidney damage
was described as patchy degeneration of cortical tubules with vascular congestion and hemorrhage
(Fairlev etal. 1934). hydropic change of proximal tubules (Kasai et al.. 2009; Kasai et al.. 2008). and as
nuclear enlargement in proximal tubule cells (Kasai et al.. 2009). Changes in serum chemistry and
urinalysis indices were also noted as evidence of renal damage. In a 13-week inhalation study of male and
87
-------
female rats (Kasai et al., 2008) kidney toxicity was only observed in female rats exposed to 3,200 ppm of
1,4-dioxane (i.e., hydropic change in the renal proximal tubules), which suggests a possible greater
susceptibility of female rats to renal damage following inhalation of 1,4-dioxane.
Other noted noncancer effects in laboratory animals included acute vascular congestion of the
lungs (Fairlev et al.. 1934): changes in relative lung weights (Kasai et al.. 2008): and decrease in body
weight gain (Kasai et al., 2009: Kasai et al., 2008). Following a 13-week exposure, higher 1,4-dioxane
plasma levels were found in female rats than male rats (Kasai et al.. 2008). 1,4-Dioxane was measured in
plasma along with systemic effects following subchronic inhalation exposure to 1,4-dioxane in rats (Kasai
et al.. 2008).
Table 4-26 Inhalation toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration
NOAEL
(ppm)
LOAEL
(ppm) Effect
Reference
Subchronic studies
Rat, mouse, rabbit,
and guinea pig
(3-6/species/group);
unknown strains
F344/DuCrj rat
(10/sex/group)
0, 1,000,2,000, 5,000, or
10,000 ppm for 7 days.
Days 1-5, two 1.5 hour
exposures; day 6, one
1.5 hour exposure; and
day 7, no exposure
0, 100,200,400, 800,
1,600, 3,200, or
6,400 ppm 6 hours/day
5 days/wk, for 13 wk
NA
NA
1,000 Renal cortical
degeneration and
hemorrhage;
hepatocellular
degeneration and necrosis
100 Respiratory epithelium:
nuclear enlargement of
epithelial cells
Fairley et al.
(1934)
Kasai et al.
(2008)
Chronic studies
Wistar rat (288/sex)
F344/DuCrj male
rat
(50/group)
111 ppm for 7hours/day,
5 days/wk, for 2 years
0, 50, 250, or 1,250 ppm
for 6 hours/day,
5 days/wk for 2 years
111
(free
standing)
N/A
NA No significant effects were
observed on BWs,
survival, organ weights,
hematology, clinical
chemistry, or
histopathology
50 Respiratory epithelium:
nuclear enlargement of
epithelial cells, atrophy,
and metaplasia
Torkelson et
al. (1974)
Kasai et al.
(2009)
4.6.2.1. 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 a single 50 ppm inhalation exposure to 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). Using the Young et al. Q978a, b) rat kinetic model, the metabolism of 1,4-dioxane in rats that
were exposed to 400, 800, 1,600, and 3,200 ppm via inhalation for 13 weeks could not be accurately
predicted due to a lack of knowledge on needed model parameters and biological processes (see Section
3.5.3 and Appendix B). It appears, following prolonged inhalation exposure to 1,4-dioxane at
concentrations up to 3,200 ppm, that metabolism is induced (Appendix 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 (Wooetal.. 1978. 1977b). 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. High doses of
1,4-dioxane were 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 CYP2B1/2 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 oxidative 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.. 1977c). 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 the
parent compound may be responsible for 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 toxicity did not occur unless
clearance pathways were saturated and elimination of 1,4-dioxane from the blood was reduced. A
dose-dependent increase of 1,4-dioxane concentration in the blood was seen, which correlated to the
observed dose-dependent increase in incidences of nasal, liver, and kidney toxicities (Kasai et al.. 2008).
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. Thus, 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.
89
-------
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 is "likely
to be carcinogenic to humans" based on evidence of carcinogenicity in several 2-year bioassays
conducted in four strains of rats, two strains of mice, and in guinea pigs (Kano et al.. 2009; Kasai et al..
2009; JBRC. 1998; Yamazaki et al.. 1994; NCI. 1978; Kocibaetal. 1974; Argus etal.. 1973; Hoch-
Ligeti and Argus. 1970; Hoch-Ligeti et al.. 1970; Argus et al.. 1965). Tissue sites where tumors have been
observed in these laboratory animals due to exposure to 1,4-dioxane include, peritoneum (Kano et al..
2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). mammary gland (Kano et al.. 2009: Kasai
et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). liver (Kano et al.. 2009: Kasai et al.. 2009). kidney
(Kasai et al.. 2009). Zymbal gland (Kasai et al.. 2009). subcutaneous (Kasai et al.. 2009). nasal tissue
(Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kocibaetal.. 1974:
Argus etal.. 1973: Hoch-Ligeti et al.. 1970). and lung (Hoch-Ligeti and Argus. 1970). 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).
A MOA 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 etal.. 1973). Dose-response and temporal data support the
occurrence of cell proliferation prior to the development of liver tumors (JBRC. 1998: 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. 1998:
Kocibaetal.. 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 4-21) from cytotoxicity and cell proliferation to eventual liver tumor formation are not
available. Also, Kociba et al. (1974) reported renal degeneration, necrosis, and regenerative proliferation
in exposed rats, but no increase in the incidence of kidney tumors, which does not support a
cytotoxicity/cell proliferation MOA.
For nasal tumors, there is a hypothesized MOA that includes metabolic induction, cytotoxicity,
and regenerative cell proliferation (Kasai et al.. 2009). The induction of CYP450 has some support from
data illustrating that following acute oral administration of 1,4-dioxane by gavage or drinking water,
CYP2E1 was inducible in nasal mucosa (Nannelli et al.. 2005). CYP2E1 mRNA was increased
approximately two- to threefold in nasal mucosa (and in the kidney, see Section 3.3) in the Nannelli et al.
(2005) study. While cell proliferation was observed following 1,4-dioxane exposure in both a 2-year
inhalation study in male rats (1,250 ppm) (Kasai et al.. 2009) and a 2-year drinking water study in male
(274 mg/kg-day) and female rats (429 mg/kg-day), no evidence of cytotoxicity in the nasal cavity was
90
-------
observed (Kasai et al.. 2009); therefore, cytotoxicity, as a key event, is not supported. Nasal lesions,
including inflammation, hyperplasia, and metaplasia, were frequently seen in inhalation studies conducted
by the NTP with no evidence of nasal carcinogenicity (Haseman and Hailey. 1997; Ward etal.. 1993).
Following a 13-week inhalation study in rats, a concentration-dependent increase of 1,4-dioxane in the
blood was observed (Kasai et al.. 2008). Studies have shown that water-soluble, gaseous irritants cause
nasal injuries such as squamous cell carcinomas (Morgan et al.. 1986). Similarly, 1,4-dioxane, which has
been reported as a miscible compound (Hawley and Lewis. 2001). also caused nasal injuries that were
concentration-dependent, including nasal tumors (Kasai et al.. 2009). Additionally, it has been suggested
that in vivo genotoxicity may contribute to the carcinogenic MOA for 1,4-dioxane (Kasai et al., 2009)
(see Section 4.7.3.6 for further discussion). Collectively, these data are insufficient to support the
hypothesized MO As.
There are no data available regarding any hypothesized MOA by which 1,4-dioxane produces
kidney, lung, peritoneal (mesotheliomas), mammary gland, Zymbal gland, and subcutis tumors.
U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) indicate that for
tumors occurring at a site other than the initial point of contact, the weight of evidence for carcinogenic
potential may apply to all routes of exposure that have not been adequately tested at sufficient doses. An
exception occurs when there is convincing information (e.g., toxicokinetic data) that absorption does not
occur by other routes. Information available on the carcinogenic effects of 1,4-dioxane via the oral route
demonstrates that tumors occur in tissues remote from the site of absorption. In addition, information on
the carcinogenic effects of 1,4-dioxane via the inhalation route in animals also demonstrates that tumors
occur at tissue sites distant from the portal of entry. Information on the carcinogenic effects of
1,4-dioxane via the inhalation and dermal routes in humans and via the dermal route in animals is absent.
If sufficient external dose is applied, it is assumed that an internal dose will be achieved regardless of the
route of exposure. Therefore, based on the observance of systemic tumors following oral and inhalation
exposure, 1,4-dioxane is "likely to be carcinogenic to humans" by all routes of exposure.
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 was limited and number of reported cases was small (Buffler et al.. 1978; Thiess et al.. 1976).
Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
pigs (Kano et al.. 2009; Kasai et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994; NCI. 1978; Kociba et al..
1974; Torkelson et al.. 1974; Argus etal.. 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 etal.. 1965). male guinea pigs (Hoch-Ligeti and Argus. 1970). male Sprague Dawley rats
(Argus etal.. 1973; Hoch-Ligeti et al.. 1970). male and female Sherman rats (Kociba etal.. 1974). female
Osborne-Mendel rats (NCI. 1978). male and female F344/DuCrj rats (Kano et al.. 2009; JBRC. 1998;
Yamazaki et al.. 1994). male and female B6C3Fi mice (NCI. 1978). and male and female Crj:BDFl mice
(Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994); and following inhalation exposure in male F344
91
-------
rats (Kasai et al.. 2009). 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: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kocibaetal.. 1974V Both
tumor types have been seen in rats and mice exposed to 1,4-dioxane via drinking water and inhalation.
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. Kasai et al. (2009) noted
evidence of liver toxicity and tumor incidences (i.e., hepatocellular adenoma) in male F344/DuCrj rats
following inhalation exposures to 1,250 ppm. Increased liver toxicities included hepatocellular necrosis,
spongiosis hepatis, and acidophilic and basophilic cell foci.
Nasal cavity tumors were also observed in Sprague Dawley rats (Argus et al.. 1973: Hoch-Ligeti
etal.. 1970). Osborne-Mendel rats (NCI. 1978). Sherman rats (Kocibaetal.. 1974). and F344/DuCrj rats
(Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). Most tumors were
characterized as squamous cell carcinomas. Nasal tumors were not elevated in B6C3F] or Crj:BDFl mice.
Kano et al. (2009) and Kasai et al. (2009) were the only studies that evaluated nonneoplastic changes in
nasal cavity tissue following prolonged exposure to 1,4-dioxane via oral and inhalation routes,
respectively.
Histopathological lesions in female F344/DuCrj rats following oral exposure to 1,4-dioxane were
suggestive of toxicity and regeneration in nasal tissue (i.e., atrophy, adhesion, inflammation, nuclear
enlargement, and hyperplasia and 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. Re-examination 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.
Histopathological lesions in male F344/DuCrj rats following exposure to 1,4-dioxane via
inhalation were also suggestive of toxicity and regeneration in nasal tissue (i.e., atrophy, inflammation,
nuclear enlargement, hyperplasia and metaplasia of the respiratory and olfactory epithelium, and
inflammation). Some of these effects occurred at lower concentrations (50 ppm and 250 ppm) than those
shown to produce nasal cavity tumors (1,250 ppm) in male rats. Nasal squamous cell carcinomas were
observed in the dorsal area of levels 1-3 of the nasal cavity and were characterized as well-differentiated
and keratinized. In two cases, invasive growth into adjacent tissue was noted, marked by carcinoma
growth out of the nose and through a destroyed nasal bone.
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
(JBRC. 1998: Yamazaki et al.. 1994) and the Kasai et al. (2009) study. Female rats dosed with
429 mg/kg-day in drinking water for 2 years also showed a statistically significant increase in mammary
92
-------
gland adenomas (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). In male rats, exposed via
inhalation, a statistically significant positive trend of mammary gland adenomas was observed by Kasai et
al. (2009). A statistically significant increase and/or trend of subcutis fibroma, Zymbal gland adenoma,
and renal cell carcinoma incidences was also observed in male rats exposed for 2 years via inhalation
(Kasai et al., 2009). A significant increase in the incidence of these tumors was not observed in other
chronic oral or inhalation bioassays of 1,4-dioxane (NCI. 1978; Kocibaet al.. 1974; Torkelson et al..
1974).
4.7.3. Mode of Action Information
The hypothesized MOAs 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 kidney, peritoneal, mammary gland, Zymbal gland,
or subcutis tumors due to the absence of any chemical specific information for these tumor types.
4.7.3.1. Identification of Key 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. 1998; 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
(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, since no in vivo or in vitro
assays have identified the toxic moiety resulting from 1,4-dioxane exposure, liver toxicity due to
metabolites cannot be ruled out. Therefore, this hypothesis is not supported. Nannelli 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 the
highly reactive 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
93
-------
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 rat hepatocyte cell
proliferation observed in several studies (Mivagawa et al., 1999; Uno et al., 1994; Goldsworthy et al..
1991; Stottetal.. 1981). No studies of tumor formation have been conducted that specifically examine
mouse liver, thus precluding any determination on whether 1,4-dioxane acts as a tumor promoter in the
mouse liver. These mechanistic studies provide evidence of cell proliferation but do not indicate whether
mitogenesis or cytotoxicity is responsible for increased cell turnover.
The doses in the hepatotoxicity studies where cytotoxicity and cell proliferation were observed
are not equivalent to the doses used in the cancer bioassays. Although Kociba et al. (1974) (noted
evidence of liver toxicity at or below the dose levels that produced liver tumors, they did not report
incidence data for these effects. Thus, a dose-response relationship is unable to be established using the
available studies linking cytotoxicity and cell proliferation observations with tumorigenesis. Additionally,
conflicting data from rat and mouse bioassays suggest that cytotoxicity may not be a required precursor
event for 1,4-dioxane-induced cell proliferation.
Figure 4-1.
Toxicokinetics
Oral absorption
of 1,4-dioxane
Metabolism by
CYP2E1 and
CYP2B1/2
HEAA
elimination in
the urine
Metabolic
saturation and
accumulation of
1,4-dioxane in
the blood
Hypothesized MOA for
Liver Tumors
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.
94
-------
4.7.3.1.2. Nasal cavity.
A possible key event in the MOA hypothesis for nasal tumors is sustained proliferation of
spontaneously transformed nasal epithelial cells, resulting in the eventual formation of nasal cavity
tumors (Kasai et al., 2009). Figure 4-2 presents a schematic represntation of possible key events in the
MOA for 1,4-dioxane nasal carcinogenicity. Cell proliferation was observed following 1,4-dioxane
exposure in both a 2-year inhalation study in male rats (1,250 ppm) (Kasai et al., 2009) and a 2-year
drinking water study in male (274 mg/kg-day) and female rats (429 mg/kg-day) (Kano et al.. 2009).
However, neither study reported evidence of cytotoxicity in the nasal cavity therefore, cytotoxicity as a
key event is not supported. Nasal lesions, including inflammation, hyperplasia, and metaplasia, were
frequently seen in inhalation studies conducted by the NTP with no evidence of nasal carcinogenicity
(Haseman and Hailey. 1997; Wardetal., 1993). Kasai et al. (2009; 2008) suggest that nasal toxicity is
related to the accumulation of the parent compound following metabolic induction at high doses up to
3,200 ppm; however, since no in vivo or in vitro assays have examined the toxic moiety resulting from
1,4-dioxane exposure, nasal toxicity due to metabolites cannot be ruled out. Nannelli et al. (2005)
demonstrated that CYP2E1 was inducible in nasal mucosa following acute oral administration of
1,4-dioxane by gavage and drinking water, which could potentially lead to an increase in the oxidative
metabolism of 1,4-dioxane and nasal toxicity. However, Nannelli et al. (2005) neither characterized this
pathway nor identified possible reactive intermediates or nasal toxicities.
95
-------
Figure 4-2.
Toxicokinetics
Inhalation of
water droplets
Oral absorption
of 1,4-dioxane
Inhalation of
1,4-dioxane
vapors
Metabolism by
CYP2EI and
CYP2B1/2
Metabolic
saturation and
accumulation of
1,4-dioxane in
the blood
HEAA
elimination in
the urine
Hypothesized MOA for
Nasal Cavity Tumors
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 4-22). 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 reported lesions indicative of cytotoxicity (Kano et al..
2008; JBRC. 1998; NCI. 1978). 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
96
-------
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; Golds-worthy 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; Kasai et al..
2009; JBRC. 1998; Yamazaki et al.. 1994; NCI. 1978; Kocibaetal. 1974). but were not elevated in two
strains of mice (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994; NCI. 1978). Irritation of the nasal
cavity of rats was indicated in studies by the observation of inflammation (Kasai et al.. 2009; Kasai et al..
2008) and also rhinitis (JBRC. 1998). The Kasai et al. (2009; 2008) studies also showed atrophy of the
nasal epithelium in rats, and the JRBC (1998) study also observed atrophy of the nasal epithelium as well
as 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; Kasai et al.. 2009; JBRC. 1998; Yamazaki et
al.. 1994). Oxidation of 1,4-dioxane metabolism by CYP450s is not supported as a key event in the MOA
hypothesis of nasal tumors. Although Nannelli et al. (2005) demonstrated that CYP2E1 was inducible in
nasal mucosa following acute oral administration of 1,4-dioxane by gavage and drinking water, the study
lacked details regarding the toxic moiety (e.g., parent compound or reactive intermediate) and resulting
nasal toxicity. Accumulation of 1,4-dioxane in blood, as a precursor event of nasal tumor formation is
also not supported because the parent compound 1,4-dioxane was only measured in one subchronic study
(Kasai et al.. 2008) and in this study no evidence of nasal cytotoxicity, cell proliferation, or incidence of
nasal tumors were reported.
4.7.3.3. Dose-Response Relationship
4.7.3.3.1. Liver
Table 4-27 presents the temporal sequence (i.e., the table columns in sequential order from
1,4-dioxane metabolism, to liver damage, cell proliferation, hyperplasia, and the formation of adenomas
and/or carcinomas) 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 (Kocibaetal.. 1974). Hepatic tumors were only observed at the
highest dose (Kocibaetal.. 1974). Hepatic regeneration was indicated in the mid- and high-dose group by
97
-------
the formation of hepatocellular hyperplastic nodules. Liver hyperplasia was also reported in rats from the
JBRC (1998) study, at or below the dose level that resulted in tumor formation (Kano et al.. 2009):
however, hepatocellular degeneration and necrosis were not reported. The liver hyperplasia reported in
JBRC (1998) was later reclassified to hepatocellular adenoma or altered hepatocellular foci (Kano et al..
2009). These results suggest that hepatic cell proliferation 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.
Table 4-27 Temporal sequence and dose-response relationship for possible key events and
liver tumors in rats and mice
Key event (time — >)
Dose (mg/kg-day) or Metabolism Cel1
Exposure (ppm) 1,4-dioxane Liver damage proliferation
Hyperplasia
Adenomas
and/or
carcinomas
Kociba et al., (1974) — Sherman rats (male and female combined)
0 mg/kg-day — a — a — a
14 mg/kg-day +b — a — a
121 mg/kg-day +b +c — a
1,307 mg/kg-day +b +c — a
a
a
+c
+c
a
a
a
+c
NCI, (1978)— male Osborne-Mendel rats
0 mg/kg-day — a — a — a
240 mg/kg-day +b — a — a
530 mg/kg-day +b — a — a
a
a
a
a
a
a
NCI, (1978)— female Osborne-Mendel rats
0 mg/kg-day — a — a — a
350 mg/kg-day +b — a — a
640 mg/kg-day +b — a — a
a
a
a
a
+c
+c
NCI, (1978)— male B6C3Fi mice
0 mg/kg-day — a — a — a
720 mg/kg-day +b — a — a
830 mg/kg-day +b — a — a
a
a
a
a
+c
+c
NCI, (1978)— female B6C3Fi mice
0 mg/kg-day — a — a — a
380 mg/kg-day +b — a — a
860 mg/kg-day +b — a — a
a
a
a
a
+c
+c
Kano et al., (2009); JBRC, (1998)— male F344/DuCrj rats
0 mg/kg-day — a — a — a
11 mg/kg-day +b — a — a
55 mg/kg-day +b — a — a
274 mg/kg-day +b +c'd — a
a
a
a
a
a
a
a
+c,e
98
-------
Table 4-27 (Continued) Temporal sequence and dose-response relationship for possible key
events and liver tumors in rats and mice
Key event (time — >)
Dose (mg/kg-day) or
Exposure (ppm)
Kano et al., (2009); JBRC,
0 mg/kg-day
18 mg/kg-day
83 mg/kg-day
429 mg/kg-day
Kano et al., (2009); JBRC,
0 mg/kg-day
49 mg/kg-day
191 mg/kg-day
677 mg/kg-day
Kano et al., (2009): JBRC,
0 mg/kg-day
66 mg/kg-day
278 mg/kg-day
964 mg/kg-day
Metabolism
1,4-dioxane
Liver damage
Cell
proliferation
Hyperplasia
Adenomas
and/or
carcinomas
(1998) — female F344/DuCrj rats
a
+ b
+ b
+ b
(1998)— male
a
+b
+b
+b
a
a
a
a
Crj:BDF1 mice
a
a
a
+c,d
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
+c,e
a
+c,e
+c,e
+c,e
(1998) — female Crj:BDF1 mice
a
+ b
+ b
+ b
Kasai et al. (2008) — F344 rats (male and
0 ppm
100 ppm
200 ppm
400 ppm
800 ppm
1,600 ppm
3,200 ppm
6,400 ppm
Kasai et al., (2009) — male
0 ppm
50 ppm
250 ppm
1,250 ppm
a
a
a
a
a
a
a
a,g
F344 rats
a
a
a
a
a
a
a
+c,d
female combined)
a
a
a
a
a
a
+f
a,g
a
a
a
+ h
a
a
a
a
a
a
a
a
a
a
a
a,g
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a,g
a
a
a
a
a
+c,e
+c,e
+c,e
a
a
a
a
a
a
a
a,g
a
a
a
+h
B— 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.
°+ Statistically significant increase noted.
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).
e+ Kano et al. (2009) reported incidence rates for hepatocellular adenomas and carcinomas.
f+ Kasai et al. (2008) reported significant incidence rates for single cell necrosis in female rats only (3,200 ppm) following a 2 year
bioassay.
9—All rats died during the first week of the 13-week bioassay (Kasai et al.. 2008).
h+ Kasai et al. (2009) reported incidence rates for centrilobular necrosis and hepatocellular adenomas in male rats (1,250 ppm).
99
-------
4.7.3.3.2. Nasal cavity.
Table 4-28 presents the temporal sequence (i.e., the table columns in sequential order from
1,4-dioxane metabolism, to nasal damage, cell proliferation, hyperplasia, and the formation of adenomas
and/or carcinomas) and dose-response relationship for possible key events in the nasal tissue
carcinogenesis of 1,4-dioxane. 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; JBRC. 1998). In
another study, dose-response information provided some support for nasal toxicity and regeneration in
nasal epithelium occurring before tumor development (Kasai et al., 2009). However, the role of
cytotoxicity as a required precursor event is not supported by data from any of the reviewed studies. The
accumulation of parent 1,4-dioxane as a key event has some support since concentration-dependent
increases were noted for 1,4-dioxane in plasma concurrent with toxicities observed that are possible
precursor events (i.e., regeneration in nasal epithelium) (Kasai et al., 2008). In a subsequent study by
Kasai et al. (2009) some of these same possible precursor events were observed at 50, 250, and 1,250 ppm
with evidence of nasal tumors at the highest concentration (1,250 ppm).
Table 4-28 Temporal sequence and dose-response relationship for possible key events and
nasal tumors in rats and mice
Key event (time —>)
Dose (mg/kg-day)
or Exposure Metabolism Nasal Cell .. . . Adenomas
and/or
carcinomas
1 IVItJiaUUIISIII IMctSctl \>CM iii- ii
(PPm) 1,4-dioxane cytotoxicity proliferation Hyperplas.a and/or
Kociba et al., (1974)—Sherman rats (male and female combined)
0 mg/kg-day
14 mg/kg-day
121 mg/kg-day
1,307 mg/kg-day
NCI, (1978)—female Osborne-Mendel rats
0 mg/kg-day
350 mg/kg-day
640 mg/kg-day
-male B6C3Fi mice
0 mg/kg-day
720 mg/kg-day
830 mg/kg-day
NCI, (1978)—female B6C3Fi mice
0 mg/kg-day
380 mg/kg-day
860 mg/kg-day
100
-------
Table 4-28 (Continued): Temporal sequence and dose-response relationship
key events and nasal tumors in rats and mice
Key
event (time — >)
or Exposure Metabolism Nasal Cell
(ppm) 1,4-dioxane cytotoxicity proliferation
Kano et al., (2009); JBRC
0 mg/kg-day
11 mg/kg-day
55 mg/kg-day
274 mg/kg-day
Kano et al., (2009); JBRC
0 mg/kg-day
18 mg/kg-day
83 mg/kg-day
429 mg/kg-day
Kano et al., (2009): JBRC
0 mg/kg-day
49 mg/kg-day
191 mg/kg-day
677 mg/kg-day
Kano et al., (2009); JBRC
0 mg/kg-day
66 mg/kg-day
278 mg/kg-day
964 mg/kg-day
Kasai et al. (2008)— F344
0 ppm
100 ppm
200 ppm
400 ppm
800 ppm
1,600 ppm
3,200 ppm
6,400 ppm
Kasai et al. (2009)— male
0 ppm
50 ppm
250 ppm
1,250 ppm
, (1998) — male F344/DuCrj rats
a a
+ b _a
+ b _a
+ b _a
, (1998)— female F344/DuCrj rats
a a
+ b _a
+ b _a
+ b _a
, (1998) — male Crj:BDF1 mice
a a
+ b _a
+ b _a
+ b _a
, (1998) — female Crj:BDF1 mice
a a
+ b _a
+ b _a
+ b _a
rats (male and female combined)
a a
+ b _a
+ b _a
+c — a
+c _a
+<= _a
+c _a
+a,b,f a,f
F344 rats
a a
+ b _a
+ b _a
+ b _a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a,f
a
a
a
+c
Hyperplasia
a
a
a
+c,d
a
a
a
+c,d
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a,f
a
a
a
+e
for possible
Adenomas
and/or
carcinomas
a
a
a
+c,d
a
a
a
+c,d
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a,f
a
a
a
+c
a— No evidence demonstrating key event.
b+ 1,4-dioxane metabolism was not evaluated as part of these studies. 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+ Kano et al. (2009) reported incidence rates for squamous cell hyperplasia (respiratory epithelium) and squamous cell
carcinomas (nasal cavity); however, information from JBRC (1998) on significant incidence of squamous cell hyperplasia was
used to create this table.
e+ Kasai et al. (2009) reported incidence rates for squamous cell hyperplasia in male rats (1,250 ppm) following a 2 year bioassay.
'All rats died during the first week of the 13 week bioassay (Kasai et al.. 2008).
101
-------
4.7.3.4. Temporal Relationship
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:BDFl 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. 1998). 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 may lead to regenerative cell proliferation and tumor formation following chronic exposure. As
discussed above, histopathological evidence of regenerative cell proliferation has been seen following
long-term exposure to 1,4-dioxane (JBRC. 1998; Kocibaetal.. 1974). Tumors occurred earlier at high
doses in both mice and rats from this study (Yamazaki. 2006); 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. Sustained nasal damage may lead to regenerative
cell proliferation and tumor formation following chronic exposure. As discussed above (Section
4.2.2.2.1), no evidence of cytotoxicity has been observed following exposure to 1,4-dioxane, despite
histopathological evidence of regenerative cell proliferation and nasal tumors at the highest exposure
concentration (Kano et al.. 2009; Kasai et al.. 2009) (see Table 4-28). Other incidences of nasal damage
102
-------
may have occurred before tumor formation; however, temporal information regarding these events was
not available (i.e., interim sacrifices were not performed).
4.7.3.5. Biological Plausibility and Coherence
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.
4.7.3.5.2. Nasal 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 MOAs 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).
but were observed in an inhalation study in F344 rats exposed to 1,250 ppm for 5 days/week for 2 years.
Two human studies of occupational exposure, ranging from 0.06 ppm to 75 ppm for Imonth up to 41
years, reported negative findings regarding increased tumor risk (Buffler et al., 1978; Thiess et al.. 1976).
It is important to note, neither nasal tumors in the human studies nor genotoxicity in human or animal
studies were evaluated following inhalation exposure to 1,4-dioxane
While there is no known MOA for 1,4-dioxane and the human studies are inconclusive regarding
tumor risk, the noted nasal tumors in rats are considered biologically plausible and relevant to humans,
since similar cell types considered to be at risk are prevalent throughout the respiratory tract of rats and
humans. In general, rats may be more susceptible to nasal lesions than humans due to differences in the
anatomy and geometry of the upper respiratory tract (e.g., larger fraction of inspired air ventilates rat
nasal cavity compared to the human) and resulting differences in absorption (e.g., rat nasal cavity is more
103
-------
efficient at scrubbing gases than human) or in local respiratory system effects; however, there is not as
much known about other respiratory tract lesions (e.g., trachea or lower respiratory tract) (U.S. EPA.
2012a. 2009a). Species differences in absorption and respiratory tract uptake for 1,4-dioxane have not
been studied, thus it still represents an area of uncertainty for this compound.
4.7.3.6. Other Possible Modes of Action
An alternate MOA could be hypothesized that 1,4-dioxane alters DNA, either directly or
indirectly (Kasai et al.. 2009). 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 agenotoxic 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 4-23. Negative findings were reported for
mutagenicity in Salmonella typhimurium, Escherichia coll, and Photobacterium phosphoreum (Mutatox
assay) (Morita and Havashi. 1998; Hellmer and Bolcsfoldi. 1992; Kwanetal.. 1990; Khudolev et al..
1987; Nestmann et al.. 1984; Haworth et al.. 1983; Stottetal. 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 (Zimmermann 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 Havashi. 1998). Lastly, in the presence of toxicity, positive results were reported
for meiotic nondisjunction in drosophila (Munoz and Barnett. 2002). DNA damage (Sinaet al.. 1983).
and cell transformation (Sheu et al.. 1988).
Additionally, 1,4-dioxane metabolism did not produce reactive intermediates that covalently
bound to DNA (Stottetal.. 1981; Woo etal.. 1977c) and DNA repair assays were generally negative
(Goldsworthy et al.. 1991; Stottetal.. 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 the Hypothesized Mode of Action
4.7.3.7.1. Liver.
The available evidence in support of any hypothesized MOA for liver tumors is not conclusive. 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.
Tumor formation has also been observed in the absence of cytotoxicity (Kano et al.. 2009; JBRC. 1998).
104
-------
Cell proliferation and tumor promotion have been shown to occur after prolonged exposure to
1,4-dioxane (Miyagawaetal.. 1999; Unoetal.. 1994; Goldsworthy et al.. 1991; Lundberg et al.. 1987;
Bulletal.. 1986; Stottetal.. 1981; Kingetal.. 1973).
4.7.3.7.2. Nasal cavity.
The available evidence in support of any hypothesized MOA for nasal tumors is not conclusive.
Nasal carcinogens are generally characterized as potent genotoxins (Ashby. 1994); however, other MOAs
have been proposed for nasal carcinogens that induce effects through other mechanisms (Kasper et al..
2007; Green et al., 2000). In the human studies evidence of nasal tumors were not assessed, nor
genotoxicity in human or animal studies following inhalation exposure to 1,4-dioxane, so the role of
genotoxicity cannot be ruled out. A MOA hypothesis involving nasal damage, cell proliferation, and
hyperplasia is possible, but data are not available to support this hypothesis. In studies that examined
nasal effects after exposure to 1,4-dioxane, at least one of these events is missing. More specifically, nasal
cavity tumors have been reported by Kasai et al. (2009) in the absence of cytotoxicity and in Kano et al.
(2009) in the absence of hyperplasia. Therefore, as per EPA's Cancer Guidelines (U.S. EPA. 2005a).
there is insufficient biological support for potential key events and to have reasonable confidence in the
sequence of events and how they relate to the development of nasal tumors following exposure to
1,4-dioxane. Using the modified Hill criteria, exposure-response and temporal relationships have not been
established in support of any hypothetical mode of carcinogenic action for 1,4-dioxane.
4.7.3.8. Relevance of the Mode of Action to Humans
Several hypothesized MOAs for 1,4-dioxane induced tumors in laboratory animals have been
discussed along with the supporting evidence for each. 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 other
tumor types (Kano et al.. 2009; Kasai et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). Human relevancy
is assumed unless information indicates otherwise (U.S. EPA. 2005a).
4.8. Susceptible Populations and Life Stages
There is no direct evidence to establish that certain populations and lifestages may be 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
105
-------
(Carpenter etal.. 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; Havashi 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 CYP2E1 genetic
polymorphism resulting in increased expression of this enzyme may be less susceptible to toxicity
following exposure to 1,4-dioxane.
Gender differences were noted in subchronic and chronic toxicity studies of 1,4-dioxane in mice
and rats (see Sections 4.6 and 4.7). No consistent pattern of gender sensitivity was identified across
studies. In a 13 week inhalation study of male and female rats (Kasai et al.. 2008) kidney toxicity, as
evidenced by hydropic change in the renal proximal tubules, was observed in female rats exposed to
3,200 ppm of 1,4-dioxane, but not male rats. This suggests a possible increased susceptibility of female
rats to renal damage following inhalation exposure to 1,4-dioxane.
106
-------
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. 1998; Drewetal.. 1978; David. 1964; KestenetaL 1939; Laugetal.. 1939; Schrenk and Yant.
1936; de Navasquez. 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. 1998; Yamazaki et al.. 1994;
NCI. 1978; Kociba et al.. 1974; Argus etal.. 1973; Argus etal.. 1965) (see Table 4-25).
Liver toxicity in the available chronic studies was characterized by necrosis, spongiosis hepatis,
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, as
well as hepatocellular regeneration as indicated by hepatocellular hyperplastic nodule formation at these
doses. 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 (1998) 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 (1998)
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. 1998); 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 (1998) 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 etal.. 1974; Argus et
al.. 1973; Argus etal.. 1965; Fairley et al.. 1934). Kociba et al. (1974) described renal 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 (Kociba et al.. 1974). Doses
107
-------
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 (1998) 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. 1998; 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 (1998) 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).
Even though the study reported by Kociba et al. (1974) had one noteworthy weakness, it had
several noted strengths, including: (1) two-year study duration; (2) use of both male and female rats and
three dose levels, 10-fold apart, plus a control group; (3) a sufficient number of animals per dose group
(60 animals/sex/dose group; and (4) the authors conducted a comprehensive evaluation of the animals
including body weights and clinical observations, blood samples, organ weights of all the major tissues,
and a complete histopathological examination of all rats. The study weakness was that the authors did not
report individual incidence data that would have allowed for a BMD analysis of this robust dataset.
5.1.2. Methods of Analysis—Including Models (PBPK, BMD, etc.)
Available human PBPK models were evaluated 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 recalibrated model predictions for blood 1,4-dioxane levels did not
adequately fit the experimental values (see Appendix B). The model structure is insufficient to capture the
apparent species difference in the blood 1,4-dioxane Vd between rats and humans. 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). Additionally, the models do not account for induction in metabolism, which
may be present in animals exposed repeatedly to 1,4-dioxane. Therefore, some other modification(s) to
the Reitz et al. (1990) PBPK model structure would be necessary to correct the PBPK models for use in
derivation of toxicity values (see Appendix B for more details).
Kociba et al. (1974) did not provide quantitative incidence or severity data for liver and kidney
degeneration and necrosis. Therefore, benchmark dose (BMD) modeling could not be performed for this
108
-------
study, and thus 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.
An alternative POD was derived using incidence data reported for cortical tubule degeneration in
the kidneys in male and female rats (NCI. 1978). 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
based on cortical tubule degeneration in the NCI (1978) study (BMDL10 of 22.3 mg/kg-day) (Table 5-2).
The BMDL10 value of 22.3 mg/kg-day from the NCI (1978) study is about double 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
0/31a 20/31b 27/33b 0/31a 0/34 10/32b
"Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this review.
blncidence significantly elevated compared to control by Fisher's Exact test (p < 0.001) performed for this review.
Source: NCI (1978).
Table 5-2 BMD and BMDL values derived from BMD modeling of the incidence 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).
109
-------
5.1.3. RfD Derivation - Including Application of Uncertainty Factors
(UFs)
The RfD of 3 x 1CT2 mg/kg-day is based on liver and kidney toxicity in rats exposed to
1,4-dioxane in the drinking water for 2 years (Kocibaet al., 1974). The Kociba et al. (1974) 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. The RfD for 1,4-dioxane is derived by dividing the
NOAEL of 9.6 mg/kg-day (Kocibaet al., 1974) by a composite UF of 300, as follows:
RfD = NOAEL / UF
9.6 mg/kg-day / 300
0.03 or 3 x lO-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 (UFA) was used to account for pharmacokinetic and
pharmacodynamic differences between rats and humans. Existing PBPK models could not be used to
derive an oral RfD for 1,4-dioxane (Appendix B).
A default interindividual variability UF of 10 (UFH) was 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 to extrapolate from a subchronic to a chronic (UFS) exposure duration was not necessary
(e.g., UFS = 1) because the RfD was derived from a study using a chronic exposure protocol.
An UF to extrapolate from a LOAEL to a NOAEL (UFL) was not necessary (e.g., UFL =1)
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 by the study authors based on liver and kidney toxicity; however, quantitative incidence data were
not reported. Several additional oral studies (of acute/short-term, subchronic, and chronic durations) were
available that support liver and kidney toxicity as the critical effect (Kano et al., 2008; JBRC. 1998; NCL
1978; Argus etal.. 1973) (Table 4-15 and Table 4-17). Although degenerative liver and kidney toxicity
was not observed in rats from the JBRC (1998) 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.
An UF of 3 for database deficiencies (UFD) was applied due to the lack of a multigeneration
reproductive toxicity study.
110
-------
5.1.4. RfD Comparison Information
PODs and candidate oral RfDs based on selected studies included in Table 4-18 are arrayed in
Figure 5-1 to Figure 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 candidate RfD is reflected in the 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 candidate 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; Argus et al.. 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 based on this endpoint 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 (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice).
JBRC (1998) 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 based on this endpoint 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) administered during gestation can produce delayed ossification of the sternebrae and
reduced fetal BWs (Giavini etal.. 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 toxicity
are degenerative liver 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 etal.. 1974) and are presented
for completeness.
Ill
-------
100
10
Rat
Mouse
Rat
"ci
•
01
001
1
•
o
• POD
HjAni mai-to-hum an
OHuman variation
EILOAELto NOAEL
DSubchronicto Chronic
^Database deficiencies
ORfD
Hepatocellulardegenerationand Increase in serum liverenz/mes: Increase in serum liver enzymes;
necrosis: NOAEL: 2 yr rat drinking water NOAEL: 2 yr mouse drinking water study NOAEL: 2 yr rat drinking water study
study (Kocibaetal,, 1974,062929} (JBRC, 1998,196240) (JBRC, 1993,196240)
Kociba et al. (1974) and JBRC (1998),
Figure 5-1. Potential points of departure (POD) based on liver toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane.
112
-------
1000
100
Rat
Rat
Rat
10
I
B
s
Q
0 1
o 01
• POD
fTTTlAnimal-lo-hurnan
I [Human variation
E23LOAEL to NOAEL
CUSubchronicto Chronic
HDatabase deficiencies
ORTD
Glomerulonephritis: LOAEL; 13 month Degeneration and necrosis of tubular Cortical tubule degeneration: BMDL10;
rat drinkingwater study (Argus et aL epithelium: NOAEL: 2 yr rat drinking 2 yr rat drinking water study (NCI. 197B.
1973.062912) water study (Kociba et al., 1974, 062936)
062929)
Argus et al. (1973): Kociba et al. (1974): NCI (1978
Figure 5-2.
Potential points of departure (POD) based on kidney toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane.
113
-------
Mouse
Rat
100
10
•&
E
I
• POD
HflAnimal-to-human
Q| Human variation
MLOAELta NOAEL
Dsubchronicto Chronic
Hoatabase deficiencies
oRfD
Nasal inflammation: NOAEL: 2 yr mouse drinking water study
(JBRC. 1998. 196240)
Nasal inflammation: NOAEL: 2 yr rat drinking water study
(JBRC. 1998. 196240)
JBRC (1998).
Figure 5-3.
Potential points of departure (POD) based on nasal inflammation with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane.
114
-------
1000
100
Rat
Kidney
Rat
Liver
Rat
Developmental
Mouse
Respiratory
10
•O
D)
ra
o
Q
001
¥
o
• POD
EEAnimal-to-human
D Human variation
^LOAELto NOAEL
dsubchronicto Chronic
B Database deficiencies
oRfD
Degeneration and necrosis Hepatocellular degeneration Delayed ossification of Nasal inflammation
of tubular epithelium; and necrosis; NOAEL; 2 yr sternebrae and reduced NOAEL;2yr mouse
NOAEL; 2 yr rat drinking rat drinking water sduy fetalbody weight; NOAEL; drinkingwaterstudy(JBRC,
water study (Kocibaetal. (Kocibaetal , 1974, rat study gestation days 6- 1998,196240)
1974,062929) 062929) 15(Giaviniet al., 1985,
062924)
Kociba et al. (1974): Giavini et al. (1985): JBRC (1998).
Figure 5-4. Potential points of departure (POD) based on organ-specific toxicity
endpoints with corresponding applied uncertainty factors and derived
candidate RfDs following chronic 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)
5.2.1. Choice of Principal Study and Candidate Critical Effect(s) with
Rationale and Justification
Two human studies of occupational exposure to 1,4-dioxane have been published (BuffleretaL
1978; Thiess et al.. 1976); however, neither study provides sufficient information and data to quantify
subchronic or chronic noncancer effects. In each study, findings were negative and deemed inconclusive
by the EPA due to the small cohort size and the limited number of reported cases (Buffler et al.. 1978;
Thiess etal. 1976).
115
-------
Four inhalation studies in animals were identified in the literature; two 13-week subchronic
studies in several species of laboratory animals (Kasai et al.. 2008; Fairlev etal.. 1934) and two 2-year
chronic studies in rats (Kasai et al., 2009; Torkelson et al.. 1974).
In the subchronic study by Fairley et al. (1934). rabbits, guinea pigs, rats, and mice
(3-6/species/group) were exposed to 1,000, 2,000, 5,000, or 10,000 ppm of 1,4-dioxane vapor for
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 kidney and liver 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. In this study, a LOAEL of
1,000 ppm for liver and kidney degeneration in rats, mice, rabbits, and guinea pigs was identified by EPA.
In the subchronic study by Kasai et al. (2008). male and female rats (10/group/sex) were exposed
to 0, 100, 200, 400, 800, 1,600, 3,200, and 6,400 ppm of 1,4-dioxane for 6 hours/day, 5 days/week for 13
weeks. This study observed a range of 1,4-dioxane-induced nonneoplastic effects across several organ
systems including the liver and respiratory tract (from the nose to the bronchus region) in both sexes and
the kidney in females. Detailed dose-response information was provided, illustrating a vapor
concentration-dependent increase of nuclear enlargement of nasal (respiratory and olfactory), trachea, and
bronchus epithelial cells (both sexes); vacuolic changes in nasal and bronchial epithelial cells (both
sexes), necrosis and centrilobular swelling of hepatocytes (both sexes); and hydropic change in the
proximal tubules of the kidney (females). The study authors determined nuclear enlargement of the nasal
respiratory epithelium as the most sensitive lesion and a LOAEL of 100 ppm was identified based on this
effect. However, it is important to note that the severity of the change (i.e., nuclear enlargement) was
similar (i.e., slight) at the four lowest tested vapor levels (i.e., 100, 200, 400 and 800 ppm) in male and
female rats; with only a moderate observation of severity noted at the two highest tested vapor levels
(i.e., 1,600 and 3,200 ppm). Additionally, nuclear enlargement may be found in any cell type responding
to microenvironmental stress or undergoing proliferation. It may also be an indicator of exposure to a
xenobiotic in that the cells are responding by transcribing mRNA. Several studies indicate that it may also
be identified as an early change in response to exposure to a carcinogenic agent (Wiemann et al.. 1999;
Enzmann et al.. 1995; Clawson et al.. 1992; Ingram and Grasso. 1987. 1985); however, its relationship to
the typical pathological progression from initiated cell to tumor is unclear. Therefore, nuclear
enlargement as a specific morphologic diagnosis is not considered an adverse effect of exposure to
1,4-dioxane.
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. A free standing NOAEL of
111 ppm was identified in this study by EPA.
116
-------
Kasai et al. (2009) reported data for groups of male F344 rats (50/group) exposed to 0, 50, 250,
and 1,250 ppm of 1,4-dioxane for 6 hours/day, 5 days/week, for 2 years. In contrast to the subchronic
Kasai et al. (2008) study, this 2-year bioassay reported more nonneoplastic effects in multiple organ
systems. Effects observed included: (1) inflammation of nasal respiratory and olfactory epithelium,
(2) squamous cell metaplasia and hyperplasia of nasal respiratory epithelium, (3) atrophy and respiratory
metaplasia of olfactory epithelium, (4) hydropic change and sclerosis in the lamina propria of nasal
cavity, (5) nuclear enlargement in proximal tubules of the kidney, in the centrilobular region of the liver,
and of the respiratory and olfactory epithelium, (6) centrilobular necrosis in the liver, and (7) spongiosis
hepatis. Some of these histopathological lesions were significantly increased compared to controls at the
lowest exposure level (50 ppm), including nuclear enlargement of respiratory and olfactory epithelium;
and atrophy and respiratory metaplasia of olfactory epithelium. Many of these histopathological lesions
were increased in a concentration-dependent manner.
Whether spongiosis hepatis/cystic degeneration represents a preneoplastic change or a
nonneoplastic change has been the subject of scientific controversy (Karbe and Kerlin. 2002; Stroebel et
al.. 1995; Bannasch et al.. 1982). Spongiosis hepatis 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. 2003; Stroebel et al.. 1995). In contrast, it 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). Following inhalation of 1,4-dioxane, spongiosis
hepatis was associated with other preneoplastic (e.g., liver foci) and nonneoplastic (e.g., centrilobular
necrosis) changes in the liver (Kasai et al.. 2009). However, the incidence rates of spongiosis hepatis and
liver tumors were highly correlated; therefore, spongiosis hepatis was considered a preneoplastic lesion
following inhalation exposure and not considered further in the noncancer analysis.
The Fairley et al. (1934) study was inadequate to characterize the inhalation risks of 1,4-dioxane
because control animals were not used, thus limiting the ability to perform statistical analysis;
additionally, no data for low-dose exposure were reported. Because Torkelson et al. (1974) identified a
free-standing NOAEL only, this study was also deemed inadequate to characterize the inhalation risks of
1,4-dioxane. A route-to-route extrapolation from the 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; Fairlev 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 (Appendix B).
Therefore, the chronic Kasai et al. (2009) study was selected as the principal study for the
derivation of the RfC. The Kasai et al. (2009) 2-year bioassay utilized 50 animals per exposure group, a
range of exposure concentrations which were based on the results of the subchronic study (Kasai et al..
2008). and thoroughly examined toxicity of 1,4-dioxane in multiple organ systems. Based on the
noncancer database for 1,4-dioxane, this study demonstrated exposure concentration-related effects for
histopathological lesions at a lower concentration (50 ppm) compared to the subchronic Kasai et al.
(2008) study. The 2-year bioassay (Kasai et al.. 2009) did not observe effects in both sexes, but the use of
117
-------
only male rats was proposed by the study authors as justified because of data illustrating the absence of
induced mesotheliomas in female rats following exposure to 1,4-dioxane in drinking water (Yamazaki et
al., 1994). Additionally, a similar pattern of effects was observed after oral exposure to 1,4-dioxane (Kano
et al.. 2009; JBRC. 1998) as was observed in the Kasai et al. (2009) 2-year inhalation study.
Incidences of nonneoplastic lesions from the Kasai et al. (2009) study that were statistically
significantly increased as compared to control were considered candidates for the critical effect. These
candidate endpoints included centrilobular necrosis of the liver, squamous cell metaplasia of the nasal
respiratory epithelium, squamous cell hyperplasia of the nasal respiratory epithelium, respiratory
metaplasia of the nasal olfactory epithelium, sclerosis in the lamina propria of the nasal cavity, and two
degenerative nasal lesions, that is, atrophy of the nasal olfactory epithelium and hydropic change in the
lamina propria (Table 5-3). Despite statistically significant increases at the low- and mid-exposure
concentrations (50 and 250 ppm, respectively), incidences of nuclear enlargement of the respiratory
epithelium (nasal cavity), olfactory epithelium (nasal cavity), and proximal tubule (kidney) were not
considered candidates for the critical effect since it is not considered by EPA to be adverse, as discussed
previously (see Section 4.6.2 and Table 4-22).
Table 5-3 Incidences of nonneoplastic lesions resulting from chronic exposure (ppm) to
1,4-dioxane considered for identification of a critical effect.
Concentration (ppm)
Species/Strain Tissue
Liver
Rat/ F344 (male)
Nasal
Endpoint
Centrilobular necrosis
Squamous cell metaplasia;
respiratory epithelium
Squamous cell hyperplasia;
respiratory epithelium
Respiratory metaplasia;
olfactory epithelium
Atrophy; olfactory epithelium
Hydropic change;
lamina propria
Sclerosis; lamina propria
0
1/50
0/50
0/50
11/50
0/50
0/50
0/50
50
3/50
0/50
0/50
34/50a
40/50a
2/50
0/50
250
6/50
7/50b
1/50
49/50a
47/50a
36/50a
22/50a
1,250
12/50a
44/50a
10/50a
48/50a
48/50a
49/50a
40/50a
ap < 0.01 by x2 test.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2009).
118
-------
5.2.2. Methods of Analysis
Benchmark dose (BMD) modeling (U.S. EPA. 2012b) was used to analyze the candidate
endpoints identified for 1,4-dioxane. Use of BMD methods involves fitting mathematical models to the
observed dose-response data and provides a BMD and its 95% lower confidence limit (BMDL) associated
with a predetermined benchmark response (BMR). For 1,4-dioxane, the selected datasets in Table 5-4
were considered as candidate critical effects and analyzed using BMD modeling to determine potential
PODs. Information regarding the degree of change in the selected endpoints that is considered
biologically significant was not available. Therefore, a BMR of 10% extra risk was selected under the
assumption that it represents a minimally biologically significant response level (U.S. EPA. 2012b).
The estimated BMDs and BMDLs based on incidences of centrilobular necrosis, squamous cell
metaplasia and hyperplasia of the respiratory epithelium, and hydropic change of lamina propria are
presented in Table 5-4. Due to lack of fit or substantial model uncertainty, BMD modeling results were
deemed inadequate for the following endpoints: atrophy (olfactory epithelium), respiratory metaplasia
(olfactory epithelium), and sclerosis (lamina propria). Consequently, for these last three endpoints, the
NOAEL/LOAEL approach was used to determine potential PODs. The detailed results of the BMD
analysis are provided in Appendix F.
5.2.3. Exposure Duration and Dosimetric Adjustments
Because an RfC assumes continuous human exposure over a lifetime, data derived from
inhalation studies in animals need to be adjusted to account for the noncontinuous exposure protocols
used in these studies. In the Kasai et al. (2009) study, rats were exposed to 1,4-dioxane for 6 hours/day, 5
days/week for 2 years. Therefore, the duration-adjusted PODs for liver and nasal lesions in rats were
calculated as follows:
hours exposed per day days exposed per week
PODADI (ppm) = POD (ppm) x -^ x — *—•—
ADJ vpp j vpp j 24 hours 7 days
RfCs are typically expressed in units of mg/m3; so PODADi (ppm) values were converted using
the chemical specific conversion factor of 1 ppm = 3.6 mg/m3 for 1,4-dioxane (Table 2-1). The following
calculation was used:
3.6 mg/m3
PODADJ (mg/m3) = PODADJ (ppm) x
The calculated PODADJ (mg/m3) values for all considered endpoints are presented in the last
column of Table 5-4.
119
-------
Table 5-4 Duration adjusted POD estimates for BMDLs (from best fitting BMDS models)
or NOAELs/LOAELs from chronic exposure to 1,4-dioxane
Endpoint
NOAEL3 LOAELb
(ppm) (ppm)
Model
BMR BMD BMDL PODADj PODADj
(%) (ppm) (ppm) (ppm) (mg/m3)
Liver Effects
Centrilobular
necrosis; Liver
Dichotomous-Hill 10
220
60
10.7
38.6
Nasal Effects
Squamous cell
metaplasia; - - Log-probit 10 218 160 28.6
respiratory epithelium
Squamous cell
hyperplasia; - - Log-probit 10 756 561 100
respiratory epithelium
Respiratory
metaplasia; olfactory — 50 — c — — — 8.9
epithelium
Atrophy; olfactory c ft Q
epithelium ~ DU ~ " ~ ~ °'a
Hydropic change;
' K a - - Log-logistic 10 69 47 8.4
lamina propria
103
361
32.2
32.2
30.2
50
250
8.9
32.2a
Sclerosis; lamina
propria
aNOAEL is identified in this assessment as the highest tested exposure dose at which there is no statistically significant effect in the
exposed group as compared to control.
bLOAEL is identified in this assessment as the lowest tested exposure dose at which there is a statistically significant effect in the
exposed group as compared to control.
°BMD modeling results are inadequate for use in deriving a POD. Therefore, the NOAEL/LOAEL approach is used to determine a
POD for these endpoints. BMD analysis for these endpoints is described in Appendix F.
dBased on the NOAEL of 50 ppm.
Based on a review of the data in Table 5-4. hepatic centrilobular necrosis was shown to be less
sensitive than the nasal effects and was not considered further as a candidate critical effect. Similarly, the
squamous cell metaplasia and hyperplasia of the respiratory epithelium yielded potential PODs that were
at least 3-fold higher than the remaining nasal effects; thus, these two effects were not considered further
as candidate critical effects. The PODs (adjusted for continuous exposure) for sclerosis of the lamina
propria, atrophy of the olfactory epithelium, and respiratory metaplasia of the olfactory epithelium were
identical (32.2 mg/m3) and similar to the PODADJ for hydropic change of the lamina propria (30.2 mg/m3).
Although the PODADi estimates for these four endpoints were either identical or similar, the magnitude of
response (i.e., increased incidence of effect) at each PODAoj for these effects varied (i.e., 0% for
sclerosis, 10% for hydropic change, 59% for respiratory metaplasia, 80% for atrophy).
As shown in Table 5-3. atrophy and respiratory metaplasia of the olfactory epithelium were the
most sensitive effects based on responses of 80 and 59% at their respective PODs of 50 ppm (LOAELs).
Increased incidences of the other nasal effects, as well as liver effects (i.e., centrilobular necrosis), were
observed at exposures of 50 ppm or greater and the magnitude of the responses at these exposures were
lower than those observed for atrophy and respiratory metaplasia of the olfactory epithelium. Typically,
120
-------
chemically-induced nasal effects include atrophy and/or necrosis, cell proliferation/hyperplasia, and
metaplasia depending on the nature of the tissue damage and level of exposure (Harkema et al. 2006;
Boorman et al.. 1990; Gaskell. 1990). However, the pathological progression of these events is uncertain
and often accompanied by an inflammatory response. Since the data do not support a continuum of
pathological events associated with respiratory tract effects, both atrophy and respiratory metaplasia of
the olfactory epithelium were selected as co-critical effects in this assessment. Additionally, these effects
were the most sensitive noncancer effects considered following inhalation of 1,4-dioxane.
For the derivation of a RfC based upon an animal study, the selected POD must be adjusted to
reflect the human equivalent concentration (HEC). The HEC was calculated by the application of a
dosimetric adjustment factor (DAF), in accordance with the U.S. EPAMethods for Derivation of
Inhalation Reference Concentrations and Application of Inhalation Dosimetry (hereafter referred to as the
RfC methodology) (U.S. EPA. 1994b). DAFs are ratios of animal and human physiologic parameters, and
are dependent on the nature of the contaminant (particle or gas) and the target site (e.g., respiratory tract
or remote to the portal-of-entry) (U.S. EPA. 1994b).
1,4-Dioxane is miscible with water and has a high blood:air partition coefficient. Typically,
highly water-soluble and directly reactive chemicals (i.e., Category 1 gases) partition predominantly into
the upper respiratory tract, induce portal-of-entry effects, and do not accumulate significantly in the
blood. 1,4-Dioxane induces effects at the portal-of-entry (i.e., respiratory tract), liver, and kidneys, and it
has been measured in the blood after inhalation exposure (Kasai et al.. 2009; Kasai etal.. 2008). The
observations of systemic (i.e., nonrespiratory) effects and measured blood levels resulting from
1,4-dioxane exposure indicate that this compound is absorbed into the bloodstream and distributed
throughout the body. Thus, 1,4-dioxane might be best described as a water-soluble and non-directly
reactive gas. Gases such as these are readily taken up into respiratory tract tissues and can also diffuse
into the blood (Medinsky and Bond. 2001). The effects observed in the olfactory epithelium may be the
result of the metabolism of 1,4-dioxane to an acid metabolite; however, for the reasons stated above, it is
unclear whether or not these effects are solely the result of portal-of-entry or systemic delivery. A similar
pattern of effects was observed after oral exposure to 1,4-dioxane (Kano et al.. 2009; JBRC. 1998).
121
-------
In consideration of the evidence described above, the human equivalent concentration (HEC) for
1,4-dioxane was calculated by the application of the appropriate dosimetric adjustment factor (DAF) for
systemic acting gases, in accordance with the U.S. EPA RfC methodology (U.S. EPA. 1994b).
The calculation of the HEC used in this assessment is as follows:
DAF = (Hb/g)A/(Hb/g)H
DAF = 1,861/1,666
DAF = 1.12
where:
(Hb/g)A = the animal bloochair partition coefficient =1,861 (Sweeney et al. 2008)
(Hb/g)H = the human blood:air partition coefficient =1,666 (Sweeney et al., 2008)
Given that the animal blood:air partition coefficient is higher than the human value resulting in a DAF>1,
a default value of 1 is substituted in accordance with the U.S. EPA RfC methodology (U.S. EPA. 1994b).
Analysis of the existing inhalation dosimetry modeling database supports the application of a DAF of 1
for a systemic acting gas (U.S. EPA. 2012a. 2009a). In addition, a robust computational fluid dynamic
(CFD) and PBPK modeling database supports the scientific rationale to apply a DAF of 1 for both portal
of entry and systemic effects irrespective of "gas categorization" (U.S. EPA. 2012a). Application of these
models to gases that have similar physicochemical properties and induce similar nasal effects as
1,4-dioxane yield estimated DAFs > 1.
Utilizing a DAF of 1, the HEC for atrophy and respiratory metaplasia of the olfactory epithelium
in male F344/DuCrj rats is calculated as follows:
PODHEc (mg/m3) = PODADj (mg/m3) x DAF
= PODADj (mg/m3) x 1.0
= 32.2 mg/m3 x 1.0
= 32.2 mg/m3
Therefore, the PODHEc of 32.2 mg/m3 for the co-critical effects of atrophy and respiratory
metaplasia of the olfactory epithelium is used for the derivation of a RfC for 1,4-dioxane.
122
-------
5.2.4. RfC Derivation- Including Application of Uncertainty Factors
(UFs)
The RfC of 3 x 1CT2 mg/m3 is based on atrophy and respiratory metaplasia of the olfactory
epithelium in male rats exposed to 1,4-dioxane via inhalation for 2 years (Kasai et al., 2009). The RfC for
1,4-dioxane is derived by dividing the PODHEc by a composite UF of 1,000.
RfC = PODHEC / UF
= 32.2 mg/m3 / 1,000
= 0.0322 or 3 X 10~2 mg/m3 (rounded to 1 significant figure)
An interspecies UF of 3 (UFA) was used for animal-to-human extrapolation to account for
pharmacodynamic differences between species. This uncertainty factor is comprised of two separate areas
of uncertainty to account for differences in the toxicokinetics and toxicodynamics of animals and humans.
In this assessment, the toxicokinetic uncertainty was accounted for by the calculation of a HEC and
application of a dosimetric adjustment factor as outlined in the RfC methodology (U.S. EPA. 1994b). As
the toxicokinetic differences are thus accounted for, only the toxicodynamic uncertainties remain, and an
UFA of 3 is retained to account for this uncertainty.
A default interindividual variability UF of 10 (UFH) was 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. However, a recent modeling study by Valcke and Krishnan (2011) assessed the impact of
exposure duration and concentration on the human kinetic adjustment factor and estimated the neonate to
adult 1,4-dioxane blood concentration ratio to be 3.2. Thus, a full factor of 10 was used to account for
differences between adults and neonates, as well as other differences in gender, age, health status, or
genetics that might result in a different disposition of, or response to, 1,4-dioxane.
An UF to extrapolate from a subchronic to a chronic (UFS) exposure duration was not necessary
(e.g., UFS = 1) because the RfC was derived from a study using a chronic exposure protocol.
An UF of 10 (UFL) was used to extrapolate from a LOAEL to a NOAEL because a LOAEL was
used as the POD. A NOAEL for atrophy and respiratory metaplasia of the olfactory epithelium was not
identified in the study by Kasai et al. (2009).
An UF of 3 for database deficiencies (UFD) was applied due to the lack of a multigeneration
reproductive toxicity study.
123
-------
5.2.5. RfC Comparison Information
Figure 5-5 presents PODs, applied UFs, and derived candidate RfCs based on each of the
endpoints from the chronic inhalation study by Kasai et al. (2009) in male rats. The PODs are based on
the BMDLio, NOAEL, or LOAEL, and appropriate unit conversions, duration, and dosimetric
adjustments were applied before applications of UFs. The predominant noncancer effects of chronic
inhalation exposure to 1,4-dioxane include nasal and liver effects. Figure 5-5 provides a graphical display
of these effects that were observed in the Kasai et al. (2009) study. The nasal effects involving the
olfactory epithelium represent the most sensitive effects.
1000
0 01
• POD
nnAnimal-to-human
DHuman variation
LOAELto NOAEL
DSubchronicto Chronic
• Database deficiencies
5 RfC
Squamous cell Squamous cell Respiratory Atrophy in the Hydropic Sclerosis of the Centrilobular
metaplasia in hyperplasiain metaplasia in nasal olfactory change in lamina propria; necrosis in the
the respiratory the respiratory theolfactory epithelium; lamina propria; NOAEL liver; BMDL10
epithelium: epithelium: epithelium: LOAEL NOAEL
NOAEL NOAEL LOAEL
Kasai et al. (2009)
Figure 5-5.
Potential points of departure (POD) for candidate endpoints with
corresponding applied uncertainty factors and derived candidate RfCs
following chronic inhalation exposure of F344 male rats to 1,4-dioxane.
5.2.6. Previous RfC Assessment
An RfC for 1,4-dioxane was not previously available on the IRIS database.
124
-------
5.3. Uncertainties in the Oral Reference Dose and Inhalation
Reference Concentration
The following discussion identifies the uncertainties associated with deriving the RfD and RfC
for 1,4-dioxane. As presented earlier in this section (see Sections 5.1.2. 5.1.3 for the RfD and Sections
5.2.2. and 5.2.3 for the RfC), the uncertainty factor approach (U.S. EPA. 2002a. 1994b) was used to
derive the RfD and RfC for 1,4-dioxane. Using this approach, the POD was divided by a set of factors to
account for uncertainties associated with a number of steps in the analysis, including extrapolation from
LOAEL to NOAEL, extrapolation from animals to humans, a diverse population of varying
susceptibilities, and to account for database deficiencies. Because information specific to 1,4-dioxane was
unavailable to fully inform these extrapolations, default factors were generally applied.
An adequate range of animal toxicology data are available for the hazard assessment of
1,4-dioxane, as described throughout the previous section (Section 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 two subchronic bioassays in rabbits, guinea pigs, mice, and rats, and
two chronic inhalation bioassays in rats. Toxicity associated with inhalation exposure to 1,4-dioxane was
observed predominately in the liver and nasal cavity. 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 a higher POD and potential RfD compared
to the same values derived from 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 BMD modeling of the dose-response. However, confidence in the NOAEL can be derived
from additional studies (JBRC. 1998: NCI. 1978: Argus etal.. 1973: Argus etal.. 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.
The RfC was derived by applying UFs to a LOAEL for atrophy and respiratory metaplasia of the
olfactory epithelium. The incidence data for the observed effects were not amenable to BMD modeling
(see Appendix F). The LOAEL for these effects was less than or equal to the LOAEL or NOAEL for
other effects observed in the Kasai et al. (2009) 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.
125
-------
Pharmacokinetic models are useful to examine species differences in pharmacokinetic processing;
however, it was determined that dosimetric adjustment using pharmacokinetic modeling 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. A DAF was used to account for pharmacokinetic differences
between rodents and humans in the derivation of the RfC; however, there was no information to inform
pharmacodynamic differences between species, so an UF of 3 was used in derivation of the RfC to
account for these uncertainties.
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 and RfC. Human variation may be larger or
smaller; however, 1,4-dioxane specific data to examine the potential magnitude of over estimation or
under estimation are unavailable.
Uncertainties in the assessment of the health hazards of 1,4-dioxane are associated with
deficiencies in reproductive toxicity information. The oral and inhalation databases lack a multigeneration
reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
1,4-dioxane (Giavini etal.. 1985). This developmental study indicates that the developing fetus may be a
target of toxicity. No developmental studies are available following inhalation to 1,4-dioxane.
5.4. Cancer Assessment
5.4.1. Choice of Study/Data - with Rationale and Justification
5.4.1.1. Oral Study/Data
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, 1998;
Yamazaki et al.. 1994; NCI. 1978; Kocibaetal.. 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 (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 carcinomas, which
126
-------
may underestimate the combined incidence of rats with adenomas or carcinomas. 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, and nasal cavity
tumors were also seen in high-dose male and female rats.
Dr. Yamazaki (JBRC) provided data in an email to Dr Stickney (SRC) on 12/18/2006 (2006) that
showed that the survival of mice in the Kano et al. (2009) study was low in all male groups (31/50, 33/50,
25/50 and 26/50 in control, low-, mid-, and high-dose groups, respectively) and particularly low in
high-dose females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-, and high-dose groups,
respectively). These deaths occurred primarily during the second year of the study. Survival at 12 months
in male mice was 50/50, 48/50, 50/50, and 48/50 in control, low-, mid-, and high-dose groups,
respectively. Female mouse survival at 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-,
mid-, and high-dose groups, respectively (Yamazaki. 2006). Furthermore, these deaths were primarily
tumor related. Liver tumors were listed as the cause of death for 31 of the 45 pretermination deaths in
high-dose female Crj:BDFl mice (Yamazaki. 2006).
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)
Tumor Incidence
Animal dose
Study Species/strain/gender (mg/kg-day)
0
Kociba et al. Sherman rats, male and 14
(1974) female combined3'13 121
1,307
0
Male Osborne-Mendel „ .n
. b ^4U
rats
530
0
Female
nch^rno-Mondol nt<=b'c
640
M^l M O7O\
v ' o
Male B6C3F-, miced 720
830
0
Female B6C3F-I miced 380
860
Liver
1/1 06h
0/110
1/106
10/661
NA
NA
NA
0/3 1h
10/301
11 /291
8/49h
19/501
28/471
0/50h
21 /481
35/371
Nasal
cavity
0/1 06h
0/110
0/106
3/66
0/33h
12/26
16/331
0/34h
10/301
8/291
NA
NA
NA
NA
NA
NA
Peritoneal
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
Mammary
gland
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
127
-------
Table 5-5 (Continued): 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)
Tumor Incidence
Animal dose
Study Species/strain/gender (mg/kg-day)
0
Male F344/DuCrj 11
ratsd'c'r'u 55
274
0
Female F344/DuCrj 18
ratsa'ej'g 83
429
K^inn ot Til fOnOQ^
0
H 49
191
677
0
d 66
278
964
Liver
3/50
4/50
7/50
39/50iik
3/50
1/50
6/50
48/50iik
23/50
31/50
37/501
40/50j'k
5/50
35/50j
41/50j
46/50j'k
Nasal
cavity
0/50
0/50
0/50
7/50k
0/50
0/50
0/50
8/50iik
0/50
0/50
0/50
1/50
0/50
0/50
0/50
1/50
Peritoneal
2/50
2/50
5/50
28/50iik
1/50
0/50
0/50
0/50
NA
NA
NA
NA
NA
NA
NA
NA
Mammary
gland
1/50
2/50
2/50
6/50k
8/50
8/50
11/50
18/50i|k
NA
NA
NA
NA
NA
NA
NA
NA
"Incidence of hepatocellular carcinoma.
blncidence of nasal squamous cell carcinoma.
Incidence of hepatocellular adenoma.
Incidence of hepatocellular adenoma or carcinoma.
Incidence (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 at p < 0.05 by Fisher's Exact test.
'Significantly different from control at p < 0.01 by Fisher's Exact test.
kp < 0.01; positive dose-related trend (Peto's test).
NA = data were not available for modeling
5.4.1.2. Inhalation Study/Data
Epidemiological studies of populations exposed to 1,4-dioxane via inhalation are not adequate for
dose-response analysis and thus derivation of an inhalation unit risk (IUR). However, two chronic
inhalation studies in animals are available and were evaluated for the potential to estimate an IUR
(Table 5-6). The chronic inhalation study conducted by Torkelson et al. (1974) in rats did not find any
treatment-related tumors; however, only a single exposure concentration was used (111 ppm 1,4-dioxane
vapor for 7 hours/day, 5 days/week for 2 years). A chronic bioassay of 1,4-dioxane by the inhalation route
reported by Kasai et al. (2009) provides data adequate for dose-response modeling and was subsequently
chosen as the study for the derivation of an IUR for 1,4-dioxane. In this bioassay, groups of 50 male F344
rats were exposed to either 0, 50, 250 or 1,250 ppm 1,4-dioxane, 6 hours/day, 5 days/week, for 2 years
(104-weeks). In male F344 rats, 1,4-dioxane produced a statistically significant increase in incidence
and/or a statistically significant dose-response trend for the following tumor types: hepatomas, nasal
squamous cell carcinomas, renal cell carcinomas, peritoneal mesotheliomas, mammary gland
128
-------
fibroadenomas, Zymbal gland adenomas, and subcutis fibromas (Kasai et al., 2009). The incidence of
adenomas and carcinomas were combined in this assessment in accordance with EPA's Guidelines on
Carcinogen Risk Assessment which notes that etiologically similar tumor types, i.e., benign and malignant
tumors of the same cell type, can be combined due to the possibility that benign tumors could progress to
the malignant form (U.S. EPA. 2005a; McConnell et al., 1986). Consistent with the oral cancer
assessment (Appendix D). the incidence of hepatic adenomas and carcinomas (combined) was used to
calculate an IUR (see Table 5-6).
Table 5-6 Incidence of liver, nasal cavity, kidney, peritoneal, and mammary gland,
Zymbal gland, and subcutis tumors in rats exposed to 1,4-dioxane via
inhalation for 2 years.
Study
Species/ Animal
strain/ Exposure
gender (ppm)
Tumor Incidence
Nasal
Mammary Zymbal
Liver0 cavityd Kidney6 Peritoneal' gland gland9 Subcutis
Torkelson
et al.
(1974f
Kasai et al.
(2009)b
Male
Wistar
rats
Female
Wistar
rats
Male
E1AA
rats
0
111
0
111
0
50
250
1,250
0/150
0/206
0/139
0/217
1/50
2/50
4/50
22/50
0/150
0/206
0/139
0/217
0/50
0/50
1/50
6/50m
0/1 501
1/2061
1/1 39j
0/21 7j
0/50
0/50
0/50
4/50
NA
NA
NA
NA
2/50
4/50
14/50"
41/50"
NA
NA
11/139k
29/2 17k
1/501
2/501
3/501
5/501
NA
NA
NA
NA
0/50
0/50
0/50
4/50
0/150
2/206
0/139
0/217
1/50
4/50
9/50"
5/50
Incidence reported based on survival to 9 months.
""Incidence reported based on survival to 12 months.
Incidence of hepatocellular adenoma or carcinoma. For Kasai et al. (2009) incidence data was provided via email from Dr. Tatsuya
Kasai (JBRC) to Dr. Reeder Sams (U.S.EPA) on 12/23/2008 (2008). Statistics were not reported. Individual incidence rates for
adenomas and carcinomas are in Table 5-8.
Incidence of nasal squamous cell carcinoma.
Incidence of renal cell carcinoma.
'incidence of peritoneal mesothelioma.
Incidence of Zymbal gland adenoma.
hlncidence of subcutis fibroma.
'Incidence of kidney fibroma.
'Incidence of kidney adenocarcinoma
Incidence of mammary gland adenoma.
'incidence of mammary gland fibroadenoma.
mTumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.05).
"Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p & 0.01).
NA = data are not available
129
-------
5.4.2. Dose-Response Data
5.4.2.1. Oral Data
Table 5-7 summarizes the incidence of hepatocellular adenoma or carcinoma in rats and mice
from the Kano et al. (2009) 2-year drinking water study. There were statistically significant increasing
trends in tumorigenic response for males and females of both species. The dose-response curve for female
mice is steep, with 70% incidence of liver tumors occurring in the low-dose group (66 mg/kg-day).
Exposure to 1,4-dioxane increased the incidence of these tumors in a dose-related manner.
A statistically significant increase in the incidence of peritoneal mesotheliomas was observed in
high-dose male rats only (28/50 rats, Table 5-5). The incidence of peritoneal mesotheliomas was lower
than the observed incidence of hepatocellular adenomas or carcinomas in male rats (Table 5-7): therefore,
the incidence of hepatocellular adenomas or carcinomas was used to derive an oral CSF for 1,4-dioxane.
Table 5-7 Incidence of hepatocellular adenomas or carcinomas in rats and mice exposed
to 1,4-dioxane in drinking water for 2 years
Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male Crj:BDF1 mice
Female Crj:BDF1 mice
Animal dose
(mg/kg-day)
0
11
55
274
0
18
83
429
0
49
191
677
0
66
278
964
Incidence of liver tumors3
3/50
4/50
7/50
39/50b'c
3/50
1/50
6/50
48/50b'c
23/50
31/50
37/50d
40/50b'c
5/50
35/50c
41/50C
46/50b'c
"Incidence of either hepatocellular adenomas or carcinomas.
bp < 0.05; positive dose-related trend (Peto's test).
°Significantly different from control at p < 0.01 by Fisher's Exact test.
Significantly different from control at p < 0.01 by Fisher's Exact test.
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009)
130
-------
5.4.2.2. Inhalation Data
Multi-tumor dose-response modeling was performed for all tumor responses from the Kasai et al.
(2009) bioassay. Kasai et al. (2009) reported tumor incidence data for male F344 rats exposed via
inhalation to 0, 50, 250, or 1,250 ppm 1,4-dioxane for 6 hours/day, 5 days/week, for 2 years (104-weeks).
Statistically significant positive dose-response trends were observed for nasal cavity squamous cell
carcinomas, hepatomas, renal cell carcinomas, peritoneal mesotheliomas, mammary gland fibroadenomas,
and Zymbal gland adenomas. Following 250 ppm 1,4-dioxane exposure, statistically significantly
elevated tumor incidences were found in two tissue types (i.e., peritoneal mesothelioma and subcutis
fibroma) compared to controls. It is important to note, for observations of subcutis fibroma, the incidence
was increased compared to controls at all concentrations, but a decrease in incidence, compared to the
mid-concentration, was noted at the highest concentration (1,250 ppm). However, a statistically
significantly decreased survival rate was noted in this exposure group by the study authors. Interim
sacrifices were not performed. Tumor incidences following 1,250 ppm inhalation exposure to 1,4-dioxane
were statistically elevated compared to controls in three tissues (i.e., nasal cavity squamous cell
carcinoma, hepatomas, and peritoneal mesothelioma). Incidence data for the tumor types reported by
Kasai et al. (2009) are summarized in Table 5-8.
Table 5-8 Incidence of tumors in F344 male rats exposed to 1,4-dioxane via inhalation for
104 weeks (6 hours/day, 5 days/week)
Animal Exposure (ppm)
Tumor Type
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma8
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Mammary gland adenoma
Zymbal gland adenoma
Subcutis fibroma
0
0/50
1/50
0/50
1/50
0/50
2/50
1/50
0/50
0/50
1/50
50
0/50
2/50
0/50
2/50
0/50
4/50
2/50
0/50
0/50
4/50
250
1/50
3/50
1/50
4/50
0/50
14/50C
3/50
0/50
0/50
9/50c
1,250
6/50a'b
21/50a'c
2/50
22/50a'c
4/50a
41/50a'c
5/50d
1/50
4/50a
5/50
Statistically significant trend for increased tumor incidence by Peto's test (p £ 0.01).
bTumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.05).
Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.01).
dStatistically significant trend for increased tumor incidence by Peto's test (p & 0.05).
Provided via email from Dr. Tatsuya Kasai (JBRC) to Dr. Reeder Sams (U.S. EPA) on 12/23/2008 (2008). Statistics were not
reported for these data by study authors, so statistical analyses were conducted by EPA.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2009) and Kasai (2008)
131
-------
5.4.3. Dose Adjustments and Extrapolation Method(s)
5.4.3.1. Oral
Human equivalent doses (HEDs) were calculated from the administered animal doses using a BW
scaling factor (BW°75) (U.S. EPA. 2011). This was accomplished using the following equation:
/animal BW [kg] \°-25
HED = animal dose (mg/kg) x =-^
^ 5/ &J Vhuman BW [kg]/
For all calculations, a human BW of 70 kg was used. HEDs for the principal study (Kano et al..
2009) are given in Table 5-9. HEDs were also calculated for supporting studies (NCI. 1978; Kociba et al..
1974) and are also shown in Table 5-9.
132
-------
Table 5-9 Calculated HEDs for the tumor incidence data used for dose-response modeling
Animal BW (g)
Study Species/strain/gender TWA
432a
Male F344/DuCrj rats 432a
432a
267a
Female F344/DuCrj rats 267a
267a
4/.ya
Male Crj:BDF1 mice 47.9a
47.9a
35.9a
Female Crj:BDF1 mice 35.9a
35.9a
325b
Kociba et al. (1974) ™^r™° H^B (comDinea) 325b
285C
470b
470b
31 Ob
31 Ob
MPI M OT'M
32b
32b
30b
30b
Animal dose HED
(mg/kg-day) (mg/kg-day)d
11
81
398
18
83
429
49
191
677
66
278
964
14
121
1,307
240
530
350
640
720
830
380
860
3.1
23
112
4.5
21
107
7.9
31
110
10
42
145
3.7
32
330
69
152
90
165
105
121
55
124
aTWA 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).
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 liver tumors is not conclusive. Therefore,
a linear low-dose extrapolation approach was used to estimate human carcinogenic risk associated with
1,4-dioxane oral exposure.
However, several of the external peer review panel members for the oral assessment (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 nonlinear extrapolation approach to estimate human
133
-------
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 4.5.1. 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 (Mivagawaetal.. 1999; Uno etal.. 1994; Goldsworthy et al., 1991; Lundberg et
al.. 1987; Bulletal. 1986; Stottetal. 1981; Kingetal.. 1973). However, key events related to the
promotion of tumor formation by 1,4-dioxane are not conclusive. 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 nonlinear approach. EPA determined that there are no data
available to inform the low-dose region of the dose response, and thus, a nonlinear approach was not
included.
Accordingly, the CSF for 1,4-dioxane was derived via a linear extrapolation from the POD
calculated by fitting a curve in BMDS to 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 (Appendix D);
however, the POD used to derive the CSF is based on the most sensitive species and target organ in the
principal study (Kano et al., 2009).
The oral CSF was calculated using the following equation:
= BMR/BMDLHED
5.4.3.2. Inhalation
In accordance with the U.S. EPA (1994b) RfC methodology, the HEC values were calculated by
the application of DAFs. As discussed in Section 5.2.3. since 1,4-dioxane is miscible with water, has a
high partition coefficient, and induces effects throughout the body of the rat, this substance was
considered to be a systemic acting gas and a DAF of 1.0 was applied. The lifetime continuous inhalation
risk for humans is defined as the slope of the line drawn from the POD through the origin, with the POD
defined as the lower 95% bound on the exposure associated with a level of extra risk near the low end of
the data range.
All PODs were converted to equivalent continuous exposure levels by multiplying by [(6
hours)/(24 hours)] x[(5 days)/(7 days)], under the assumption of equal cumulative exposures leading to
equivalent outcomes.
134
-------
Given the multiplicity of tumor sites observed in animals, basing the IUR on one tumor site may
underestimate the carcinogenic potential of 1,4-dioxane via inhalation. Also, simply pooling the counts of
animals with one or more tumors (i.e., counts of tumor bearing animals) would tend to underestimate the
overall risk for tumors observed at independent sites and ignores potential differences in the
dose-response relationships across the sites (NRC. 1994; Bogen. 1990). NRC (1994) has also noted that
the assumption of independence across tumor types is not likely to produce substantial error in the risk
estimates unless tumors across multiple sites are known to be biologically dependent.
The U.S. EPA's BMDS (v2.2 beta) MS_Combo program was utilized as a computational
approach to calculating the dose associated with a specified composite risk under the assumption of
independence of tumors. The best fitting BMDS multistage model was determined for each individual
tumor type as shown in Section 5.4.4.2 and Appendix G. These models account for spontaneous tumor
generation in controls. The Guidelines for Carcinogen Risk Assessment recommend calculation of an
upper bound to account for uncertainty in the estimate (U.S. EPA. 2005a). Complete details of this
analysis are included in Appendix G. In addition, Bayesian MCMC computations were conducted as
described by Kopylev et al. (2009) using WinBugs (Spiegelhalter et al. 2003). For uncertainty
characterization, MCMC methods have the advantage of providing information about the full distribution
of risk and/or BMDs, which can be used in generating a confidence bound. This MCMC approach, which
builds on the re-sampling approach recommended by Bogen (1990). also provides a distribution of the
combined potency across sites. This supporting analysis was completed in addition to the MS_Combo
analysis and additional details are included in Appendix G.
Several hypothesized MOA(s) have been proposed for liver and nasal tumors, although these
MOA(s) are not supported by the available data (see Sections 4.7.3.3 and 4.7.3.4). Specifically, tumors
occur in rodent models in the absence of data to identify hypothesized key events (e.g., cytotoxicity).
Also, studies evaluating the kinetics of 1,4-dioxane suggest that liver carcinogenicity is related to the
accumulation of the parent compound following metabolic saturation; however, data are not available to
determine the toxic moiety (i.e., parent compound and/or metabolite(s)) (see Section 3.3 and 4.7.3.1.1).
For kidney, lung, peritoneal (mesotheliomas), mammary gland, Zymbal gland, and subcutis tumors, there
are no available data regarding any hypothesized carcinogenic MOA(s) for 1,4-dioxane.
The 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 MOA of the carcinogen and the shape of the cancer dose-response curve. The linear
extrapolation approach is used as a default option if the mode of carcinogenic action is not identified. A
nonlinear extrapolation approach can be used for cases with sufficient data to ascertain the mode of action
and to conclude that it is not linear at low doses. Also, nonlinear extrapolation having significant
biological support may be presented in addition to a linear approach when the available data and weight
of evidence support a nonlinear approach. In the case of 1,4-dioxane, there is insufficient biological
support to identify key events and to have reasonable confidence in the sequence of events and how they
relate to the development of tumors following exposure to 1,4-dioxane; thus, the data are not strong
enough to ascertain the mode of action applying the Agency's mode of action framework (U.S. EPA.
135
-------
2005a). Therefore, EPA concluded that a default linear extrapolation should be utilized to estimate the
cancer risk estimates for inhalation and oral exposure to 1,4-dioxane.
IUR estimates were calculated using the following equation:
IUR = BMR/BMCLHEc
5.4.4. Oral Slope Factor and Inhalation Unit Risk
5.4.4.1. Oral Slope Factor
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 drinking water (Kano et al.. 2009; NCI.
1978; Kocibaetal. 1974) (Table 5-5). Animal doses were used for BMD modeling, and then RED BMD
and BMDL values were calculated using BW3/4 scaling employing animal TWA body weights
(Table 5-10) and a human BW of 70 kg. For all models, a BMR of 10% extra risk was employed. BMDs
and BMDLs from all models are reported, and the model outputs and plots corresponding to the
best-fitting models are shown (Appendix D). When the best-fitting model is not a multistage model, the
multistage model output and plot are also provided (Appendix D). A summary of the BMD modeling
results for the Kano et al. (2009). NCI (1978). and Kociba et al. (1974) studies is shown in Table 5-10.
136
-------
Table 5-10 BMD RED and BMDLHED values from best-fit 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
Kano et al.
(2009)
Kociba et al.
(1974)
NCI (1978)
Gender/strain/species
Male F344/DuCrj ratsb
Female F344/DuCrj ratsc
Male Crj:BDF1 miced
Female Crj:BDF1 miced
Female Crj:BDF1 miced'e
Female Crj:BDF1 miced'f'h
Female F344/DuCrj rats9
Male F344/DuCrj rats9
Male F344/DuCrj ratsb
Female F344/DuCrj ratsd
Male and female (combined)
Sherman rats9
Male and female (combined)
Sherman ratsb
Male Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female B6C3F-I micec
Male B6C3Fi mice'
BMDHEDa BMDLHEDa Oral CSF
Tumor type (mg/kg-day) (mg/kg-day) (mg/kg-day)"1
Hepatocellular
carcinoma
Nasal
squamous ceil
carcinoma
Peritoneal
mesothelioma
Mammary
gland adenoma
Nasal
squamous cell
carcinomas
Hepatocellular
carcinoma
Nasal
squamous ceil
carcinomas
Hepatocellular
adenoma
Hepatocellular
adenoma or
carcinoma
17.43
19.84
5.63
0.83
3.22e
7.51f
94.84
91.97
26.09
40.01
448.24
290.78
16.10
40.07
28.75
23.12
87.98
14.33
14.43
2.68
0.55
2.12e
4.95f
70.23
68.85
21.39
20.35
340.99
240.31
10.66
25.82
18.68
9.75
35.67
7.0 x
6.9 x
3.7 x
1.8x
1.4 x
1.0x
1.4 x
1.5x
4.7 x
4.9 x
2.9 x
4.2 x
9.4 x
3.9 x
5.4 x
1.0 x
2.8 x
10'3
ID'3
1C'2
10'1
1C'1
10'1
ID'3
10'3
ID'3
ID'3
1C'4
1C'4
10'3
ID'3
ID'3
ID'2
ID'3
aValues associated with a BMR of 10% unless otherwise noted.
bProbit model, slope parameter not restricted.
""Multistage model, degree of polynomial = 2.
dLog-logistic model, slope restricted a 1.
eValues associated with a BMR of 30%.
Values associated with a BMR of 50%.
Multistage model, degree of polynomial =3.
hSee BMDS model output Figure D-12.
'Gamma model.
The multistage model did not provide an adequate fit (as determined by/>-value < 0.1, and % p >
10.11) to the data for the incidence of hepatocellular adenoma or carcinoma in female mice (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).
137
-------
The predicted BMD10 and BMDL10 for the female mouse data are presented in Table 5-10, 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 BMD10 and BMDL10 for female rat mammary
tumors and male peritoneal tumors obtained from the log-logistic and probit models, respectively, are
presented in Table 5-10.
A comparison of the BMD and BMDL estimates derived for rats and mice from the Kano et al.
(2009). NCI (1978). and Kociba et al. (1974) studies (Table 5-10) indicates that female mice are more
sensitive to liver carcinogenicity induced by 1,4-dioxane compared to other species or tumor types.
Therefore, 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:
0.50
CSF = 4.95 mg/kg - day (BMDL50 HED for female mice) = °-10 (mg/kg ~ d&y)
Calculation of a CSF for 1,4-dioxane is based upon the dose-response data for the most sensitive
species and gender.
5.4.4.2. Inhalation Unit Risk
As stated in Section 5.4.2.2. multiple tumor types have been observed in rats following inhalation
exposure to 1,4-dioxane. These data have been used to develop IUR estimates for 1,4-dioxane. The
multistage cancer models available in the BMDS (version 2.1.1) were fit to the incidence data for each
tumor type observed in rats exposed to 1,4-dioxane via inhalation (Kasai et al.. 2009) to determine the
degree (e.g., 1st, 2nd, or 3rd) of the multistage model that best fit the data (details in Appendix G). In
contrast to the oral slope factor analysis, suitable multistage model fits were obtained for all of the
datasets included in the inhalation unit risk analysis. Then, the best fitting models for each endpoint were
used in the BMDS (version 2.2Beta) MS_Combo program to estimate a total tumor BMC and BMCL10. A
Bayesian MCMC analysis was also performed using WinBUGS to calculate the total tumor risk and it
yielded similar results (see Appendix G). A summary of the BMDS model predictions for the Kasai et al.
(2009) study is shown in Table 5-11. Experimental exposure concentrations were used for BMD
modeling and then continuous human equivalent exposures were calculated by adjusting for duration of
exposure (Table 5-11) and applying an appropriate DAF (see Section 5.2.3). In accordance with the U.S.
EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). the BMCLio (lower bound on the
concentration estimated to produce a 10% increase in tumor incidence over background) was estimated
for the dichotomous incidence data and the results of the model that best characterized the cancer
incidences were selected. BMCs and BMCLs from all models are reported, and the output and plots
corresponding to the best-fitting model are shown (Appendix G).
The IUR estimates are provided in Table 5-11. Human equivalent risks estimated from the
individual rat tumor sites ranged from 2 x 10"7 to 2 x 10"6 ((ig/m3)"1 (rounded to one significant figure).
138
-------
The highest IUR (2 x 10~6 ((ig/m3)"1) corresponded to peritoneal mesotheliomas in male rats, and the
lowest IUR (2 x 10~7 (jig/m3)"1) corresponded to renal cell carcinoma and Zymbal gland adenomas in male
rats. The MS_Combo analysis yielded an IUR estimate of 5 x io~6 (jig/m3)"1.
Table 5-11 Dose-response modeling summary results for male rat tumors associated with
inhalation exposure to 1,4-dioxane for 2 years
Point of Departure0
Tumor Type3
Multistage
Model
Degree"
Nasal cavity squamous cell .
carcinoma
Hepatocellular adenoma
carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland
fibroadenoma
Zymbal gland adenoma
Subcutis fibroma
1
3
1
1
3
1
Bioassay Exposure
Concentration (ppm)
BMC10
1,107
252.8
1,355
82.21
1,635
1,355
141.8
BMCL10
629.9
182.3
1,016
64.38
703.0
1,016
81.91
HEC
(mg/m3)d
BMC10
712.3
162.7
872
52.89
1,052
872
91.21
BMCL10
405.3
117.3
653.7
41.42
452.4
653.7
52.70
IUR
Estimate6
(ug/m3y1
2.5 x 10'7
8.5 x 10"7
1.5 x 10"7
2.4 x 10'6
2.2 x 10"7
1.5 x 10"7
1.9 x 10'6
BMDS MS_Combo Total
Tumor Analysis'
40.4
30.3
26.0
19.5
5.0 x 10'6
"Tumor incidence data from Kasai et al. (2009).
bBest-fitting multistage model degree (p>0.1, lowest AIC). See Appendix G for modeling details.
°BMC = Concentration at specified extra risk (benchmark dose); BMCL = 95% lower bound on concentration at specified extra risk.
dHuman continuous equivalent estimated by multiplying exposures by [(6 hours)/(24 hours) x (5 days)/(7 days) x molecular weight of
1,4-dioxane]/ 24.45.
eThe inhalation unit risk (ug/m3)"1 was derived from the BMCLio, the 95% lower bound on the concentration associated with a 10%
extra cancer risk. Specifically, by dividing the BMR (0.10) by the BMCL10. Thus, representing an upper bound, continuous
lifetime exposure estimate of cancer potency.
'Results in this table are from the BMDS MS_Combo model (see model output in Appendix G. Section G.3). Additionally, Bayesian
analysis using WinBUGS was performed and yielded similar results (see Appendix G. Section G.4).
Given the multiplicity of tumor sites, basing the inhalation unit risk on one tumor site may
underestimate the carcinogenic potential of 1,4-dioxane. Consistent with recommendations of the NRC
(1994) and the EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). the total risk and
upper bound risk for all tumor sites in male F344 rats was estimated. This estimate of total risk describes
the risk of developing any combination of the tumor types considered. As shown in Table 5-11. the
resulting inhalation unit risk for all tumor types in male F344 rats was 5 x 10~6 (jig/m3)"1. Consideration of
all tumor sites approximately doubled the unit risk compared to the highest unit risk associated with any
individual tumor type, 2 x 10~6 ((ig/m3)"1 for male peritoneal mesotheliomas.
139
-------
The HEC BMCL10 for the combined tumor estimate in male rats was chosen as the POD and the
lURof 5 x io~6 ((ig/m3)"1 was calculated as follows:
IUR (mg/m3)-1 = 19j5°rcg/m3 = 0.005 (mg/m3)-1
lUR^g/m3)-1 = 0.005 (mg/m3)-1 x ^j- = 5 x 10~6
IUR ([ig/m3)-1 = SxlO
Based on the analysis discussed above, the recommended upper bound estimate on human extra
cancer risk from continuous lifetime inhalation exposure to 1,4-dioxane is 5 x 10~6 ((ig/m3)"1. The IUR
reflects the exposure-response relationships for the multiple tumor sites in male F344 rats.
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 evidence or no data from human epidemiology studies (U.S. EPA. 1986aV) 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. An inhalation unit risk was not previously derived.
5.5. Uncertainties in Cancer Risk Values
In this assessment, extrapolation of high-dose data from laboratory animals to estimate potential
risks to human populations from low-dose exposure to 1,4-dioxane has engendered some uncertainty in
the results. Several types of uncertainty may be considered quantitatively, but other important
uncertainties can only be considered qualitatively. Thus, an overall integrated quantitative uncertainty
analysis is not presented. However, the sources of uncertainty and their potential impacts on the
assessment are described below and in Table 5-12.
140
-------
5.5.1. Sources of Uncertainty
5.5.1.1. Choice of Low-Dose Extrapolation Approach
The possibilities for the low-dose extrapolation of tumor risk from exposure to 1,4-dioxane, or
any chemical, are linear or 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
mesotheliomas of the peritoneal cavity (Kano et al. 2009; Kasai et al.. 2009; JBRC. 1998; NCI. 1978;
Kocibaetal. 1974). 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 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 its metabolite interacts with cells initiating 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 this model has a generic biological motivation based on long-established biologically-
based 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.
The multistage cancer model provided adequate fits for the tumor incidence data following a
2-year inhalation exposure to 1,4-dioxane by male rats (Kasai et al., 2009). In the studies evaluated for the
oral cancer assessment (Kano et al.. 2009; NCI. 1978; KocibaetaL. 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 via ingestion (Appendix
D for details). The multistage model did not provide an adequate fit for the female mouse liver tumor
dataset based upon the following (U.S. EPA. 2012b):
• Goodness-of-fit/>-value was less than 0.10 indicating statistically significant lack of fit;
• Akaike's Information Criterion (AIC) was larger than other acceptable models;
• Observed data deviated substantially from the fitted model, as measured by their standardized
%2 residuals (i.e., residuals with values greater than an absolute value of one).
By default, the BMDS software imposes 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
141
-------
incidence data for female mouse liver adenomas or carcinomas, even after dropping the highest dose
group.
The log-logistic model was selected because it was the only model that provided an adequate fit
to the female mouse liver tumor 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 BMDL50HED was estimated by applying
appropriate parameter constraints to the selected model, consistent with the BMD Technical Guidance
Document (U.S. EPA. 2012R
The human equivalent oral CSFs estimated from tumor datasets with statistically significant
increases ranged from 4.2 x 10"4to 1.0 x 10"1 permg/kg-day (Table 5-10). arange of about three orders of
magnitude, with the upper and lower 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 carcinogenicity. 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 and IUR is unknown.
5.5.1.3. Cross-Species Scaling
For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied
(U.S. EPA. 2011) to address toxicological equivalence of internal doses between each rodent species and
humans, consistent with the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). It is assumed
that equal risks result from equivalent constant lifetime exposures.
Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
However, since similar cell types are prevalent throughout the respiratory tract of both rats and humans,
the tumors are considered biologically plausible and relevant to humans.
5.5.1.4. 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 following oral exposure, there is a reasonably small
142
-------
degree of uncertainty at the 50% excess incidence level (the POD for linear low-dose extrapolation), as
indicated by the proximity of the BMDLHED (4.95 mg/kg-day) to the BMDHED (7.51 mg/kg-day). For the
multistage model applied for the male rat inhalation dataset, there is a reasonably small degree of
uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
5.5.1.5. Bioassay 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 (rats and mice) with a sufficient number
(N=50) 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 (NCL 1978; KocibaetaL 1974) were available and were
fully considered for the derivation of the oral CSF.
The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
Three dose levels plus an untreated control group were examined following exposure to 1,4-dioxane via
inhalation for 2 years.
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 shown 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])
resulted in a considerable and significant increase in combined liver adenomas and carcinomas observed.
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.
Dr. Yamazaki (JBRC) provided in an email to Dr. Stickney (SRC) on 12/18/2006 (2006) that the
survival of mice in the Kano et al. (2009) study was particularly low in high-dose females (29/50, 29/50,
17/50, and 5/50 in control, low-, mid-, and high-dose groups, respectively). These deaths occurred
primarily during the second year of the study. Female mouse survival at 12 months was 50/50, 50/50,
48/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively (Yamazaki. 2006).
Furthermore, these deaths were primarily tumor related. Liver tumors were listed as the cause of death for
1/21, 2/21, 8/33, and 31/45 of the pretermination deaths in control, low-, mid- and, high-dose female
Crj:BDFl mice (Yamazaki. 2006). Therefore, because a number of the deaths in female mice were
attributed to liver tumors, this endpoint and species was still considered to be relevant for this analysis;
however, the high mortality rate does contribute uncertainty. Additionally, the oral CSF may actually be
larger if the survival adjusted tumor data were available.
143
-------
Additionally, the incidence of hepatocellular adenomas and carcinomas in historical controls was
evaluated with the data from Kano et al. (2009). Katagiri et al. (1998) summarized the incidence of
hepatocellular adenomas and carcinomas in control male and female BDF1 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 BDF1. These incidence rates are near the historical control values, and thus are appropriate
for consideration in this assessment.
Male F344 rat data were used to estimate risk following inhalation of 1,4-dioxane. Kano et al.
(2009) showed that male rats were more sensitive than female rats to the effects of 1,4-dioxane following
oral administration; therefore, male rats were chosen to be studies in the 2-year bioassay conducted by the
same laboratory (Kasai et al.. 2009). The sensitivity and tumorigenic response of female rats or male or
female mice following inhalation of 1,4-dioxane is unknown. Since female mice were the most sensitive
gender and species examined in the Kano et al. (2009) oral study, female mice may also be more sensitive
to the inhalation of 1,4-dioxane, which would result in a greater risk.
5.5.1.7. Relevance 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 re levant to humans due to exposure to 1,4-dioxane.
The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
rats. There is no information on 1,4-dioxane to indicate that the observed rodent tumors are not relevant to
humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
rodents and humans. In the absence of information to indicate otherwise and considering similar cell types
are prevalent throughout the respiratory tract of rats and humans, the nasal, liver, renal, peritoneal,
mammary gland, Zymbal gland and subcutis tumors were considered relevant to humans.
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.
144
-------
Table 5-12 Summary of uncertainty in the 1,4-dioxane cancer risk estimation
Consideration/
approach
Low-dose
extrapolation
procedure
Dose metric
Potential Impact
Departure from EPA's
Guidelines for
Carcinogen Risk
Assessment POD
paradigm, if justified,
could 1 or t unit risk
an unknown extent
Alternatives could
t or J, CSF by an
Decision
Log-logistic model
to determine POD,
for CSF;
Combined tumor
modeling for IUR;
linear low-dose
extrapolation from
POD
Used administered
exposure
Justification
A linear low-dose extrapolation approach was used to
estimate human carcinogenic risk associated with
1,4-dioxane exposure. Where data are insufficient to
ascertain the MOA, EPA's 2005 Guidelines for
Carcinogen Risk Assessment recommend application
of a linear low-dose extrapolation approach.
Experimental evidence supports a role for metabolism
in toxicity, but it is unclear if the parent compound,
unknown extent
metabolite or both contribute to 1,4-dioxane toxicity.
Cross-species
scaling
Alternatives could
1 or f CSF [e.g.,
3.5-fold I (scaling by
BW) or T twofold
(scaling by BW067 )]
BW075 (default
approach)
There are no data to support alternatives.
BW075 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 CSF (Kano et al
t or I cancer potency 2009):
by an unknown extent
IUR (Kasai et al
Alternative bioassays were available and considered
for derivation of oral CSF and inhalation IUR.
2009)
Species /gender
combination
Human
relevance of
mouse tumor
data
Human
population
variability in
metabolism and
response/
sensitive
subpopulations
Human risk could
1 or t, depending on
relative sensitivity
If rodent tumors
proved not to be
relevant to humans,
unit risk would not
apply i.e., could
Risk t or I to an
unknown extent
Female mouse
(CSF);
Male rat (IUR)
Mouse liver
adenomas and
carcinomas are
relevant to humans
(basis for CSF).
Rat tumors at
multiple sites are
relevant to humans
(basis for IUR)
Considered
qualitatively
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. No female mouse data were
available for derivation of the IUR.
1,4-dioxane is a multi-site carcinogen in rodents and
the MOA(s) is unknown; carcinogenicity observed in
the rodent studies is considered relevant to human
exposure.
No data to support range of human
variability/sensitivity, including whether children are
more sensitive.
145
-------
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, kidney, and nasal.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. 1998: Drewetal. 1978: David. 1964: Kestenetal.. 1939: Laugetal.. 1939:
Schrenk and Yant. 1936: deNavasquez. 1935: Fairlev 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.. 2008: JBRC. 1998: NCL 1978: Argus etal.. 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 etal.. 1974). Kidney damage was characterized by degeneration of the
cortical tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCI. 1978: Kociba etal.. 1974:
Argus etal.. 1973: Argus etal.. 1965: Fairlev etal.. 1934). In chronic inhalation studies conducted in rats,
nasal and liver toxicity were the primary noncancer health effects. Degeneration of nasal tissue
(i.e., metaplasia, hyperplasia, atrophy, hydropic change, and vacuolic change) and preneoplastic cell
proliferation were observed in the nasal cavity following inhalation exposure to 1,4-dioxane for 2 years
(Kasai et al.. 2009). Liver toxicity was described as necrosis of the centrilobular region and preneoplastic
changes were noted as well.
Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
pigs (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: NCI. 1978: Kociba etal.. 1974: Torkelson et al..
1974: Argus etal.. 1973: Hoch-Ligeti and Argus. 1970: Hoch-Ligeti et al.. 1970: Argus etal.. 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 and following inhalation exposure in
rats. Nasal (squamous cell carcinomas), peritoneal, mammary, Zymbal gland, and subcutaneous tumors
were also observed in rats, but were not seen in mice. With the exception of the NCI (1978) study, the
146
-------
incidence of nasal cavity tumors was generally lower than that of tumors observed in other tissues of the
same study population.
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is "likely
to be carcinogenic to humans" based on evidence of multiple tissue carcinogenicity in several 2-year
bioassays conducted in three strains of rats, two strains of mice, and in guinea pigs (Kano et al.. 2009;
Kasai et al. 2009; JBRC. 1998; NCI. 1978; Kocibaetal.. 1974; Argus etal. 1973; Hoch-Ligeti and
Argus. 1970; Hoch-Ligeti et al.. 1970; Argus etal.. 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 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 for liver
tumorsjnvolving 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 etal.. 1973). Some dose-response and temporal evidence support the occurrence of cell
proliferation prior to the development of liver tumors (JBRC. 1998; Kocibaetal.. 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. Data from rat and mouse bioassays (JBRC. 1998; 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.. 2009; JBRC.
1998; NCI. 1978). Data regarding a plausible dose response and temporal progression from cytotoxicity
to cell proliferation and eventual liver tumor formation are not available. Hypothesized MO As by which
1,4-dioxane induces tumors in other organ systems such as the respiratory system lack supporting data
(see Section 4.7.3).
6.2. Dose Response
6.2.1. Noncancer/Oral
The RfD of 3 x io~2 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 (Kocibaetal.. 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
147
-------
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 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/lnhalation
The RfC of 3 x 10~2 mg/m3 was derived based on co-critical effects of olfactory epithelium
atrophy and respiratory metaplasia in rats exposed for 2 years to 1,4-dioxane via inhalation (Kasai et al.
2009). This study was chosen as the principal study because it provides an adequate study design and the
most sensitive measure of adverse effects by 1,4-dioxane. The POD was derived using the LOAEL for
olfactory epithelium atrophy and respiratory metaplasia in male rats (Kasai et al.. 2009). A composite UF
of 1,000 was applied, consisting of factors of 10 for a LOAEL-to NOAEL extrapolation, 10 for
interindividual variability, 3 for animal-to-human extrapolation, and 3 for database deficiencies.
The overall confidence in the RfC is medium. Confidence in the principal study (Kasai et al..
2009) is medium. Confidence in the database is medium due to the lack of supporting studies and a
multigeneration reproductive toxicity study. Reflecting medium confidence in the principal study and
medium confidence in the database, the confidence in the RfC is medium.
6.2.3. Cancer
Under EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is
"likely to be carcinogenic to humans" by all routes of exposure. This descriptor is based on evidence of
carcinogenicity from animal studies.
6.2.3.1. 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 not
conclusive (see Section 4.7.3 for a more detailed discussion of 1,4-dioxane's hypothesized MOAs).
Therefore, based on the U.S. EPA's 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
148
-------
dose-response data from the range of observation and converting it to a HED (BMDL50 HED of
4.95 mg/kg-day).
The uncertainties associated with the quantitation of the oral CSF are discussed below.
6.2.3.2. Inhalation
The IUR for 1,4-dioxane of 5 x 10~6 ((ig/m3)"1 was based on a chronic inhalation study conducted
by Kasai et al. (2009). Statistically significant increases in tumor incidence and positive dose-response
trends were observed at multiple sites in the male rat including the nasal cavity (squamous cell
carcinoma), liver (adenoma), peritoneal (mesothelioma), and the subcutis (fibroma). Statistically
significant dose-response trends were also observed in the kidney (carcinoma), mammary gland
(fibroadenoma), and the Zymbal gland (adenoma). The available data indicate that the MOA(s) by which
1,4-dioxane induces tumors in rats is not conclusive (see Section 4.7.3 for a more detailed discussion of
1,4-dioxane's hypothesized MOAs). Therefore, based on the EPA's Guidelines for Carcinogen Risk
Assessment (U.S. EPA. 2005a). a linear low dose extrapolation was used. A combined tumor BMD
approach (see Section 5.4.3.2 and Appendix G for details) was used to calculate the POD for the total
tumor risk following inhalation of 1,4-dioxane. The POD was calculated by curve fitting the animal
experimental dose-response data from the range of observation and converting it to a continuous human
equivalent exposure.
The uncertainties associated with the quantitation of the IUR are discussed below.
6.2.3.3. Choice of Low-Dose Extrapolation Approach
The possibilities for the low-dose extrapolation of tumor risk from exposure to 1,4-dioxane, or
any chemical, are linear or 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
mesotheliomas 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 available for the remaining tumor types. In Section 4.7.3. hypothesized MOAs were
explored for 1,4-dioxane. Te available evidence in support of the hypothesized MOAs for 1,4-dioxane is
not conclusive. In the absence of a MOA(s) for the observed tumor types associated with 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
149
-------
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.
The multistage cancer model provided adequate fits for the tumor incidence data following a
2-year inhalation exposure to 1,4-dioxane by male rats (Kasai et al.. 2009). However, in the studies
evaluated for the oral cancer assessment (Kano et al.. 2009; NCI. 1978; Kocibaetal.. 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 via ingestion (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. 2012b):
• Goodness-of-fit/7-value was less than 0.10 indicating statistically significant lack of fit;
• AIC was larger than other acceptable models;
• Observed data deviated substantially from the fitted model, as measured by their standardized
%2 residuals (i.e., residuals with values greater than an absolute value of one).
By default, the BMDS software imposes 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, even after dropping the highest dose
group.
The log-logistic model was selected because it was the only model that provided an adequate fit
to the female mouse liver tumor 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 BMD Technical Guidance
Document (U.S. EPA. 2012b).
The human equivalent oral CSF estimated from liver tumor datasets with statistically significant
increases ranged from 4.2 x 10"4to 1.0 x 10"1 per mg/kg-day, arange of about three orders of magnitude,
with the upper and lower extremes coming from the combined male and female data for hepatocellular
carcinomas (Kocibaetal., 1974) and the female mouse liver adenoma and carcinoma dataset (Kano et al.,
2009).
6.2.3.4. 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.
150
-------
6.2.3.5. Cross-Species Scaling
For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied to
address toxicological equivalence of internal doses between each rodent species and humans, consistent
with the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). It is assumed that equal risks
result from equivalent constant lifetime exposures.
Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
6.2.3.6. 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 following oral exposure, there is a reasonably small
degree of uncertainty at the 50% excess incidence level (the POD for linear low-dose extrapolation), as
indicated by the proximity of the BMDLHED (4.95 mg/kg-day) to the BMDHED (7.51 mg/kg-day) . For the
multistage model applied for the male rat inhalation dataset, there is a reasonably small degree of
uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
6.2.3.7. Bioassay 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 (rats and mice) with a sufficient number (N=50)
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; Kocibaetal.. 1974) were available and were fully
considered for the derivation of the oral CSF.
The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
Three dose levels plus an untreated control group were examined following exposure to 1,4-dioxane via
inhalation for 2 years.
6.2.3.8. Choice of Species/Gender
The oral CSF for 1,4-dioxane was derived 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, from 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
151
-------
carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day [animal dose] 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
[animal dose] or 10 mg/kg-day [HED]) could refine and reduce uncertainty for the oral CSF.
Male F344 rat data were used to estimate risk following inhalation of 1,4-dioxane. Kano et al.
(2009) showed that male rats were more sensitive than female rats to the effects of 1,4-dioxane following
oral administration; therefore, male rats were studied in the 2-year bioassay conducted by the same
laboratory (Kasai et al., 2009). The sensitivity and tumorigenic response of female rats or male or female
mice following inhalation of 1,4-dioxane is unknown. Since female mice were the most sensitive gender
and species examined in the Kano et al. (2009) study, female mice may also be more sensitive to the
inhalation of 1,4-dioxane which would result in a greater risk.
6.2.3.9. Relevance to Humans
The oral CSF was 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 following exposure to
1,4-dioxane are not plausible in humans. Liver adenomas and carcinomas were considered plausible
outcomes in humans due to exposure to 1,4-dioxane.
The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
rats. There is no information on 1,4-dioxane to indicate that the observed rodent tumors are not relevant to
humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
rodents and humans.
6.2.3.10. 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 MOA
for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.
152
-------
REFERENCES
ACGIH (American Conference of Governmental Industrial Hygienists). (2011). 1,4-dioxane. Threshold limit
values for chemical substances and physical agents and biological exposure indices. In Documentation of the
threshold limit values and biological exposure indices (27th ed.). Cincinnati, OH.
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
ExpTher 292: 228-237.
Andersen. ME: Clewell HJ. Ill: Gargas. ML: Smith. FA: Reitz. RH. (1987). Physiologically based
pharmacokinetics and the risk assessment process for methylene chloride. Toxicol Appl Pharmacol 87: 185-
205. http://dx.doi.org/10.1016/0041-008X(87)90281-X
Argus. MF: Arcos. JC: Hoch-Ligeti. C. (1965). Studies on the carcinogenic activity of protein-denaturing agents:
Hepatocarcinogenicity of dioxane. JNatl Cancer Inst 35: 949-958.
Argus. MF: Sohal RS: Bryant. GM: Hoch-Ligeti. C: Arcos. JC. (1973). Dose-response and ultrastructural
alterations in dioxane carcinogenesis. Influence of methylcholanthrene on acute toxicity. Eur J Cancer 9:
237-243. http://dx.doi.org/10.1016/0014-2964(73)90088-l
Ashbv. J. (1994). The genotoxicity of 1,4-dioxane. Mutat Res 322: 141-142. http://dx.doi.org/10.1016/0165-
1218(94)00022-0
Atkinson. R. (1989). Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic
compounds. JPhys ChemRef Data 1: 1-246.
ATSDR (Agency for Toxic Substances and Disease Registry). (2005). Health consultation. 1,4-Dioxane in
private drinking water near Naval Air Station Whidbey Island, Ault Field.
http://www.docstoc.com/docs/27599091/Health-Consultation
ATSDR (Agency for Toxic Substances and Disease Registry). (2012). Toxicological profile for 1,4 dioxane
[ATSDR Tox Profile]. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service.
http://www.atsdr.cdc.gov/toxprofiles/tp.asp?id=955&tid=199
Bannasch. P. (2003). Comments on R. Karbe and R.L. Kerlin (2002) Cystic degeneration/spongiosis hepatis
(Toxicol Pathol 30 (2), 216-227) [Letter]. Toxicol Pathol 31: 566-570.
http://dx.doi.org/10.1080/01926230390224700
Bannasch. P: Moore. MA: Klimek. F: Zerban. H. (1982). Biological markers of preneoplastic foci and neoplastic
nodules in rodent liver. Toxicol Pathol 10: 19-34. http://dx.doi.org/10.1177/019262338201000204
Barber. H. (1934). Haemorrhagic nephritis and necrosis of the liver from dioxan poisoning. Guy's Hosp Rep 84:
267-280.
Bogen. KT. (1990). Uncertainty in environmental health risk assessment. New York, NY: Garland Publishing.
Boorman. GA: Morgan. KT: Uriah. LC. (1990). Nose, larynx and trachea. In GA Boorman; SL Eustis; MR
Elwell; WF MacKenzie (Eds.), Pathology of the Fischer rat: Reference and atlas (pp. 315-337). San Diego,
CA: Academic Press.
Braun. WH: Young. JD. (1977). Identification of beta-hydroxyethoxyacetic acid as the major urinary metabolite
of 1,4-dioxane in the rat. Toxicol Appl Pharmacol 39: 33-38. http://dx.doi.org/10.1016/0041-
008X(77)90174-0
Bronaugh. RL. (1982). Percutaneous absorption of cosmetic ingredients. In P Frost; SN Horwitz (Eds.),
Principles of cosmetics for the dermatologist (pp. 277-284). St. Louis, MO: C.V. Mosby.
Brown. RP: Delp. MD: Lindstedt SL: Rhomberg. LR: Beliles. RP. (1997). Physiological parameter values for
physiologically based pharmacokinetic models [Review]. Toxicol Ind Health 13: 407-484.
Buffler. PA: Wood. SM: Suarez. L: Kilian. DJ. (1978). Mortality follow-up of workers exposed to 1,4-dioxane. J
Occup Environ Med 20: 255-259.
153
-------
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: Arutvunyan. AV: Stepanov. MG: Oparina. TI: Prokopenko. VM. (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: 832-836.
CAA. Clean Air Act as amended by Pub. L. No. 101-549. section 604: Phase-out of production and
consumption of class I substances. 42 USC § 767 Ic (1990).
Cal/EPA (California Environmental Protection Agency). (2000). Determination of noncancer chronic reference
exposure levels: Appendix D3. Chronic toxicity summary. 1,4-Dioxane (pp. 189-195). Sacramento, CA:
California Environmental Protection Agency, Office of Environmental Health Hazard Assessment.
http://oehha.ca.gov/air/hot spots/2008/AppendixD3final.pdf#page=189
Cal/EPA (California Environmental Protection Agency). (2008). Technical support document for noncancer
RELs. Acute toxicity summary. 1,4-dioxane (pp. 80-84). Sacramento, CA: California Environmental
Protection Agency, Office of Environmental Health Hazard Assessment.
http://oehha.ca.gov/air/hot spots/2008/AppendixD2final.pdf#page=80
Cal/EPA (California Environmental Protection Agency). (2013). Proposition 65 list of chemicals: Chemicals
known to the state to cause cancer or reproductive toxicity. Sacramento, CA: California Environmental
Protection Agency, Office of Environmental Health Hazard Assessment.
http://www.oehha.ca.gov/prop65/prop65 list/files/P65single072613.pdf
Carpenter. SP: Lasker. JM: Raucv. 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.
CDPH (California Department of Public Health). (2011). 1,4-Dioxane for Drinking Water Systems.
http://www.cdph.ca.gOv/certlic/drinkingwater/Pages/l.4-dioxane.aspx
Clark. B: Furlong. JW: Ladner. A: Slovak. AJM. (1984). Dermal toxicity of dimethyl acetylene dicarboxylate,
N-methyl pyrrolidone, triethylene glycol dimethyl ether, dioxane and tetralin in the rat. IRCS Med Sci 12:
296-297.
Clawson. GA: Blankenship. LJ: Rhame. JG: Wilkinson. PS. (1992). Nuclear enlargement induced by
hepatocarcinogens alters ploidy. Cancer Res 52: 1304-1308.
Commonwealth of Massachusetts. (2012). Standards and guidelines for contaminants in Massachusetts drinking
water. Commonwealth of Massachusetts, Executive Office of Energy and Environmental Affairs,
Department of Environmental Protection, Office of Research and Standards.
http://www.mass.gov/dep/water/dwstand.pdf
Connecticut (Connecticut Department of Public Health). (2012). Fact Sheet: 1,4-dioxane in well water [Fact
Sheet]. Hartford, CT: Connecticut Department of Public Health. Environmental & Occupational Health
Assessment Program. http://www.ct.gov/dph/lib/dph/environmental health/eoha/pdf/1 4 dioxane.pdf
David. H. (1964). Electron-microscopic findings in dioxan-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: Richter. P: Stevens. YW. (1996). Health evaluation of 1,4-dioxane [Review].
Toxicol Ind Health 12: 1-43.
Drew. RT: Patel JM: Lin. FN. (1978). Changes in serum enzymes in rats after inhalation of organic solvents
singly and in combination. Toxicol Appl Pharmacol 45: 809-819. http://dx.doi.org/10.1016/0041-
008X(78)90172-2
Enzmann. H: Kuhlem. C: Loser. E: Bannasch. P. (1995). Dose dependence of diethylnitrosamine-induced
nuclear enlargement in embryonal turkey liver. Carcinogenesis 16: 1351-1355.
http://dx.doi.0rg/10.1093/carcin/16.6.1351
Ernstgard. L: Iregren. A: Sjogren. B: Johanson. G. (2006). Acute effects of exposure to vapours of dioxane in
humans. Hum Exp Toxicol 25: 723-729. http://dx.doi.org/10.1177/0960327106073805
154
-------
EWG (Environmental Working Group). (2012). EWG research shows 22 percent of all cosmetics may be
contaminated with cancer-causing impurity. Available online at http://www.ewg.org/news/news-
releases/2007/02/08/ewg-research-shows-22-percent-all-cosmetics-mav-be-contaminated-cancer
Fairlev. A: Linton. EC: Ford-Moore. AH. (1934). The toxicity to animals of 1:4 dioxan. J Hyg 34: 486-501.
http://dx.doi.org/10.1017/S0022172400043266
FDA (U.S. Food and Drug Administration). (2006). Food additives permitted for direct addition to food for
human consumption; glycerides and polyglycides. In Code of Federal Regulations (pp. 75-76). (21 CFR
172.736). Food and Drug Administration.
http://edocket.access.gpo.gov/cfr 2006/aprqtr/pdf/21cfrl72.736.pdf
Fisher. J: Mahle. D: Bankston. L: Greene. R: Gearhart J. (1997). Lactational transfer of volatile chemicals in
breast milk. Am Ind Hyg Assoc J 58: 425-43 L http://dx.doi.org/10.1080/15428119791012667
Franke. C: Studinger. G: Berger. G: Bohling. S: Bruckmann. U: Cohors-Fresenborg. D: Johncke. U. (1994). The
assessment of bioaccumulation. Chemosphere 29: 1501-1514. http://dx.doi.org/10.1016/0045-
6535(94)90281-X
Frantik. E: Hornychova. M: Horvath. M. (1994). Relative acute neurotoxicity of solvents: Isoeffective air
concentrations of 48 compounds evaluated in rats and mice. EnvironRes 66: 173-185,
http://dx.doi.org/10.1006/enrs.1994.1053
Galloway. SM: Armstrong. MJ: Reuben. C: Gorman. S: Brown. B: Cannon. C: Bloom. AD: Nakamura. F:
Ahmed. M: Duk. S: Rimpo. J: Margolin. BH: Resnick. MA: Anderson. B: Zeiger. E. (1987). Chromosome
aberrations and sister chromatid exchanges in Chinese hamster ovary cells: Evaluations of 108 chemicals
[Review]. Environ Mol Mutagen 10: 1-175. http://dx.doi.org/10.1002/em.2850100502
Gaskell BA. (1990). Nonneoplastic changes in the olfactory epithelium-- experimental studies [Review].
Environ Health Perspect 85: 275-289.
Giavini. E: Vismara. C: Broccia. ML. (1985). Teratogenesis study of dioxane in rats. Toxicol Lett 26: 85-88.
http://dx.doi.org/10.1016/0378-4274(85)90189-4
Goldberg. ME: Johnson. HE: Pozzanj UC: Smyth. HF. Jr. (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, http://dx.doi.org/10.1080/00028896409342606
Goldsworthy. TL: Monticello. TM: Morgan. KT: Bermudez. E: Wilson. DM: Ja'ckh. R: BE. B. (1991).
Examination of potential mechanisms of carcinogenicity of 1,4-dioxane in rat nasal epithelial cells and
hepatocytes. Arch Toxicol 65: 1-9. http://dx.doi.org/10.1007/BF01973495
Green. T: Lee. R: Moore. RB: Ashbv. J: Willis. GA: Lund. VJ: MJL. C. (2000). Acetochlor-induced rat nasal
tumors: Further studies on the mode of action and relevance to humans. Regul Toxicol Pharmacol 32: 127-
133. http://dx.doi.org/10.1006/rtph.2000.1413
Grosjean. D. (1990). Atmospheric chemistry of toxic contaminants. 2. Saturated aliphatics: Acetaldehyde,
dioxane, ethylene glycol ethers, propylene oxide. J Air Waste Manag Assoc 40: 1522-1531.
Guilmette. RA: Cheng. YS: Griffith. WC. (1997). Characterising the variability in adult human nasal airway
dimensions. Ann Occup Hyg 41: 491-496.
Hall WC. (1990). Peritoneum, retroperitoneum, mesentery and abdominal cavity. In GA Boorman; SL Eustis;
MR Elwell; CA Montgomery, Jr.; WF MacKenzie (Eds.), Pathology of the Fischer rat (pp. 63-69). San
Diego, CA: Academic Press.
Hansch. C; Leo. A; Hoekman. D. (1995). Exploring QSAR: Hydrophobic, electronic, and steric constants.
Washington, DC: American Chemical Society.
Harkema. JR; Carey. SA; Wagner. JG. (2006). The nose revisited: A brief review of the comparative structure,
function, and toxicologic pathology of the nasal epithelium [Review]. Toxicol Pathol 34: 252-269.
Haseman. JK; Hailev. JR. (1997). An update of the National Toxicology Program database on nasal carcinogens.
MutatRes 380: 3-11. http://dx.doi.org/10.1016/S0027-5107(97)00121-8
Haseman. JK; Hailev. JR; Morris. RW. (1998). Spontaneous neoplasm incidences in Fischer 344 rats and
B6C3F1 mice in two-year carcinogenicity studies: A National Toxicology Program update. Toxicol Pathol
26: 428-441. http://dx.doi.org/10.1177/019262339802600318
155
-------
Haseman. JK: Huff. J: Boorman. GA. (1984). Use of historical control data in carcinogenicity studies in rodents.
Toxicol Pathol 12: 126-135. http://dx.doi.org/10.1177/019262338401200203
Hawlev. GG: Lewis. RJ. Sr. (2001). Hawley's condensed chemical dictionary. In GG Hawley; RJ Lewis, Sr.
(Eds.), (14 ed.). New York, NY: John Wiley & Sons.
Haworth. S: Lawlor. T: Mortelmans. K: Speck. W: Zeiger. E. (1983). Salmonella mutagenicity test results for
250 chemicals. EnvironMutagen 5: 3-142. http://dx.doi.org/10.1002/em.2860050703
Havashi. S: Watanabe. J: Kawaiiri. 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.
http://dx.doi.org/10.1016/6l65-1161(92)90043-L
Hoch-Ligeti. C: Argus. MF. (1970). Effect of carcinogens on the lung of guinea pigs. In P Nettlesheim; MG
HannaJr; JW DeatherageJr (Eds.), Morphology of Experimental Respiratory Carcinogenesis: Proceedings of
a Biology Division, Oak Ridge National Laboratory, Conference held in Gatlinburg, Tennessee, May 13-16,
1970 (pp. 267-279). Oak Ridge, TN: United States Atomic Energy Comission, Division of Technical
Information, http://www.ntis.gov/search/product.aspx?ABBR=CONF700501
Hoch-Ligeti. C: Argus. MF: Arcos. JC. (1970). Induction of carcinomas in the nasal cavity of rats by dioxane. Br
J Cancer 24: 164-167.
HSDB (Hazardous Substances Data Bank). (2007). 1,4-Dioxane. Bethesda, Maryland: National Library of
Medicine, National Toxicology Program, Hazardous Substances Data Bank.
Huang. CY: Huang. KL: Cheng. TJ: Wang. JD: Hsieh. LL. (1997). The GST Tl and CYP2E1 genotypes are
possible factors causing vinyl chloride induced abnormal liver function. Arch Toxicol 71: 482-488.
http://dx.doi.org/10.1007/s002040050416
IARC (International Agency for Research on Cancer). (1999). 1,4-Dioxane. In Re-evaluation of some organic
chemicals, hydrazine and hydrogen peroxide (pp. 589-602). Lyon, France.
http://monographs.iarc.fr/ENG/Monographs/vol71/mono71-25.pdf
ICRP (International Commission on Radiological Protection). (1975). Report of the task group on reference
man: ICRP publication 23. New York, NY: International Commission of Radiological Protection, Pergamon
Press. http://dx.doi.org/10.1016/0146-6453(80)90047-0
ICRP (International Commission on Radiological Protection). (2002). Basic anatomical and physiological data
for use in radiological protection: Reference values (pp. 1-277). (ISSN 0146-6453, EISSN 1872-969X, ICRP
Publication 89). New York, NY: Pergamon Press. http://dx.doi.org/10.1016/S0146-6453(03)00002-2
Ingram. AJ: Grasso. P. (1985). Nuclear enlargement--an early change produced in mouse epidermis by
carcinogenic chemicals applied topically in the presence of a promoter. JAppl Toxicol 5: 53-60.
Ingram. AJ: Grasso. P. (1987). Nuclear enlargement produced in mouse skin by carcinogenic mineral oils. J
Appl Toxicol 7: 289-295.
JBRC (Japan Bioassay Research Center). (1998). Two-year studies of 1,4-dioxane inF344 rats and BDF1 mice
(drinking water). Kanagawa, Japan.
Johnstone. RT. (1959). Death due to dioxane? AMA Arch Ind Health 20: 445-447.
Kanada. M: Miyagawa. M: Sato. M: Hasegawa. H: Honma. T. (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.
http://dx.doi.org/10.2486/indhealth.32.145
Kano. H: Umeda. Y: Kasai T: Sasaki. T: Matsumoto. M: Yamazaki. K: Nagano. K: Arito. H: Fukushima. S.
(2009). Carcinogenicity studies of 1,4-dioxane administered in drinking-water to rats and mice for 2 years.
Food Chem Toxicol 47: 2776-2784. http://dx.doi.0rg/10.1016/i.fct.2009.08.012
Kano. H: Umeda. Y: Saito. M: Senoh. H: Ohbavashi. H: Aiso. S: Yamazaki. K: Nagano. K: Fukushima. S.
(2008). Thirteen-week oral toxicity of 1,4-dioxane in rats and mice. J Toxicol Sci 33: 141-153.
http://dx.doi.org/10.2131/its.33.141
156
-------
Karbe. E: Kerlin. RL. (2002). Cystic degeneration/spongiosis hepatis in rats. Toxicol Pathol 30: 216-227.
http://dx.doi.org/10.1080/019262302753559551
Kasai T. (2008). 1,4-Dioxane toxicity studies, [personal communication].
Kasai T: Kano. H: Umeda. Y: Sasaki. T: Ikawa. N: Nishizawa. T: Nagano. K: Arito. H: Nagashima. H:
Fukushima. S. (2009). Two-year inhalation study of carcinogenicity and chronic toxicity of 1,4-dioxane in
male rats. Inhal Toxicol 21: 889-897. http://dx.doi.org/10.1080/08958370802629610
Kasai. T: Saito. M: Senoh. H: Umeda. Y: Aiso. S: Ohbavashi. H: Nishizawa. T: Nagano. K: Fukushima. S.
(2008). Thirteen-week inhalation toxicity of 1,4-dioxane in rats. Inhal Toxicol 20: 961-971.
http://dx.doi.org/10.1080/08958370802105397
Kasper. P: Uno. Y: Mauthe. R: Asano. N: Douglas. G: Matthews. E: Moore. M: Mueller. L: Nakajima. M:
Singer. T: Speit G. (2007). Follow-up testing of rodent carcinogens not positive in the standard genotoxicity
testing battery: IWGT workgroup report [Review]. MutatRes 627: 106-116.
http://dx.doi.0rg/10.1016/i.mrgentox.2006.10.007
Katagiri. T: Nagano. K: Aiso. S: Senoh. H: Sakura. Y: Takeuchi. T: Okudaira. M. (1998). A pathological study
on spontaneous hepatic neoplasms in BDF1 mice. J Toxicol Pathol 11: 21-25.
http://dx.doi.org/10.1293/tox.ll.21
Kesten. HP: Mulinos. MG: Pomerantz. L. (1939). Pathologic effects of certain glycols and related compounds.
Arch Pathol 27: 447-465.
Khudolev. W: Mizgireuv. I: Pliss. GB. (1987). The study of mutagenic activity of carcinogens and other
chemical agents with Salmonella typhimurium assays: Testing of 126 compounds. Arch Geschwulstforsch
57: 453-462.
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: 67-71.
http://dx.doi.org/10.1016/0304-3835(90)90012-M
Knoefel PK. (1935). Narcotic potency of some cyclic acetals. J Pharmacol Exp Ther 53: 440-444.
Kociba. RJ: McCollister. SB: Park. C: Torkelson. TR: Gehring. PJ. (1974). 1,4-dioxane. I. Results of a 2-year
ingestion study in rats. Toxicol Appl Pharmacol 30: 275-286. http://dx.doi.org/10.1016/0041-
008X(74)90099-4
Kociba. RJ: Torkelson. TR: Young. JD: Gehring. PJ. (1975). 1,4-Dioxane: Correlation of the results of chronic
ingestion and inhalation studies with its dose-dependent fate in rats. In Proceedings of the 6th Annual
Conference on Environmental Toxicology. Wright-Patterson Air Force Base, OH: Wright-Patterson Air
Force Base, Air Force Systems Command, Aerospace Medical Division, Aerospace Medical Research
Laboratory, http://www.ntis.gov/search/product.aspx?ABBR= ADA024899
Koissi. N: Shah. NH: Ginevan. B: Eck. WS: Roebuck. BD: Fishbein. JC. (2012). Lactone metabolite common to
the carcinogens dioxane, diethylene glycol, and N-nitrosomorpholine: aqueous chemistry and failure to
mediate liver carcinogenesis in the F344 rat. ChemRes Toxicol 25: 1022-1028.
http://dx.doi.org/10.1021/tx3000076
Kopylev. L: John Fox. J: Chen. C. (2009). Combining risks from several tumors using Markov Chain Monte
Carlo. In RM Cooke (Ed.), Uncertainty Modeling in Dose Response (1 ed., pp. 197-205). Hoboken, NJ: John
Wiley & Sons.
Kurl. RN: Poellinger. L: Lund. J: Gustafsson. JA. (1981). Effects of dioxane onRNA synthesis in the rat liver.
Arch Toxicol 49: 29-33. http://dx.doi.org/10.1007/BF00352068
Kwan. KK: Dutka. BJ: Rao. SS: Liu. D. (1990). Mutatox test: A new test for monitoring environmental
genotoxic agents. Environ Pollut 65: 323-332. http://dx.doi.org/10.1016/0269-7491(90)90124-U
Laug. EP: Calvery. HO: Morris. HJ: Woodard. G. (1939). The toxicology of some glycols and derivatives. J Ind
Hyg Toxicol 21: 173-201.
Lesage. S: Jackson. RE: Priddle. MW: Riemann. PG. (1990). Occurrence and fate of organic solvent residues in
anoxic groundwater at the Gloucester landfill, Canada. Environ Sci Technol 24: 559-566.
http://dx.doi.org/10.1021/es00074a016
157
-------
Leung. HW: Paustenbach. DJ. (1990). Cancer risk assessment for dioxane based upon a physiologically-based
pharmacokinetic approach. Toxicol Lett 51: 147-162.
Lewandowski. TA: Rhomberg. LR. (2005). A proposed methodology for selecting a trichloroethylene inhalation
unit risk value for use in risk assessment [Review]. Regul Toxicol Pharmacol 41: 39-54.
http://dx.doi.0rg/10.1016/i.yrtph.2004.09.003
Lewis. RJ. Sr. (2000). Sax's Dangerous Properties of Industrial Materials (10 ed.). New York, NY: John Wiley &
Sons, Inc.
Lide. DR. (2000). CRC handbook of chemistry and physics. In DR Lide (Ed.), (81 ed., pp. 3-46). Boca Raton,
FL: CRC Press.
Liu. Y: Johnson. MR: Matida. EA: Kherani. S: Marsan. J. (2009). Creation of a standardized geometry of the
human nasal cavity. J Appl Physiol 106: 784-795. http://dx.doi.org/10.1152/iapplphysiol.90376.2008
Lundberg. I: Ekdahl. M: Kronevi T: Lidums. V: Lundberg. S. (1986). Relative hepatotoxicity of some industrial
solvents after intraperitoneal injection or inhalation exposure in rats. Environ Res 40: 411-420.
http://dx.doi.org/10.1016/S0013-9351(86)80116-5
Lundberg. I: Hogberg. J: Kronevi. T: Holmberg. B. (1987). Three industrial solvents investigated for tumor
promoting activity in the rat liver. Cancer Lett 36: 29-33. http://dx.doi.org/10.1016/0304-3835(87)90099-1
Lyman. W: Reehl. W: Rosenblatt. D. (1990). Handbook of chemical property estimation methods:
Environmental behavior of organic compounds. In WJ Lyman; WF Reehl; DH Rosenblatt (Eds.).
Washington, DC: American Chemical Society.
Maine CDC (Maine Center for Disease Control and Prevention). (2012). Maximum exposure guidelines (MEGs)
for drinking water. Maine Department of Human Services.
http://www.maine.gov/dhhs/mecdc/environmental-health/eohp/wells/documents/megtableoct2012.pdf
Marzulli. FN; 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. http://dx.doi.org/10.1016/0015-6264(81)90530-7
McConnell EE; Solleveld. HA; Swenberg. JA; Boorman. GA. (1986). Guidelines for combining neoplasms for
evaluation of rodent carcinogenesis studies. J Natl Cancer Inst 76: 283-289.
McFee. AF; Abbott MG; Gulati DK; Shelby. MD. (1994). Results of mouse bone marrow micronucleus studies
on 1,4-dioxane. MutatRes 322: 145-148.
McGregor. DB; Brown. AG; Howgate. S; McBride. D; Riach. C; Caspary. WJ. (1991). Responses of the
L5178Y mouse lymphoma cell forward mutation assay. V: 27 coded chemicals. Environ Mol Mutagen 17:
196-219. http://dx.doi.org/10.1002/em.2850170309
Medinsky. MA; Bond. JA. (2001). Sites and mechanisms for uptake of gases and vapors in the respiratory tract
[Review]. Toxicology 160: 165-172. http://dx.doi.org/10.1016/S0300-483X(00)00448-0
Meylan. WM; Howard. PH; Boethling. RS; Aronson. D; Printup. H; Gouchie. S. (1999). Improved method for
estimating bioconcentration/bioaccumulation factor from octanol/water partition coefficient. Environ Toxicol
Chem 18: 664-672. http://dx.doi.org/10.1002/etc.5620180412
Mikheev. MI; Gorlinskava Ye. P; Solowova. TV. (1990). The body distribution and biological action of
xenobiotics. J Hyg Epidemiol Microbiol Immunol 34: 329-336.
Mirkova. ET. (1994). Activity of the rodent carcinogen 1,4-dioxane in the mouse bone marrow micronucleus
assay. MutatRes 322: 142-144.
Miyagawa. M; Shirotori. T; Tsuchitani. M; Yoshikawa. K. (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.
Morgan. KT; Patterson. PL; Gross. EA. (1986). Responses of the nasal mucociliary apparatus of F-344 rats to
formaldehyde gas. Toxicol Appl Pharmacol 82: 1-13. http://dx.doi.org/10.1016/0041-008X(86)90431-X
Morita. T. (1994). No clastogenicity of 1,4 dioxane as examined in the mouse peripheral blood micronucleus
test. Mammalian Mutagenicity Study Group Communications 2: 7-8.
158
-------
Morita. T: Havashi. 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.
http://dx.doi.org/10.1002/(SICI)1098-2280(1998)32:3<269::AID-EM10>3.0.CO:2-8
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. http://dx.doi.org/10.1007/BF01683603
Munoz. ER: Barnett BM. (2002). The rodent carcinogens 1,4-dioxane and thiourea induce meiotic non-
disjunction in Drosophila melanogaster females. MutatRes 517: 231-238. http://dx.doi.org/10.1016/S1383-
5718(02)00083-9
Nannelli. A: De Rubertis. A: Longo. V: Gervasi PG. (2005). Effects of dioxane on cytochrome P450 enzymes in
liver, kidney, lung and nasal mucosa of rat. Arch Toxicol 79: 74-82. http://dx.doi.org/10.1007/s00204-004-
0590-z
NAS (National Academy of Sciences). (2003). Polysorbate 20. In Food chemicals codex (5th ed.). Washington,
DC. http://www.nap.edu/catalog.php7record id= 10731
NCI (National Cancer Institute). (1978). Bioassay of 1,4-dioxane for possible carcinogenicity. (78-1330
NCICGTR-80). Bethesda, MD. http://ntp.niehs.nih.gov/ntp/htdocs/LT rpts/tr080.pdf
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: Bothwell PD: Harrington. TR. (1984). Mutagenicity in a modified
Salmonella assay of fabric-protecting products containing 1,1,1-trichloroethane. Environ Mol Mutagen 6: 71-
80. http://dx.doi.org/10.1002/em.2860060109
New Hampshire DBS (New Hampshire Department of Environmental Services). (2011). Environmental fact
sheet: 1,4-dioxane and drinking water [Fact Sheet]. (WD-DWGB-3-24). Concord, NH.
http://des.nh.gov/organization/commissioner/pip/factsheets/dwgb/documents/dwgb-3-24.pdf
NIOSH (National Institute for Occupational Safety and Health). (2004). NIOSH pocket guide to chemical
hazards: Dioxane. Cincinnati, OH. http://www.cdc.gov/niosh/npg/npgd0237.html
NIOSH (National Institute for Occupational Safety and Health). (2010). Dioxane. Atlanta, GA.
http://www.cdc.gov/niosh/npg/npgd0237.html
NRC (National Research Council). (1983). Risk assessment in the federal government: Managing the process.
Washington, DC: National Academies Press, http://www.nap.edu/openbook.php7record id=366&page=Rl
NRC (National Research Council). (1994). Science and judgment in risk assessment. Washington, DC: National
Academy Press. http://www.nap.edu/openbook.php?isbn=030904894X
NRC (National Research Council). (2009). Science and decisions: Advancing risk assessment. Washington, DC:
National Academies Press, http://www.nap.edu/catalog/12209.html
NRC (National Research Council). (2011). Review of the Environmental Protection Agency's draft IRIS
assessment of formaldehyde. Washington, DC: National Academies Press.
http://www.nap.edu/catalog/13142.html
NTP (National Toxicology Program). (2011). 1,4-dioxane. In Report on carcinogens, twelfth edition (pp. 176-
178). U.S. Department of Health and Human Services, Public Health Service, National Toxicology Program.
http://ntp.niehs.nih.gov/ntp/roc/twerfth/rocl2.pdf
O'Neil MJ: Smith. A: Heckelman. PE: Obenchain. JR: Gallipeau. JR: D'Arecca. MA. (2001). The Merck index:
An encyclopedia of chemicals, drugs, and biologicals. In MJ O'Neil; A Smith; PE Heckelman; JR
Obenchain; JR Gallipeau; MA D'Arecca (Eds.), (13th ed.). Whitehouse Station, NJ: Merck & Co., Inc.
OSHA (Occupational Safety & Health Administration). (2004a). Air contaminants: occupational safety and
health standards for shipyard employment. In Occupational Safety and Health Administration Code of
Federal Regulations. (29 CFR 1915.1000). Washington, DC: U.S. Department of Labor.
http://www.osha.gov/pls/oshaweb/owadisp.show document?ptable=STANDARDS&pid=10286
OSHA (Occupational Safety & Health Administration). (2004b). Appendix A. Safety and health regulations for
construction: Gases, vapors, fumes, dusts, and mists. In Occupational Safety and Health Administration Code
of Federal Regulations. (29 CFR 1926.55, Appendix A). Washington, DC: U.S. Department of Labor.
http://www.osha.gov/pls/oshaweb/owadisp.show document?ptable=STANDARDS&pid=10629
159
-------
OSHA (Occupational Safety & Health Administration). (2004c). Table Z-l: Limits for air contaminants.
Occupational safety and health standards. In Occupational Safety and Health Administration Code of Federal
Regulations. (29 CFR 1910.1000). Washington, DC: U.S. Department of Labor.
http://www.osha.gov/pls/oshaweb/owadisp.show document?ptable=STANDARDS&pid=9992
Overton. JH: Kimbell. JS: Miller. FJ. (2001). Dosimetry modeling of inhaled formaldehyde: The human
respiratory tract. Toxicol Sci 64: 122-134.
Park. JH: Hussam. A: Couasnon. P: Fritz. D: Carr. PW. (1987). Experimental reexamination of selected partition
coefficients from Rohrschneider's data set. Anal Chem 59: 1970-1976.
http://dx.doi.org/10.1021/ac00142a016
Platz. J: Sehested. J: Mogelberg. T: Nielsen. OJ: Wallington. TJ. (1997). Atmospheric chemistry of 1,4-dioxane.
Faraday Trans 1 93: 2855-2863. http://dx.doi.org/10.1039/a700598i
Pozzani. UC: Weil CS: Carpenter. CP. (1959). The lexicological 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. Am Ind Hyg Assoc J 20: 364-369.
http://dx.doi.org/10.1080/00028895909343733
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. http://dx.doi.org/10.1016/0041-
008X(84)90064-4
Reitz. RH: McCroskev. PS: Park. CN: Andersen. ME: Gargas. ML. (1990). Development of a physiologically
based pharmacokinetic model for risk assessment with 1,4-dioxane. Toxicol Appl Pharmacol 105: 37-54.
http://dx.doi.org/10.1016/0041-008X(90)90357-Z
Rosenkranz. HS: Klopman. G. (1992). 1,4-dioxane: Prediction of in vivo clastogenicity. MutatRes 280: 245-
251.http://dx.doi.org/10.1016/0165-1218(92)90054-4
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 CD-I mice. Mutat Res 586: 28-
37. http://dx.doi.0rg/10.1016/i.mrgentox.2005.05.007
Sato. K. (1989). Glutathione transferases as markers of preneoplasia and neoplasia. Adv Cancer Res 52: 205-
255.
Schrenk. HH: Yant WP. (1936). Toxicity of dioxan. J Ind Hyg Toxicol 18: 448-460.
Sheu. CW: Moreland. FM: Lee. JK: Dunkel. VC. (1988). In vitro BALB/3T3 cell transformation assay of
nonoxynol-9 and 1,4-dioxane. EnvironMol Mutagen 11: 41-48. http://dx.doi.org/10.1002/em.2850110106
Silverman. L: Schulte. HF: First. MW. (1946). Further studies on sensory response to certain industrial solvent
vapors. J Ind Hyg Toxicol 28: 262-266.
Sina. JF: Bean. CL: Dysart. GR: Taylor. VI: Bradley. MO. (1983). Evaluation of the alkaline elution/rat
hepatocyte assay as a predictor of carcinogenic/mutagenic potential. Mutat Res Environ Mutagen Relat Subj
113: 357-391. http://dx.doi.org/10.1016/0165-1161(83)90228-5
Smyth. HF. Jr: Seaton. J: Fischer. L. (1941). The single dose toxicity of some glycols and derivatives. J Ind Hyg
Toxicol 23: 259-268.
Spicer. CW: Gordon. SM: Holdren. MW: Kelly. TJ: Mukund. R. (2002). Hazardous air pollutant handbook :
measurements, properties, and fate in ambient air. Boca Raton, FL: CRC Press.
http://www.crcnetbase.com/doi/booyiO.1201/9781420032352
Spiegelhalter. D: Thomas. A: Best. N: Lunn. D. (2003). WinBugs version 1.4 user manual. Cambridge, UK:
MRC Biostatistics Unit, http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/manuall4.pdf
Sticknev. JA: Sager. SL: Clarkson. JR: Smith. LA: Locev. BJ: Bock. MJ: Hartung. R: Olp. SF. (2003). An
updated evaluation of the carcinogenic potential of 1,4-dioxane. Regul Toxicol Pharmacol 38: 183-195.
http://dx.doi.org/10.1016/S0273-2300(03)00090-4
Stoner. GD: Conran. PB: Greisiger. EA: Stober. J: Morgan. M: Pereira. MA. (1986). Comparison of two routes
of chemical administration on the lung adenoma response in strain A/J mice. Toxicol Appl Pharmacol 82: 19-
31.http://dx.doi.org/10.1016/0041-008X(86)90433-3
160
-------
Stott WT: Quasi 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.
http://dx.doi.org/10.1016/0041-008X(91)90232-4
Stroebel P: Mayer. F: Zerban. H: Bannasch. P. (1995). Spongiotic pericytoma: A benign neoplasm deriving
from the perisinusoidal (Ito) cells in rat liver. Am J Pathol 146: 903-913.
Surprenant KS. (2002). Dioxane. In Ullmann's Encyclopedia of Industrial Chemistry (6th ed.). Weinheim,
Germany: Wiley-VCH Verlag. http://dx.doi.org/10.1002/14356007.a08 545
Sweeney. LM: Thrall KD: Poet TS: Corlev. RA: Weber. TJ: Locev. BJ: Clarkson. J: Sager. S: Gargas. ML.
(2008). Physiologically based pharmacokinetic modeling of 1,4-dioxane in rats, mice, and humans. Toxicol
Sci 101: 32-50. http://dx.doi.org/10.1093/toxsci/kfm251
Takano. R: Muravama. N: Horiuchi. K: Kitajima. M: Shono. F: Yamazaki. H. (2010). Blood concentrations of
1,4-dioxane in humans after oral administration extrapolated from in vivo rat pharmacokinetics, in vitro
human metabolism, and physiologically based pharmacokinetic modeling. J Health Sci 56: 557-565.
http://dx.doi.org/10.1248/ihs.56.557
Thiess. AM: Tress. E: Fleig. I. (1976). Arbeitsmedizinische Untersuchungsergebnisse von Dioxan-exponierten
Mitarbeitern [Industrial-medical investigation results in the case of workers exposed to dioxane].
Arbeitsmedizin, Sozialmedizin, Umweltmedizin 11: 35-46.
Thurman. GB: Simms. BG: Goldstein. AL: Kilian. DJ. (1978). The effects of organic compounds used in the
manufacture of plastics on the responsivity of murine and human lymphocytes. Toxicol Appl Pharmacol 44:
617-641. http://dx.doi.org/10.1016/004l-008X(78)90269-7
Tinwell H: Ashbv. J. (1994). Activity of 1,4-dioxane in mouse bone marrow micronucleus assays. Mutat Res
322: 148-150.
Torkelson. TR: Leong. BKJ: Kociba. PJ: Richter. WA: Gehring. PJ. (1974). 1,4-Dioxane. II. Results of a 2-year
inhalation study in rats. Toxicol Appl Pharmacol 30: 287-298. http://dx.doi.org/10.1016/0041-
008X(74)90100-8
U.S. Army Public Health Command. (2010). Studies on metabolism of 1,4-dioxane. (Toxicology Report No. 87-
XE-08WR-09). Aberdeen Proving Ground, MD: U.S. Army Environmental Command.
U.S. Congress. (2011). Consolidated Appropriations Act, 2012. (Pub. L. No. 112-74; 125 STAT. 786). 112th
U.S. Congress. http://www.gpo.gov/fdsys/pkg/PLAW-112publ74/pdf/PLAW-112publ74.pdf
U.S. EPA (U.S. Environmental Protection Agency). (1986a). Guidelines for carcinogen risk assessment [EPA
Report]. (EPA/630/R-00/004). Washington, DC.
http://epa.gov/raf/publications/pdfs/CA%20GUIDELINES 1986.PDF
U.S. EPA (U.S. Environmental Protection Agency). (1986b). Guidelines for mutagenicity risk assessment [EPA
Report]. (EPA/630/R-98/003). Washington, DC. http://www.epa.gov/iris/backgrd.html
U.S. EPA (U.S. Environmental Protection Agency). (1986c). Guidelines for the health risk assessment of
chemical mixtures [EPA Report]. (EPA/630/R-98/002). Washington, DC.
http ://cfpub. epa. gov/ncea/cfm/recordisplav. cfm?deid=22567
U.S. EPA (U.S. Environmental Protection Agency). (1988). Recommendations for and documentation of
biological values for use in risk assessment [EPA Report]. (EPA/600/6-87/008). Cincinnati, OH.
http ://cfpub. epa. gov/ncea/cfm/recordisplav. cfm?deid=3 4855
U.S. EPA (U.S. Environmental Protection Agency). (1991). Guidelines for developmental toxicity risk
assessment [EPA Report]. (EPA/600/FR-91/001). Washington, DC: U.S. Environmental Protection Agency,
Risk Assessment Forum, http://www.epa.gov/raf/publications/guidelines-dev-toxicitv-risk-assessment.htm
U.S. EPA (U.S. Environmental Protection Agency). (1994a). Interim policy for particle size and limit
concentration issues in inhalation toxicity studies [EPA Report]. Washington, DC.
http ://cfpub. epa. gov/ncea/cfm/recordisplav. cfm?deid= 186068
U.S. EPA (U.S. Environmental Protection Agency). (1994b). Methods for derivation of inhalation reference
concentrations and application of inhalation dosimetry [EPA Report]. (EPA/600/8-90/066F). Research
Triangle Park, NC. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=71993
161
-------
U.S. EPA (U.S. Environmental Protection Agency). (1995). The use of the benchmark dose approach in health
risk assessment [EPA Report]. (EPA/630/R-94/007). Washington, DC.
http://www.epa.gov/raf/publications/useof-bda-healthrisk.htm
U.S. EPA (U.S. Environmental Protection Agency). (1996). Guidelines for reproductive toxicity risk assessment
[EPA Report]. (EPA/630/R-96/009). Washington, DC.
http://www.epa.gov/raf/publications/pdfs/REPRO51 .PDF
U.S. EPA (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicity risk assessment [EPA
Report]. (EPA/630/R-95/001F). Washington, DC.
http://www.epa.gov/raf/publications/pdfs/NEUROTOX.PDF
U.S. EPA (U.S. Environmental Protection Agency). (2000a). Benchmark dose technical guidance document
[external review draft] [EPA Report]. (EPA/630/R-00/001). Washington, DC.
http://www.epa.gov/raf/publications/benchmark-dose-doc-draft.htm
U.S. EPA (U.S. Environmental Protection Agency). (2000b). Science policy council handbook: Risk
characterization [EPA Report]. (EPA 100-B-00-002). Washington, D.C.: Office of Science Policy, Office of
Research and Development, http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2000c). Supplementary guidance for conducting health risk
assessment of chemical mixtures [EPA Report]. (EPA/630/R-00/002). Washington, DC.
http ://cfpub. epa. gov/ncea/cfm/recordisplav. cfm?deid=205 3 3
U.S. EPA (U.S. Environmental Protection Agency). (2002a). A review of the reference dose and reference
concentration processes [EPA Report]. (EPA/630/P-02/002F). Washington, DC: Risk Assessment Forum,
U.S. Environmental Protection Agency. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=51717
U.S. EPA (U.S. Environmental Protection Agency). (2002b). Toxic Substances Control Act (TSCA) Inventory
Update Database. Available online at http://www.epa.gov/iur/ (accessed February 22. 2010).
U.S. EPA (U.S. Environmental Protection Agency). (2005a). Guidelines for carcinogen risk assessment [EPA
Report]. (EPA/630/P-03/001F). Washington, DC: Risk Assessment Forum.
http://www.epa.gov/cancerguidelines/
U.S. EPA (U.S. Environmental Protection Agency). (2005b). Supplemental guidance for assessing susceptibility
from early-life exposure to carcinogens [EPA Report] (pp. 1125-1133). (EPA/630/R-03/003F). Washington,
DC. http://www.epa.gov/cancerguidelines/guidelines-carcinogen-supplement.htm
U.S. EPA (U.S. Environmental Protection Agency). (2006a). A framework for assessing health risk of
environmental exposures to children [EPA Report]. (EPA/600/R-05/093F). Washington, DC.
http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=158363
U.S. EPA (U.S. Environmental Protection Agency). (2006b). Peer review handbook (3rd edition) [EPA Report].
(EPA/100/B-06/002). Washington, DC. http://www.epa.gov/peerreview/
U.S. EPA (U.S. Environmental Protection Agency). (2009a). Status report: Advances in inhalation dosimetry of
gases and vapors with portal of entry effects in the upper respiratory tract [EPA Report]. (EPA/600/R-
09/072). Research Triangle Park, NC. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=212131
U.S. EPA (U.S. Environmental Protection Agency). (2009b). Toxicological review of 1,4-dioxane (CAS No.
123-91-1) in support of summary information on the Intergrated Risk Information System (IRIS) [External
Review Draft] [EPA Report] (pp. 1-276). (EPA/635/R-09/005). Washington, DC.
http ://cfpub. epa. gov/ncea/cfm/recordisplav. cfm?deid= 199330
U.S. EPA (U.S. Environmental Protection Agency). (2010). Toxicological review of 1,4-Dioxane (CAS No. 123-
91-1) in support of summary information on the Integrated Risk Information System (IRIS) [EPA Report].
(EPA-635/R-09-005-F). Washington, DC. http://www.epa.gov/iris/toxreviews/0326tr.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2011). Recommended use of body weight 3/4 as the default
method in derivation of the oral reference dose [EPA Report]. (EPA/100/R11/0001). Washington, DC.
http://www.epa.gov/raf/publications/interspecies-extrapolation.htm
U.S. EPA (U.S. Environmental Protection Agency). (2012a). Advances in inhalation gas dosimetry for
derivation of a reference concentration (rfc) and use in risk assessment [EPA Report]. (EPA/600/R-12/044).
Washington, DC. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=244650
162
-------
U.S. EPA (U.S. Environmental Protection Agency). (2012b). Benchmark dose technical guidance. (EPA/100/R-
12/001). Washington, DC: Risk Assessment Forum.
http://www.epa.gov/raf/pubrications/pdfs/benchmarkdose guidance.pdf
U.S. EPA (U.S. Environmental Protection Agency). (2012c). EPA announces NAS' review of IRIS Assessment
development process. Available online at
http://vosemite.epa.gOv/opa/admpress.nsf/0/lce2a7875daf093485257a000054df547OpenDocument
U.S. EPA (U.S. Environmental Protection Agency). (2013a). 1,4-Dioxane PBPK model code in support of IRIS
assessment.
U.S. EPA (U.S. Environmental Protection Agency). (2013b). Toxic release inventory. 2011 TRI national
analysis basic data files. Available online at http://www2.epa.gov/toxics-release-inventorv-tri-prograrn/2011-
tri-national-analysis-basic-data-files
U.S. EPA (U.S. Environmental Protection Agency). (2013c). Toxicological review of 1,4-Dioxane (with
inhalation update) (CAS No. 123-91-1) in support of summary information on the Integrated Risk
Information System (IRIS) [EPA Report]. (EPA-635/R-11/003-F). Washington, DC.
U.S. EPA (U.S. Environmental Protection Agency). (2013d). WinBUGS model code in support of 1,4-dioxane
IRIS assessment.
UNEP (United Nations Environment Programme). (2000). The Montreal Protocol on substances that deplete the
ozone layer. Nairobi, Kenya: United Nations Environment Programme, Ozone Secretariat.
http://www.google.com/url?sa=t&source=web&cd=l&ved=OCBIQFjAA&url=http%3A%2F%2Fwww.unep.
org%2Fozone%2Fpdfs%2Fmontreal-protocol2000.pdf&ei=-c89TPXON9PRngf-i-
idDg&usg=AFOiCNH4OH15inPn5XFcYTvblPPRDZu-fO&sig2=aaSaM nuOlXlHc409kBvgw
Uno. Y: Takasawa. H: Miyagawa. M: Inoue. Y: Murata. T: Yoshikawa. K. (1994). An in vivo-in vitro replicative
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: 189-205.
http://dx.doi.org/10.1016/0165-1218(94)90046-9
Valcke. M: Krishnan. K. (2011). Assessing the impact of the duration and intensity of inhalation exposure on the
magnitude of the variability of internal dose metrics in children and adults. Inhal Toxicol 23: 863-877.
http://dx.doi.org/10.3109/08958378.2011.609918
van Delft JH: van Agen. E: van Breda. SG: Herwijnen. MH: Staal. YC: Klemjans. JC. (2004). Discrimination of
genotoxic from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25: 1265-1276.
http://dx.doi.org/10.1093/carcin/bghl08
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.
http://dx.doi.org/10.1111/i. 1432-1033.1996.0476z.x
Ward. JM: Uno. H: Kurata. Y: Weghorst CM: Jang. JJ. (1993). Cell-proliferation not associated with
carcinogenesis in rodents and humans [Review]. Environ Health Perspect 101: 125-135.
http://dx.doi.org/10.2307/3431855
Watanabe. J: Havashi. 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: Ram. PA: Agrawal AK: Shapiro. BH. (1991). Interpulse interval in circulating
growth hormone patterns regulates sexually dimorphic expression of hepatic cytochrome P450. PNAS 88:
6868-6872.
WHO (World Health Organization). (2005). 1,4-Dioxane in drinking water. (WHO/SDE/WSH/05.08/120).
Geneva, Switzerland.
Wiemann. C: Enzmann. H: Loser. E: Schluter. G. (1999). Nonlinearity of nuclear enlargement in hepatocytes
induced by the carcinogen N'-nitrosomorpholine inOvo. Cancer Detect Prev 23: 485-495.
Wirth. W: Klimmer. O. (1936). [On the toxicology of organic solvents. 1,4 dioxane (diethylene dioxide)]. Archiv
fuer Gewerbepathologie und Gewerbehygiene 17: 192-206.
163
-------
Wolfe. NL: Jeffers. PM. (2000). Hydrolysis. InRS Boethling; D Mackay (Eds.), Handbook of property
estimation methods for chemicals: Environmental and health sciences (pp. 311-333). Boca Raton, FL: Lewis
Publishers, http://dx.doi.org/10.1201/9781420026283.chl3
Wolford. ST: Schroer. RA: Gohs. FX: Gallo. PP: Brodeck. M: Falk. HB: Ruhren. R. (1986). Reference range
data base for serum chemistry and hematology values in laboratory animals. J Toxicol Environ Health A 18:
161-188. http://dx.doi.org/10.1080/15287398609530859
Woo. YT: Arcos. JC: Argus. MF: Griffin. GW: K. N. (1977a). Structural identification of p-dioxane-2-one as the
major urinary metabolite of p-dioxane. Naunyn Schmiedebergs Arch Pharmacol 299: 283-287.
http://dx.doi.org/10.1007/BF00500322
Woo. YT: Argus. MF: Arcos. JC. (1977b). Metabolism in vivo of dioxane: Effect of inducers and inhibitors of
hepatic mixed-function oxidases. Biochem Pharmacol 26: 1539-1542. http://dx.doi.org/10.1016/0006-
2952(77)90431-2
Woo. YT: Argus. MF: Arcos. JC. (1977c). Tissue and subcellular distribution of 3H-dioxane in the rat and
apparent lack of microsome-catalyzed covalent binding in the target tissue. Life Sci21: 1447-1456.
http://dx.doi.org/10.1016/0024-3205(77)90199-0
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.
Yamamoto. S: Ohsawa. M: Nishizawa. T: Saito. A: Kasai T: Noguchj T: Nagano. K: Matsushima. T. (2000).
Long-term toxicology study of 1,4-dioxane in the F344 rats by multiple-route exposure (drinking water and
inhalation) [Abstract]. J Toxicol Sci 25: 347.
Yamamoto. S: Urano. K: Koizumi. H: Wakana. S: Hioki. K: Mitsumori. K: Kurokawa. Y: Havashi. Y: T. N.
(1998a). 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.
Yamamoto. S: Urano. K: Nomura. T. (1998b). Validation of transgenic mice harboring the human prototype c-
Ha-rasgene as a bioassay model for rapid carcinogenicity testing [Review]. Toxicol Lett 102-103: 473-478.
http://dx.doi.org/10.1016/S0378-4274(98)00341-5
Yamazakj K. (2006). Correspondence between Kazunori Yamazaki and Julie Stickney.
Yamazakj K: Ohno. H: Asakura. M: Narumi. A: Ohbavashi. H: Fujita. H: Ohnishi. M: Katagiri. T: Senoh. H:
Yamanouchi. K: Nakavama. E: Yamamoto. S: Noguchi. T: Nagano. K: Enomoto. M: Sakabe. H. (1994).
Two-year toxicological and carcinogenesis studies of 1,4-dioxane in F344 rats and BDF1 mice. In K Sumino;
S Sato; NG Shinkokai (Eds.), Proceedings: Second Asia-Pacific Symposium on Environmental and
Occupational Health 22-24 July, 1993: Kobe (pp. 193-198). Kobe, Japan: Kobe University School of
Medicine, International Center for Medical Research.
Yant WP: Schrenk. HH: Waite. CP: Patty. FA. (1930). Acute response of guinea pigs to vapors of some new
commercial organic compounds: VI. Dioxan. Public Health Rep 45: 2023-2032.
Yasuhara. A: Shiraishi. H: Nishikawa. M: Yamamoto. T: Uehiro. T: Nakasugi. O: Okumura. T: Kenmotsu. K:
Fukuj H: Nagase. M: Ono. Y: Kawagoshi. Y: Baba. K: Noma. Y. (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. http://dx.doi.org/10.1016/S0021-9673(97)00078-2
Yasuhara. A: Tanaka. Y: Tanabe. A: Kawata. K: Katami. T. (2003). Elution of 1,4-dioxane from waste landfill
sites. Bull Environ Contam Toxicol 71: 641-647. http://dx.doi.org/10.1007/s00128-003-8917-7
Yoon. JS: Mason. JM: Valencia. R: Woodruff. RC: Zimmering. S. (1985). Chemical mutagenesis testing in
Drosophila. IV. Results of 45 coded compounds tested for the National Toxicology Program. Environ
Mutagen7: 349-367. http://dx.doi.org/10.1002/em.286007Q310
Young. JD: Braun. WH: Gehring. PJ. (1978a). The dose-dependent fate of 1,4-dioxane in rats. J Environ Pathol
Toxicol 2: 263-282. http://dx.doi.org/10.1080/15287397809529693
Young. JD: Braun. WH: Gehring. PJ. (1978b). Dose-dependentfate of 1,4-dioxane in rats(b). J Toxicol Environ
Health A 4: 709-726. http://dx.doi.org/10.1080/15287397809529693
164
-------
Young. JD: Braun. WH: Gehring. PJ: Horvath. BS: Daniel RL. (1976). 1,4-Dioxane andbeta-
hydroxyethoxyacetic acid excretion in urine of humans exposed to dioxane vapors. Toxicol Appl Pharmacol
38: 643-646. http://dx.doi.org/10.1016/0041-008X(76)90195-2
Young. JD: Braun. WH: Rampy. LW: Chenoweth. MB: Blau. GE. (1977). Pharmacokinetics of 1,4-dioxane in
humans. J Toxicol Environ Health 3: 507-520. http://dx.doi.org/10.1080/15287397709529583
Zimmermann. FK: Mayer. VW: Scheel I: Resnick. MA. (1985). Acetone, methyl ethyl ketone, ethyl acetate,
acetonitrile and other polar aprotic solvents are strong inducers of aneuploidy in Saccharomyces cerevisiae.
MutatRes 149: 339-351. http://dx.doi.org/10.1016/0027-5107(85)90150-2
165
-------
APPENDIX A. SUMMARY OF EXTERNAL PEER
REVIEW AND PUBLIC COMMENTS AND
DISPOSITION
The Toxicological Review ofl,4-Dioxane has undergone two formal external peer reviews
performed by scientists in accordance with EPA guidance on peer review (U.S. EPA. 2006b. 2000b). The
first peer review focused on the toxicity following oral exposure to 1,4-dioxane. For completeness, the
inhalation data were added to the assessment and the combined document was submitted for a second
peer review and public comment - with a request for reviewers to focus on the inhalation portion of the
assessment.
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 for both the oral assessment and inhalation update. 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. EPA also received scientific comments from the public. Public comments are posted to
the federal docket at www.regulations.gov: search for docket ID Nos. EPA-HQ-ORD-2009-0210 for the
oral assessment1 and EPA-HQ-ORD-2011-0390 for the inhalation assessment2. A summary of these
public comments and EPA's responses are included in separate sections of this appendix.
A.1. External Peer Review Panel Comments -- Oral Assessment
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.
1 Public comments on the draft 1,4-dioxane Toxicological Review (oral assessment) posted to www.regulations.gov
can be found at the following URL: http://www.regulations.gov/#!docketDetail:D=EPA-HQ-ORD-2009-0210
2 Public comments on the draft 1,4-dioxane Toxicological Review (inhalation update) posted to
www.regulations.gov can be found at the following URL: http://www.regulations.gov/#!docketDetail:D=EPA-HQ-
ORD-2011-0390
A-l
-------
A.1.1. 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 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.
1977b) 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-2
-------
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 (2012) 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.
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 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
(1998) report.
A-3
-------
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 glomerulonephritis?
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 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 noncancer 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 noncancer 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.
A-4
-------
4. Please comment on the identification and characterization of sources of uncertainty in Section
5. and Section 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 Section 5_ and Section 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 noncancer 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 (1998). It is assumed the
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 4.6.2.1 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.
A.1.2. 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.
A-5
-------
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 (1998) 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 (1998)
and NCI (1978). which was presented in Section 5.1. The effects reported by JBRC
(1998) and NCI (1978) were consistent with what was observed by Kociba et al. (1974)
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 (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.
A-6
-------
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 aNOAEL (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 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 Bl 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.
A-7
-------
4. EPA evaluated the PBPK and empirical models available to describe kinetics following
inhalation of 1,4-dioxane (Reitzetal. 1990; Young etal.. 1978a. b; Young etal.. 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
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 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
BW°7. 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 U.S. EPA 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
recalibrated 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) (Figure B-9 and Figure B-10). The
A-8
-------
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-16). 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 (Figure B-12 and Figure 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, 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 U.S. EPA 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 the U.S. EPA during the evaluation of the model and should
have been described as a clearance constant (first-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.
A-9
-------
(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 (Appendix B).
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.
• 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.
• Seven reviewers stated that the uncertainty factor of 10 for interindividual variability
(intraspecies) is fully supportable.
• 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.
Response. The preferred approach to interspecies scaling is the use of a PBPK model;
however, the PBPK models available for 1,4-dioxane are not suitable for use in this
health assessment as outlined elsewhere. Another approach that has been commonly
implemented in the cancer assessments is the use of body weight scaling based on body
A-10
-------
surface area (BW3/4 scaling). It is not standard practice to apply BW3/4 scaling in
noncancer assessments at this time. The current default approach used by the Agency
when PBPK models are not available for extrapolation is the application of an UFA of 10,
which was implemented in this assessment.
The absence of a multigenerational reproductive study is why the uncertainty factor for
database deficiencies (UFD) was retained; however, it was reduced from 10 to 3. In the
text in Section 5.1.3 text was included to clearly state 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.
A.1.3. Carcinogenicity of 1,4-dioxane and derivation of an oral slope
factor
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
U.S. EPA's 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.
A-ll
-------
2. Evidence indicating the mode of action of carcinogenicity of 1,4-dioxane was considered.
Several hypothesized MOAs were evaluated within the Toxicological 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 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; this reviewer also pointed
out that EPA should not rule out a metabolite as the toxic moiety.
One reviewer stated this was outside of his/her 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. One of these reviewers also
suggested that a nonlinear cancer risk assessment model should be utilized.
One reviewer suggested adding more text to the summary statement to fully reflect the
available MOA information which should 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(s), 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 U.S. EPA's
Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 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. In accordance with the U.S.
EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). the absence of
A-12
-------
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.
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 nonlinear 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. 1998) 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 (1998) study as the
principal study for the development of an OSF. However, two reviewers that agreed with
the choice of JBRC (1998) 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 (1998) 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 External Peer Review draft of the Toxicological Review of
1,4-Dioxane was released (U.S. EPA. 2009b). the cancer portion of the study conducted
by the JBRC laboratory was published in the peer-reviewed literature as Kano et al.
(2009). This manuscript was reviewed by EPA. EPA determined that the data published
by Kano et al. (2009) should be included in the assessment of 1,4-dioxane for several
reasons: (1) while the JBRC (1998) was a detailed laboratory report, it was not
A-13
-------
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 manuscript included
additional information such as body weight growth curves and means and standard
deviations of estimated dose for both rats and mice of both sexes. Thus, the Toxicological
Review was updated to reflect the inclusion of the data from Kano et al. (2009). and
Appendix E was added for a clear and transparent display of the data included in the
multiple reports.
In response to the peer reviewers, 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. Text was also added to Section 4.2.1.2.6
regarding the choice of high dose selection as included in the Kano et al. (2009)
manuscript. Additional discussion regarding the mortality rates was also added to Section
5.4.1 in selection of the critical study for the oral cancer assessment. 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 (Kano et al.. 2009) and this was also added to the text in Section 4.2.1.2.6.
4. Combined liver tumors (adenomas and carcinomas) in female CjrBDFl mice from the JBRC
(1998) 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 CjrBDFl mice should be used for the
derivation of the OSF. Five of these reviewers agreed with the rationale for the selection
of the female CjrBDFl 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 CjrBDFl male and female mice.
One reviewer commented that the scientific basis for the selection of female CjrBDFl
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
A-14
-------
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 U.S. EPA 2005 Guidelines for Carcinogen Risk 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 nonlinear 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.
A-15
-------
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
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 U.S. EPA Guidelines for
Carcinogen Risk Assessment (U.S. EPA. 2005a) 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 U.S. EPA Guidelines for Carcinogen Risk Assessment (2005a)
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 deficiencies 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 BMDLi0,which was more than 15-fold lower than
the response at the lowest dose (66 mg/kg-day), was reconsidered in response to a public
comment. BMDs and BMDLs were calculated using a BMR of 30 and 50% extra risk
(BMD30, BMDL30, BMD50, and BMDL50). A BMR of 50% was used as it resulted in a
BMDL closest to the response level at the lowest dose tested in the bioassay.
A.2. Public Comments - Oral Assessment
Comments on the Toxicological Review of 1,4-Dioxane submitted by the public for the external
peer review of the oral toxicity values are summarized below in the following categories: Oral reference
dose for 1,4-dioxane, carcinogenicity of 1,4-dioxane, PBPK modeling, and other comments.
A-16
-------
A.2.1. Oral reference dose (RfD) for 1,4-dioxane
Comment: An UF for database deficiencies is not necessary because of considerable
evidence showing no reproductive or developmental effects from 1,4-dioxane exposure.
Response: Due to the lack of a multigenerational reproductive study for 1,4-dioxane an
UF of 3 was retained for database deficiencies. Without clear evidence showing a lack of
reproductive or developmental effects in a multigenerational reproductive study, there is
still uncertainty in this area.
A.2.2. Carcinogenicity of 1,4-dioxane
Comment: Using liver tumors as the basis for the oral CSF is more appropriate than nasal
tumors (1988 IRIS assessment of 1,4-dioxane); however, the use of mouse liver tumor
data is inappropriate because it is inconsistent with other liver models both quantitatively
and in the dose-response pattern. High mortality rates in the study are also a limitation.
Liver tumor data from rats should be used instead, which represents a better animal
model for 1,4-dioxane carcinogenicity assessment.
Response: Even though the dose-response is different for mice and rats, the female mice
were considered to be appropriate for the carcinogenicity assessment for several reasons.
The female mouse liver tumors from the Kano et al. (2009) report were found to be the
most sensitive species and endpoint. Section 4.2.1.2.6 was updated to include additional
information on mortality rates. The majority of the animals lived past 52 weeks (only 4
females died prior to 52 weeks, 2 in each the mid- and high-dose groups). The cause of
death in the female mice that died between 1 and 2 years was attributed to liver tumors.
Comment: The OSF was based on the most sensitive group, Crj:BDFl mice; however
BDF1 mice have a high background rate of liver tumors. The incidence of liver tumors in
historical controls for this gender/species should be considered in the assessment.
Sensitivity of the test species/gender as well as other criteria should be considered in the
selection of the appropriate study, including internal and external validity as outlined in
Lewandowski and Rhomberg (2005). The female Crj:BDFl mice had a low survival rate
that should be considered in the selection of the animal model for 1,4-dioxane
carcinogenicity.
Response. Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas
and carcinomas in control male and female BDF1 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 BDF1. These incidence rates are near the
historical control values and thus are appropriate for consideration in this assessment.
A-17
-------
Additional text regarding these historical controls was added to the study description in
Section 4.2.1.2.6.
Comment: Low-dose linear extrapolation for the oral CSF is not appropriate nor justified
by the data. The weight of evidence supports a threshold (nonlinear) MOA when
metabolic pathway is saturated at high doses. Nonlinear extrapolations should be
evaluated and presented for 1,4-dioxane. Oral CSFs should be derived and presented
using both the BW3/4 scaling as well as available PBPK models to extrapolate across
species.
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. The oral CSF for 1,4-dioxane was derived using BW3/4
scaling for interspecies extrapolation. The PBPK and empirical models available for
1,4-dioxane were evaluated and found not to be adequate for use in this assessment,
described in detail in Appendix B.
Comment: The POD for the BDF1 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 U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a).
Response. The comment is correct that the animal BMDLio was more than 15-fold lower
than the response at the lowest dose (66 mg/kg-day) in the bioassay. BMDs and BMDLs
were calculated using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and
BMDL50). 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 geometric mean of the oral cancer slope factors (as done with B[a]P &
DDT) should have been used instead of relying on the female BDF1 mouse data, since a
MOA could not be determined for 1,4-dioxane.
Response. In accordance with the BMD Technical Guidance Document (U.S. EPA.
2012b) averaging tumor incidence is not a standard or default approach. Averaging the
tumor incidence response diminishes the effect seen in the sensitive species/gender.
Comment. EPA should critically reexamine the choice of JBRC (1998) as the principal
study since it has not been published or peer-reviewed. A transcript of e-mail
correspondence should be provided.
Response. JBRC (1998) was published as conference proceedings as Yamazaki et al.
(1994) and recently in the peer-reviewed literature as Kano et al. (2009). Additional study
information was also gathered from the authors (Yamazaki. 2006) and is available upon
request from the IRIS Hotline. The peer-reviewed and published data from Kano et al.
A-18
-------
(2009) was incorporated into the final version of the Toxicological Review of
1,4-Dioxane.
Comment. The WOE does not support a cancer descriptor of likely to be carcinogenic to
humans determination, but rather suggestive human carcinogen at the high dose levels
used in rodent studies seems more appropriate for the following reasons: 1) lack of
conclusive human epidemiological data; 2) 1,4-dioxane is not mutagenic; and 3) evidence
at high doses it would act via cell proliferation MOA.
Response: A cancer 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 was chosen. Also, mesotheliomas of the peritoneum, mammary, and nasal
tumors have been observed in rats. The Agency agrees that human epidemiological
studies are inconclusive. The evidence at any dose is insufficient to determine a MOA.
A.2.3. PBPK Modeling
Comment. EPA should have used and considered PBPK models to derive the oral
toxicity values (rat to human extrapolation) rather than relying on a default method. 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). This model
was determined not to be appropriate for interspecies extrapolation. Additionally, see
response to the external peer review panel comment B4.
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.
A-19
-------
A.2.4. Other Comments
Comment: EPA should consider the Kasai et al. (2009; 2008) studies for inhalation and
MOA relevance.
Response: The 13 week and 2-year inhalation studies by Kasai et al. (2009; 2008) were
published late in the development stage of this assessment. The IRIS Program will
evaluate these recently published 1,4-dioxane inhalation data for the potential to derive
an RfC in a separate assessment.
Comment: 1,4-Dioxane is not intentionally added to cosmetics and personal care
products - correct sentence on page 4.
Response: This oversight was corrected in the document.
A.3. External Peer Review Panel Comments - Inhalation Update
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. In cases where comments were
made regarding the oral assessment for 1,4-dioxane, those comments are noted, considered, and changes
were made to the oral assessment as appropriate; however this was not intended to be a second peer
review of the oral assessment finalized in 2010 (U.S. EPA. 2010).
A.3.1. General Charge Questions
1. Is the Toxicological Review logical, clear and concise? Has EPA clearly presented and
synthesized the scientific evidence for noncancer and cancer health effects from exposure to
1,4-dioxane viainhalation?
Comment. Four reviewers agreed that the Toxicological Review of 1,4-dioxane was
logical, clear, and concise. Two reviewers commented that the majority of the
Toxicological Review was logical, clear, and concise, but provided several
recommendations to improve the document. The specific recommendations included:
(1) documentation of literature search terms, (2) description of the severity of the lesions
observed by Kasai et al. (2008) should be included in the main body of the text,
(3) clarification of the toxicological significance of nuclear enlargement with clear
differentiation between study author and EPA's conclusions regarding this endpoint,
(4) improvement of Table 4-27 and Table 4-28 as they do not readily demonstrate
A-20
-------
temporal relationships of interest, (5) removal of repetitive text, (6) reduction of
unnecessary text in the mode of action analysis, (7) correction of inconsistencies between
oral and inhalation approaches to derive the reference values, (8) the addition of
information on ambient exposures to 1,4-dioxane, and (9) improve the writing of the text
of Section 4.6.2 and expand Section 4.6.2.1 to focus on the possibility that the parent
compound is the toxic moiety.
Additionally, one reviewer made reference to a public comment noting an error in the
PBPK model code in the description of the slowly perfused tissue. This reviewer
suggested the code be corrected and provided in the assessment. However, the reviewer
did agree with the conclusion that the existing PBPK models are inadequate to
perform route-to-route and cross-species extrapolation of animal studies.
Response. (1) Additional information was provided in Section i regarding the literature
search strategy employed for 1,4-dioxane. (2) The severity of the nasal lesions observed
by Kasai et al. (2008) was included in Table 4-17; no additional language was added to
the text as the data is presented clearly in tabular format. (3) With regards to nuclear
enlargement, additional search of the literature and consulation with an Agency
pathologist revealed that nuclear enlargement may be found in any cell type responding
to microenvironmental stress or undergoing proliferation. It may also be an indicator of
exposure to a xenobiotic in that the cells are responding by transcribing mRNA. Several
studies indicate that it may also be identified as an early change in response to exposure
to a carcinogenic agent (Wiemann et al.. 1999; Enzmann et al.. 1995; Clawson et al..
1992; Ingram and Grasso. 1987. 1985); however, its relationship to the typical
pathological progression from initiated cell to tumor is unclear. Therefore, nuclear
enlargement as a specific morphologic diagnosis was not considered an adverse effect of
exposure to 1,4-dioxane. Clarifying text was added to the document regarding the
uncertainty surrounding this reported observation to Sections 4.2.1.1.3. 4.2.1.2.6.
4.2.2.1.2. 4.2.2.2.2. and 5.2.1. (4) Table 4-27 and Table 4-28 were described in more
depth in their accompanying sections to describe their content and the temporal nature.
(5)/(6) The Agency continues to evaluate and incorporate recommendations made by the
NAS that should streamline (i.e., reduce redundancy), strengthen and improve
transparency within the IRIS documents. The NAS recommendations implemented in this
document are described in APPENDIX I. (7) There are necessary differences in the
derivation of oral and inhalation reference values, discussed in Section 5.4.4. and
clarified in Section 5.4.4.2. For instance, the oral slope factor derivation does not use the
multistage model, whereas the inhalation unit risk derivation does. This is due to a lack of
a suitable multistage model being identified for the female mouse liver tumor data used to
derive the oral slope factor, whereas appropriate multistage model fits were obtained for
the tumor data used to derive the inhalation unit risk. This departure resulted in a
necessary and significant difference in approaches. (8) While it is important for risk
assessors to understand ambient exposure levels in utilization of IRIS reference values,
A-21
-------
ambient exposure levels are dependent upon location and media and thus are not included
in IRIS assessments. In the context of the overall risk assessment paradigm, IRIS
documents provide the hazard identification information and the dose-response analysis
in support of the derivation of reference values for the chemical of interest. (9) The
suggestions made by the reviewer to improve the writing and summaries in 4.6.2 were all
incorporated. 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. As the reviewer notes, and is already stated in the toxicological review, it is
possible that the parent compound is the toxic moiety; however, the section was not
rewritten with a focus on the parent compound. Regarding the PBPK model, the code
errors identified by a public commenter and referenced by a member of the peer review
panel were corrected (discussed further in response to public comments, below).
Additionally, the model equations have been available in Appendix B of previous version
of the toxicological review released. In this final version, however, the model code is not
provided in the text, but is available electronically via HERO, along with the executable
.m script files (U.S. EPA. 2013a).
2. Please identify any additional peer-reviewed studies from the primary literature that should be
considered in the assessment of noncancer and cancer health effects from exposure to
1,4-dioxane via inhalation.
Comment: Four reviewers stated they were unaware of any additional studies available to
add to the inhalation toxicity evaluation of 1,4-dioxane. One reviewer provided additional
general references pertaining to dose extrapolation for the derivation of the RfC
specifically regarding the default values used for the human extrathoracic surface area
and minute ventilation. Another reviewer provided some general references related to
evaluation of tumors and mode of action, along with a few 1,4-dioxane specific papers.
The 1,4-dioxane specific papers suggested for consideration were:
a. Takano, T, Murayama, N, Horiuchi, K, Kitajima, M, Shono, F. (2010). Blood
concentrations of 1,4-dioxane in humans after oral administration extrapolated
from in vivo rat pharmacokinetics, in vitro human metabolism, and
physiologically based pharmacokinetic modeling. J Health Sci 56: 557-565.
(Note: The reviewer noted that this paper is not likely to be useful in the
assessment; however, a short summary should be added to the appropriate section
in the toxicological review)
b. U.S. Army Public Health Command (2010). Studies on Metabolism of
1,4-Dioxane, Toxicology Report No. 87-08 WR-09, Aberdeen Proving Ground,
MD.
c. WHO (World Health Organization). (2005). 1,4-Dioxane in Drinking Water,
WHO/SDE/WSH/05.08/120, Geneva.
A-22
-------
Response: Reference (a) above was evaluated for the utility of the described PBPK
model in predicting toxicokinetics of 1,4-dioxane in rats and humans. A summary of
Takano et al. (2010) and an evaluation of the model was added to Section 3.5.2.5.
Reference (b) was cited as supporting information regarding the metabolites of
1,4-dioxane in Section 3.3. Reference (c) is a report produced by an organization other
than the U.S. EPA and was considered during development of this assessment; however,
the Agency performed an independent analysis of the scientific informa available for
1,4-dioxane and did not cite this document. Toxicity values and classifications for
1,4-dioxane reported by other agencies were added to Appendix H.
The additional general references pertaining to dose extrapolation for the derivation of
the RfC specifically regarding the default values used for the human extrathoracic surface
area and minute ventilation were related to the inclusion of the alternative RfC
calculation in Appendix G. This appendix was removed following external peer review.
See response to charge question 4 (see Section A.3.2X below, relating to the RfC for
more details.
A.3.2. Inhalation reference concentration (RfC) for 1,4-dioxane
1. A 2-year inhalation bioassay in male rats (Kasai et al.. 2009) was selected as the basis for the
derivation of the RfC. Please comment on whether the selection of this study is scientifically
supported and clearly described. If a different study is recommended as the basis for the RfC,
please identify this study and provide scientific support for this choice.
Comment: Four reviewers agreed that the selection of the 2-year bioassay in male rats
(Kasai et al.. 2009) as the critical study used for the derivation of the RfC was
scientifically justified. Two reviewers also agreed with the aforementioned, but stated
that decision not to collect female rat data for the 2-year bioassay was not scientifically
supported by the study authors (Kasai et al.. 2009). especially given that the 13-week
bioassay (Kasai et al.. 2008) showed female rats more responsive than male rats
following inhalation exposure. More specifically, the two reviewers highlighted that one
of the selected critical effects (atrophy of the olfactory epithelium) was observed in
female rats and not male rats following 13 weeks of exposure to 1,4-dioxane vapors, thus
making the female rat more responsive to 1,4-dioxane following inhalation exposure.
Response: The Agency did not conclude that the available data supports the female rats
as definitively more responsive than male rats following 13 weeks of exposure to
1,4-dioxane vapors. BMD analysis of the incidence of olfactory atrophy in female rats
from the Kasai et al. (2008) study provides a BMCLio of 65 ppm (fit with the
Dichotomous Hill model). Application of a total UF of 1,000 would yield an RfC of
0.065 ppm compared to an RfC of 0.05 ppm calculated from the 2 year bioassay. A
review of the pathological observations also does not indicate that females are
A-23
-------
definitively more responsive to 1,4-dioxane exposure. Of the lesions noted, most were
considered to be of the lowest severity grade. Of these lesions, equivalent responses were
observed between males and females and in some cases greater in females and in others
greater in males. Thus, information to suggest that females are more responsive than
males is currently lacking. Additionally, in accordance with the weight-of-evidence
framework described in the Methods for Derivation of Inhalation Reference
Concentrations and Application of Inhalation Dosimetry (U.S. EPA. 1994b). the
selection of the 2-year bioassay in male rats as the critical study is justified. Furthermore,
an uncertainty factor of 3 for an incomplete database was applied. This uncertainty factor
is intended to account for the inability of any single laboratory animal study to adequately
address all possible adverse outcomes in humans. Therefore, in consideration of the data
presented in each of the studies as well as the difference in the study durations (13 versus
104 wks), the selection of the 2-year bioassay in male rats as the critical study is justified.
2. Atrophy and respiratory metaplasia of the olfactory epithelium in male rats were concluded by
EPA to be adverse effects and were selected as co-critical effects for the derivation of the RfC.
Please comment on whether the selection of these co-critical effects and their characterization
is scientifically supported and clearly described. If a different health endpoint is recommended
as the critical effect for deriving the RfC, please identify this effect and provide scientific
support for this choice.
Comment: Four reviewers agreed with the selection of co-critical effects in the derivation
of the RfC and stated that the selection was scientifically supported and clearly described.
The remaining two reviewers also agreed with the selection of co-critical effects in the
derivations of the RfC; however, they provided suggestions on how to strengthen the
justification for EPA's decision or improve clarity. These reviewers suggested EPA
(1) provide further justification for why nuclear enlargement was not considered as a
critical effect and (2) clearly state the criteria for selection of the critical effect. One
reviewer also noted inconsistency between the oral and inhalation assessments regarding
the consideration of spongiosis hepatis as a nonneoplastic lesion and potential critical
effect.
Response. In response to reviewer comments, EPA further investigated nuclear
enlargement. As stated in response to inhalation assessment general charge question 1
(Section A.3.1). nuclear enlargement may be found in any cell type responding to
microenvironmental stress or undergoing proliferation. It may also be an indicator of
exposure to a xenobiotic in that the cells are responding by transcribing mRNA. Several
studies indicate that it may also be identified as an early change in response to exposure
to a carcinogenic agent (Wiemann et al.. 1999; Enzmann et al.. 1995; Clawson et al..
1992; Ingram and Grasso. 1987. 1985); however, its relationship to the typical
pathological progression from initiated cell to tumor is unclear. Therefore, consideration
and selection of this response as a critical endpoint would not be supported by the
available scientific information. Clarifying text was added to the document regarding
A-24
-------
nuclear enlargement as noted in response to charge question Al, and specifically in
Section 5.2.1 as to why it was not considered as a critical effect.
Additional clarifying text was added to Section 5.2.3 regarding the use of respiratory
metaplasia and atrophy of the olfactory epithelium as co-critical effects, noting that they
were the most sensitive effects considered following inhalation of exposure to
1,4-dioxane. EPA agrees there was inconsistency in way spongiosis hepatis was
considered between the oral and inhalation assessments. Spongiosis hepatis was removed
from the list of candidate critical effects in the inhalation assessment. However, whether
spongiosis hepatis/cystic degeneration represents a preneoplastic change or a
nonneoplastic change has been the subject of scientific controversy (Karbe and Kerlin.
2002; Stroebel et al.. 1995; Bannasch et al.. 1982). Spongiosis hepatis 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. 2003; Stroebel et al.. 1995). In contrast, it 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). Following
inhalation of 1,4-dioxane, spongiosis hepatis was associated with other preneoplastic
(e.g., liver foci) and nonneoplastic (e.g., centrilobular necrosis) changes in the liver
(Kasai et al.. 2009). Additionally, the incidence rates of spongiosis hepatis and liver
tumors were highly correlated; therefore, spongiosis hepatis was considered a
preneoplastic lesion following inhalation exposure and not considered further in the
noncancer analysis. This justification was added to the document in Section 5.2.1.
3. Benchmark dose (BMD) modeling methodology (U.S. EPA. 2012b) was used to analyze the
candidate endpoints identified for 1,4-dioxane. However, due to poor fit or substantial model
uncertainty, BMD model results were inadequate for the following nasal lesions: atrophy
(olfactory epithelium), respiratory metaplasia (olfactory epithelium), and sclerosis (lamina
propria). Consequently, the NOAEL/LOAEL approach was used to identify the POD for
derivation of the RfC. Please comment on whether this approach is scientifically supported and
clearly described.
Comment: Six reviewers agreed that the use of the NOAEL/LOAEL approach in the
derivation of the RfC is scientifically supported and clearly described.
Response. EPA agrees with the reviewers regarding the use of the NOAEL/LOAEL
approach in the derivation of the RfC, no changes were made to the document.
The human equivalent concentration (HEC) for 1,4-dioxane was calculated by the
application of the dosimetric adjustment factor (DAF) for systemic acting gases
(i.e., Category 3 gases), in accordance with the U.S. EPA RfC methodology (U.S. EPA.
1994b). This conclusion was based upon a number of factors, including the low reactivity
of 1,4-dioxane, and the occurrence of systemic effects following oral and inhalation
A-25
-------
exposure to 1,4-dioxane. However, since 1.4-dioxane is water soluble and induces effects
in portal-of-entry tissues, an alternative calculation of the HEC for 1,4-dioxnae based on
the application of the corresponding DAF for the portal-of entry acting gases
(i.e., Categrory 1) is provided in Appendix G.
4. Please comment on EPA's conclusion that 1.4-dioxane is a Category 3 gas, and the resulting
application of the corresponding dosimetric adjustment factor (DAF) in deriving the RfC. If a
different approach is recommended in the derivation of the RfC, please identify this approach
and provide scientific support for the proposed changes.
Comment: All of the reviewers thought the approach used in the main body of the
document was reasonable and consistent with the Agency's current definitions and
approaches, as well as the effects observed. Two reviewers thought the inclusions of an
alternative approach in Appendix G. was reasonable. Two other reviewers noted
problems with the outcome of the default calculation used in the alternative approach.
Two reviewers thought the lesions seen in the inhalation study may represent portal-of-
entry responses; one of these reviewers thought additional text should be added to the
document.
Response. Since the reviewers were in agreement with the extrapolation approach
employed and described in the main body of the document, Appendix G in the external
peer review draft that demonstrated the application of the Agency's default method for
deriving an RfC for category 1 gases was removed. The alternative approach used default
ratios of ventilation rate and surface areas cited and often used in accordance with the
Agency's RfC Methods (U.S. EPA. 1994b). which are also supported by several sources
including ICRP (2002). Guilmette et al. (1997). and Liu et al. (Liu et al.. 2009).
The text corresponding to the dosimetric extrapolation approach applied for 1,4-dixoane
has been revised for clarity and transparency; however, no changes to the quantitative
approach were made. EPA agrees that 1,4-dioxane induces portal of entry effects.
1,4-Dioxane is miscible with water and has a high blood:air partition coefficient. Unlike
typical highly water soluble and reactive portal-of-entry acting gases, 1,4-dioxane also
induces lower respiratory tract and systemic effects and has been measured in the blood
after inhalation exposure. Thus, it is difficult to determine what contribution circulating
1,4-doxane makes to the portal-of-entry effects observed. Therefore, for the purposes of
dosimetric extrapolation, 1,4-dioxane was treated as a systemic acting gas and a DAF of
1 was applied. In addition, a robust CFD and PBPK modeling database supports the
scientific rationale to apply of DAF of 1 for both portal of entry and systemic effects
irrespective of "gas categorization" (U.S. EPA. 2012a).
A-26
-------
5. Please comment on the rationale for the selection of the UFs applied to the POD for the
derivation of the RfC. Are the UFs appropriate based on A Review of the Reference Dose and
Reference Concentration Processes [(U.S. EPA. 2002a); Section 4.4.5;
www.epa.gov/iris/backgrd.html1 and clearly described? If changes to the selected UFs are
proposed, please identify and provide scientific support for the proposed changes.
Comment: Four reviewers agreed with the selection and justification of the UFs applied
to the POD for the derivation of the RfC. One of these reviewers, however, suggested that
it be noted that the reproductive toxicity and teratogenicity indices monitored in rats by
Giavini et al. (1985) were unremarkable. Two reviewers agreed with the selection of the
UFs but requested clarification of the justification for the database uncertainty factor.
One reviewer further questioned the reliability of the UF of 10 to extrapolate to a
NOAEL given the lack of an exposure group below 50 ppm where one of the critical
effects was noted with an incidence rate of 80% (olfactory epithelium), and the lack of
female rats exposure in the 2 year bioassay despite evidence of increased responsiveness
to 1,4-dioxane vapors following inhalation as compared to the male rat in a 13 week
bioassay. Additionally, one reviewer debated the application of the UF of 10 for
individual differences among human subjects given that dosimetric differences for
particles among human subjects is often 1.3 rather than 3.
Response. In accordance with U.S. EPA (2002a). the database was characterized and
applied to the derivation of the RfC. Giavini et al. (1985) administered 1,4-dioxane by
gavage in water to pregnant rats. The authors found statistically significant changes in
fetal body weight and reduced ossification of the sternebrae at the highest dose group;
however, the lack of a multigenerational reproductive study warrants the use of a 3 for
UFD. As outlined in detail in response to the inhalation assessment charge question Bl,
the available data do not support female rats as definitively more responsive than male
rats following 13 weeks of exposure to 1,4-dioxane vapors. A recent modeling study by
Valcke and Krishnan (2011) assessed the impact of exposure duration and concentration
on the human kinetic adjustment factor and estimated the neonate to adult 1,4-dioxane
blood concentration ratio to be 3.2. Thus, a full factor of 10 was used to account for
differences between adults and neonates, as well as other differences in gender, age,
health status, or genetics that might result in a different disposition of, or response to,
1,4-dioxane.
A-27
-------
A.3.3. Carcinogenicity of 1,4-dioxane and derivation of an inhalation
unit risk
1. Under EPA's Guidelines for Carcinogen Risk Assessment [(U.S. EPA. 2005a): Section 2.5;
www.epa.gov/iris/backgrd.html1. the draft IRIS assessment characterizes 1,4-dioxane as "likely
to be carcinogenic to humans" by all routes of exposure. Please comment on whether this
characterization of the human cancer potential of 1,4-dioxane is scientifically supported and
clearly described.
Comment: Five out of six reviewers agreed with the characterization that 1,4-dioxane is
"likely to be carcinogenic to humans." However, one of these reviewers suggested a more
transparent application of the criteria to the inhalation cancer data to classify the
compound as "likely" would be beneficial. One reviewer disagreed with the cancer
classification of "likely to be carcinogenic to humans" and suggested that it should be
classified as a "possible human carcinogen". This reviewer provided several arguments as
a basis for a different classification: 1) no evidence of increased cancer incidence in
humans exposed to 1,4-dioxane in the limited number of epidemiology studies, 2)
negative in vivo and in vitro genotoxicity experiments suggesting that 1,4-dioxane is, at
most, a weak genotoxicant, 3) data demonstrating observed tumors in rodents occur
following high chronic exposures, 4) the parent compound is the proximate irritant,
cytotoxicant, and carcinogenic moiety, and 5) conclusions and classifications by other
organizations (i.e., German Commission for the Health Hazards of Chemical Compounds
in the Work Area, ACGIH, IARC and WHO).
Response: Five of the six reviewers agreed with the characterization of "likely to be
carcinogenic to humans" and no change was made to this conclusion in the final
Toxicological Review. With respect to the one reviewer who suggested applying the
criteria more transparently to the inhalation data alone; when considering the
characterization of the carcinogenic potential for a compound, the available data across
all exposure routes is first considered. If, for example, portal of entry effects are
observed for one route of exposure and not the other, or there is evidence that a chemical
is not absorbed from a particular route of exposure, then separate cancer descriptors may
be used to describe the cancer potential. In the case of 1,4-dioxane, the tumors that were
observed in animals were systemic and independent of the route of exposure.
The one reviewer that disagreed with the classification provided a suggested
classification that appears to be based on earlier 1986 U.S. EPA cancer classification
terminology. As summarized in Section 4.7.1. the available human studies with small
cohorts and limited number of reported cases are inconclusive. The Agency agrees with
the reviewer that the majority of the genotoxicity studies are negative, suggesting
1,4-dioxane is not genotoxic (Section 4.5.1). and that tumors have been observed in
rodents following chronic exposure (summarized in Section 4.7.2). A lack of data to
A-28
-------
determine the toxic moiety (e.g., parent compound, intermediate, or terminal metabolite),
does not impact the Agency's cancer classification.
2. The draft assessment concludes that there is insufficient information to identify the mode(s) of
carcinogenic action for 1,4-dioxane. Please comment on whether this determination is
appropriate and clearly described. If it is judged that a mode of action can be established for
1,4-dioxane, please identify the mode of action and its scientific support (i.e., studies that
support the key events, and specific data available to inform the shape of the exposure-
response curve at low doses).
Comment: Five out of six reviewers agreed with EPA's conclusion that there is
insufficient scientific information to establish the mode(s) of carcinogenic action for
1,4-dioxane. However, one of these reviewers suggested integrating the sequence of
events for a possible mode of action described in a public comment into the body of the
Toxicological Review. Another one of these five reviewers provided several examples of
places in the toxicological review that could use clarification of study limitations and
consideration of pertinent data: impact of 1,4-dioxane volatility on in vitro and skin/paint
study results; mechanistic section needs more discussion and analysis of a potential
genotoxic mode of action; critical deficiencies in the database should be noted in the
discussion of cytotoxicity/cell proliferation mode of action; examine dose-response
relationships for effects seen in the 13-week studies and how they may predict tumor
incidence; the lack of mouse liver initiation-promotion studies should be noted; and data
do not support statements regarding metabolic saturation and subsequent toxicity. One of
the six reviewers disagreed with EPA's conclusion that there is insufficient information
to identify a MOA for 1,4-dioxane. This reviewer commented that data clearly support a
cytotoxicity/inflammation/ regenerative hyperplasia MOA with a dose threshold, citing
the Kociba et al. (1974). Kano et al. (2008). and Kasai et al. (2009: 2008) studies.
Response: The Agency agrees with five of the six reviewers that there is insufficient
evidence to establish a carcinogenic MOA for 1,4-dioxane. As seen in responses to the
public comments regarding the carcinogenicity of 1,4-dioxane (Section A.4.2). the
sequence of events proposed by the public commenter are not supported by the available
data. These key events for the hypothesized MOA are visualized in Figure 4-1 of the
Toxicological Review.
The available data do not clearly support a cytotoxic/inflammation/regenerative
hyperplasia MOA (Section 4.7.3). Specifically, the studies referenced by the reviewer
(Kasai et al.. 2009; Kano et al.. 2008; 2008; Kociba etal.. 1974) do not examine
cytotoxicity or regenerative cell proliferation in the nasal cavity. Further, the existing data
examine a small number of exposures and timepoints. Kasai et al. (2009) suggests either
genotoxic or cytotoxic MOA for 1,4-dioxane, but their data do not provide sufficient
evidence to conclude one way or the other. Furthermore, there is no evidence of
cytotoxicity in the nasal cavity in the Kasai et al. (2009; 2008) studies. Additionally,
A-29
-------
evidence of cytotoxicity in one tissue type, does not dictate that cytotoxicity will be
present in all tissues at the same dose. Thus, the database does not provide evidence for
each stage of a regenerative hyperplasia MOA.
A number of changes were made as a result of the specific comments made regarding
clarity and study limitations. Regarding the volatility of 1,4-dioxane and reliability of the
negative in vitro studies and skin paint studies, text was added to Section 4.5.1 noting the
four negative in vitro studies that reported using closed systems and to Section 4.2.3
regarding the reliability of the data from unoccluded versus occluded skin paint
initiation/promotion studies. Text was revised in Section 4.5.1 to state clearly that half of
the studies showed 1,4-dioxane was not genotoxic; however, data are not sufficient to
support a genotoxic MOA and no additional discussion regarding this MOA was added to
the document. Text was added to Section 4.7.3 noting deficiencies in the database
surrounding a cytotoxicity/cell proliferation MOA. As a result of the peer review
comment, the noncancer effects were reexamined in detail and how they may relate to the
cancer effects seen. An attempt was made to create new tables showing the noncancer
and cancer effects across the dose and time; however, these tables were found to
introduce more confusion. Therefore, only clarifying text was added (Sections 4.7.1.
4.7.3.1.2. and 4.7.3.3) regarding the noncancer effects and their relation to the cancer
effects and the temporal sequence of events, as well as clarifying the. In response to
another comment from the reviewer, a statement was added to Section 4.7.3.1.1 to clearly
state that no studies have been conducted to specifically examine the mouse liver, thus
precluding any determination on whether 1,4-dioxane acts as a tumor promoter in the
mouse liver. A thorough review of statements in the document pertaining to metabolic
saturation and its relation to toxicity was performed in response to the reviewers
comment. Several changes were made throughout the document (e.g., Section 3.5.1.
4.6.2.1. and 4.7.3.7.1) clarifying relationships observed (or not) between metabolic
saturation and toxicity. In general metabolic saturation was observed in single dose
studies (Young et al.. 1978a. b). We agree with the reviewer that a single dose study does
not provide adequate information to support metabolic saturation following repeated
long-term exposures, and that since 1,4-dioxane induces P450 enzymes it is likely to
enhance metabolic elimination in long-term exposure scenarios. Additional kinetic
information is needed to determine if metabolic saturation is a precursor to atoxic effect.
Kociba et al. (Kocibaetal.. 1975) that stated toxicity was only observed after metabolism
was saturated did not present data for repeated doses to support this conclusion.
A-30
-------
3. A two-year inhalation cancer bioassay in male rats (Kasai et al., 2009) was selected as the basis
for the derivation of the inhalation unit risk (IUR). Please comment on whether the selection of
this study is scientifically supported and clearly described. If a different study is recommended
as the basis for the IUR, please indentify this study and provide scientific support for this
choice.
Comment: Five of the six reviewers agreed that the use of the two year inhalation cancer
bioassay in male rats Kasai et al. (2009) is the most appropriate study to use for the
derivation of the IUR. Five of the six reviewers also stated the selection was clearly
described and justified or supported within the toxicological review. The other reviewer
neither disagreed or agreed with the selection of the study; however, the reviewer noted
that the Kasai et al. (2009) study is the only comprehensive inhalation study available for
this chemical, because the other study by Torkelson et al. (1974) used only one dose and
did not perform histology on the nasal tissues.
Response. No dissenting opinions or comments warranting additional justification were
provided by the external review panel regarding selection of the principal study for
derivation of the IUR. Thus, no changes were made to the assessment related to the
selection and justification of the Kasai et al. (2009) study for derivation of the IUR.
4. The incidence of hepatocellular adenomas and carcinomas, nasal cavity squamous cell
carcinoma, renal cell carcinoma, peritoneal mesothelioma, mammary gland fibroadenoma,
Zymbal gland adenoma, and subcutis fibroma were selected to serve as the basis for the
derivation of the IUR. Please comment on whether this selection is scientifically supported and
clearly described. If a different health endpoint is recommended for deriving the IUR, please
identify this endpoint and provide scientific support for this choice.
Comment: Five of the six reviewers agreed with EPA's choice to combine these tumor
types for derivation of the IUR, noting the statistically significant tumor incidence rates
and the dose related increase in tumors. One of the five reviewers that agreed with the
approach questioned if data are available to fully justify the pooling of certain tumor
types. One of these five reviewers noted that the mice were more sensitive than rats to the
hepatocarcinogenic effects of 1,4-dioxane following drinking water exposure. Thus, since
mice were not included in a 2-year inhalation cancer bioassay, the IUR may be
underestimated and this should be noted as a source of uncertainty qualitatively and a
quantitatively. This reviewer suggested a quantitative adjustment to the IUR by
multiplying the IUR by the ratio of hepatocellular neoplasms in male rats: female mice
from the oral study. The sixth reviewer disagreed with combining all of these tumor
types, arguing that Zymbal gland tumors are limited to male rats; and peritoneal
mesothelioma, subcutis fibroma, and mammary fibroadenoma are typical spontaneous
tumors in F344 rats (Haseman et al.. 1998: Hall. 1990).
Response: In agreement with five of the six reviewers, the Agency retained the
combination of the tumor types with statistically significant incidence rates different from
A-31
-------
control or a statistically determined dose-related trend in the combined tumor analysis for
the derivation of the IUR. Data were not available to establish whether the tumor types
were biologically dependent, thus independence was assumed and is not expected to
produce substantial error in the risk estimates (NRC. 1994). It is acknowledged that
Zymbal gland tumors do not occur in humans due to the lack of a Zymbal gland;
however, site concordance is not always assumed for animals and humans (U.S. EPA.
2005a) because events leading to Zymbal gland tumors may occur at other sites in
humans. Additional text was added to Sections 5.5.1.6 and 6.2.3.8 to address the possible
underestimation of the carcinogenic inhalation potential of 1,4-dioxane since female mice
were the most sensitive following oral administration and were not included in the 2-year
inhalation cancer bioassay. While the uncertainties were noted qualitatively, a
quantitative adjustment was not performed on the IUR as this is not a standard approach
conducted by the agency. The sixth reviewer raised objections to using peritoneal
mesothelioma, subcutis fibroma, and mammary fibroadenoma as the reviewer
characterized them as "very commonly observed, spontaneous tumors in control F344
rats." The study authors used untreated, clean air exposed rats as an experimental control
to account for any possible spontaneous tumors that may arise. Furthermore, the Agency
accounts for the background rate in controls when using the multistage cancer model.
5. The IUR was derived based on multiple carcinogenic effects observed in rats exposed to
1,4-dioxane via inhalation. A Bayesian approach was used to estimate a BMDLio associated
with the occurrence of these multiple tumors, and then a linear low-dose extrapolation from
this POD was performed to derive the IUR. Additionally, for comparative purposes only, a
total tumor analysis was performed with the draft BMDS (version 2.2Beta) MSCombo model
that yielded similar results (see Appendix FT). Please comment on whether these approaches for
deriving the IUR have been clearly described and appropriately conducted?
Comment: Two reviewers commented that the approaches were clearly described and
appropriately conducted; however, the methods to quantitate cancer risk are outside of
their areas of expertise. Four of the reviewers commented that both methods, Bayesian
and BMDS, are clearly described and appear appropriately conducted since both methods
yielded similar results. However, one of these four reviewers noted that additional
information to reproduce the Bayesian analysis should be provided. Another of these four
reviewers noted that IUR estimates may actually be larger since survival was
significantly reduced in the high exposure group and that the cancer dose-response
modeling did not use survival adjusted data. One reviewer commented that the limitations
and assumptions related to the risk of developing any combination of the tumor types is
not well documented in the toxicological review. Additionally, one reviewer noted that
the total tumor approach was not utilized in the derivation of the oral CSF and
recommended a total tumor analysis for male and female rats exposed to 1,4-dioxane in
drinking water. One reviewer did not support the Agency's default use of Haber's Law to
make adjustments for the exposure duration in the derivation of the IUR (or RfC). This
A-32
-------
reviewer suggested additional examination of the 1,4-dioxane data to gain insights into a
and (3, if possible to further describe uncertainties associated with this duration
adjustment.
Response. Overall, the reviewers were in support of the quantitative approaches to the
multitumor analysis for the derivation of the IUR. As a result of the public comments
regarding the documentation and reproducibility of the Bayesian WinBUGS approach
(Kopylev et al. 2009: Spiegelhalter et al.. 2003). and the fact that the BMDS MS_Combo
model has completed peer review since the draft of this assessment was released, the
transparent, reproducible MS_Combo approach is now considered the primary approach
for derivation of the IUR and the Bayesian WinBUGS approach is a supporting analysis
with details in Appendix G (external peer review draft. Appendix H). Additional details
on the WinBUGS analysis was added to the appendix and the model code was made
available via HERO (U.S. EPA. 2013d). Using MS_Combo approach as the primary
approach did not result in any quantitative changes to the IUR.
As stated in response to general charge question 1, similar methods to analyze the total
tumor risk were not available at the time of the completion of the oral assessment.
Additionally, the multistage model did not provide adequate fit for female mouse liver
tumor data and was not used in derivation of the oral slope factor, whereas the inhalation
unit risk derivation does utilize the multistage model. However, in response to the
reviewer's comment, the male and female rat data were analyzed using the BMDS
MS_Combo model. BMDLHEc values for male rat and female rat combined tumors were
determined to be 7.59 and 11.26 mg/kg-day, respectively. Using a BMR of 0.1 oral CSFs
of 0.013 and 0.0088 (mg/kg-day)"1 were calculated for the male and female rat data,
respectively. Thus the combined tumor analysis for the oral assessment does not impact
the selection of the gender/species or overall oral CSF for 1,4-dioxane.The Agency
concurs with the reviewer who states that the IUR estimates may actually be larger if
survival adjusted data were used and this was noted in Section 5.5.1.6. However, day of
death data were not available in the Kasai (2009) study, thus this analysis cannot be
performed.
Data are not available to move away from the default value of 1 for a and |3 in the C x T
duration adjustment approach for inhalation exposure. Two, 13-week subchronic studies
in laboratory animals (Kasai et al.. 2008; Fairlev etal.. 1934) and two, 2-year chronic
studies in rats (Kasai et al.. 2009; Torkelson et al.. 1974) were identified; however, these
data did not report the severity of the lesions for multiple timepoints.
A-33
-------
A.4. Public Comments - Inhalation Update
The Toxicological Review ofl,4-Dioxane (with Inhalation Update) was released fora 60-day
public comment period in September 2011. A listening session was scheduled in October 2011; however,
no participants registered to speak, so the listening session was cancelled. EPA received written public
comments on the draft assessment from Toxicology Excellence for Risk Assessment (TERA) and joint
comments from the National Association of Manufacturers (NAM) the Aerospace Industries Association
(AIA) provided by ARCADIS. The major comments received have been synthesized and paraphrased
below. EPA's responses to the comments and information regarding how the assessment has been
revised, where applicable, are included.
A.4.1. Inhalation reference concentration (RfC) for 1,4-dioxane
Comment: The use of 3 for the database uncertainty factor (UFD) based on the lack of a
multigenerational reproductive study is not warranted. Statistically significant changes in
fetal weight and ossified sternebrae reported by Giavini et al. (1985) are not
lexicologically significant. No effects were seen on reproductive organs in the oral or
inhalation subchronic and chronic studies (Kano et al., 2009; Kasai et al., 2009; Kano et
al.. 2008; Kasai et al.. 2008; NCI. 1978; Kocibaetal.. 1974; Torkelson et al.. 1974). For
these reasons the UFD should be reconsidered in the derivation of the RfC.
Response: Giavini et al. (1985) administered 1,4-dioxane by gavage in water to pregnant
rats. The authors found statistically significant changes in fetal body weight at the highest
dose group and reduced ossification of the sternebrae. The other studies were not
designed to examine reproductive or developmental outcomes, and thus cannot be used to
infer the reproductive/developmental toxicity of 1,4-dioxane. While Torkelson et al.
(1974) did examine the testes and uterus for gross histopathological changes (e.g., tumor)
and did not find increased incidence of tumors, this does not indicate that 1,4-dioxane
may not be a developmental toxicant. The study of reproductive organs in subchronic and
chronic studies is not a replacement for a multigeneration reproductive/developmental
study. A UFD of 3 was used for the oral assessment and was retained for the inhalation
assessment due to the lack of a multigenerational reproductive study.
A.4.2. Carcinogenicity of 1,4-dioxane
Comment. Low dose linearity should not have been assumed to derive the proposed IUR
since sufficient data exist to support a cytotoxic-proliferative mode of action (MOA)
based generally on the following arguments: 1,4-dioxane is neither mutagenic nor an
initiator, but it can act as a promoter, "literature indicates that 1,4-dioxane is a weak
genotoxic carcinogen", Kasai et al. (2009) characterized the MOA as "cytotoxic-
A-34
-------
proliferative". Additionally, the Agency's statement that there is insufficient evidence to
support any hypothesized MOA is not supported by the "open literature and the data
summarized and interpreted in the draft TR". Histopathology results for the nasal
cavity/olfactory epithelium, liver, and kidney from Kasai et al. (2009) clearly indicate
that cytotoxicity precedes tumor development.
Response. The Kasai et al. (2009) study does not provide evidence of cytotoxicity in the
nasal cavity. Kasai et al. (2009) suggest either a genotoxic or cytotoxic MOA for
1,4-dioxane, but their data do not provide sufficient evidence for one hypothesis over the
other. There is no evidence of cytotoxicity in the Kasai et al. (2009; 2008) study. For
instance, inflammation by itself is not direct evidence of cytotoxicity. For the liver and
kidney, Kasai et al. (2009) provide direct evidence of cytotoxicity including clinical
pathology (liver) and histopathology (liver and kidney) data. Additionally, evidence of
cytotoxicity in one tissue type, does not dictate that cytotoxicity will be present in all
tissues at the same dose.
Due to a lack of information to inform the MOA, the Agency used the default linear
extrapolation approach per the EPA Guidelines for Carcinogen Risk Assessment (U.S.
EPA. 2005a). Specifically, the Guidelines state that "nonlinear approaches generally
should not be used in cases where the mode of action has not been ascertained" and that
linear extrapolation will be used as the default in these cases.
It is important to note that five of the six members on the independent expert peer review
panel for this draft assessment agreed with EPA's conclusions regarding the weight of
evidence in support of a linear approach to derive the IUR, and all reviewers, including
the public commenters, supported EPA's decision to use the Kasai et al. (2009) study as
the basis for determining the IUR.
Comment. 1,4-Dioxane dose not cause mutagenicity, initiation, or DNA repair.
1,4-Dioxane dose cause promotion and DNA replication. Occurrence of respiratory
tumors in rodents may be caused by 1,4-dioxane exceeding the metabolic capacity of the
tissue. 1,4-Dioxane does cause liver tumors and liver toxicity precedes tumors in time in
both sexes of rats and mice, and precedes tumors in dose in both sexes of rats. Liver
toxicity indicated by biochemical measures does occur at similar tumorigenic doses in
mice; however histopathological indication of liver toxicity does not appear to precede
tumors in either sex of mice. EPA needs to show the liver hyperplasia noted in Kano et
al. (2009) in Appendix E of the draft toxicological review. 1,4-Dioxane does cause dose-
dependent nasal toxicity as indicated in the histological analyses at all time points in both
sexes of rats and mice and this toxicity precedes tumors in time and dose. It is
hypothesized that 1,4-dioxane causes liver and nasal tumors in rats and mice through a
regenerative hyperplasia MOA, which demonstrates a threshold. The applicability of this
MOA to other tumor types is unknown, so a separate, default linear extrapolation may be
appropriate for those tumor types.
A-35
-------
Response. The Agency's determination that the MOA has not been established is
supported by five of the six external peer reviewers. The samples associated with liver
hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-
examined according to updated criteria for liver lesions and were afterwards classified as
either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009),
therefore there are no liver hyperplasia incidence data from Kano et al. (2009) to report in
Appendix E as the commenter suggests.
Due to a lack of information to substantiate the MOA, the Agency used the default linear
extrapolation approach per the EPA Guidelines for Carcinogen Risk Assessment (U.S.
EPA. 2005a). Specifically, the Guidelines state that "nonlinear approaches generally
should not be used in cases where the mode of action has not been ascertained" and that
linear extrapolation will be used as the default in these cases.
Comment. Peritoneal mesotheliomas found in male rats, but not female counterparts, is
likely due to the occurrence of tunica vaginalis mesotheliomas in male rats. Rats are
much more sensitive to developing mesotheliomas from the tunica vaginalis than
humans.
Response. The etiology and origin of the peritoneal mesotheliomas reported in Kano et
al. (2009) and Kasai et al. (2009) are unknown. The commenter indicated a range of
considerations including human sensitivity and / or relevance for the peritoneal
mesotheliomas observed in male rats (Kano et al.. 2009; Kasai et al.. 2009). The EPA
Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) state that all tumor types
are to be analyzed in a dose-response assessment followed by a synthesis that considers,
among other things, human relevance of each tumor type. In the absence of scientific
information to evaluate the human relevance of peritoneal mesotheliomas observed in
male rats exposed to 1,4-dioxane EPA is required to implement the approaches from the
guidance (U.S. EPA. 2005a). EPA concluded there continues to be uncertainty as to the
etiology, origin, and species sensitivity of the peritoneal mesotheliomas found in the rats,
and the tumor is relevant to humans and evaluated in the cancer assessment.
Comment. EPA should document a complete MOA evaluation for each relevant tumor
type by including a discussion on what is known about the key events in each tissue.
Response. MOA information available for tumors associated with exposure to
1,4-dioxane was evaluated in the Toxicological Review (Section 4.7.3). The MOA by
which 1,4-dioxane produces liver, nasal, kidney, peritoneal (mesotheliomas), mammary
gland, Zymbal gland, and subcutis 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 identify whether 1,4-dioxane or one of its metabolites is responsible for the
observed effects. Thus, it is not possible to document a complete MOA in any tissue. This
conclusion is supported by five of the six external reviewers.
A-36
-------
Comment. The parameters necessary to reproduce the total tumor analysis using the
Bayesian method (WinBUGS) are not provided; the analysis is poorly documented; and
the rationale for application of the analysis is incomplete.
Response. The BMDS (version 2.2Beta) MS_Combo approach for total tumor analysis
that was also included in support of the WinBUGS approach in the draft toxicological
review, is now highlighted as the main approach in the body of the document. The
MS_Combo approach uses the U.S. EPA's Benchmark Dose Software and is a
transparent, reproducible approach that provided similar to the output from the complex
WinBUGS analysis. The WinBUGS analysis is still included in this toxicological review
as a supporting analysis in Appendix G. Additional details on the WinBUGS analysis was
included in Appendix G and the model code made available via HERO (U.S. EPA.
2013d). Using MS_Combo approach as the primary approach did not result in any
quantitative changes to the IUR.
Comment. The requirements for scientific data to support a MOA appear too stringent.
EPA should provide guidance on what would be considered sufficient scientific evidence
to determine a MOA.
Response. 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. The data would fit the criteria described in the U.S. EPA's
Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a).
Comment. The attribution of some tumor types to exposure to 1,4-dioxane is
questionable based on statistics, including subcutis fibromas and Zymbal Gland
adenomas. There is also uncertainty surrounding the origin of the tumors reported in the
Kasai et al. (2009) study (e.g., may be the result of metastatic deposition), and hence the
assumption of biological independence among the tumor types included in the total tumor
analysis is not supported. Thus, the pooling of tumor types for derivation of the IUR in
the draft TR leads to overestimation of the actual carcinogenicity, and only tumor types
with statistically significant differences in incidence rate compared to control animals
should be used. Additionally, the highest dose used in the Kasai et al. (2009) study
exceeds the maximum tolerated dose (MTD) and should be excluded from the dose-
response analysis to derive the IUR.
Response. The commenter suggested that Zymbal Gland adenomas should not be
considered related to 1,4-dioxane exposure because the incidence rate at the highest dose
group was not statistically different from control; however, the Peto test did find a
statistically significant increasing trend. Tumor types were included in the analysis if they
showed a statistical difference from control or a statistically significant trend was evident.
Zymbal Gland adenomas were included in the analysis because the Guidelines for
Carcinogen Risk Assessment (U.S. EPA. 2005a) do not require site concordance and a
statistically significant dose-response trend was observed for these tumors. Similarly,
A-37
-------
subcutis fibromas were included in the total tumor analysis because a statistically
significant difference was seen in the mid dose group. The rationale for inclusion of
tumors in the multitumor analysis is described in Section 5.4.4.2. Additional scientific
information would be required to evaluate the hypothesis that the tumors "may be the
result of metastatic deposition."
The Kasai et al. (2009) study demonstrates that the high dose used in determining the
IUR is below the MTD for 1,4-dioxane. Kasai et al. (2009) state that the highest exposure
concentration (1,250 ppm) used in the 2 year study was found to fulfill established
criteria such that the highest dose should not exceed the MTD. Additionally, Kasai et al.
(2008) state that the MTD is likely higher than the 111 ppm reported by Torkelson et al.
(1974). The 3,200 ppm high dose in the 13 week Kasai et al. (2008) study is higher than
the 1,250 ppm dose used in the 2 year bioassay (Kasai et al.. 2009). and no overt toxicity
was reported at the 3,200 ppm exposure level.
A.4.3. PBPK modeling
Comment: PBPK models of sufficient quality are available and should have been used to
reduce uncertainty in both the oral and inhalation assessments. Technical errors were
identified in the PBPK analysis that should be addressed and the use of the models should
be reevaluated for both the oral and inhalation assessment.
Response. The model code errors noted in the public comments were addressed as noted
below; however, the changes did not significantly impact model predictions nor the
overall decision on model use in the assessment.
Comment. The permeation constant to describe the slowly perfused (diffusion-limited)
tissue compartment was improperly used in the PBPK model.
Response. If one assumes that the exiting venous concentration is at equilibrium with the
tissue, then the diffusion-limited tissue mass balance could be described as was shown in
the model code. It does slowly transport in/out of the tissue while having the property that
the tissue moves toward equilibrium with the blood, so it is empirically correct, though it
is acknowledged that this was not the most common way to code this compartment.
Therefore, to be up-to-date with current modeling practices, the blood flow to the slowly
perfused tissues (QS) was used instead of the diffusion limited constant (SPDC) change
was made to the model code; however, this had very minimal quantitative impact on
model output. Additionally, the fraction of fat and slowly perfused tissue compartments
was updated to be more similar to the values used in the values used in the published
models (see Table B-l).
A-38
-------
Comment. The metabolism of 1,4-dioxane in misused a zero order rate constant in the
equation.
Response. The metabolic constant was correctly used in the model code as a first order
rate constant; however, it was incorrectly described in the text and code comments as
zero-order. The description of the rate constant was corrected in the text and the model
code to be clear it is a first-order rate constant.
Comment. The model description for the urinary excretion of HEAA is not adjusted to
the ratio of the molecular weights, thus under predicting the concentration of HEAA in
urine.
Response. The reviewer is correct that the molecular weight was not accounted for, and
since the model mass units are in milligrams, the urinary excretion was corrected to
account for the mass conversion to HEAA. The corrected model predicts the human
urinary HEAA early time points well and over predicts the latter time points (694 mg
versus 621 mg) - See Appendix B. Following all updates to the model, metabolic
parameters were re-optimized and the plots and predictions updated in Appendix B.
These changes improved the model fits, but the model predictions of blood 1,4-dioxane
were still 4- to 7-fold lower than the data.
Comment. Complete model code (including all .m and .csl files) should be included for
the public and reviewers to use. It should be clear what model code was used to generate
each figure in the appendix.
Response. New practice within NCEA for transparency is to make the model code
accessible via the Health and Environmental Research Online (HERO) database. The
model code is now available via the online database and has been removed from the
appendix (U.S. EPA.2013a).
Comment. Although the Young et al. (1977) paper does have value in the model
development process, there are issues with the study design and exposure estimation, so it
should not be used to dismiss the use of the PBPK model for the assessment.
Response. In the absence of evidence to the contrary, the Agency cannot discount the
human blood kinetic data published by Young et al. (1977). As the commenter noted, the
liquids likely absorbed some 1,4-dioxane; however, if the volume of air they extract is
much less than the volume inhaled by a subject in an hour, then they won't contribute
much to the overall absorption. Thus, this reason presented by the commenter is not
sufficient for the Agency to discount the data for model validation.
A-39
-------
A.4.4. Other comments
Comment. There are other relevant data that are missing from this assessment. Reports
that should be referenced include: Takano et al. (2010). J Health Sci 56(5): 557-565 and
Department of the Army (2010) Toxicology Report No., 87-XE-08WR-09, Studies on
Metabolism of 1,4-dioxane.
Response. These same references were mentioned by a member of the independent
external peer review panel - refer to the response to the inhalation assessment update
general charge question #2 above. Briefly, Takano et al. (2010) was evaluated and added
to the assessment in Section 3.5.2.5. The Army study was added to Section 3.3 of the
toxicological review.
A-40
-------
APPENDIX B. EVALUATION OF EXISTING
PHARMACOKINETIC MODELS FOR 1,4-DIOXANE
B.1. 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, (3-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 (Takano et al., 2010; Fisher et al., 1997; Leung and Paustenbach. 1990;
Reitzetal.. 1990).
In developing 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 (Young etal., 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 recalibrated
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; Young et al.. 1977) and recalibration and exercising of the Reitz et al. (1990) PBPK model
using parameter values identified by Leung and Paustenbach (1990) and Sweeney et al. (2008). as well as
optimized values, to determine the potential utility of the models for 1,4-dioxane for interspecies and
route-to-route extrapolation.
B-l
-------
B.2. Implementation of the Empirical Models in acsIX
The scope of this effort consisted of implementation of the Young et al. (1978a. b; 1977)
empirical rat and human models using acsIX, version 3.0.2.1 (Aegis Technologies, Huntsville, AL).
Using the model descriptions and equations given in Young et al. Q978a, 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). All
model files are available electronically via HERO (U.S. EPA. 2013a).
B.2.1. 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. Pulmonary absorption is described by a first-order rate constant (kn^)- The
metabolism of 1,4-dioxane and subsequent appearance of HEAA is described by Michaelis-Menten
kinetics governed by a maximum rate (Vmax, mg/hr) and affinity constant (Km, mg). The elimination of
both 1,4-dioxane and HEAA were described with 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 was determined by dividing the amount of
1,4-dioxane in blood by a volume of distribution (Vd) of 0.301 L, which was the average Vd determined
from the i.v. dose studies.
Inhalation (k
i.v. admin
/
dt Km+Dioxbody
-kexDiox^
dt Km + Diox,
body
Exhaled (65%
kexDioxbody J*
> Urine (35%N
kme xHEAA
•>• Urine
Figure B-l. Schematic representation of empirical model for 1,4-dioxane in rats.
Figure B-2 illustrates the Young et al. (1977) human empirical model for 1,4-dioxane. 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(dlox) and kme(HEAA), respectively. Pulmonary
absorption is described similar to the approach used in the rat empirical model. Blood concentrations of
1,4-dioxane and HEAA are calculated as instantaneous amount (mg) divided by volume of distribution
(Vd): Vd(dlox) or Vd(HEAA) (104 and 480 mL/kg BW, respectively [calculated by Young et al. (1977)1).
B-2
-------
Inhalation
Dioxane
'd(Diox) X
HEAA
*
-------
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.2.3. Results
The acslX implementation of the Young et al. (1978a. b) rat empirical model is in good
agreement with the 1,4-dioxane blood levels from the i.v. experiments and the model output reported in
the published paper (Figure B-3). However, the acslX version predicts urinary HEAA following i.v. dose
to reach a maximum sooner than the measured and predicted levels reported in the paper (Figure B-4).
These discrepancies may be due, at least in part, to the reliance in the acslX implementation on a constant,
standard urine volume rather than experimental measurements of urine volume, which may have been
different from the assumed value and may have varied overtime. Unreported model parameters (e.g., lag
times for appearance of excreted HEAA in bladder urine) may also contribute to the discrepancy.
10000
acslX version - Young et al
(197Sa.b) empirical node I
30 40
Time(hrs)
Source:
Left panel: Data points from Young et al. (1978a. b), and lines generated from EPA's acslX implementation of the Young et al.
(1978a. b) empirical rat model.
Right panel: Reprinted with permission of Taylor & Francis, Young et al. (1978a. b). The lines in the figure on the right are best fit
lines, and do not represent empirical rat model simulations.
Figure B-3. Output of 1,4-dioxane blood level data from the acslX implementation (left)
and published (right) empirical rat model simulations of i.v. administration
experiments.
B-4
-------
10.000
1.4-Dioxane
HEAA(mgEq/L)
• 1.4-Dioxane
D HEAAingEq/L;
20 30
Time (hrs)
o 11 ( i i t I i i i i _
0 5 10 IS M » 3D 35 « « 60
Hourt
Source:
Left panel: Data points from Young et al. d978a. b), and lines generated from EPA's acsIX implementation of the Young et al.
(1978a. b) empirical rat model.
Right panel: Reprinted with permission of Taylor & Francis, Young et al. (1978a. b). The lines in the figure on the right are best fit
lines, and do not represent empirical rat model simulations.
Figure B-4. Output of HEAA urine level data from acslXtreme implementation of the
empirical rat model (left) and published (right) data following i.v.
administration experiments.
The Young et al. (1978a. b) report did not provide model predictions for the 50-ppm inhalation
experiment. However, the acsIX implementation produces blood 1,4-dioxane predictions that are similar
to the reported observations (Figure B-5). As with the urine data from the i.v. experiment, the amount of
HEAA in urine predicted using the acsIX implementation was approximately threefold lower than the
observations However, this prediction is the amount of HEAA excreted overtime and does not rely on an
estimate of urine volume to calculate, thus the reason for the discrepancy is likely due unreported model
parameters (e.g., lag times for appearance of excreted HEAA in bladder urine) or to more complex
kinetics than described using this simple model structure.
B-5
-------
D
acslXversion - Young et al.
il973a, b) empirical model
r'OLing et al. -]197Sa. b)
observations
—25
>-.
acslX version - Young et al.
(1973a. b) empirical model
Young etcil. (197Sa. b)
observations
-------
3000
2500
£ 2000
-a 1500
1000
500
• Male Delta
D Female Data
D Moclel prediction
400 GOO 1.500
Inhalation Exposure (pprr)
5200
Source: Male and female data digitized from Kasai et al. (2008). Model prediction from EPA's acsIX implementation of the Young et
al. (1978a. b) empirical rat model.
Figure B-6. acsIX predictions of blood 1,4-dioxane levels using the Young et al. (1978a,
b) model compared with data from Kasai et al. (2008).
Inhalation data for a single exposure level (50 ppm) are available for humans. The acsIX
predictions of the blood 1,4-dioxane observations are similar to the predictions reported in Young et al.
(1977) (Figure B-7). Limited blood HEAA data were reported (n = 2-3 individuals), 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. Young et al. (1977) only compared model
simulations to blood 1,4-dioxane in their report.
Data for cumulative urinary HEAA amounts are provided in Young et al. (1977). and no
analytical problems associated with these data were reported. The acsIX prediction of the HEAA kinetics
profile is similar to the observations (Figure B-8). Unlike urinary HEAA observations in the rat, human
observations were reported as cumulative amount produced, negating the need for urine volume data.
Therefore, discrepancies between model predictions and experimental observations were reduced.
B-7
-------
1,4-Dioxane predicted - acsIX version
Young (1977) empirical model
— — HEAA predicted - acsIX version Young
(1977)empirical model
• HEAA observed
Time(hrs)
Source:
Left panel: Data points from Young et al. (1977). and lines generated from EPA's acsIX implementation of the Young et al. (1977)
empirical human model.
Right panel: Reprinted with permission of Taylor & Francis, Young et al. (1977). The lines in the figure on the right are best fit lines,
and do not represent empirical human model simulations.
Figure B-7. Output of 1,4-dioxane and HEAA blood concentrations from the acsIX
implementation of the empirical human model (left) and published (right)
data of a 6-hour, 50-ppm inhalation exposure.
700 -
"w
V 600 -
1
C 500 -
1
1
D
^^acslx version- Young et al. (1977)
empirical model
D Observed
0 5 10 15 20 25
Time(hrs)
Source: Data points from Young et al. (1977), and lines generated from EPA's acsIX implementation of the Young et al. (1977)
empirical human model.
Figure B-8. Observations and acsIX predictions of the cumulative amount of HEAA in
human urine following a 6-hour, 50-ppm inhalation exposure.
B-8
-------
B.2.4. Conclusions for Empirical Model Implementation
The empirical models described by Young et al. Q978a, b; 1977) for rats and humans were
implemented using acslX. The models were modified to allow for user-defined inhalation exposures by
addition of a first-order rate constant for pulmonary uptake of 1,4-dioxane, fitted to the inhalation data.
No modifications were made to describe oral absorption as adequate data are not available for parameter
estimation. The acslX predictions of 1,4-dioxane in the blood are similar to the published data and
simulations of 6-hour, 50 ppm inhalation exposures in rats (Figure B-5) and humans (Figure B-7) and 3 to
1,000 mg/kg i.v. doses in rats (Figure B-3). However, the acslX version predicts lower urinary HEAA
amounts and concentrations in rats appearing earlier than either the Young et al. (1978a. b) model
predictions or the experimental observations (Figure B-4 and Figure B-5). The lower predicted urinary
HEAA concentrations in the acslXtreme implementation for rats are likely due to use of default values for
urine volume in the absence of measured volumes. The reason for the differences in time-to-peak levels
or amount of HEAA in urine is unknown, but may be the result of an unreported adjustment by Young et
al. (1978a, b) in model parameter values or more complex kinetics than can be described with this model
structure. Additionally, the acslX implementation of the Young et al. (1978a. b) model failed to provide
adequate fit to blood data reported following subchronic inhalation of 1,4-dioxane in rats at the two high
doses (Kasai et al.. 2008V
For humans, Young et al. (1977) did not report model predictions of urinary HEAA levels. The
urinary HEAA levels predicted by acslX approximated the observations reasonably well (Figure B-8).
while the blood HEAA did not (Figure B-7). However, unlike the situation in rats, these urine data are not
dependent on urine volumes (observations were reported as cumulative HEAA amount rather than HEAA
concentration). Presently, there is no explanation for the lack of fit of the empirical model to the blood
HEAA data. Since no blood HEAA model fits were shown in Young et al. (1977), it is unclear if the
discrepancy is in the original model or only in the acslX implementation.
B.3. Initial Evaluation of the PBPK Models
The PBPK model of Reitz et al. (1990) was selected for further evaluation of its potential
application in this assessment. The model was not sufficient as published, and thus was recalibrated using
measured values for cardiac and alveolar flow rates and tissue:air partition coefficients (Sweeney et al..
2008; Leung and Paustenbach. 1990). The predictions of blood and urine levels of 1,4-dioxane and
HEAA, respectively, from the recalibrated model were compared with the empirical model predictions of
the same dosimeters to determine whether the recalibrated 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. Additionally, the models were tested using first-order metabolism (instead of Michaelis-
Menten saturable metabolism) to determine if better model predictions could be generated.
B-9
-------
B.3.1. Initial Recalibration of the Reitz et al. PBPK Model
Concern regarding adjustments made to some of the parameter values in Reitz et al. (1990)
prompted a recalibration 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 liverair value in order to attain an adequate fit to the observations. This approach increases
uncertainty in these parameter values, and in the utilization of the model for extrapolation. Therefore, the
model was recalibrated 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.3.2. Flow Rates
The cardiac output of 30 L/hr/kg074 (Table B-l) reported by Reitz et al. (Reitz etal.. 1990) is
approximately double the mean resting value of 14 L/hr/kg0 74 reported in the widely accepted
compendium of Brown et al. (1997). Resting cardiac output was reported to be 5.2 L/min (or 14
L/hr/kg074), while strenuous exercise resulted in a flow of 9.9 L/min (or 26 L/hr/kg074) (Brown et al..
1997). Brown et al. (1997) also cite the ICRP (1975) as having a mean respiratory minute volume of
7.5 L/min, which results in an alveolar ventilation rate of 6.86 L/min (assuming 8.5% lung dead space,
(Overton etal.. 2001)). or 17.7 L/min/kg074. Again, this is roughly half the value of 30 L/hr/kg074
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/hr/kg074 is not consistent with the experimental conditions
being simulated.
A minute volume of 7.5 L/min (or 17 L/hr/kg074) was used in the acslX implementation of the
Young et al. (1977) model for volunteers having a mean BW of 84 kg and fit the blood 1,4-dioxane data
reasonably well. Based on these findings, the cardiac output and alveolar ventilation rates of 17.0 and
17.7 L/hr/kg074were biologically plausible forthe 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 forthe model recalibration.
B-10
-------
Table B-l Human PBPK model parameter values published in literature and values used
by EPA in this assessment for 1,4-dioxane
Parameter (Abbreviation)
Body weight (BW)
Cardiac output (QCC)a
Alveolar ventilation (QPCf
Reitz et al.
(1990)
70
30
30
Leung and
Paustenbach
(1990)
84.1
15
15
Sweeney et al.
(2008)
70
13
13
EPAb
84.1
17.0
17.7
Fractional Blood Flows
Liver (QLC)
Fat (QFC)
Richly perfused (QRC)
Slowly perfused (QSC)
0.25
0.05
0.52
0.18
0.25
0.05
0.51
0.19
0.227
0.052
0.472P
0.249
0.25
0.05
0.52P
0.18
Fractional Tissue Volumes
Liver (VLC)
Fat (VFC)
Richly perfused (VRC)
Slowly perfused (VSC)
Blood (VBC)
Unperfused tissue (VUC)
0.031
0.231
0.037
0.561
0.05
-
0.04
0.20
0.05
0.62
-
-
0.033
0.214
0.1 66q
0.437
0.079
0.071
0.04
0.20
0.05q
0.57
0.05
0.09
Partition Coefficients (PCs)
Blood:air(PB)
Fatair(PFA)
Liverair(PLA)
Rapidly perfused tissue:air (PRA)
Slowly perfused tissue:air (PSA)
3,650C
851
1,557
1,557
1.5571
1,825±94d
(77 = 14)
851 ± 118d
(n=8)
1,557± 114d
(77=4)
1,557g
997 ± 254d
(n=6)
1,666 ±287
(n=36)
865e
1,862±739f
(77 = 14)
560 ± 175h
(77=7)
1,348±290f
(77=7)
1,825
851
1,557
1,557
260j'm
B-ll
-------
Table B-l (Continued): Human PBPK model parameter values published in literature and
values used by EPA in this assessment for 1,4-dioxane
Parameter (Abbreviation)
Reitz et al.
(1990)
Leung and
Paustenbach
(1990)
Sweeney et al.
(2008)
EPAb
Metabolic Constants
Maximum rate for 1,4-dioxane
metabolism (Vmaxc; mg/hr-kg BW07)
Metabolic affinity constant (Km; mg/L)
HEAA urinary elimination rate constant
(kme, hour-1)
12.5"
3.00
0.56
13.3°
15
-
54, 75
or192fe
29, 32,
or147!
0.35
5.8j
5.3j
0.30j
aL/hr/kg BW074
bValues utilized by EPA in this assessment. Body weight was mean weight reported by Young et al. (1977).
°Doubled from experimental value (1,825) to obtain better fit to human data (Reitz et al.. 1990).
dLeung as Paustenbach (1990) did not state if the values were reported ± standard deviation or standard error.
eAverage of Reitz et al. (1990) rat value and mouse value determined by Sweeney et al. (2008).
'Assumed equal to the measurement for rat tissue determined by Sweeney et al. (2008).
9Assumed equal to liverair partition coefficient.
hAssumed equal to mouse kidney determined by Sweeney et al. (2008).
'Authors reported poor fits to the venous blood data for rats and humans when the experimentally determined muscle:air partition
coefficient was used (value not reported) and had improved fits of the data when the partition coefficient for liverair was used.
'Obtained by model optimization.
kUsed parallelogram scaling approach based on scaled in vitro data to give a range of values referred to by the authors as
"minimum, representative, and maximum."
'Scaled rat in vitro data according to in vitro human:rat ratios to give a similar range as Vmax, referred to by the authors as
"minimum, representative, and maximum."
"Value used in Figure B-11. estimated 4-fold lower value than Leung as Paustenbach (1990) because recalibrated model was
predictions were 4- to 7-fold lower than the data; however, this parameter value is not considered "biologically plausible."
"Reported in manuscript as 6.55 mg/hr-kg BW°86. Converted to mg/hr-kg BW07 for consistency.
"Reported in manuscript as 6.55 mg/hr-kg BW°86. Converted to mg/hr-kg BW07 for consistency.
"Calculated from QRC=1-(QFC+QSC+QLC)
Calculated from VRC=1-(VLC+VFC+VSC+VBC+VUC)
B.3.3. 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 used vial equilibration
techniques for experimental determinations. The values reported in Leung and Paustenbach (1990) were
also used, at least as starting points, by Reitz et al. (1990). Leung and Paustenbach (1990) reported mean
values and an indication of variance (it was not clear if the values were standard deviations or standard
errors) for human blood:air, rat blood:air, rat liverair, rat muscle:air (e.g., slowly perfused tissue:air), and
rat fat:air (Table B-l). They assumed the rapidly perfused tissue:air partition coefficient was equal to the
value for the liver and that all human tissue partition coefficients were equivalent to the rat, except where
the separate determination was made for human blood: air partition coefficient.
Sweeney et al. (2008) experimentally determined partition coefficients for blood:air (mouse, rat,
and human), liverair (mouse and rat), fatair (mouse), richly perfused tissue:air (mouse), and slowly
perfused tissue:air (mouse). Values for human tissue:air partition coefficients for the model were
B-12
-------
estimated as averages of rat and mouse values (liverair, fatair, and slowly perfused tissue:air) or set
equal to the mouse value (richly perfused:air set equal to mouse kidney:air partition coefficient) (Sweeney
et al.. 2008). For example, the human fatair partition coefficient, used an average (851) of the Reitz et al.
(1990) rat value (851) and their experimentally determined mouse value (879) (Sweeney et al.. 2008).
For the PBPK model implementation, tissue:blood partition coefficients for each compartment
were determined by dividing the tissue:air partition coefficients by the blood:air partition coefficient.
B.3.4. Calibration Method
The PBPK model was recalibrated three times using the physiological values selected by EPA
(current assessment, Table B-l) and the (1) partition coefficients of Leung and Paustenbach (1990).
(2) Sweeney et al. (2008). and (3) biologically plausible values based on these two publications,
separately. For each calibration, the metabolic parameters VmaxC and Km, were simultaneously fit (using
the parameter estimation tool provided in the acslX software) to the output of 1,4-dioxane blood
concentrations generated by the acslX 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 that
were validated against the experimental observations were used to provide a more robust data set for
model fitting, since 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.3.5. Results
Results of the model recalibration are provided in Table B-2. The recalibrated values for VmaxC
and kme associated with the Leung and Paustenbach (1990) or Sweeney et al. (2008) tissue:air partition
coefficients are very similar. Plots of predicted and experimentally observed blood 1,4-dioxane and
urinary HEAA levels are shown in Figure B-9 and Figure B-10 for Leung and Paustenbach (1990) and
Sweeney et al. (2008) partition coefficients. Neither recalibration resulted in an adequate fit to the blood
1,4-dioxane data from the empirical model output or the experimental observations. Recalibration using
either the Leung and Paustenbach (1990) or Sweeney et al. (2008) partition coefficients resulted in blood
1,4-dioxane predictions that were 4- to 7-fold lower than empirical model predictions or observations.
The refitted values for kme resulted in HEAA levels in urine that were very similar to the
empirical model output (compare Figure B-7. Figure B-9. and Figure B-10). which was not surprising,
given the fitting of a single parameter to the data.
Model outputs of the blood 1,4-dioxane and urinary HEAA levels using the EPA suggested
(Table B-2) parameters are shown in Figure B-11. To obtain these improved fits, a very low value for the
slowly perfused tissue:air partition coefficient (22) was used. The value was 4- to 6-fold lower than the
B-13
-------
measured values reported in Leung and Paustenbach (1990) and Sweeney et al. (2008). and 7-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 (e.g., 2- to 3-fold lower than the observations), the value used for the
slowly perfused tissue partition coefficient is not supported by laboratory data.
Table B-2 PBPK metabolic and elimination parameter values resulting from recalibration
of the human model using alternative values for physiological flow rates3 and
tissuerair partition coefficients
Source of Partition Coefficients
Leung and Paustenbach
(1990)
Sweeney et al.
(2008)
EPA
Maximum rate for 1,4-dioxane metabolism (Vmaxc)
4.9
4.0
5.8
Metabolic affinity constant (Km)'
1.8
0.78
5.3
HEAA urinary elimination rate constant (kme)'
0.27
0.25
0.30
aCardiac output = 17.0 L/hr/kg BW074, alveolar ventilation = 17.7 L/hr/kg BW074
bmg/hr/kg BW07
°mg/L
"hour"1
1
o
Q
o
_0
m
-PBPK predicted
Empirical predicted
1.4-Dioxane observed
X
^
~ 5'JO
O
£ 300
<
jjj 200
•3
£
3
u 100
^—PBPK predicted
— Empirical predicted
D Observed
6 8
Time(hrs)
10 15
Timefhrs)
Source: Data points from Young et al. (1977). Dotted lines generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Solid lines generated from EPA's implementation of Reitz et al. (1990) human PBPK model using partition
coefficient values from Leung and Paustenbach (1990).
Figure B-9. Human predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-dioxane
exposure and recalibration of the PBPK model with tissue:air partition
coefficient values from Leung and Paustenbach (1990).
B-14
-------
o
a
— — Empirical pn
D 1.4-Dioxane observed
_§
01 ijMM
I
"S 400
E 300 -
Time (hrs)
PBPK predicted
— ^ Enpirical predicted
D Observed
Time (hrs)
Source: Data points from Young et al. (1977). Dotted lines generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Solid lines generated from EPA's implementation of Reitz et al. (1990) human PBPK model using partition
coefficient values from Sweeney et al. (2008).
Figure B-10. Human predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-dioxane
exposure and recalibration of the PBPK model with tissue:air partition
coefficient values from Sweeney et al. (2008).
B-15
-------
ited
~ ™ Empirical predicted
D 1 4-Dio>:ane observec
^—PBPKpre dieted
— Empirical predicts c
D HEAAobserved
Time(hrs)
10 15
Time(hrs)
Source: Data points from Young et al. (1977). Dotted lines generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Solid lines generated from EPA's implementation of Reitz et al. (1990) human PBPK model using partition
coefficient values from EPA estimated biologically plausible parameters (see Table B-1).
Figure B-ll. Human predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-dioxane
exposure, using EPA biologically plausible parameters.
B.3.6. Conclusions for PBPK Model Implementation
Recalibration of the human PBPK model was performed using experiment-specific values for
cardiac output and alveolar ventilation (Young etal.. 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 4- to 7-fold
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 (Figure B-9 and Figure B-10).
Output from the model using biologically plausible physiological parameter values (Table B-1).
Figure B-ll shows that application of a value for the slowly perfused tissue: air partition coefficient,
which is 6-fold lower than the measured value reported by Leung and Paustenbach (1990). results in
closer agreement of the predictions to observations. Thus, model recalibration 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.
The Sweeney et al. (2008) PBPK model consisted of compartments for fat, liver, slowly perfused,
and other well perfused tissues. Lung and stomach compartments were used to describe the route of
exposure, and an overall volume of distribution compartment was used for calculation of urinary
excretion levels of 1,4-dioxane and its metabolite, HEAA. Metabolic constants (VmaxC and Km) for the rat
PBPK model were derived by optimization data from an i.v. exposure of 1,000 mg/kg data (Young et al..
B-16
-------
1978a. b) for induced metabolism. For uninduced metabolism data generated by i.v. exposures to 3, 10,
30, and 100 mg/kg were used (Young etal. 1978a. b). Data generated from the 300 mg/kg i.v. exposure
were not used to estimate VmaxC and Km. The best fitting values for VmaxC to estimate the blood data from
the Young et al. (1978a. b) study using the Sweeney et al. (2008) model resulted in VmaxC values of 12.7,
10.8, 7.4 mg/kg-hr07; suggesting a gradual dose dependent increase in metabolic rate with dose. These
estimates were for a range of doses between 3 and 1,000 mg/kg i.v. dose. Although the Sweeney et al.
(2008) model utilized two values for VmaxC (induced and uninduced), the PBPK model does not include
dose-dependent function description of the change of Vmax for i.v. doses between 100 and 1,000 mg/kg.
PBPK model outputs were compared with other data not used in fitting model parameters by visual
inspection. The model predictions gave adequate match to the 1,4-dioxane exhalation data after a
1,000 mg/kg i.v. dose. 1,4-Dioxane exhalation was overpredicted by a factor of about 3 for the 10 mg/kg
i.v. dose. Similarly, the simulations of exhaled 1,4-dioxane after oral dosing were adequate at
1,000 mg/kg, and 100 mg/kg (within 50%), but poor at 10 mg/kg (model overpredicted by a factor of
five). The fit of the model to the human data (Young etal.. 1977) was also problematic (Sweeney et al..
2008). Using physiological parameters of Brown et al. (1997) and measured partitioning parameters
(Sweeney et al.. 2008; Leung and Paustenbach. 1990) with no metabolism, measured blood 1,4-dioxane
concentrations reported by Young et al. (1977) could not be achieved using the reported exposure
concentrations. Inclusion of any metabolism further decreased predicted blood concentrations. If
estimated metabolism rates were used with the reported exposure concentration, urinary metabolite
(HEAA) excretion was underpredicted (Sweeney et al.. 2008). Thus, the models were inadequate to use
for rat to human extrapolation.
B.3.7. Sensitivity Analysis
A sensitivity analysis of the Reitz et al. (1990) model was performed, using the EPA values listed
in Table B-l, 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.3.8. 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:
f'(x) * Ax f(x)
B-17
-------
where * 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.3.9. Results
The sensitivity coefficients for the seven most influential model parameters at 1 and 4 hours of
exposure are shown in Figure B-12. The three parameters with the highest sensitivity coefficients in
descending order are alveolar ventilation (QPC), the blood:air partition coefficient (PB), and the slowly
perfused tissue:air partition coefficient (PSA). 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.
0.
QPC
-
PB
-
fe PSA
0
a QSC
ro
°- QCC
-
VmaxC
K
Sensitivity Coefficients: CV - 1hr
31 0.10 1.
I
I
I
I
I
I
DO
0.
QPC
-
PB
-
oi PSA
0
m V ,,
ro
^ Km
PRA
QSC
Sensitivity Coefficients: CV - 4 hr
31 0.10 1.
I
I
I
I
I
I
DO
Figure B-12. 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.4. 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),
B-18
-------
recalibrating 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.4.1. 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.4.2. 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:air 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.3.4. 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 recalibrated 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.4.3. Results
The predicted time courses for a 6-hour, 50-ppm inhalation exposure for the recalibrated human
PBPK model with mean (central tendency) and ± 2 standard deviations from the mean values for partition
coefficients are shown in Figure B-l3 for the Leung and Paustenbach (1990) values and Figure B-l4 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
B-19
-------
plausible values from Leung and Paustenbach (1990) or Sweeney et al. (2008). the model predictions are
still at least 2- to 4-fold lower than either the empirical model output or the experimental observations.
The range of possible urinary HEAA predictions approximate 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.
e
o
'«
5
o
o
4
o
_0
01
1 I
D Young etal. U977!Observed Data
— — Young etal. (1977) Empirical Model
— — PC- Upper
PC- Central
--- PC-Lower
\x
\\
o
J 500
Cl
>
^ 200
E
=1
O
D Young etal. (1977! Observed Data
Young etal. (1977) Empirical Model
PC-Upper
PC-Central
— — — PC- Lower
Time (hrs)
Time (hrs)
Source: Data points from Young et al. (1977). Red dotted line generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Blue lines generated from EPA's implementation of Reitz et al. (1990) human PBPK model using partition
coefficient values (solid blue line = mean partition coefficients; dotted blue lines = upper and lower boundaries on partition
coefficients) from Leung and Paustenbach (1990).
Figure B-13. Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients from Leung & Paustenbach
(1990) with empirical model predictions and experimental observations for
human blood 1,4-dioxane concentrations (left) and amount of HEAA in
human urine (right) from a 6-hour, 50-ppm inhalation exposure.
B-20
-------
c
o
u
o
5
o
J>
CD
a Younget al. 1.19771 Observed Data
_ _ Younget al. •!;<"• Empirical Mod
— — PC-Upper
PC-Central
— — — PC-Lower
-------
BAA. Alternative Model Parameterization
Since the PBPK model does not predict the experimental observations of Young et al. (1977)
when parameterized by biologically plausible values, an exercise was performed to explore alternative
parameters and values capable of producing an adequate fit of the data. Since the metabolism of
1,4-dioxane appears to be linear in humans for a 50-ppm exposure (Young et al.. 1977). the parameters
VmaxC and Km were replaced by a first-order, non-saturable metabolism rate constant, kLC. This rate
constant was fitted to the experimental blood 1,4-dioxane data using partition coefficient values of
Sweeney et al. (2008) to minimize the Vd (i.e., maximize the blood 1,4-dioxane levels). The resulting
model predictions are shown in Figure B-15. As before, the maximum blood 1,4-dioxane levels were
approximately sevenfold lower than the observed values.
I
I
o
0
4
D Younget
— ~ Young et al. (1977i Empirical Model
kLC (1.2) -filled
Time(hrs)
Source: Data points from Young et al. (1977). Red dotted line generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Blue solid line generated from EPA's implementation of Reitz et al. (1990) human PBPK model using
partition coefficient values from Sweeney et al. (2008) and a first-order metabolism rate constant (kLC = 1.2 hr"1) instead of saturable
metabolism.
Figure B-15. Predictions of human blood 1,4-dioxane concentration following calibration
of a first-order metabolism rate constant, kLc (1.2 hr"1), to the experimental
data.
A recalibration was performed using only the data from the exposure phase of the experiment,
such that the elimination data did not influence the initial metabolism and tissue distribution. The model
predictions from this exercise are shown in Figure B-16. These predictions are more similar to the
observations made during the exposure phase of the experiment; however, this is achieved at greatly
reduced elimination rate and hence under predictions of urinary HEAA (compare Figure B-ll and
Figure B-16).
B-22
-------
I
flS
I
o
u
.2
0
4
Younget al. il977i Observed Data
. Young et al. (1977) Empirical Model
-kLC i'0.li-fitted
Time (hrs)
Source: Data points from Young et al. (1977). Red dotted line generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Blue solid line generated from EPA's implementation of Reitz et al. (1990) human PBPK model using
partition coefficient values from Sweeney et al. (2008) and a first-order metabolism rate constant (kLC = 0.1 hr"1) instead of saturable
metabolism.
Figure B-16. Predictions of blood 1,4-dioxane concentration following calibration of a
first-order metabolism rate constant, kLc(0.1 hr"1), to only the exposure
phase of the experimental data.
Finally, the model was recalibrated by simultaneously fitting kLC and the slowly perfused
tissue:air partition (PSA) 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 a very low value of
10 (compared to experimentally determined values, see Table B-l). The resulting model predictions,
however, were closer to the observations (Figure B-17). 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 kLC) 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 and poor predictions of urinary elimination of HEAA.
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-23
-------
E
O
0
D Young et al. >1977i Observed Data
— — rounget al. (1377| Empirical Mndel
kLC t;o.2Si and PSA ilOi fitted
Time (hrs)
Source: Data points from Young et al. (1977). Red dotted line generated from EPA's acsIX implementation of Young et al. (1977)
empirical human model. Blue solid line generated from EPA's implementation of Reitz et al. (1990) human PBPK model, where the
first-order metabolism rate constant (kLc = 0.28 hr"1) and slowly perfused partition coefficient (PSA = 10) were simultaneously fit to
the data.
Figure B-17. Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a first-order metabolism rate constant ( RLC = 0.28 hr"1) and
slowly perfused tissuerair partition coefficient (PSA= 10) to the
experimental data.
B.5. 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
(Figure B-3. Figure B-4. Figure B-5. Figure B-7. and Figure B-8). 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. Q978a, 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 closer to the observed
data 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, they were
not modified to describe oral exposures due to a lack of adequate human or animal data for
parameterization. Additionally, the inhalation Young et al. (1977) model did not provide adequate fits to
the subchronic exposure plasma levels of 1,4-dioxane in rats using the data from the Kasai et al. (2008)
study, which is likely due to the absence of a model description for metabolic induction.
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
B-24
-------
plausible values for the model parameters. The recalibrated model predictions for blood 1,4-dioxane did
not adequately fit the experimental values using measured tissue:air partition coefficients from Leung and
Paustenbach (1990) or Sweeney et al. (2008) (Figure B-9 and Figure B-10). Use of a slowly perfused
tissue:air partition coefficient 4- to 7-fold lower than measured values produces exposure-phase
predictions that are much closer to observations, but does not replicate the elimination kinetics
(Figure B-16). Recalibration of the model with upper bounds on the tissue:air partition coefficients results
in predictions that are still 2- to 4-fold lower than empirical model prediction or observations
(Figure B-13 and Figure B-14). 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-17). Artificially low values for the other tissue:air
partition coefficients are not expected to improve the model fit, because blood 1,4-dioxane is less
sensitive to these parameters than it is to VmaxC and Km. This suggests that the model structure is
insufficient to capture the apparent species difference in the blood 1,4-dioxane Vd between rats and
humans. 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). Additionally, the models do not account
for induction in metabolism, which may be present in animals exposed repeatedly to 1,4-dioxane.
Therefore, some other modification(s) to the Reitz et al. (1990) model structure may be necessary.
Sweeney et al. (2008) PBPK model provided an overall improvement on previous models; however, the
Sweeney et al. (2008) inhalation model predictions of animal and human data were still problematic.
B.6. acsIX Model Code
The PBPK acsIX model code is made available electronically through EPA's Health and
Environmental Research Online (HERO) database. All model files may be downloaded in a zipped
workspace from HERO (U.S. EPA. 2013a).
B-25
-------
APPENDIX C. DETAILS OF BMD ANALYSIS FOR
ORAL RFD FOR 1,4-DIOXANE
C.1. 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)
0
0/31 a
240
20/31 b (65%)
530
27/33b (82%)
Females (mg/kg-day)
0
0/31 a
350
0/34
640
10/32b(31%)
Statistically 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 %2 goodness-of-fit test, several models in the software provided adequate fits
to the data for the incidence of cortical tubule degeneration in male and female rats (% p > 0.1)
(Table C-2). Comparing across models, a better fit is indicated by a lower AIC value (U.S. EPA. 2012b).
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/kg-day for this endpoint. The Weibull model provided the best fit to the data for
female rats (Table C-2, Figure C-2) 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
-------
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
Model
AIC
Scaled
Residual of
p-valuea Interest
BMD10
(mg/kg-day)
BMDL10
(mg/kg-day)
Male
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage (2 degree)d
Probit
Weibullb
Quantal-Linear
74.458
89.0147
75.6174
74.168
74.458
88.782
74.458
74.458
0.6514
0.0011
1
0.7532
0.6514
0.0011
0.6514
0.6514
0
-1.902
0
0
0
-1.784
0
0
28.80
88.48
20.85
51.41
28.80
87.10
28.80
28.80
22.27
65.84
8.59
38.53
22.27
66.32
22.27
22.27
Female
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage (2 degree)d
Probit
Weibullb
Quantal-Linear
41.9712
43.7405
41.7501
43.7495
48.1969
43.7405
41.75
52.3035
0.945
0.9996
0.9999
0.9997
0.1443
0.9997
0.9999
0.03
0.064
0
0
0
-1.693
0
0
-2.086
524.73
617.44
591.82
584.22
399.29
596.02
596.45
306.21
437.08
471.92
447.21
436.19
297.86
456.42
452.36
189.49
a p-Value from the x 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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
Data from NCI (1978).
C-2
-------
LogProbit Model with 0.95 Confidence Level
•
I
o
13
(0
0.8
0.6
0.4
0.2
100
200
300
400
500
dose
14:4902/01 2010
Data points obtained from NCI (1978).
Figure C-l. BMD Log-probit model of cortical tubule degeneration incidence data for
male rats exposed to 1,4-dioxane in drinking water for 2 years.
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File: C:\14DBMDS\lnp_nci_mrat_cortdeg_Lnp-BMR10-restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnp_nci_mrat_cortdeg_Lnp-BMR10-restrict.plt
Mon Feb 01 14:49:17 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 = 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
User has chosen the log transformed model
C-3
-------
Default Initial (and Specified) Parameter Values
background = 0
intercept = -5.14038
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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 NA
intercept -5.22131 0.172682 -5.55976 -4.88286
slope 1 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -35.8087 3
Fitted model -36.084 1 0.550629 2 0.7593
Reduced model -65.8437 1 60.07 2 <.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
= 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
-------
Weibull Model with 0.95 Confidence Level
•a
a>
•s
o
'•s
CO
0.5
0.4
0.3
0.2
0.1
Weibull
BMDL
BMD
100
200
14:2012/042009
Data points obtained from NCI (1978).
300
dose
400
500
600
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.
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/xpower)]
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
-------
Default Initial (and Specified) Parameter Values
Background = 0.015625
Slope = 1.55776e-010
Power = 3.33993
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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.$
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 NA
Slope 1.15454e-051 1.#QNAN 1.#QNAN 1.#QNAN
Power 18 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -19.8748 3
Fitted model -19.875 1 0.000487728 2 0.9998
Reduced model -32.1871 1 24.6247 2 <.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
ChiA2 = 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 = 452.359
C-6
-------
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 (%2
p > 0.1) and AIC values, are reported for each endpoint (U.S. EPA. 2012b). 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-1. The
data and BMD modeling results are presented separately for each dataset as follows:
• Hepatic adenomas and carcinomas in female F344 rats (Table D-2 and Table D-3:
Figure D-l)
• Hepatic adenomas and carcinomas in male F344 rats (Table D-4 and Table D-5; Figure D-2
and Figure 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; Figure D-6 and
Figure D-7)
o Peritoneal mesotheliomas in male F344 rats (Table D-10; Figure D-8 and Figure D-9)
• Hepatic adenomas and carcinomas in female BDF1 mice (Table D-ll. Table D-l2. and
Table D-13; Figure D-10. Figure D-ll. Figure D-l2. and Figure D-l3)
• Hepatic adenomas and carcinomas in male BDF1 mice (Table D-l4 and Table D-l5;
Figure D-14 and Figure D-15)
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-l6)
D-l
-------
• 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-17)
o BMDS dose-response modeling results for incidence of hepatocellular carcinoma in
male and female Sherman rats (combined) (Kociba etal.. 1974) exposed to
1,4-dioxane in drinking water for 2 years (Table D-18; Figure D-16 and Figure 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; Figure D-19 and Figure 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; Figure D-21 and Figure 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; Figure D-23 and Figure D-24)
• Incidence of hepatocellular adenoma or carcinoma in male and female B6C3Fi 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 (NCL 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; Figure D-26 and Figure D-27).
D.1. General Issues and Approaches to BMDS Modeling
D.1.1. 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 may
develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a: McConnell et al..
1986). 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
-------
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 BMD Technical
Guidance Document (U.S. EPA. 2012b) were applied as follows:
• p-value for goodness-of-fit > 0.10
• AIC smaller than other acceptable models
• %2 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 in Table D-l.
D-3
-------
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.
Table D-l Recommended models for rodents exposed to 1,4-dioxane in drinking water
(Kano et al., 2009).
selection BM°a BMDI-a BMDHEDa BMDLHEDa
Endpoint criterion Model Type AIC p-value mg/kg-day mg/kg-day mg/kg-day mg/kg-day
Female F344 Rat
TuSrars L°vvestAIC 5 91.5898 0.4516 79.83 58.09 19.84 14.43
Mammary
Gland LowestAIC Log-Logistic 194.151 0.8874 161.01 81.91 40.01 20.35
Tumors
Cavity LowestAIC Multistage 42.6063 0.9966 381.65 282.61 94.84 70.23
Tumors <3 de9ree)
Male F344 Rat
Tumors LowestAIC Probit 147.787 0.9867 62.20 51.12 17.43 14.33
Peritoneal
Meso- LowestAIC Probit 138.869 0.9148 93.06 76.32 26.09 21.39
thelioma
Nasal .. ...
Cavity LowestAIC Multistage 24.747 0.9989 328.11 245.63 91.97 68.85
— (3 degree)
Tumors v a '
Female BDF1 Mouse
Hepatic LowestAIC Log-Logistic 176.214 0.1421 5.54 3.66 0.83 0.55
Tumors BMR 50% Log-Logistic 176.214 0.1421 49.88b 32.93b 7.51b 4.951
Male BDF1 Mouse
b
Tumors LowestAIC Log-Logistic 248.839 0.3461 34.78 16.60 5.63 2.68
"Values for BMR 10% unless otherwise noted.
bBMR 50%.
Data from Kano et al., (2009).
D-4
-------
D.2. Female F344 Rats: Hepatic Carcinomas and Adenomas
The incidence data for hepatic carcinomas and adenomas in female F344 rats (Kano et 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).
Dose (mg/kg-day)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
0
3
0
3
47
50
18
1
0
1
49
50
83
6
0
6
44
50
429
48
10
48
2
50
Source: Kano et al. (2009).
Note that the incidence of rats with adenomas, with carcinomas, and with either adenomas or
carcinomas, are monotone non-decreasing functions of dose except for 3 female rats in the control group.
These data therefore appear to be appropriate for dose-response modeling using BMDS.
The results of the BMDS modeling for the entire suite of models are presented in Table D-3.
D-5
-------
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
Gamma
Logistic
Log-Logistic
Log-Probitb
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)0
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
93.1067
91.7017
93.102
93.0762
114.094
91.5898
93.2682
91.8786
93.2255
114.094
4,458.37
p-value
0.3024
0.4459
0.3028
0.3074
0.0001
0.4516
0.2747
0.3839
0.2825
0.0001
NCd
BMDio
mg/kg-day
89.46
93.02
88.34
87.57
25.58
79.83
92.81
85.46
92.67
25.58
NCd
BMDLio
mg/kg-day
62.09
71.60
65.52
66.19
19.92
58.09
59.31
67.84
59.89
19.92
NCd
x2a
0.027
0.077
0.016
0.001
-1.827
-0.408
0.077
-0.116
0.088
-1.827
0
BMDlOHED
mg/kg-day
22.23
23.12
21.95
21.76
6.36
19.84
23.06
21.24
23.03
6.36
0
BMDLio HED
mg/kg-day
15.43
17.79
16.28
16.45
4.95
14.43
14.74
16.86
14.88
4.95
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.
Data from Kano et al. (2009).
D-6
-------
Multistage Cancer Model with 0.95 Confidence Level
T3
"O
O
2
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
100
150
200 250
dose
300
350
400
450
07:20 10/262009
Data points obtained from 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_kano2009_frat_hepato_adcar_Msc-BMR10-2poly.plt
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/xl-beta2*dose/x2 ) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
D-7
-------
Background = 0.0281572
Beta(l) = 0
Beta(2) = 1.73306e-005
Asymptotic Correlation Matrix of Parameter Estimates (*** The model parameter(s)
-Beta(l)have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Background Beta(2)
Background 1 -0.2
Beta(2) -0.2 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0362773 * * *
Beta(l) 0 * * *
Beta(2) 1.65328e-005 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -42.9938 4
Fitted model -43.7949 2 1.60218 2 0.4488
Reduced model -120.43 1 154.873 3 <.0001
AIC: 91.5898
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0363 1.814 3.000 50 0.897
18.0000 0.0414 2.071 1.000 50 -0.760
83.0000 0.1400 7.001 6.000 50 -0.408
429.0000 0.9540 47.701 48.000 50 0.202
ChiA2 = 1.59 d.f. = 2 P-value = 0.4516
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 79.8299
BMDL = 58.085
BMDU = 94.0205
Taken together, (58.085 , 94.0205) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00172161
D-8
-------
D.3. Male F344 Rats: Hepatic Carcinomas and Adenomas
The data for hepatic adenomas and carcinomas in male F344 rats (Kano et al., 2009) are shown in
Table D-4.
Table D-4 Data for hepatic adenomas and carcinomas in male F344 rats (Kano et al.,
2009).
Dose (mg/kg-day)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
0
3
0
3
47
50
11
4
0
4
46
50
55
7
0
7
43
50
274
32
14
39
11
50
Source: Kano et al. (2009).
Note that the incidence of rats with hepatic adenomas, carcinomas, and with either adenomas or
carcinomas are monotone non-decreasing functions of dose. These data therefore appear to be appropriate
for dose-response modeling using BMDS.
The results of the BMDS modeling for the entire suite of models tested using the data for hepatic
adenomas and carcinomas for male F344 rats are presented in Table D-5.
D-9
-------
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
Gamma
Logistic
Log-Logistic
Log-Probitb
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
149.884
147.813
149.886
149.913
152.836
149.814
149.772
147.787
149.856
152.836
4441.71
p-value
0.7257
0.9749
0.7235
0.6972
0.0978
0.8161
0.9171
0.9867
0.7576
0.0978
NCd
BMDio
mg/kg-day
62.41
68.74
62.10
61.70
23.82
61.68
63.62
62.20
62.63
23.82
NCd
BMDLio
mg/kg-day
30.79
55.39
34.61
37.49
18.34
28.26
27.49
51.12
30.11
18.34
NCd
x2a
-0.03
0.097
-0.021
-0.003
-0.186
-0.063
-0.024
-0.05
-0.039
-0.186
0
BMDlOHED
mg/kg-day
17.49
19.27
17.41
17.29
6.68
17.29
17.83
17.43
17.56
6.68
0
BMDLio HED
mg/kg-day
8.63
15.53
9.70
10.51
5.14
7.92
7.71
14.33
8.44
5.14
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.
Data from Kano et al. (2009).
D-10
-------
Probit Model with 0.95 Confidence Level
0.8
0.6
.2 0.4
o
0.2
250
dose
07:32 10/26 2009
Data points obtained from Kano et al. (2009).
Figure D-2. Probit BMD model for the combined incidence of hepatic adenomas and
carcinomas 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_hepato_adcar_Prb-BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-BMR10.plt
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
background = 0 Specified
intercept = -1.51718
slope = 0.00831843
Asymptotic Correlation Matrix of Parameter Estimates
D-ll
-------
(*** The model parameter (s) -background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
intercept slope
intercept 1 -0.69
slope -0.69 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
intercept 1.53138 0.160195 -1.84535 -1.2174
slope 0.00840347 0.000976752 0.00648907 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
= 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-12
-------
Multistage Cancer Model with 0.95 Confidence Level
o
'o
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
BMDL
BMD
0
50
100
150
200
250
07:32 10/26 2009
Data points obtained from Kano et al. (2009).
dose
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_kano2009_mrat_hepato_adcar_Msc-BMR10-3poly.plt
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/xl-beta2*dose/x2-beta3*dose/x3) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-13
-------
Default Initial Parameter Values
Background = 0.0623822
Beta(l) = 0.00142752
Beta(2) = 0
Beta(3) = 5.14597e-008
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(2)have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
Background Beta(l) Beta(3)
Background 1 -0.67 0.58
Beta(l) -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(l) 0.001449 * * *
Beta(2) 0 * * *
Beta(3) 5.11829e-008 * * *
Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -71.8804 4
Fitted model -71.8858 3 0.0107754 1 0.9173
Reduced model -115.644 1 87.528 3 <.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
ChiA2 = 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-14
-------
D.4. F344 Rats: Tumors at Other Sites
The data for tumors at sites other than the liver in male and female F344 rats (Kano et al., 2009)
are shown in Table D-6. Note that the incidence of rats with these endpoints are monotone non-decreasing
functions (except female peritoneal mesotheliomas). These data therefore appear to be appropriate for
dose-response modeling using BMDS.
Table D-6 Data for significant tumors at other sites in male and female F344 rats (Kano et
al., 2009).
Dose (mg/kg-day)
Female
Tumor site and type
Nasal cavity squamous cell carcinoma
Peritoneal mesothelioma
Mammary gland adenoma
Total number per group
0
0
1
6
50
18
0
0
7
50
83
0
0
10
50
429
7
0
16
50
0
0
2
0
50
Male
11
0
2
1
50
55
0
5
2
50
274
3
28
2
50
Source: Kano et al., (2009).
The results of the BMDS modeling for the entire suite of models are presented in Table D-7
through Table D-10 for tumors in the nasal cavity, mammary gland, and peritoneal cavity.
D-15
-------
Table D-7 BMDS dose-response modeling results for the incidence of nasal cavity tumors
in female F344 ratsa (Kano et al., 2009).
Model
Gamma
Logistic
Log-Logistic
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)d
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
44.4964
44.4963
44.4963
44.4963
45.6604
43.0753
42.6063
44.4963
44.4963
45.6604
46.4963
p-value
1
1
1
1
0.6184
0.9607
0.9966
1
1
0.6184
0.9997
BMDio
mg/kg-day
403.82
421.54
413.69
400.06
375.81
366.07
381.65
414.11
414.86
375.81
413.96
BMDLio
mg/kg-day
269.03
351.74
268.85
260.38
213.84
274.63
282.61
333.31
273.73
213.84
372.57
x2b
0
0
0
0
0.595
0.109
0.021
0
0
0.595
1.64x10'8
BMDlOHED
mg/kg-day
100.35
104.75
102.80
99.42
93.39
90.97
94.84
102.91
103.09
93.39
102.87
BMDLio HED
mg/kg-day
66.85
87.41
66.81
64.71
53.14
68.24
70.23
82.83
68.02
53.14
92.58
"Nasal 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.
Data from Kano et al. (2009).
D-16
-------
Multistage Cancer Model with 0.95 Confidence Level
I
a
o
t3
CO
0.3
0.25
0.2
0.15
0.1
0.05
Multistage'Cancer '
Linear extrapolation
BMDL
BMD
100 150 200 250 300 350
dose
400
0 50
07:28 10/26 2009
Data points obtained from Kano et al. (2009).
Figure D-4. Multistage BMD model (3 degree) for nasal cavity tumors in female
F344 rats.
450
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_kano2009_frat_nasal_car_Msc-BMR10-3poly.plt
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/xl-beta2*dose/x2-beta3*dose/x3) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
D-17
-------
Beta(3) = 1.91485e-009
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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(l) 0 * * *
Beta(2) 0 * * *
Beta(3) 1.89531e-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 -20.2482 4
Fitted model -20.3031 1 0.109908 3 0.9906
Reduced model -30.3429 1 20.1894 3 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
ChiA2 = 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 = 381.651
BMDL = 282.609
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-18
-------
Table D-8 BMDS dose-response modeling results for the incidence of nasal cavity tumors
in male F344 ratsa (Kano et al., 2009).
Model
Gamma
Logistic
Log-Logistic
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)d
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
26.6968
26.6968
26.6968
26.6968
26.0279
24.9506
24.747
26.6968
26.6968
26.0279
28.6968
p-value
1
1
1
1
0.8621
0.988
0.9989
1
1
0.8621
0.9994
BMDio
mg/kg-day
299.29
281.06
288.31
303.06
582.49
365.19
328.11
287.96
288.00
582.49
290.52
BMDLio
mg/kg-day
244.10
261.29
245.29
238.86
256.43
242.30
245.63
257.01
246.36
256.43
261.47
x2b
0
0
0
0
0.384
0.073
0.015
0
0
0.384
6.25x10'5
BMDlOHED
mg/kg-day
83.89
78.78
80.81
84.94
163.28
102.37
91.97
80.72
80.73
163.28
81.44
BMDLio HED
mg/kg-day
68.42
73.24
68.75
66.95
71.88
67.92
68.85
72.04
69.06
71.88
73.29
"Nasal 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.
Data from Kano et al. (2009).
D-19
-------
Multistage Cancer Model with 0.95 Confidence Level
o
i
d
o
••§
to
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMID
50
100
07:34 10/26 2009
150
dose
200
250
300
Data points obtained from 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_kano2009_mrat_nasal_car_Msc-BMR10-3poly.plt
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/xl-beta2*dose/x2-beta3*dose/x3) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
D-20
-------
Beta(3) = 3.01594e-009
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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(l) 0 * * *
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
ChiA2 = 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-21
-------
Table D-9 BMDS dose-response modeling results for the incidence of mammary gland
adenomas in female F344 rats (Kano et al., 2009).
Model
Gamma
Logistic
Log-Logistic13
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
194.222
194.475
194.151
195.028
194.222
194.222
194.222
194.441
194.222
194.222
197.916
p-value
0.8559
0.7526
0.8874
0.5659
0.8559
0.8559
0.8559
0.7656
0.8559
0.8559
NCd
BMDio
mg/kg-day
176.66
230.35
161.01
270.74
176.66
176.66
176.66
223.04
176.65
176.65
94.06
BMDLio
mg/kg-day
99.13
159.73
81.91
174.66
99.13
99.13
99.13
151.60
99.13
99.13
14.02
x2a
0.465
0.612
0.406
-0.075
0.465
0.465
0.465
0.596
0.465
0.465
3.49x10'5
BMDlOHED
mg/kg-day
43.90
57.24
40.01
67.28
43.90
43.90
43.90
55.43
43.90
43.90
23.37
BMDLio HED
mg/kg-day
24.63
39.69
20.35
43.41
24.63
24.63
24.63
37.67
24.63
24.63
3.48
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
Data from Kano et al. (2009).
D-22
-------
Log-Logistic Model with 0.95 Confidence Level
o
OS
0.5
0.4
0.3
0.2
0.1
400
45(
11:31 02/01 2010
Data points obtained from Kano et al. (2009).
Figure D-6. Log-Logistic 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_kano2009_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-siope*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-23
-------
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
background intercept
background 1 -0.53
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 Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -94.958 4
Fitted model -95.0757 2 0.235347 2 0.889
Reduced model -98.6785 1 7.4409 3 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
= 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-24
-------
Multistage Cancer Model with 0.9b Confidence Level
T3
3
O
I
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
0 50 100
07:27 10/262009
Data points obtained from Kano et al. (2009).
150
200 250
dose
300
350
400
450
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_maitim_ad_Msc-BMR10-lpoly. (d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_maitim_ad_Msc-BMR10-lpoly.plt
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/xl)]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.136033
D-25
-------
Beta(l) = 0.000570906
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.58
Beta(l) -0.58 1
Parameter Estimates
95.0% Wald Confidence Interval Variable Estimate Std. Err. Lower Conf. Limit Upper
Conf. Limit
Background .133161 * * *
Beta(l) 0.000596394 * * *
Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -94.958 4
Fitted model -95.111 2 0.305898 2 0.8582
Reduced model -98.6785 1 7.4409 3 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-26
-------
Table D-10 BMDS dose-response modeling results for the incidence of peritoneal
mesotheliomas in male F344 rats (Kano et al., 2009).
Model
Gamma
Logistic
Log-Logistic
Log-Probitb
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
140.701
139.016
140.699
140.69
140.826
140.747
140.747
138.869
140.709
140.826
2992
p-value
0.9189
0.8484
0.9242
0.9852
0.3617
0.8135
0.8135
0.9148
0.8915
0.3617
NCd
BMDio
mg/kg-day
73.52
103.52
72.56
70.29
41.04
77.73
77.73
93.06
74.77
41.04
NCd
BMDLio
mg/kg-day
35.62
84.35
36.37
52.59
30.51
35.43
35.43
76.32
35.59
30.51
NCd
x2a
0.018
0.446
0.014
0.001
-1.066
0.067
0.067
0.315
0.027
-1.066
0
BMDio HED
mg/kg-day
20.61
29.02
20.34
19.70
11.50
21.79
21.79
26.09
20.96
11.50
0
BMDLio HED
mg/kg-day
9.98
23.65
10.19
14.74
8.55
9.93
9.93
21.39
9.97
8.55
0
"Maximum 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.
Data from Kano et al. (2009).
D-27
-------
Probit Model with 0.95 Confidence Level
O
I
C
O
'•§
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 -
100
150
200
250
dose
07:41 10/26 2009
Data points obtained from 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_kano2009_mrat_peri_meso_Prb-BMR10.plt
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
background = 0 Specified
intercept = -1.73485
slope = 0.00692801
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
D-28
-------
intercept slope
intercept 1 -0.75
slope -0.75 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
intercept -1.73734 0.18348 -2.09695 -1.37772
slope 0.00691646 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
ChiA2 = 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-29
-------
Multistage Cancer Model with 0.95 Confidence Level
T3
-------
Default Initial Parameter Values
Background = 0.0358706
Beta(l) = 0.000816174
Beta(2) = 7.47062e-006
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l) Beta(2)
Background 1 -0.67 0.59
Beta(l) -0.67 1 -0.98
Beta(2) 0.59 -0.98 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0366063 * * *
Beta(l) 0.000757836 * * *
Beta(2) 7.6893e-006 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -67.3451 4
Fitted model -67.3733 3 0.056567 1 0.812
Reduced model -95.7782 1 56.8663 3 <.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
ChiA2 = 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 = 35.4296
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-31
-------
D.5. Female BDF1 Mice: Hepatic Carcinomas and Adenomas
Data for female BDF1 mouse hepatic carcinomas and adenomas are shown in Table D-ll. 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 BDF1 mice (Kano et al.,
2009).
Dose (mg/kg-day)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
0
5
0
5
45
50
66
31
6
35
15
50
278
20
30
41
9
50
964
3
45
46
4
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 BDF1 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 BDF1 mice. The
log-logistic model provided the best-fit to the data as indicated by the AIC and/"-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). see Figure D-10. Thus,
the log-logistic model was run for BMRs of 30 and 50%. The output from these models is shown in
Figure D-ll and Figure 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 (964 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-32
-------
Table D-12 BMDS dose-response modeling results for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice (Kano et al., 2009).
Model
Gamma
Logistic
Log-Logistic13
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
203.331
214.951
176.214
198.354
203.331
203.331
203.331
217.671
203.331
203.331
7300.48
p-value
0
0
0.1421
0
0
0
0
0
0
0
NCd
BMDio
mg/kg-day
26.43
58.05
5.54
26.37
26.43
26.43
26.43
69.89
26.43
26.43
NCd
BMDLio
mg/kg-day
19.50
44.44
3.66
19.57
19.50
19.50
19.50
56.22
19.50
19.50
NCd
x2a
-2.654
3.201
-0.121
-1.166
-2.654
-2.654
-2.654
3.114
-2.654
-2.654
0
BMDio HED
mg/kg-day
3.98
8.74
0.83
3.97
3.98
3.98
3.98
10.5
3.98
3.98
0
BMDLio HED
mg/kg-day
2.94
6.69
0.55
2.95
2.94
2.94
2.94
8.46
2.94
2.94
0
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
bBest-fitting model, lowest AIC value.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
Data from Kano et al. (2009).
Table D-13 BMDS Log-Logistic dose-response modeling results using BMRs of 10, 30, and
50% for the combined incidence of hepatic adenomas and carcinomas in female
BDF1 mice (Kano et al., 2009).
BMR
10%
30%
50%
AIC
176.214
176.214
176.214
p-value
0.1421
0.1421
0.1421
BMD
mg/kg-day
5.54
21.38
49.88
BMDL
mg/kg-day
3.66
14.11
32.93
x2a
-0.121
-0.121
0
BMDHED
mg/kg-day
0.83
3.22
7.51
BMDI-HED
mg/kg-day
0.55
2.12
4.95
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
Data from Kano et al. (2009).
D-33
-------
Log-Logistic Model with 0.95 Confidence Level
•
I
o
13
(0
0.8
0.6
0.4
0.2
Log-Logistic
BMDLJBMD
200
400
600
800
1000
dose
11:2605/122010
Data points obtained from Kano et al. (2009).
Figure D-10. Log-Logistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in female BDF1 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_kano2009_fmouse_hepato_adcar_Lnl-BMR10-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR10-Restrict.p
It
Wed May 12 11:26:35 2010
BMDS Model Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*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
D-34
-------
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.1
intercept = -4.33618
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
background intercept
background 1 -0.32
intercept -0.32 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0.105265 * * *
intercept -3.90961 * * *
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 -86.107 2 3.6029 2 0.1651
Reduced model -131.248 1 93.8853 3 <.0001
AIC: 176.214
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.1053 5.263 5.000 50 -0.121
66.0000 0.6149 30.743 35.000 50 1.237
278.0000 0.8639 43.194 41.000 50 -0.905
964.0000 0.9560 47.799 46.000 50 -1.240
ChiA2 = 3.90 d.f. = 2 P-value = 0.1421
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 5.54218
BMDL = 3.65848
D-35
-------
Log-Logistic Model with 0.95 Confidence Level
•
I
o
13
(0
0.8
0.6
0.4
0.2
Log-Logistic
EJyiDLpBMD
200
400
600
800
1000
dose
11:2605/122010
Data points obtained from Kano et al. (2009).
Figure D-ll. Log-Logistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in female BDF1 mice with a BMR of 30%.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR30-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR30-Restrict.p
It
Wed May 12 11:26:36 2010
BMDS Model Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*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
D-36
-------
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.1
intercept = -4.33618
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
background intercept
background 1 -0.32
intercept -0.32 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0.105265 * * *
intercept -3.90961 * * *
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 -86.107 2 3.6029 2 0.1651
Reduced model -131.248 1 93.8853 3 <.0001
AIC: 176.214
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.1053 5.263 5.000 50 -0.121
66.0000 0.6149 30.743 35.000 50 1.237
278.0000 0.8639 43.194 41.000 50 -0.905
964.0000 0.9560 47.799 46.000 50 -1.240
ChiA2 = 3.90 d.f. = 2 P-value = 0.1421
Benchmark Dose Computation
Specified effect =0.3
Risk Type = Extra risk
Confidence level = 0.95
BMD = 21.377
BMDL = 14.1113
D-37
-------
Log-Logistic Model with 0.95 Confidence Level
•
I
o
13
(0
0.8
0.6
0.4
0.2
Log-Logistic
BMDL
BMD
200
400
600
800
1000
dose
11:2605/122010
Data points obtained from Kano et al. (2009).
Figure D-12. Log-Logistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in female BDF1 mice with a BMR of 50%.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR50-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR50-Restrict.p
It
Wed May 12 11:26:36 2010
BMDS Model Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*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
D-38
-------
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.1
intercept = -4.33618
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
background intercept
background 1 -0.32
intercept -0.32 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0.105265 * * *
intercept -3.90961 * * *
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 -86.107 2 3.6029 2 0.1651
Reduced model -131.248 1 93.8853 3 <.0001
AIC: 176.214
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.1053 5.263 5.000 50 -0.121
66.0000 0.6149 30.743 35.000 50 1.237
278.0000 0.8639 43.194 41.000 50 -0.905
964.0000 0.9560 47.799 46.000 50 -1.240
ChiA2 = 3.90 d.f. = 2 P-value = 0.1421
Benchmark Dose Computation
Specified effect = 0.5
Risk Type = Extra risk
Confidence level = 0.95
BMD = 49.8797
BMDL = 32.9263
D-39
-------
Multistage Cancer Model with 0.95 Confidence Level
•
I
o
13
(0
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
nMDL||BMD
200
400
600
800
1000
dose
11:2605/122010
Data points obtained from Kano et al. (2009).
Figure D-13. Multistage BMD model (1 degree) for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_fmouse_hepato_adcar_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_fmouse_hepato_adcar_Msc-BMR10-lpoly.plt
Wed May 12 11:26:31 2010
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]
The parameter betas are restricted to be positive
Dependent variable = 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
D-40
-------
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 = 0.51713
Beta(l) = 0.00201669
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.65
Beta(l) -0.65 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.265826 * * *
Beta(l) 0.00398627 * * *
* - 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 -99.6653 2 30.7195 2 2.1346928e-007
Reduced model -131.248 1 93.8853 3 <.0001
AIC: 203.331
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.2658 13.291 5.000 50 -2.654
66.0000 0.4357 21.783 35.000 50 3.770
278.0000 0.7576 37.880 41.000 50 1.030
964.0000 0.9843 49.213 46.000 50 -3.651
ChiA2 = 35.65 d.f. = 2 P-value = 0.0000
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 26.4309
BMDL = 19.5045
BMDU = 37.5583
Taken together, (19.5045, 37.5583) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00512702
D-41
-------
D.6. Male BDF1 Mice: Hepatic Carcinomas and Adenomas
Data for hepatic carcinomas and adenomas in male BDF1 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 BDF1 mice (Kano et al.,
2009).
Dose (mg/kg-day)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
0
9
15
23
27
50
49
17
20
31
19
50
191
23
23
37
13
50
677
11
36
40
10
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 BDF 1 mice are presented in Table D-15.
D-42
-------
Table D-15 BMDS dose-response modeling results for the combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice (Kano et al., 2009).
Model
Gamma
Logistic
Log-Logistic13
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
250.551
251.187
248.839
252.244
250.551
250.551
250.551
251.326
250.551
250.551
250.747
p-value
0.1527
0.112
0.3461
0.0655
0.1527
0.1527
0.1527
0.1048
0.1527
0.1527
NCd
BMDio
mg/kg-day
70.99
91.89
34.78
133.53
70.99
70.99
70.99
97.01
70.99
70.99
11.60
BMDLio
mg/kg-day
44.00
61.98
16.60
78.18
44.00
44.00
44.00
67.36
44.00
44.00
1.63
x2a
0.605
0.529
0.656
0.016
0.605
0.605
0.605
0.518
0.605
0.605
-1.25x10'5
BMDlOHED
mg/kg-day
11.48
14.86
5.63
21.60
11.48
11.48
11.48
15.69
11.48
11.48
1.88
BMDLio HED
mg/kg-day
7.12
10.02
2.68
12.64
7.12
7.12
7.12
10.90
7.12
7.12
0.26
"Maximum 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.
Data from Kano et al. (2009).
D-43
-------
Log-Logistic Model with 0.95 Confidence Level
s
d
o
••§
OS
0.9
0.8
0.7
0.6
0.5
0.4
0.3
Log-Logistic
3MDL BMD
0
100 200 300 400
dose
500
600
700
07:30 10/262009
Data points obtained from Kano et al. (2009).
Figure D-14. Log-Logistic BMD model for the combined incidence of hepatic adenomas
and carcinomas in male BDF1 mice.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_mmouse_hepato_adcar_Lnl-BMR10-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_mmouse_hepato_adcar_Lnl-BMR10-Restrict.p
It
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 =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 = 0.46
D-44
-------
intercept = -5.58909
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix )
background intercept
background 1 -0.69
intercept -0.69 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0.507468 * * *
intercept -5.74623 * * *
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 -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-45
-------
Multistage uancer Model witn u.aa uontidence Level
I
d
o
t3
cc
0.9
0.8
0.7
0.6
0.5
0.4
0.3
Multistage Cancer
Linear extrapolation
BMDL
BMD
100
200
300
400
500
600
700
07:30 10/262009
Data points obtained from Kano et al. (2009).
dose
Figure D-15. Multistage BMD model (1 degree) for the combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mmouse_hepato_adcar_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mmouse_hepato_adcar_Msc-BMR10-lpoly.plt
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/xl)]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.573756
D-46
-------
Beta(l) = 0.00123152
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.58
Beta(l) -0.58 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.545889 * * *
Beta(l) 0.00148414 * * *
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 -123.275 2 3.80413 2 0.1493
Reduced model -128.859 1 14.9718 3 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
-------
D.7. BMD Modeling Results from Additional Chronic Bioassays
Data and BMDS modeling results for the additional chronic bioassays (NCI. 1978; Kociba et al.,
1974) were evaluated for comparison with the Kano et al. (2009) study. These results are presented in the
following sections.
The BMDS dose-response modeling estimates and HEDs that resulted are presented in 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
Kociba et al., (1974) Male and Female (combined) Sherman
Hepatic
Tumors3
Nasal
Cavity
Tumors'3
NCI, (1978)
Hepatic
Turners0
Nasal
Cavity
Tumorsb
NCI, (1978)
Nasal
Cavity
Tumorsb
NCI, (1978)
Hepatic
Tumorsd
NCI, (1978)
Hepatic
Tumorsd
Lowest AIC Probit 84.3126 0.606
Lowest AIC Multistage 26.4156 0.9999
(3 degree)
Female Osborne-Mendel Rats
Lowest AIC Log-Logistic 84.2821 0.7333
Lowest AIC Log-Logistic 84.2235 0.2486
Male Osborne-Mendel Rats
Lowest AIC Log-Logistic 92.7669 0.7809
Female B6C3F! Mice
~' « -" <
model v a '
Male B6C3F1 Mice
Lowest AIC Gamma 177.539 0.7571
BMDio BMDLio BMDioHED
mg/kg-day mg/kg-day mg/kg-day
Rats
1113.94 920.62 290.78
1717.16 1306.29 448.24
111.46 72.41 28.75
155.32 100.08 40.07
56.26 37.26 16.10
160.68 67.76 23.12
601.69 243.92 87.98
BMDLio HED
mg/kg-day
240.31
340.99
18.68
25.82
10.66
9.75
35.67
Incidence of hepatocellular carcinoma.
""Incidence of nasal squamous cell carcinoma.
Incidence of hepatocellular adenoma.
Incidence of hepatocellular adenoma or carcinoma.
Data from Kociba et al., (1974) and NCI, (1978).
D-48
-------
D.7.1. Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma
(Kocibaetal., 1974)
The incidence data for hepatocellular carcinoma and nasal squamous cell carcinoma are presented
in Table D-17. The predicted BMDioHED and BMDLio HED values are also presented in Table D-18 and
Table D-19 for hepatocellular carcinomas and nasal squamous cell carcinomas, 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)
0
14
121
1,307
Incidence of hepatocellular
carcinoma3
1/1 06b
0/110
1/106
10/66d
Incidence of nasal squamous cell
carcinoma3
0/1 06C
0/110
0/106
3/66d
aRats surviving until 12 months on study.
bp < 0.001; positive dose-related trend (Cochran-Armitage test).
°p < 0.01; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: Reprinted with permission of Elsevier; Kociba et al. (1974).
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
Gamma
Logistic
Log-Logistic
Log-Probitb
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
86.2403
84.3292
86.2422
84.4246
85.1187
86.2868
86.2868
84.3126
86.2443
85.1187
1503.63
p-value
0.3105
0.6086
0.3103
0.5977
0.3838
0.3109
0.3109
0.606
0.3104
0.3838
NCd
BMDio
mg/kg-day
985.13
1148.65
985.62
1036.97
940.12
1041.72
1041.72
1113.94
998.33
940.12
NCd
BMDLio
mg/kg-day
628.48
980.95
611.14
760.29
583.58
628.56
628.56
920.62
629.93
583.58
NCd
x2a
-0.005
-0.004
-0.005
-0.011
0.279
-0.006
-0.006
-0.005
-0.005
0.279
0
BMDlOHED
mg/kg-day
257.15
299.84
257.28
270.68
245.40
271.92
271.92
290.78
260.60
245.40
0
BMDLio HED
mg/kg-day
164.05
256.06
159.53
198.46
152.33
164.07
164.08
240.31
164.43
152.33
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.
Data from Kociba et al. (1974).
D-49
-------
Probit Model with 0.95 Confidence Level
o
'o
0.25
0.2
0.15
0.1
0.05
Probit
BMDL
BMD
0
200
400
11:54 10/27 2009
Data points obtained from Kociba et al. (1974).
600
dose
800
1000
1200
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.plt
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(.
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 =250
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
Asymptotic Correlation Matrix of Parameter Estimates
is the cumulative normal
D-50
-------
(*** The model parameter(s) -background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
intercept slope
intercept 1 -0.82
slope -0.82 1
Parameter Estimates
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 Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -39.3891 4
Fitted model -40.1563 2 1.53445 2 0.4643
Reduced model -53.5257 1 28.2732 3 <.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 = 1,113.94
BMDL = 920.616
D-51
-------
Multistage Cancer Model with 0.95 Confidence Level
T3
-------
Default Initial Parameter Values
Background = 0.000925988
Beta(l) = 0.000124518
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.44
Beta(l) -0.44 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0038683 * * *
Beta(l) 0.000112071 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -39.3891 4
Fitted model -40.5594 2 2.34056 2 0.3103
Reduced model -53.5257 1 28.2732 3 <.0001
AIC: 85.1187
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
ChiA2 = 1.92 d.f. = 2 P-value = 0.3838
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 940.124
BMDL = 583.576
BMDU = 1,685 .88
Taken together, (583.576, 1685.88) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000171357
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.
Model
Gamma
Logistic
Log-Logistic
Log-Probitb
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)0
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
28.4078
28.4078
28.4078
28.4078
27.3521
26.4929
26.4156
28.4078
28.4078
27.3521
30.4078
p-value
1
1
1
1
0.9163
0.9977
0.9999
1
1
0.9163
0.9997
BMDio
mg/kg-day
1,572.09
1,363.46
1,464.77
1,644.38
3,464.76
1,980.96
1,717.16
1,419.14
1,461.48
3,464.76
1,465.77
BMDLio
mg/kg-day
1,305.86
1,306.67
1,306.06
1,305.49
1,525.36
1,314.37
1,306.29
1,306.44
1,306.11
1,525.35
1319.19
x2a
0
0
0
0
0.272
0.025
0.002
0
0
0.272
5.53x10'7
BMDio HED
mg/kg-day
410.37
355.91
382.35
429.24
904.42
517.10
448.24
370.44
381.50
904.42
382.62
BMDLio HED
mg/kg-day
340.87
341.09
340.93
340.78
398.17
343.10
340.99
341.03
340.94
398.17
344.35
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
°Best-fitting model.
Data from Kociba et al. (1974).
D-54
-------
Multistage Cancer Model with 0.95 Confidence Level
"0
I
d
O
'•o
cc
0.14
0.12
0.1
0.08
0.06
0.04
0.02
Multistage Cancer
Linear extrapolation
BMDL
BM:
0 200
06:25 10/272009
Data points obtained from Kociba et al. (1974).
400
600
800 1000
dose
1200
1400
1600
1800
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.plt
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/xl-beta2*dose/x2-beta3*dose/x3) ]
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 =250
Relative Function Convergence has been set to: le-008
D-55
-------
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
Beta(3) = 2.08414e-011
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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(l) 0 * * *
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
ChiA2 = 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 = 1,717.16
BMDL = 1,306.29
BMDU = 8,354 .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-56
-------
D.7.2. Nasal Cavity Squamous Cell Carcinoma and Liver
Hepatocellular Adenoma in Osborne-Mendel Rats (NCI. 1978)
The incidence data for hepatocellular adenoma (female rats) and nasal squamous cell carcinoma
(male and female rats) are presented in Table D-20. The log-logistic model adequately fit both the male
and female rat nasal squamous cell carcinoma data, as well as female hepatocellular adenoma incidence
data. For all endpoints and genders evaluated in this section, compared to the multistage models, the
log-logistic model had a higher/)-value, as well as both a lower AIC and lower BMDL. The results of the
BMDS modeling for the entire suite of models are presented in Table D-21 through Table 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 drinking water.
Effect
Male rat
Nasal cavity squamous cell carcinoma
Female rat
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
aTumor incidence values were adjusted for mortality (NCI,
0
0/33C
0
0/34C
0/3 1C
1978).
Animal Dose (mg/kg-day)a
240b
12/26d
350
10/30d
10/30d
530
16733d
640
8/29d
11/29d
Group not included in statistical analysis by NCI (1978) because the dose group was started a year earlier without
appropriate controls.
°p & 0.001; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: NCI (1978).
D-57
-------
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
drinking water for 2 years.
Model
Gamma
Logistic
Log-Logistic13
Log-Probit
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
AIC
84.6972
92.477
84.2821
85.957
84.6972
84.6972
91.7318
84.6972
84.6972
p-value
0.5908
0.02
0.7333
0.3076
0.5908
0.5908
0.0251
0.5908
0.5908
BMDio
mg/kg-day
132.36
284.09
111.46
209.47
132.36
132.36
267.02
132.36
132.36
BMDLio
mg/kg-day
94.06
220.46
72.41
160.66
94.06
94.06
207.18
94.06
94.06
x2a
0
1.727
0
1.133
0
0
1.7
0
0
BMDlOHED
mg/kg-day
34.144
73.29
28.75
54.04
34.14
34.14
68.88
34.14
34.14
BMDLio HED
mg/kg-day
24.26
56.87
18.68
41.45
24.26
24.26
53.44
24.26
24.26
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
Data from NCI (1978).
D-58
-------
Log-Logistic Model with U.ab Confidence Level
o
I
d
o
cc
0.5
0.4
0.3
0.2
0.1
Log-Logistic
BMDL
BMD
0 100
06:32 10/272009
Data points obtained from NCI (1978).
200
300
dose
400
500
600
Figure D-19. Log-Logistic 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-BMR10-Restrict.plt
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 =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
intercept = -6.62889
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
D-59
-------
(*** The 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 -6.91086 * * *
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 -40.8343 3
Fitted model -41.141 1 0.613564 2 0.7358
Reduced model -50.4308 1 19.1932 2 <.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
ChiA2 = 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-60
-------
Multistage Cancer Model with 0.95 Confidence Level
o
t3
CO
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
100
200
300
dose
400
500
600
06:32 10/272009
Data points obtained from 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.plt
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/xl)]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-61
-------
Default Initial Parameter Values
Background = 0.0385912
Beta(l) = 0.000670869
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
Beta (1)
Beta(l) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 0.00079602 * * *
* - 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
ChiA2 = 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-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.
Model
Gamma
Logistic
Log-Logistic13
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
AIC
84.7996
92.569
84.2235
87.3162
84.7996
84.7996
91.9909
84.7996
84.7996
p-value
0.1795
0.0056
0.2486
0.0473
0.1795
0.1795
0.0064
0.1795
0.1795
BMDio
mg/kg-day
176.28
351.51
155.32
254.73
176.28
176.28
328.46
176.28
176.28
BMDLio
mg/kg-day
122.27
268.75
100.08
195.76
122.27
122.27
251.31
122.27
122.27
x2a
1.466
2.148
0
1.871
1.466
1.466
2.136
1.466
1.466
BMDlOHED
mg/kg-day
45.47
90.68
40.07
65.71
45.47
45.47
84.73
45.47
45.47
BMDLio HED
mg/kg-day
31.54
69.33
25.82
50.50
31.54
31.54
64.83
31.54
31.54
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
°Slope restricted > 1.
Data from NCI (1978).
D-63
-------
Log-Logistic Model with 0.95 Confidence Level
T3
= 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
User has chosen the log transformed model
D-64
-------
Default Initial Parameter Values
background = 0
intercept = -6.64005
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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 -7.24274 * * *
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 -39.7535 3
Fitted model -41.1117 1 2.71651 2 0.2571
Reduced model -47.9161 1 16.3252 2 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
ChiA2 = 2.78 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-65
-------
Multistage Cancer Model with 0.95 Confidence Level
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
600
06:30 10/272009
Data points obtained from 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.plt
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*doseAl) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
D-66
-------
Background = 0.0569154
Beta(l) = 0.00042443
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
Beta (1)
Beta(l) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 0.000597685 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -39.7535 3
Fitted model -41.3998 1 3.29259 2 0.1928
Reduced model -47.9161 1 16.3252 2 0.0002851
AIC: 84.7996
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 34 0.000
350.0000 0.1888 6.607 10.000 35 1.466
640.0000 0.3179 11.125 8.000 35 -1.134
ChiA2 = 3.44 d.f. = 2 P-value = 0.1795
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-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.
Model
Gamma
Logistic
Log-Logistic13
Log-Probitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
AIC
93.6005
103.928
92.7669
95.0436
93.6005
93.6005
103.061
93.6005
93.6005
p-value
0.5063
0.0061
0.7809
0.2373
0.5063
0.5063
0.0078
0.5063
0.5063
BMDio
mg/kg-day
73.94
179.05
56.26
123.87
73.94
73.94
168.03
73.94
73.94
BMDLio
mg/kg-day
54.724
139.26
37.26
95.82
54.72
54.72
131.61
54.72
54.72
x2a
0
2.024
0
1.246
0
0
2.024
0
0
BMDlOHED
mg/kg-day
21.17
51.25
16.10
35.46
21.16
21.16
48.10
21.17
21.17
BMDLio HED
mg/kg-day
15.66
39.86
10.66
27.43
15.66
15.66
37.67
15.66
15.66
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
°Slope restricted > 1.
Data from NCI (1978).
D-68
-------
Log-Logistic Model with 0.95 Confidence Level
T3
= 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
User has chosen the log transformed model
Default Initial Parameter Values
background = 0
intercept = -6.08408
D-69
-------
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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 Log (likelihood) # Param' s Deviance Test d.f. P-value
Full model -45.139 3
Fitted model -45.3835 1 0.488858 2 0.7832
Reduced model -59.2953 1 28.3126 2 <.0001
AIC: 92.7669
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size 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
ChiA2 = 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 = 37.256
D-70
-------
Multistage Cancer Model with 0.95 Confidence Level
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
0
06:28 10/27 2009
Data points obtained from NCI (1978).
100
200
300
400
500
dose
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.
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.plt
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/xl)]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0578996
Beta(l) = 0.00118058
D-71
-------
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
Beta(l)
Beta(l) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 0.00142499 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -45.139 3
Fitted model -45.8002 1 1.32238 2 0.5162
Reduced model -59.2953 1 28.3126 2 <.0001
AIC: 93.6005
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 33 -0.000
240.0000 0.2896 9.558 12.000 33 0.937
530.0000 0.5301 18.024 16.000 34 -0.695
Chi^2 = 1.36 d.f. = 2 P-value = 0.5063
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-72
-------
D.7.3. Hepatocellular Adenoma or Carcinoma in B6C3Fi Mice (NCI.
1978)
The incidence data for hepatocellular adenoma or carcinoma in male and female mice are
presented in Table D-24. The 2-degree polynomial model (betas restricted > 0) was the lowest degree
polynomial that provided an adequate fit to the female mouse data (Figure D-25). while the gamma model
provided the best fit to the male mouse data (Figure D-26). The results of the BMDS modeling for the
entire suite of models are presented in Table D-25 and Table 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
0
8/49b
720
19/50d
830
28/47°
Female mouse Animal Dose (mg/kg-day)a
0
0/50b
380
21/48C
860
35/37c
"Tumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
°p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (1978).
Table D-25 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
Gamma
Logistic
Log-Logistic
Log-Probitb
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)0
Probit
Weibull
Quantal-Linear
AIC
85.3511
89.1965
85.3511
85.3511
89.986
85.3511
88.718
85.3511
89.986
p-value
1
0.0935
1
1
0.0548
1
0.1165
1
0.0548
BMD10
mg/kg-day
195.69
199.63
228.08
225.8
49.10
160.68
188.24
161.77
49.10
BMDL10
mg/kg-day
105.54
151.35
151.16
150.91
38.80
67.76
141.49
89.27
38.80
x2a
0
0.675
0
0
0
0
-1.031
0
0
BMD10HED
mg/kg-day
28.16
28.72
32.82
32.49
7.06
23.12
27.08
23.28
7.065
BMDL10HED
mg/kg-day
15.19
21.78
21.75
21.71
5.58
9.75
20.36
12.84
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.
Data from NCI (1978).
D-73
-------
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
o
t3
CO
0.8
0.6
0.4
0.2
0 100
06:36 10/27 2009
Data points obtained from NCI (1978).
200
300
400 500
dose
600
700
800
900
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.plt
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/xl-beta2*dose/x2 ) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
D-74
-------
Default Initial Parameter Values
Background = 0
Beta(l) = 2.68591e-005
Beta(2) = 3.91383e-006
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
Beta(l) Beta(2)
Beta(l) 1 -0.92
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(l) 2.686e-005 * * *
Beta(2) 3.91382e-006 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -40.6756 3
Fitted model -40.6756 2 3.20014e-010 1 1
Reduced model -91.606 1 101.861 2 <.0001
AIC: 85.3511
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 = 186.587
Taken together, (67.7635, 186.587) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00147572
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.
Model
Gammab
Logistic
Log-Logistic
Log-Probitd
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
AIC
177.539
179.9
179.443
179.443
180.618
179.483
179.984
179.443
180.618
p-value
0.7571
0.1189
NCC
NCC
0.0762
0.1554
0.1128
NCC
0.0762
BMDio
mg/kg-day
601.69
252.66
622.39
631.51
164.29
354.41
239.93
608.81
164.29
BMDLio
mg/kg-day
243.92
207.15
283.04
305.44
117.37
126.24
196.90
249.71
117.37
x2a
-0.233
0.214
0
0
0.079
0.124
0.191
0
0.079
BMDlOHED
mg/kg-day
87.98
36.94
91.01
92.34
24.02
51.82
35.08
89.02
24.02
BMDLio HED
mg/kg-day
35.67
30.29
41.39
44.66
17.16
18.46
28.79
36.51
17.16
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
"Value unable to be calculated (NC: not calculated) by BMDS.
dSlope restricted > 1.
Data from NCI (1978).
D-76
-------
Gamma Multi-Hit Model with 0.95 Confidence Level
T3
=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
Default Initial (and Specified) Parameter Values
Background = 0.17
Slope = 0.000671886
Power = 1.3
D-77
-------
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 -0.52
Slope -0.52 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.160326 0.0510618 0.060247 0.260405
Slope 0.0213093 0.000971596 0.019405 0.0232136
Power 18 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log (likelihood) # Param' s Deviance Test d.f. P-value
Full model -86.7213 3
Fitted model -86.7693 2 0.096042 1 0.7566
Reduced model -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
= 0.10 d.f. = 1 P-value = 0.7571
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 601. 692
BMDL = 243.917
D-78
-------
Multistage Cancer Model with 0.95 Confidence Level
o
t5
CO
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
100
200
300
06:34 10/272009
Data points obtained from NCI (1978).
400
dose
500
600
700
800
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_iranouse_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-BMR10-2poly.plt
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/xl-beta2*dose/x2) ]
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 =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.131156
Beta(l) = 0
D-79
-------
Beta(2) = 9.44437e-007
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(l) have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
Background Beta(2)
Background 1 -0.72
Beta(2) -0.72 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.1568 * * *
Beta(l) 0 * * *
Beta(2) 8.38821e-007 * * *
Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -86.7213 3
Fitted model -87.7413 2 2.04001 1 0.1532
Reduced model -96.715 1 19.9875 2 <.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
ChiA2 = 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 = 126.241
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-80
-------
APPENDIX E. COMPARISON OF SEVERAL DATA
REPORTS FOR THE JBRC 2-YEAR 1,4-DIOXANE
DRINKING WATER STUDY
As described in detail in Section 4.2.1.2.6 of this Toxicological Review of 1,4-Dioxane, the JBRC
conducted a 2-year drinking water study on the effects of 1,4-dioxane in both sexes of rats and mice. The
results from this study have been reported three times, once as conference proceedings (Yamazaki et al..
1994). once as a detailed laboratory report (JBRC. 1998). and once as a published manuscript (Kano et
al.. 2009). After the External Peer Review draft of the Toxicological Review of 1,4-Dioxane (U.S. EPA.
2009b) had been released, the Kano et al. (2009) manuscript was published; thus, minor changes to the
Toxicological Review of 1,4-Dioxane occurred.
The purpose of this appendix is to provide a clear and transparent comparison of the reporting of
this 2-year 1,4-dioxane drinking water study. The variations included: (1) the level of detail on dose
information reported; (2) categories for incidence data reported (e.g., all animals or sacrificed animals);
and (3) analysis of non- and neoplastic lesions. Even though the data contained in the reports varied, the
differences were minor and did not did not significantly affect the qualitative or quantitative cancer
assessment.
Tables contained within this appendix provide a comparison of the variations in the reported data
(Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). Table E-l and Table E-2 show the histological
nonneoplastic findings provided for male and female F344 rats, respectively. Table E-3 and Table E-4
show the histological nonneoplastic findings provided for male and female F344 rats, respectively.
Table E-3 and Table E-4 show the histological neoplastic findings provided for male and female F344
rats, respectively. Table E-5 and Table E-6 show the histological nonneoplastic findings provided for
male and female F344 rats, respectively. Table E-7 and Table E-8 show the histological neoplastic
findings provided for male and female Crj:BDFl mice, respectively.
E-l
-------
Table E-l Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in male
F344 rats
Male
Effect F344 Rats
Nasal respiratory epithelium;
nuclear enlargement
Nasal respiratory epithelium;
squamous cell metaplasia
Nasal respiratory epithelium;
squamous cell hyperplasia
Nasal gland; proliferation
Nasal olfactory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
respiratory metaplasia
Nasal olfactory epithelium;
atrophy
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
0 200
JBRC (1998)d
Drinking
1,000 5,000
0
Kano et al. (2009)
water concentration (ppm)
200 1,000
5,000
0 200
1,000 5,000
Calculated Dose (Intake [mg/kg-day])bc
Not
Not
Not
0/50 0/50
Not
0/50 0/50
Not
0/50 0/50
Not
Not
Not
Not
Not
Not
Not
reported
reported
reported
0/50 31/50
reported
0/50 2/50
reported
0/50 5/50
reported
reported
reported
reported
reported
reported
reported
Control
(0)
0/50
0/40
0/50
0/40
0/50
0/40
8-24 41-121
(16) (81)
0/50 0/50
0/45 0/35
0/50 0/50
0/45 0/35
0/50 0/50
0/45 0/35
209-586
(398)
26/50
12/228
31/50
15/228
2/50
1/22
Not reported
Not reported
0/50
0/40
12/50
10/40
0/50
0/40
0/50 5/50
0/45 4/35
11/50 20/50
11/45 17/35
0/50 0/50
0/45 0/35
38/50
20/228
43/50
22/22e
36/50
17/228
0 11 ±1
0/50 0/50
Not
0/50 0/50
Not
0/50 0/50
Not
Not
Not
0/50 0/50
Not
Not
Not
Not
Not
55 ±3 274 ±18
0/50 26/508
reported
0/50 31 /508
reported
0/50 2/50
reported
reported
reported
5/50 38/508
reported
reported
reported
reported
reported
E-2
-------
Table E-l (Continued): Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water
study in male F344 rats
Male
Effect F344 Rats
Lamina propria; hydropic
change
Lamina propria; sclerosis
Nasal cavity; adhesion
Nasal cavity; inflammation
Hyperplasia; liver9
Spongiosis hepatis; liver
Clear cell foci; liver9
Acidophilic cell foci; liver3
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
0 200
JBRC (1998)d
Drinking
1,000 5,000
0
Kano et al. (2009)
water concentration (ppm)
200
1,000
5,000
0 200
1,000
5,000
Calculated Dose (Intake [mg/kg-day])bc
Not
Not
Not
Not
Not
Not
Not
Not
Not
3/50 2/10
Not
reported
reported
reported
reported
reported
reported
reported
reported
reported
10/50 24/50
reported
12/50 20/50 25/50 40/50
Not
Not
Not
Not
Not
reported
reported
reported
reported
reported
Control
(0)
0/50
0/40
0/50
0/40
0/50
0/40
0/50
0/40
3/50
3/40
12/50
12/40
3/50
3/40
8-24
(16)
0/50
0/45
0/50
0/45
0/50
0/45
0/50
0/45
2/50
2/45
20/50
20/45
3/50
3/45
Not
Not
41-121
(81)
0/50
0/35
1/50
1/35
0/50
0/35
0/50
0/35
10/50
9/35f
25/50
21/35f
9/50
9/35f
reported
reported
209-586
(398)
46/50
20/228
44/50
20/228
48/50
21/228
13/50
7/22e
24/50
12/228
40/50
21/228
8/50
7122s
0 11 ±1
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
3/50 3/50
Not
12/50 8/50
Not
55 ±3
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
9/50
reported
7/50
reported
274 ±18
8/50
5/50
E-3
-------
Table E-l (Continued): Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water
study in male F344 rats
Male
Effect F344 Rats
Basophilic cell foci; liver9
Mixed-cell foci; liver9
Nuclear enlargement; kidney
proximal tubule
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC (1998)d
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000
5,000
0 200 1,000
5,000
Calculated Dose (Intake [mg/kg-day])bc
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control 8-24 41-121
(0) (16) (81)
7/50 11/50 6/50
7/40 11/45 6/35
2/50 8/50 14/50
2/40 8/45 14/358
0/50 0/50 0/50
0/40 0/45 0/35
209-586
(398)
16/50
8/22f
13/50
22/22e
50/50
22/22e
0 11 ±1 55 ±3
7/50 11/50 8/50
274 ±18
16/50f
Not reported
2/50 8/50 14/50e
13/508
Not reported
Not reported
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) forthe animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dJBRC (1998) did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
fp < 0.05 by x2 test.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-examined according to updated criteria for liver lesions and were afterwards
classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009).
E-4
-------
Table E-2 Nonneoplastic lesions:
F344 rats
Comparison of histological findings reported for the 2-year JBRC drinking water study in female
Female
Effect F344 Rats
Nasal respiratory
epithelium; nuclear
enlargement
Nasal respiratory
epithelium; squamous cell
metaplasia
Nasal respiratory
epithelium; squamous cell
hyperplasia
Nasal gland; proliferation
Nasal olfactory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
respiratory metaplasia
Nasal olfactory epithelium;
atrophy
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC (1998)bd
Drinking
0 200 1,000 5,000
0
Calculated
Not reported
Not reported
Not reported
0/50 0/50 0/50 35/50
Not reported
0/50 0/50 0/50 5/50
Not reported
0/50 0/50 0/50 11/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control
(0)
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
2/50
1/38
0/50
0/38
water concentration
200 1,000
Kano et al. (2009)
(ppm)
5,000
0 200 1,000 5,000
Dose (Intake [mg/kg-day])b'c
12-29 56-149
(21) (103)
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 28/50
0/37 24/38e
0/50 2/50
0/37 1/38
0/50 1/50
0/37 1/38
307-720
(514)
13/50
7/24e
35/50
18/248
5/50
4/24f
11/50
8/24e
39/50
22/24e
42/50
24/24e
40/50
22/24e
0 18 ±3 83 ±14 429 ±69
0/50 0/50 0/50 13/508
Not reported
0/50 0/50 0/50 35/508
Not reported
0/50 0/50 0/50 5/50
Not reported
Not reported
Not reported
0/50 0/50 28/508 39/508
Not reported
Not reported
Not reported
Not reported
Not reported
E-5
-------
Table E-2 (Continued) Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study
in female F344 rats
Female
Effect F344 Rats
Lamina propria; hydropic
change
Lamina propria; slerosis
Nasal cavity; adhesion
Nasal cavity; inflammation
Liver; hyperplasia9
Liver; spongiosis hepatis
Liver; cyst formation
Liver; clear cell focig
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC (1998)bd
Drinking
0 200 1,000 5,000
0
Calculated
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
3/50 2/50 11/50 47/50
Not reported
0/50 0/50 1/50 20/50
Not reported
Not reported
Not reported
Not reported
Not reported
Control
(0)
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
3/50
2/38
0/50
0/38
0/50
0/38
water concentration
200 1,000
Kano et al. (2009)
(ppm)
5,000
0 200 1,000 5,000
Dose (Intake [mg/kg-day])b'c
12-29 56-149
(21) (103)
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 1/50
0/37 1/38
2/50 11/50
2/37 9/38
0/50 1/50
0/37 1/38
1/50 1/50
1/37 0/38
307-720
(514)
46/50
23/24e
48/50
23/24e
46/50
24/24e
15/50
7/24e
47/50
24/24e
20/50
14/248
8/50
5/24f
Not reported
Not reported
0 18 ±3 83 ±14 429 ±69
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
1/50 1/50 5/50 4/50
Not reported
E-6
-------
Table E-2 (Continued) Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study
in female F344 rats
Female
Effect F344 Rats
Liver; acidophilic cell foci9
Liver; basophilic cell foci9
Liver; mixed-cell foci9
Kidney proximal tubule;
nuclear enlargement
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC (1998)bd
Drinking
0 200 1,000 5,000
0
Calculated
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control
(0)
water concentration
200 1,000
Kano et al. (2009)
(ppm)
5,000
0 200 1,000
5,000
Dose (Intake [mg/kg-day])b'c
12-29 56-149
(21) (103)
307-720
(514)
Not reported
Not reported
Not reported
Not reported
1/50
1/38
0/50
0/38
1/50 3/50
1/37 3/38
0/50 6/50
0/37 6/38
11/50
7/24f
39/50
22/24e
0 18 ±3 83 ±14
1/50 1/50 1/50
429 ± 69
1/50
Not reported
23/50 27/50 31/50
8/50e
Not reported
1/50 1/50 3/50
11/50f
Not reported
Not reported
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dJBRC (1998) did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
fp < 0.05 by x2 test.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-examined according to updated criteria for liver lesions and were afterwards
classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009).
E-7
-------
Table E-3 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in male
F344 rats
Male
Effect F344 Rats
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day])b|C
Not reported
Control 8-24 41-121 209-586
(0) (16) (81) (398)
0 11 ±1 55 ±3 274 ±18
Nasal cavity
Squamous cell carcinoma
Sarcoma NOS
Rabdomyosarcoma
Esthesioneuroepithelioma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50 0/50 0/50 3/50
Not reported
0/50 0/50 0/50 2/50
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 3/50e
Not reported
0/50 0/50 0/50 2/50
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50
3/50e
Not reported
0/50 0/50 0/50
2/50
Not reported
0/50 0/50 0/50
1/50
Not reported
0/50 0/50 0/50
1/50
Not reported
E-8
-------
Table E-3 (Continued) Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in
male F344 rats
Male
Effect F344 Rats
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day])bc
Not reported
Control 8-24 41-121 209-586
(0) (16) (81) (398)
0 11 ±1 55 ±3 274 ±18
Liver
Hepatocellular adenoma'
Hepatocellular carcinoma
Hepatocellular adenoma or
carcinoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50 2/50 4/50 24/50
Not reported
0/50 0/50 0/50 14/50
Not reported
Not reported
Not reported
0/50 2/50 4/49 24/50d'e
Not reported
0/50 0/50 0/49 14/50d'e
Not reported
0/50 2/50 4/49 33/50d'e
Not reported
3/50 4/50 7/50
32/50d'e
Not reported
0/50 0/50 0/50
14/50d'e
Not reported
3/50 4/50 7/50
39/50d'e
Not reported
Tumors at other sites
Peritoneum mesothelioma
Subcutis fibroma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
2/50 2/50 5/50 28/50
Not reported
5/50 3/50 5/50 12/50
Not reported
2/50 2/50 5/50 28/50d'e
Not reported
5/50 3/50 5/50 12/508
Not reported
2/50 2/50 5/50
28/50d'e
Not reported
5/50 3/50 5/50
12/508
Not reported
E-9
-------
Table E-3 (Continued) Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in
male F344 rats
Male
Effect F344 Rats
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day])bc
Not reported
Control 8-24 41-121 209-586
(0) (16) (81) (398)
0 11 ±1 55 ±3 274 ±18
Tumors at other sites (Continued)
Mammary gland fibroadenoma
Mammary gland adenoma
Mammary gland fibroadenoma
or adenoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
1/50 1/50 0/50 4/50
Not reported
0/50 0/50 0/50 0/50
Not reported
Not reported
Not reported
1/50 1/50 0/50 4/50e
Not reported
Not reported
Not reported
Not reported
Not reported
1/50 1/50 0/50
4/50e
Not reported
0/50 1/50 2/50
2/50
Not reported
1/50 2/50 2/50
6/50e
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dp < 0.01 by Fisher's Exact test.
eSignificantly increased by Peto test for trend p < 0.01.
The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-examined according to updated criteria for liver lesions and were afterwards
classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009).
E-10
-------
Table E-4 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in female F344
rats
Female
Effect F344 Rats
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200
1,000 5,000
Calculated Dose (Intake [mg/kg-day])b c
Not Reported
Control 56-149 307-720
(0) 12-29(21) (103) (514)
0 18 ±3
83 ± 14 429 ± 69
Nasal cavity
Squamous cell carcinoma
Sarcoma NOS
Rabdomyosarcoma
Esthesioneuroepithelioma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50 0/50 0/50 7/50
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 7/50d'f
Not reported
Not reported
Not reported
Not reported
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 7/50e'f
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 1/50
Not reported
E-ll
-------
Table E-4 (Continued): Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in
female F344 rats
Female
Effect F344 Rats
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200
1,000 5,000
Calculated Dose (Intake [mg/kg-day])b c
Not Reported
Control 56-149 307-720
(0) 12-29(21) (103) (514)
0 18 ±3
83 ± 14 429 ± 69
Liver
Hepatocellular adenoma9
Hepatocellular carcinoma
Hepatocellular adenoma or
carcinoma9
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
1/50 0/50 5/50 38/50
Not reported
0/50 0/50 0/50 10/50
Not reported
Not reported
Not reported
1/50 0/50 5/50 38/50e'f
Not reported
1/50 0/50 0/50 10/50e'f
Not reported
1/50 0/50 5/50 40/50e'f
Not reported
3/50 1/50 6/50
48/50e'f
Not reported
0/50 0/50 0/50
10/50e'f
Not reported
3/50 1/50 6/50
48/50e'f
Not reported
Tumors at other sites
Peritoneum mesothelioma
Subcutis fibroma
Mammary gland
fibroadenoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
1/50 0/50 0/50 0/50
Not reported
0/50 2/50 1/50 0/50
Not reported
3/50 2/50 1/50 3/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
1/50 0/50 0/50
0/50
Not reported
0/50 2/50 1/50
0/50
Not reported
3/50 2/50 1/50
3/50
Not reported
E-12
-------
Table E-4 (Continued): Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in
female F344 rats
Female
Effect F344 Rats
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200
1,000 5,000
Calculated Dose (Intake [mg/kg-day])b c
Not Reported
Control 56-149 307-720
(0) 12-29(21) (103) (514)
0 18 ±3
83 ± 14 429 ± 69
Tumors at other sites (Continued)
Mammary gland adenoma
Mammary gland
fibroadenoma
or adenoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
6/50 7/50 10/50 16/50
Not reported
Not reported
Not reported
6/50 7/50 10/50 16/50d'f
Not reported
Not reported
Not reported
6/50 7/50 10/50
16/50d'f
Not reported
8/50 8/50 11/50
18/50d'f
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) forthe animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dp < 0.05 by Fisher's Exact test.
ep < 0.01 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994) and JBRC (1998) were re-examined according to updated criteria for liver lesions and were afterwards
classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009).
E-13
-------
Table E-5 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in male
CrjrBDFl mice
Male
Crj:BDF1
Effect Mice
Nasal respiratory epithelium; nuclear
enlargement
Nasal olfactory epithelium; nuclear
enlargement
Nasal olfactory epithelium; atrophy
Nasal cavity; inflammation
Tracheal epithelium; atrophy
Tracheal epithelium; nuclear
enlargement
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC (1998)M
0 500 2,000 8,000
0
Drinking
500
Calculated
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Kano et al. (2009)
water concentration (ppm)
2,000
Dose (Intake
Control 37-94 144-358
0 (66) (251)
0/50
0/31
0/50
0/31
0/50
0/31
1/50
1/31
0/50
0/31
0/50
0/31
0/50
0/33
0/50
0/33
0/50
0/33
2/50
1/33
0/50
0/33
0/50
0/33
0/50
0/25
9/50
7/25e
1/50
0/25
1/50
1/25
0/50
0/25
0/50
0/25
8,000
0 500 2,000 8,000
[mg/kg-day])b'c
451-1,086
(768)
31/50
19/268
49/50
26/26e
48/50
26/26e
25/50
15/268
42/50
24/26e
17/50
12/268
191 ±
0 49 ±5 21 677 ±74
0/50 0/50 0/50 31/508
Not reported
0/50 0/50 9/50e 49/508
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
E-14
-------
Table E-5 (Continued): Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water
study in male Crj:BDFl mice
Male
Crj:BDF1
Effect Mice
Bronhcial epithelium; nuclear
enlargement
Bronchial epithelium; atrophy
Lung/bronchial; accumlation of foamy
cells
Liver; angiectasis
Kidney proximal tubule; nuclear
enlargement
Testis; mineralization
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC (1998)M
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0
500
Calculated
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control 37-94
0 (66)
0/50
0/31
0/50
0/31
1/50
1/31
2/50
2/31
0/50
0/31
40/50
28/31
0/50
0/33
0/50
0/33
0/50
0/33
3/50
2/33
0/50
0/33
42/50
30/33
2,000
Dose (Intake
144-358
(251)
0/50
0/25
0/50
0/25
0/50
0/25
4/50
3/25
0/50
0/25
38/50
24/25f
8,000
0 500 2,000 8,000
[mg/kg-day])b'c
451-1,086
(768)
41/50
24/26e
43/50
26/26e
27/50
22/26e
16/50
8/26f
39/50
22/26e
34/50
21/26f
191 ±
0 49 ±5 21 677 ±74
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Dose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) forthe animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dJBRC (1998) did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
fp < 0.05 by x2 test.
E-15
-------
Table E-6 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in female
CrjrBDFl mice
Female
Crj:BDF1
Effect Mice
Nasal respiratory epithelium;
Nuclear enlargement
Nasal olfactory epithelium;
Nuclear enlargement
Nasal respiratory epithelium;
Atrophy
Nasal olfactory epithelium;
Atrophy
Nasal cavity; Inflammation
Tracheal epithelium; Atrophy
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC
(1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500
2,000 8,000
0
500
2,000
Calculated Dose (Intake
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
Control 45-109
0 (77)
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
2/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
192-454
(323)
0/50
0/17
41/50
17/178
0/50
0/17
1/50
0/17
7/50
5/1 T
2/50
1/17
8,000
0 500
[mg/kg-day])b'c
759-1,374
(1,066)
41/50
5/5e
33/50
1/5
26/50
1/5
42/50
5/5e
42/50
5/5e
49/50
5/5e
0 66 ±10
0/50 0/50
Not
0/50 0/50
Not
Not
Not
Not
Not
Not
Not
Not
Not
2,000 8,000
278 ± 40 964 ± 88
0/50 41/508
reported
41/508 33/508
reported
reported
reported
reported
reported
reported
reported
reported
reported
E-16
-------
Table E-6 (Continued): Nonneoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water
study in female Crj:BDFl mice
Female
Crj:BDF1
Effect Mice
Bronhcial epithelium; Nuclear
enlargement
Bronchial epithelium; Atrophy
Lung/bronchial; Accumlation of
foamy cells
Kidney proximal tubule;
Nuclear enlargement
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC
(1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500
2,000 8,000
0
500
2,000
Calculated Dose (Intake
Not
Not
Not
Not
Not
Not
Not
Not
Not
reported
reported
reported
reported
reported
reported
reported
reported
reported
Control 45-109
0 (77)
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
1/50
1/29
0/50
0/29
1/50
1/29
0/50
0/29
192-454
(323)
22/50
13/178
7/50
3/17
4/50
3/17
0/50
0/17
8,000
0 500
[mg/kg-day])b'c
759-1,374
(1,066)
48/50
5/5e
50/50
5/5e
45/50
5/5e
8/50
0/5
0 66 ±10
Not
Not
Not
Not
Not
Not
Not
Not
2,000 8,000
278 ± 40 964 ± 88
reported
reported
reported
reported
reported
reported
reported
reported
aDose rates mg/kg-day]) were not provided in Yamazaki et al. (1994). Drinking water concentrations (ppm) of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
""Statistical analysis was not performed for data on 'All animals' in the JBRC (1998) report.
°JBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
dKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
ep & 0.01 by chi-square test.
E-17
-------
Table E-7 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in male
CrjrBDFl mice
Male
Effect Crj:BDF1
Mice
Yamazaki et al. (1994)a
0 500 2,000 8,000
Not reported
JBRC (1998)b
Drinking water concentration (ppm)
0 500 2,000 8,000
Kano et al. (2009)
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])b'c
Control 37-94 144-358 451-1,086
0 (66) (251) (768)
0 49 ±5 191 ±21 677 ±74
Nasal cavity
Esthesioneuroepithelioma
Adenocarcinoma
All Animals
Sacrificed animals
All Animals
Sacrificed animals
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 1/50
Not reported
Not reported
Not reported
0/50 0/50 0/50
1/50
Not reported
0/50 0/50 0/50
0/50
Not reported
Liver
Hepatocellular adenomas
Hepatocellular carcinomas
All Animals
Sacrificed animals
All Animals
Sacrificed animals
7/50 16/50 22/50 8/50
Not reported
15/50 20/50 23/50 36/50
Not reported
7/50 16/50 22/508 8/50
Not reported
15/50 20/50 23/50 36/50d'e
Not reported
9/50 17/50 23/508 11/50
Not reported
15/50 20/50 23/50 36/50e'f
Not reported
E-18
-------
Table E-7 (Continued): Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in
male Crj:BDFl mice
Male
Effect Crj:BDF1
Mice
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0 500 2,000 8,000
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])bc
Not reported
Control 37-94 144-358 451-1,086
0 (66) (251) (768)
0 49 ±5 191 ±21 677 ±74
Liver (Continued)
Either adenoma
or carcinoma
All Animals
Sacrificed animals
Not reported
Not reported
21/50 31/50 37/50 39/50d'e
Not reported
23/50 31/50 37/50d
40/50e'f
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) forthe animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA. 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dp < 0.05 by Fisher's Exact test.
eSignificantly increased by Peto test for trend p < 0.01.
fp < 0.01 by Fisher's Exact test.
E-19
-------
Table E-8 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in female
CrjrBDFl mice
Female
Crj:BDF1
Effect Mice
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0 500 2,000 8,000
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])b'c
Not reported
Control 45-109 192-454 759-1,374
0 (77) (323) (1,066)
0 66 ±10 278 ±40 964 ± 88
Nasal Cavity
Esthesioneruoepithelioma
Adenocarcinoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 1/50
Not reported
Not reported
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50
0/50
Not reported
0/50 0/50 0/50
1/50
Not reported
Liver
Hepatocellular adenomas
Hepatocellular carcinomas
All animals
Sacrificed
animals
All animals
Sacrificed
animals
4/50 30/50 20/50 2/50
Not reported
0/50 6/50 30/50 45/50
Not reported
4/50 30/50d 20/50d 2/50e
Not reported
0/50 6/50f 30/50d 45/50d'g
Not reported
5/50 31/50d 20/50d
3/50
Not reported
0/50 6/50f 30/50d
45/50d'g
Not reported
E-20
-------
Table E-8 (Continued): Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC drinking water study in
female Crj:BDFl mice
Female
Crj:BDF1
Effect Mice
Yamazaki et al. (1994)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0 500 2,000 8,000
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])b c
Not reported
Control 45-109 192-454 759-1,374
0 (77) (323) (1,066)
0 66 ±10 278 ±40 964 ± 88
Liver (Continued)
Either adenoma
or carcinoma
All animals
Sacrificed
animals
Not reported
Not reported
4/50 34/50d 41/50d 46/50d'g
Not reported
5/50 35/50d 41/50d
46/50d'g
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations (ppm) of 1,4-dioxane were used to identify the dose groups. Statistical test results were not
reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) forthe animals; and the midpoint of the range (shown in parentheses) was used in the external peer review draft of this
document (U.S. EPA, 2009b).
°Kano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the calculation of 1,4-dioxane toxicity values (U.S. EPA. 2013c.
2010).
dp < 0.01 by Fisher's Exact test.
eSignificantly decreased by Cochran-Armitage test for trend p < 0.05
fp < 0.05 by Fisher's Exact test.
Significantly increased by Peto test for trend p < 0.01
E-21
-------
APPENDIX F. DETAILS OF BMD ANALYSIS FOR
INHALATION RFC FOR 1,4-DIOXANE
F.1. Centrilobular Necrosis of the Liver
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-l. for centrilobular necrosis of the liver in male F344/DuCrj rats exposed
to 1,4-dioxane vapors for 2 years (Kasai et al., 2009). Doses associated with a BMR of a 10% extra risk
were calculated.
Table F-l Incidence of centrilobular necrosis of the liver in male F344/DuCrj rats exposed
to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
1/50
(2%)
50
3/50
(6%)
250
6/50
(12%)
1,250
12/50a
(24%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
As assessed by the %2 goodness-of-fit test, several models in the software provided adequate fits
to the incidence data of centrilobular necrosis of the liver in male rats (£ p > 0.1) (Table F-2). Comparing
across adequately fitting models, the BMDL estimates were not within threefold difference of each
other. Therefore, in accordance with EPA BMD technical guidance (U.S. EPA, 2012b). the adequately
fitting model that resulted in the lowest BMDL was selected as appropriate for deriving a POD which was
the Dichotomous-Hill model. BMDS modeling results for all dichotomous models are shown in Table F-2
and the model plot (Figure F-l) and output for the selected Dichotomous-Hill model are included
immediately after the table.
F-l
-------
Table F-2 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for centrilobular necrosis of the liver in male F344/DuCrj rats
exposed to 1,4-dioxane vapors (Kasai et al., 2009)
Model
AIC
p-valuea
Scaled
Residual of
Interest
BMD10
(ppm)
BMDL10
(ppm)
Male
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hill0' e
129.692
131.043
129.465
132.067
129.692
130.889
129.692
129.692
130.404
0.5099
0.2794
0.568
0.1645
0.5099
0.2992
0.5099
0.5099
0.7459
0.786
-0.142
0.676
-0.175
0.786
-0.167
0.786
0.786
-0.179
502.444
794.87
453.169
801.17
502.445
756.192
502.461
502.461
219.51
308.113
609.269
258.687
539.489
308.112
567.169
308.113
308.113
59.5598
a ,, i , ., 2 j f f-i i i f iu 1 t J J 1 \/ 1 ,-n 1 ' rl' t th t th ^l I h'lVt ^l t f f II
significant lack of fit, and thus a different model should be chosen.
bPower restricted to > 1.
°Slope restricted to a 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest BMDL.
Data from Kasai et al. (2009).
F-2
-------
Dichotomous-Hill Model with 0.95 Confidence Level
0.4
0.35
0.3
-a 0.25
a>
"o
O
t3
ro
0.2
0.15
0.1
0.05
Dichotomous-Hill
BMDL
BMD
200
400
16:1801/122011
600
dose
800
1000
1200
Data points obtained from Kasai et al. (2009).
Figure F-l. BMD Dichotomous Hill model of centrilobular necrosis incidence data for
male rats exposed to 1,4-dioxane vapors for 2 years.
Dichotomous Hill Model. (Version: 1.2; Date: 12/11/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl_Centr_necrosis_liver_Dhl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl_Centr_necrosis_liver_Dhl-BMR10-Restrict.pit
Wed Jan 12 16:34:41 2011
BMDS_Model_Run
The form of the probability function is:
P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
where: 0 <= g < 1, 0= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
v = -9999
F-3
-------
g = -9999
intercept = -8.08245
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -slope have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
v g intercept
v 1 -0.25 -0.89
g -0.25 1 0.016
intercept -0.89 0.016 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
v 0.311077 0.156196 0.00493876 0.617216
g 0.0709966 0.0662298 -0.0588115 0.200805
intercept -6.06188 1.34538 -8.69878 -3.42498
slope 1 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -62.1506 4
Fitted model -62.2022 3 0.103279 1 0.7479
Reduced model -69.3031 1 14.305 3 0.002518
AIC: 130.404
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0221 1.104 1.000 50 -0.100
50.0000 0.0522 2.612 3.000 50 0.247
250.0000 0.1285 6.423 6.000 50 -0.179
1250.0000 0.2372 11.861 12.000 50 0.046
ChiA2 = 0.10 d.f. = 1 P-value = 0.7459
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 219.51
BMDL = 59.5598
F-4
-------
F.2. Squamous Cell Metaplasia
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-3. for squamous cell metaplasia of the respiratory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
a 10% extra risk were calculated.
Table F-3 Incidence of squamous cell metaplasia of the respiratory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50
0/50
250
7/50b
(14%)
1,250
44/50a
(88%)
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
Source: Kasai et al. (2009).
For incidence of squamous cell metaplasia in F344/DuCrj male rats, the logistic and probit
models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1; see Table F-4), and thus
should not be considered further for identification of a POD. All of the remaining models exhibited
adequate fit. The BMDL estimates for all appropriately fitting models were within threefold
difference of each other, indicating that BMDL selection should be made based on model fit (U.S.
EPA. 2012b). As assessed by the AIC, the Log-probit model provided the best fit to the squamous cell
metaplasia data for male rats (Table F-4. Figure F-3). and could be used to derive a POD for this
endpoint.
F-5
-------
Table F-4 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for squamous cell metaplasia of the respiratory epithelium in
male F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai et al., 2009)
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gammab
Logistic
Log-logistic0
Log-probitc'e
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hillc
81.687
89.4148
81.5252
81.23
82.6875
87.9361
82.1236
92.9215
83.1888
0.8682
0.0464
0.9142
0.9894
0.6188
0.0779
0.7679
0.0198
0.9995
0.24
1.806
0.131
0.032
0.605
1.681
0.33
-1.76
0
218.38
370.443
218.218
217.79
231.294
337.732
218.435
87.682
240.867
150.329
288.535
158.293
159.619
141.025
268.424
145.383
68.8015
161.945
a p-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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest AIC.
Data from Kasai et al. (2009).
F-6
-------
LogProbit Model with 0.95 Confidence Level
T3
"O
<
d
O
0.8
0.6
0.4
0.2
1200
13:11 01/13 2011
Data points obtained from Kasai et al. (2009).
Figure F-2. BMD Log-probit model of squamous cell metaplasia of the respiratory
epithelium incidence data for male rats exposed to 1,4-dioxane vapors for
2 years.
Probit Model. (Version: 3.2; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnp_squ_cell_meta_re_Lnp-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnp_sgu_cell_meta_re_Lnp-BMR10-Restrict.pit
Thu Jan 13 13:11:09 2011
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
User has chosen the log transformed model
Default Initial (and Specified) Parameter Values
background = 0
intercept = -6.76507
F-7
-------
slope = 1.09006
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
intercept slope
intercept 1 -0.99
slope -0.99 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0 NA
intercept -8.86173 1.2226 -11.258 -6.46548
slope 1.40803 0.193057 1.02965 1.78642
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -38.5944 4
Fitted model -38.615 2 0.041197 2 0.9796
Reduced model -113.552 1 149.916 3 <.0001
AIC: 81.23
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 50 0.000
50.0000 0.0004 0.020 0.000 50 -0.141
250.0000 0.1384 6.922 7.000 50 0.032
1250.0000 0.8808 44.038 44.000 50 -0.017
Chi^2 = 0.02 d.f. = 2 P-value = 0.9894
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 217.79
BMDL = 159.619
F-8
-------
F.3. Squamous Cell Hyperplasia
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-5. for squamous cell hyperplasia of the respiratory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
a 10% extra risk were calculated.
Table F-5 Incidence of squamous cell hyperplasia of the respiratory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0 50 250 1,250
0/50 0/50 1/50 10/50a
(2%) (20%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
For incidence of squamous cell hyperplasia in F344/DuCrj male rats, the logistic, probit, and
quantal-linear models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1; see
Table F-6). and thus should not be considered further for identification of a POD. All of the remaining
models exhibited adequate fit. The BMDL estimates for all appropriately fitting models were within
threefold difference of each other, indicating that BMDL selection should be made based on model
fit (U.S. EPA, 2012b). As assessed by the AIC, the Log-probit model provided the best fit to the
squamous cell hyperplasia data for male rats (Table F-6. Figure F-3 and subsequent textual model output),
and could be used to derive a POD for this endpoint.
F-9
-------
Table F-6 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for squamous cell hyperplasia of the respiratory epithelium in
male F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai et al., 2009)
Model
AIC
p-valuea
Scaled
Residual of
Interest
BMD10
(ppm)
BMDL10
(ppm)
Male
Gammab
Logistic
Log-logistic0
Log-probitc'e
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hillc
81.687
89.4148
81.5252
81.23
82.6875
87.9361
82.1236
92.9215
83.1888
0.8682
0.0464
0.9142
0.9894
0.6188
0.0779
0.7679
0.0198
0.9995
0.24
1.806
0.131
0.032
0.605
1.681
0.33
-1.76
0
218.38
370.443
218.218
217.79
231.294
337.732
218.435
87.682
240.867
150.329
288.535
158.293
159.619
141.025
268.424
145.383
68.8015
161.945
a p-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 a 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest AIC.
Data from Kasai et al. (2009).
F-10
-------
LogProbit Model with 0.95 Confidence Level
T3
"O
<
d
O
0.35
0.3
0.25
0.2
0.15
0.1
0.05
1200
13:25 01/13 2011
Data points obtained from Kasai et al. (2009).
Figure F-3. BMD Log-probit model of squamous cell hyperplasia of the respiratory
epithelium incidence data for male rats exposed to 1,4-dioxane vapors for
2 years.
Probit Model. (Version: 3.2; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnp_squ_cell_hyper_re_Lnp-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnp_sgu_cell_hyper_re_Lnp-BMR10-Restrict.pit
Thu Jan 13 13:25:05 2011
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
User has chosen the log transformed model
Default Initial (and Specified) Parameter Values
background = 0
F-ll
-------
intercept = -7.75604
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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 NA
intercept -7.90911 0.186242 -8.27414 -7.54408
slope 1 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -29.9221 4
Fitted model -30.2589 1 0.673572 3 0.8794
Reduced model -42.5964 1 25.3487 3 <.0001
AIC: 62.5177
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 50 0.000
50.0000 0.0000 0.002 0.000 50 -0.040
250.0000 0.0085 0.424 1.000 50 0.889
1250.0000 0.2182 10.911 10.000 50 -0.312
ChiA2 = 0.89 d.f. = 3 P-value = 0.8282
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 755.635
BMDL = 560.86
F-12
-------
F.4. Respiratory Metaplasia
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-7. for respiratory metaplasia of the olfactory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
a 10% extra risk were calculated.
Table F-7 Incidence of respiratory metaplasia of the olfactory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
11/50
(22%)
50
34/50
(68%)
250
49/50 a
(98%)
1,250
48/50a
(96%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
As assessed by the %2 goodness-of-fit test, no models in the software provided adequate fits to the
data for the incidence of respiratory metaplasia of the olfactory epithelium in male rats (%2/> > 0.1)
(Table F-8). However, given that first non-control dose had a response level substantially above the
desired BMR (i.e., 10%), the use of BMD methods included substantial model uncertainty. The model
uncertainty associated with this dataset is related to low-dose extrapolation and consistent with BMD
Technical Guidance Document (U.S. EPA, 2012b) all available dichotomous models in the Benchmark
Dose Software (version 2.1.2) were fit to the incidence data shown in Table F-9 with the highest dose
group omitted. As assessed by the %2 goodness-of-fit test, the logistic, log-logistic, log-probit, and probit
models all exhibited a statistically significant lack of fit (i.e., %2 /"-value < 0.1; see Table F-9). and thus
should not be considered further for identification of a POD. The BMDL estimates for all appropriately
fitting models were within threefold difference of each other, indicating that BMDL selection should be
made based on model fit (U.S. EPA. 2012b). The AIC values for gamma, multistage, quantal-linear, and
Weibull models in Table F-9 are equivalent and the lowest and, in this case, essentially represent the same
model. Therefore, consistent with the Benchmark Dose Technical Guidance (U.S. EPA. 2012b). any of
them with equal AIC values (gamma, multistage, quantal-linear, or Weibull) could be used to identify a
POD for this endpoint. The model plot for the gamma model (Figure F-4) and output are included
immediately after the table.
F-13
-------
Table F-8 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for respiratory metaplasia of olfactory epithelium in male
F344/DuCrj rats (Kasai et al., 2009) exposed to 1,4-dioxane vapors
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hillc
179.68
191.339
152.72
161.267
179.68
198.785
179.68
179.68
150.466
0
0
0.0285
0
0
0
0
0
NA
-2.07
1.788
0.039
-0.39
-2.07
1.479
-2.07
-2.07
0
17.4082
34.2946
4.05465
14.3669
17.4082
61.4378
17.4082
17.4082
38.8552
12.3829
24.5917
1.90233
10.3023
12.3829
45.9091
12.3829
12.3829
31.4727
ap-Value from the x 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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
Data from Kasai et al. (2009).
F-14
-------
Table F-9 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for respiratory metaplasia of olfactory epithelium with high dose
group dropped in male F344/DuCrj rats (Kasai et al., 2009) exposed to
1,4-dioxane vapors
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gammab'e
Logistic
Log-logistic0
Log-probitc
Multistage
(2degree)d'e
Probit
Weibullb
Quantal-Linear6
129.463
133.583
131.182
131.182
129.463
136.121
129.463
129.463
0.5815
0.0119
NA
NA
0.5815
0.0066
0.5815
0.5815
-0.106
-1.031
0
0
-0.106
-1.511
-0.106
-0.106
6.46848
12.5197
14.2075
12.2114
6.46847
15.2883
6.46847
6.46847
4.73742
9.34421
3.77044
7.80131
4.73742
11.6855
4.73742
4.73742
a p-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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
eBold indicates best-fit models based on lowest AIC.
Data from Kasai et al. (2009).
F-15
-------
Gamma Multi-Hit Model with 0.95 Confidence Level
T3
"O
<
d
O
0.8
0.6
0.4
0.2
Gamma Multi-Hit
EiMDL BMD
0 50
16:24 01/13 2011
Data points obtained from Kasai et al. (2009).
100
150
200
250
dose
Figure F-4. BMD Gamma model of respiratory metaplasia of olfactory epithelium
incidence data for male rats exposed to 1,4-dioxane vapors for 2 years.
Gamma Model. (Version: 2.15; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/gam_resp_meta_no high dose_Gam-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/gam_resp_meta_no high dose_Gam-BMR10-Restrict.pit
Thu Jan 13 16:24:15 2011
BMDS_Model_Run
The form of the probability function is:
P[response]= background+(1-background)*CumGamma[siope*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 = 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 = 0.230769
Slope = 0.022439
Power = 1.3
F-16
-------
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 -0.33
Slope -0.33 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.226249 0.0588535 0.110898 0.3416
Slope 0.0162883 0.00320976 0.00999729 0.0225793
Power 1 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -62.5908 3
Fitted model -62.7313 2 0.280907 1 0.5961
Reduced model -99.1059 1 73.0301 2 <.0001
AIC: 129.463
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.2262 11.312 11.000 50 -0.106
50.0000 0.6573 32.865 34.000 50 0.338
250.0000 0.9868 49.341 49.000 50 -0.422
ChiA2 = 0.30 d.f. = 1 P-value = 0.5815
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 6.46848
BMDL = 4.73742
F-17
-------
F.5. Atrophy
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-10. for atrophy of the olfactory epithelium in male F344/DuCrj rats
exposed to 1,4-dioxane vapors for 2 years (Kasai et al.. 2009). Doses associated with a BMR of a 10%
extra risk were calculated.
Table F-10 Incidence of atrophy of the olfactory epithelium in male F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50
40/50 a
(80%)
250
47/50 a
(94%)
1,250
48/50a
(96%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
As assessed by the %2 goodness-of-fit test, the gamma, logistic, log-probit, multistage, probit,
Weibull, and quantal-linear models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1;
see Table F-ll). and thus should not be considered further for identification of a POD. The BMDL
estimates for all appropriately fitting models were within threefold difference of each other, indicating
that BMDL selection should be made based on model fit (U.S. EPA. 2012b). As assessed by the AIC, the
Log-logistic model provided the best fit to the atrophy data for male rats (Table F-ll. Figure F-5). and
could be used to derive a POD for this endpoint. However, given that first non-control dose had a
response level substantially above the desired BMR (i.e., 10%), the use of BMD methods included
substantial model uncertainty.
F-18
-------
Table F-ll Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for atrophy of olfactory epithelium in male F344/DuCrj rats
(Kasai et al., 2009) exposed to 1,4-dioxane vapors
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gammab
Logistic
Log-logistic0'6
Log-probitc
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hillc
159.444
190.692
93.9074
117.337
159.444
200.626
159.444
159.444
95.5314
0
0
0.3023
0
0
0
0
0
1
0
4.342
0
0
0
3.943
0
0
0
9.93187
33.9373
1.67195
9.42745
9.9319
61.9146
9.9319
9.9319
2.93951
8.14152
25.4454
1.01633
7.20318
8.14152
47.107
8.14152
8.14152
0.544697
a p-Value from the x 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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest AIC.
Data from Kasai et al. (2009).
F-19
-------
Log-Logistic Model with 0.95 Confidence Level
T3
ill
•S
t
C
o
^
\L
0.8
0.6
0.4
0.2
Log-Logistic
EMDL3MD
200
400
600
800
1000
1200
09:53 01/14 2011
Data points obtained from Kasai et al. (2009).
Figure F-5. BMD Log-Logistic model of atrophy of olfactory epithelium incidence data
for male rats exposed to 1,4-dioxane vapors for 2 years.
Logistic Model. (Version: 2.13; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl_atrophy_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl_atrophy_Lnl-BMR10-Restrict.plt
Fri Jan 14 09:53:22 2011
BMDS_Model_Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*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
intercept = -3.48908
slope = 1
F-20
-------
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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 -2.71122 * * *
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 -44.7657 4
Fitted model -45.9537 1 2.37596 3 0.4981
Reduced model -126.116 1 162.701 3 <.0001
AIC: 93.9074
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 50 0.000
50.0000 0.7687 38.433 40.000 50 0.525
250.0000 0.9432 47.161 47.000 50 -0.099
1250.0000 0.9881 49.405 48.000 50 -1.833
Chi^2 = 3.65 d.f. = 3 P-value = 0.3023
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1. 67195
BMDL = 1.01633
F-21
-------
F.6. Hydropic Change
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-12. for hydropic change of the lamina propria in the nasal cavity of male
F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (Kasai et al. 2009). Doses associated with a
BMR of a 10% extra risk were calculated.
Table F-12 Incidence of hydropic change of the lamina propria in the nasal cavity of
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50
2/50
(4%)
250
36/50 a
(72%)
1,250
49/50a
(98%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al., (2009).
For incidence of hydropic change of the lamina propria in F344/DuCrj male rats, the gamma,
logistic, multistage, probit, Weibull, and quantal-linear models all exhibited a statistically significant lack
of fit (i.e., x2/"-value < 0.1; see Table F-13). and thus should not be considered further for identification
of a POD. The BMDL estimates for all appropriately fitting models were within threefold difference
of each other, indicating that BMDL selection should be made based on model fit (U.S. EPA. 2012b).
As assessed by the AIC, the log-logistic model provided the best fit to the hydropic change of the lamina
propria data for male rats (Table F-13. Figure F-6 and subsequent text output), and could be used to
derive a POD of for this endpoint.
F-22
-------
Table F-13 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for hydropic change of the lamina propria in the nasal cavity of
male F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai et al., 2009)
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gammab
Logistic
Log-logistic0'6
Log-probitc
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hillc
98.3441
117.957
90.5388
91.5881
99.3482
136.585
100.225
99.3482
91.8937
0.0002
0
0.6819
0.3458
0.0256
0
0.0033
0.0256
1
-1.321
-1.143
-0.333
-0.538
-2.411
-2.099
-1.899
-2.411
0
51.979
89.2909
68.5266
63.0852
28.7899
92.6118
39.1371
28.7899
73.1032
28.7632
70.6131
46.7808
44.5657
22.6831
74.3784
23.9762
22.6831
49.2687
ap-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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest AIC.
Data from Kasai et al. (2009).
F-23
-------
Log-Logistic Model with 0.95 Confidence Level
o
2
0.8
0.6
0.4
0.2
Log-Logistic
BMDL
BMD
0 200
10:3001/142011
Data points obtained from Kasai et al. (2009).
400
600
dose
800
1000
1200
Figure F-6. BMD Log-logistic model of hydropic change of lamina propria (nasal
cavity) incidence data for male rats exposed to 1,4-dioxane vapors for
2 years.
Logistic Model. (Version: 2.13; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl_hydrpic_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl_hydrpic_Lnl-BMR10-Restrict.plt
Fri Jan 14 10:30:47 2011
BMDS_Model_Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*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
intercept = -11.5745
slope = 2.19638
F-24
-------
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
intercept slope
intercept 1 -0.99
slope -0.99 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
background 0 * * *
intercept -12.1316 * * *
slope 2.3501 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -42.9468 4
Fitted model -43.2694 2 0.645129 2 0.7243
Reduced model -136.935 1 187.976 3 <.0001
AIC: 90.5388
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 50 0.000
50.0000 0.0503 2.515 2.000 50 -0.333
250.0000 0.6994 34.969 36.000 50 0.318
1250.0000 0.9903 49.515 49.000 50 -0.744
ChiA2 = 0.77 d.f. = 2 P-value = 0.6819
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 68.5266
BMDL = 46.7808
F-25
-------
F.7. Sclerosis
All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
incidence data shown in Table F-14. for sclerosis of the lamina propria in the nasal cavity of male
F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (Kasai et al. 2009). Doses associated with a
BMR of a 10% extra risk were calculated.
Table F-14 Incidence of sclerosis of the lamina propria in the nasal cavity of F344/DuCrj
rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50
0/50
250
22/50 a
(44%)
1,250
40/50a
(80%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
As assessed by the %2 goodness-of-fit test, all models with the exception of the dichotomous-hill
model, exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1; see Table F-15). and thus
should not be considered further for identification of a POD. Since the dichotomous-hill model provided
the only fit to the sclerosis of the lamina propria data for male rats as assessed by the %2 goodness-of-fit
test (Table F-15. Figure F-7 and subsequent text output), it could be considered to derive a POD for this
endpoint; however, the model output warned that the BMDL estimate was "imprecise at best".
F-26
-------
Table F-15 Goodness-of-fit statistics and BMDio and BMDLio values from models fit to
incidence data for sclerosis of the lamina propria in the nasal cavity of male
F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai et al., 2009)
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibullb
Quantal-Linear
Dichotomous-Hillc'e
134.416
161.562
130.24
127.784
132.436
159.896
132.436
132.436
124.633
0.0123
0
0.0683
0.0829
0.0356
0
0.0356
0.0356
0.9994
-1.89
4.542
-1.579
-0.995
-1.949
4.619
-1.949
-1.949
0
75.4489
244.217
86.3863
109.558
71.9719
231.856
71.9719
71.9719
206.74
57.6938
196.446
52.4762
88.1232
57.6471
191.419
57.6471
57.6471
167.46
ap-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 a 1.
°Slope restricted to > 1.
dBetas restricted to > 0.
eModel output warned that the BMDL estimate was "imprecise at best".
Data from Kasai et al. (2009).
F-27
-------
Dichotomous-Hill Model with 0.95 Confidence Level
T3
"O
<
d
O
0.8
0.6
0.4
0.2
Dichotomous-Hill
BMDL BMD
0 200
10:53 01/14 2011
Data points obtained from Kasai et al. (2009).
400
600
dose
800
1000
1200
Figure F-7. BMD Log-logistic model of sclerosis of lamina propria (nasal cavity)
incidence data for male rats exposed to 1,4-dioxane vapors for 2 years.
Dichotomous Hill Model. (Version: 1.2; Date: 12/11/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl_sclerosis_Dhl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl_sclerosis_Dhl-BMR10-Restrict.plt
Fri Jan 14 10:53:28 2011
BMDS_Model_Run
The form of the probability function is:
P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
where: 0 <= g < 1, 0= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
v = -9999
g = -9999
intercept = -11.4511
F-28
-------
slope = 1.86444
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -g have been estimated at a boundary point, or have been
specified by the user, and do not appear in the correlation matrix)
v intercept slope
v 1 0.00074 -0.00078
intercept 0.00074 1 -1
slope -0.00078 -1 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
v 0.8 0.0565686 0.689128 0.910872
g 0 NA
intercept -62.1804 4133.38 -8163.46 8039.1
slope 11.2979 748.603 -1455.94 1478.53
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -59.3166 4
Fitted model -59.3166 3 1.23973e-006 1 0.9991
Reduced model -123.82 1 129.007 3 <.0001
AIC: 124.633
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 50 0.000
50.0000 0.0000 0.000 0.000 50 -0.001
250.0000 0.4400 22.000 22.000 50 0.000
1250.0000 0.8000 40.000 40.000 50 -0.000
ChiA2 = 0.00 d.f. = 1 P-value = 0.9994
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 206.74
Warning: BMDL computation is at best imprecise for these data
BMDL = 167.46
F-29
-------
APPENDIX G. DETAILS OF BMD ANALYSIS FOR
INHALATION UNIT RISK FOR 1,4-DIOXANE
Multistage cancer models available in the Benchmark Dose Software (BMDS) (version 2.2beta)
were fit to the incidence data for hepatocellular carcinoma and/or adenoma, nasal cavity squamous cell
carcinoma, renal cell carcinoma, peritoneal mesothelioma, and mammary gland fibroadenoma, Zymbal
gland adenoma, and subcutis fibroma in rats exposed to 1,4-dioxane vapors for 2 years (Kasai et al..
2009). Concentrations associated with a benchmark response (BMR) of a 10% extra risk were calculated.
BMCio and BMCLio values from the best fitting model, determined by adequate global- fit (y£p > 0.1)
and AIC values, are reported for each endpoint (U.S. EPA. 2012b). Given the multiplicity of tumor sites,
basing the IUR on one tumor site will underestimate the carcinogenic potential of 1,4-dioxane.
Multitumor BMD analysis was conducted using BMDS (version 2.2beta) MS_Combo program; model
output is shown in Section G.3. Additionally, a Bayesian analysis was performed using WinBUGS
(Spiegelhalter et al.. 2003). freeware developed by the MRC Biostatistical Unit, Cambridge, United
Kingdom (available at http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml) and reported in
detail in Section G.3. The results of both analyses were comparable and resulted in equivalent lURs.
A summary of the BMDS model predictions for the Kasai et al. (2009) study are shown in
Table G-l.
G.1. General Issues and Approaches to BMDS and Multitumor
Modeling
G.1.1. Combining Data tumor types
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 may
develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a: McConnell et al..
1986). In the same manner and was done for the oral cancer assessment (Appendix D). 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.
The remaining of the tumor types were assumed to occur independently.
G.1.2. Summary
The BMDS models recommended to calculate rodent BMCio and BMCLio values for individual
tumor types and combined tumor analysis are summarized in Table G-l. The first order multistage models
for most tumor types were selected because they resulted in the lowest AIC values; however, for renal cell
G-l
-------
carcinoma and Zymbal gland adenoma, the lowest AIC model was not the first order model. In BMDS,
the third order model resulted in the lowest AIC (first (1°)-, second (2°)-, and third (3°)-degree models
were evaluated); however, using the MCMC approach in WinBUGS, the third order (3°) multistage
model did not converge while the second order(2°) model did converge. Thus, for renal cell carcinoma
and Zymbal gland adenoma, the second order (2°) multistage model was used in both the MCMC
(WinBugs) approach and the BMDS (Version 2.2 beta) MS_Combo approach for direct comparison of
results. These results are shown below in Table G-l.
Table G-l Summary of BMCio and BMCLio model results for individual tumor types and
combined tumor analysis for male rats exposed to 1,4-dioxane vapors (Kasai et
al., 2009)
Endpoint
Nasal squamous cell carcinoma
Hepatocellular
adenoma/carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Zymbal gland adenoma
Subcutis fibroma3
BMDS Version 2.2beta MS_Combo
WinBUGS multitumor analysis'3
aHigh-dose dropped. See Section G.2.6
Multistage
Model
Degree
First (1°)
First (1°)
Third (3°)
First (1°)
First (1°)
Third (3°)
First (1°)
for details.
AIC
49.03
127.9
29.99
155.4
86.29
29.99
89.2
p-value
0.9607
0.6928
0.9984
0.8509
0.7904
0.9984
0.5245
X2 Residual
of Interest
0.176
-0.763
0.017
-0.204
-0.149
0.017
0.537
BMC10
(ppm)
1,107.04
252.80
1,355.16
82.21
1,635.46
1,355.16
141.762
40.4
39.2
BMCL10
(ppm)
629.95
182.26
16.15
64.38
703.03
16.15
81.9117
30.3
31.4
bln MCMC approach, the simulations for the four-parameter third order(3°) multistage model did not converge for renal cell
carcinomas and Zymbal gland adenomas. Second order (2°) multistage model was used instead.
Data from Kasai et al. (2009).
G.2. BMDS Model Output for Multistage Cancer Models for Individual
Tumor Types
For tumor incidence data reported in the Kasai et al. (2009) 2-year inhalation bioassay, multistage
cancer models of first (1°)-, second (2°)-, and third (3°)degrees were implemented BMDS (Version
2.2Beta). Incidence data used for BMD analysis are shown in Table G-2. Tumor incidence for mammary
gland adenoma was excluded from this analysis since only 1 tumor of this type was found across all
doses.
G-2
-------
Table G-2 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane vapor by
whole-body inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
Effect
Nasal squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Zymbal gland adenoma
Subcutis fibroma
0 (clean air)
0/50
1/50
0/50
1/50
0/50
2/50
1/50
0/50
1/50
50
0/50
2/50
0/50
2/50
0/50
4/50
2/50
0/50
4/50
250
1/50
3/50
1/50
4/50
0/50
14/50a
3/50
0/50
9/50a
1,250
6/50b'c
21/50a'c
2/50
23/50a'c
4/50c
41/50a'c
5/50d
4/50c
5/50
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
°p < 0.01 by Peto's test for dose-related trend.
dp < 0.05 by Peto's test for dose-related trend.
Provided via email from Dr. Tatsuya Kasaito (JBRC) Dr. Reeder Sams (U.S. EPA) on 12/23/2008 (2008).
Statistics were not reported for these data by study authors, so statistical analyses were conducted by EPA.
Source: Reprinted with permission of Informa Healthcare; Kasai et al. (2009) and Kasai (2008)
G.2.1. Nasal Squamous Cell Carcinoma
The incidence data for nasal squamous cell carcinoma were monotonic non-decreasing functions
of dose; therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
BMDS modeling for the multistage cancer model for first (1°)-, second (2°)-, and third (3°)-degree
polynomials are shown in Table G-3. The first (l°)-degree polynomial was the best fitting model based on
AIC. The plot (Figure G-l) and model output for the first (l°)-degree model are shown below.
Table G-3 BMDS Multistage cancer dose-response modeling results for the incidence of
nasal squamous cell carcinomas in male rats exposed to 1,4-dioxane vapors for
2-years (Kasai et al., 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
49.0308
50.8278
50.8278
p-value
0.9607
0.9087
0.9087
X2 Residual of
Interest
0.176
-0.021
-0.021
BMC10
(ppm)
1,107.04
1,086.94
1,086.94
BMCL10
(ppm)
629.95
642.43
642.43
"Best-fitting model based on AIC.
Data from Kasai et al. (2009).
G-3
-------
Multistage Cancer Model with 0.95 Confidence Level
o
'•G
ro
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMD
200 400 600 800
dose
1000 1200
10:26 11/172010
Data points obtained from Kasai et al. (2009).
Figure G-l. Multistage model (First (l°)-degree) for male rat nasal squamous cell
carcinomas.
MS_COMBO. (Version: 1.4; Date: 10/20/2010)
Input Data File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.(d)
Gnuplot Plotting File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.pit
Wed Nov 17 10:57:55 2010
BMDS_Model_Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]
The parameter betas are restricted to be positive
Dependent variable = 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
G-4
-------
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0.000104666
Asymptotic Correlation Matrix of Parameter Estimates
(***The model parameter(s) -Background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix )
Beta (1)
Beta(l) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 9.51733e-005 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -23.2482 4
Fitted model -23.5154 1 0.534383 3 0.9113
Reduced model -30.3429 1 14.1894 3 0.002658
AIC: 49.0308
Log-likelihood Constant 20.493267595834471
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0 50 0.000
50.0000 0.0047 0.237 0 50 -0.488
250.0000 0.0235 1.176 1 50 -0.164
1,250.0000 0.1122 5.608 6 50 0.176
Chi^2 = 0.30 d.f. = 3 P-value = 0.9607
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1107.04
BMDL = 629.948
BMDU = 2215.11
Taken together, (629.948, 2215.11) is a 90% two-sided confidence interval for the BMD
G-5
-------
G.2.2. Hepatocellular Adenoma and Carcinoma
The incidence data for the occurrence of either hepatocellular adenoma or carcinoma were
combined for this analysis as explained in Section G. 1.1. The incidence data were monotonic
non-decreasing functions of dose; therefore, these data are appropriate for dose-response modeling using
BMDS. The results of the BMDS modeling for the multistage cancer model for first-, second-, and third-
degree polynomials are shown in Table G-4. The Ist-degree polynomial was the best fitting model based
on AIC. The plot (Figure G-2) and model output for the Ist-degree model are shown below.
Table G-4 BMDS Multistage cancer dose-response modeling results for the incidence of
either hepatocellular adenoma or carcinoma in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai et al., 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
127.86
129.157
129.131
p-value
0.6928
0.7636
0.8
X2 Residual of
Interest
-0.763
-0.094
-0.068
BMCio
(ppm)
252.80
377.16
397.426
BMCLio
(ppm)
182.26
190.28
190.609
aBest-fitting model based on AIC.
Data from Kasai et al. (2009).
G-6
-------
Multistage Cancer Model with 0.95 Confidence Level
o
'•G
ro
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
200 400 600 800
dose
1000 1200
10:24 11/172010
Data points obtained from Kasai et al. (2009).
Figure G-2. Multistage model (First-degree (1°)) for male rat hepatocellular adenomas
and carcinomas.
MS_COMBO. (Version: 1.4; Date: 10/20/2010)
Input Data File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.(d)
Gnuplot Plotting File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.pit
Wed Nov 17 10:57:55 2010
BMDS_Model_Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = 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
G-7
-------
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 = 0.00480969
Beta(l) = 0.0004548
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.53
Beta(l) -0.53 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0170678 * * *
Beta(l) 0.000416776 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -61.5341 4
Fitted model -61.9302 2 0.792109 2 0.673
Reduced model -82.7874 1 42.5066 3 <.0001
AIC: 127.86
Log-likelihood Constant 55.486699676972215
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0171 0.853 1 50 0.160
50.0000 0.0373 1.867 2 50 0.099
250.0000 0.1143 5.716 4 50 -0.763
1,250.0000 0.4162 20.810 22 50 0.342
Chi^2 = 0.73 d.f. = 2 P-value = 0.6928
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 252.799
BMDL = 182.256
BMDU = 371.457
Taken together, (182.256, 371.457) is a 90% two-sided confidence interval for the BMD
G-8
-------
G.2.3. Renal Cell Carcinoma and Zymbal Gland Adenoma
The incidence data for renal cell carcinomas and Zymbal gland adenomas were the same. These
data were monotonic non-decreasing functions of dose; therefore, these data are appropriate for
dose-response modeling using BMDS. The results of the BMDS modeling for the multistage cancer
model for first (1°)-, second (2°)- and third-degree (3°) polynomials are shown in Table G-5. The
third-degree (3°) polynomial was the best fitting model based on AIC; however, when conducting the
multitumor analysis, WinBUGS was unable to converge using the third-degree (3°) model. Thus, the
second degree (2°) model was used in the multitumor analyses. The plots (Figure G-3 and Figure G-4)
and model outputs for both the second (2°)- and third-degree (3°) models are shown below.
Table G-5 BMDS Multistage cancer dose-response modeling results for the incidence of
renal cell carcinomas and Zymbal gland adenomas in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai et al., 2009)
X2 Residual of BMCio BMCLio
Polynomial Degree AIC p-value Interest (ppm) (ppm)
(1°) First 31.6629 0.8004 0.446 1,974.78 957.63
(2°) Second 30.2165 0.9817 0.085 1,435.28 999.44
(3°) Third3 29.9439 0.9984 0.017 1,355.16 1,016.15
"Best-fitting model based on AIC.
Data from Kasai et al. (2009).
G-9
-------
Multistage Cancer Model with 0.95 Confidence Level
0.2
0.15
T3
£
o
c
o
'
o
(0
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMID
0 200
10:1702/102011
Data points obtained from Kasai et al. (2009).
400
600 800
dose
1000
1200
1400
Figure G-3. Multistage model (Second-degree (2°)) for male rat renal cell carcinomas
and Zymbal gland adenomas.
Multistage Cancer Model. (Version: 1.9; Date: 05/26/2010)
Input Data File: C:/Documents and
Settings/emclanah/Desktop/BMD_14D_Cancer/Data/msc_Kasai2009_renal_Msc2-BMR10.(d)
Gnuplot Plotting File: C:/Documents and
Settings/emclanah/Desktop/BMD_14D_Cancer/Data/msc_Kasai2009_renal_Msc2-BMR10.plt
Thu Feb 10 10:17:39 2011
BMDS_Model_Run
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2) ]
The parameter betas are restricted to be positive
Dependent variable = 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
G-10
-------
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 5.40386e-008
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background -Beta(l) have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
Beta (2)
Beta(2) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 0 * * *
Beta(2) 5.11454e-008 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -13.9385 4
Fitted model -14.1082 1 0.339554 3 0.9524
Reduced model -19.6078 1 11.3387 3 0.01003
AIC: 30.2165
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 50 0.000
50.0000 0.0001 0.006 0.000 50 -0.080
250.0000 0.0032 0.160 0.000 50 -0.400
1250.0000 0.0768 3.840 4.000 50 0.085
ChiA2 = 0.17 d.f. = 3 P-value = 0.9817
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1,435.28
BMDL = 999.44
BMDU = 3,666.87
Taken together, (999.44 , 3,666.87) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000100056
G-ll
-------
Multistage Cancer Model with 0.95 Confidence Level
0.2
0.15
0.1
o
'•G
ro
0.05
Multistage Cancer
Linear extrapolation
BMID
200
400
600
800
1000
1200
1400
dose
10:28 11/172010
Data points obtained from Kasai et al. (2009).
Figure G-4. Multistage model (Third-degree (3°)) for male rat renal cell carcinomas.
MS_COMBO. (Version: 1.4; Date: 10/20/2010)
Input Data File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.(d)
Gnuplot Plotting File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.pit
Wed Nov 17 10:57:55 2010
BMDS_Model_Run
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-
beta3*doseA3) ]
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 = 250
G-12
-------
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
Beta(3) = 4.2804e-011
Asymptotic Correlation Matrix of Parameter Estimates
(*** The 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(l) 0 * * *
Beta(2) 0 * * *
Beta(3) 4.23353e-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 -13.9385 4
Fitted model -13.9719 1 0.0669578 3 0.9955
Reduced model -19.6078 1 11.3387 3 0.01003
AIC: 29.9439
Log-likelihood Constant 12.347138085809094
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0 50 0.000
50.0000 0.0000 0.000 0 50 -0.016
250.0000 0.0007 0.033 0 50 -0.182
1250.0000 0.0794 3.968 4 50 0.017
Chi^2 = 0.03 d.f. = 3 P-value = 0.9984
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1,355.16
BMDL = 1,016.15
BMDU = 3,393.6
Taken together, (1016.15, 3393.6 ) is a 90% two-sided confidence interval for the BMD
G-13
-------
G.2.4. Peritoneal Mesothelioma
The incidence data for peritoneal mesotheliomas were monotonic non-decreasing functions of
dose; therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
BMDS modeling for the multistage cancer model for 1st, 2nd, and 3rd-degree polynomials are shown in
Table G-6. The Ist-degree polynomial was the best fitting model based on AIC. The plot (Figure G-5)
and model output for the Ist-degree model are shown below.
Table G-6 BMDS Multistage cancer dose-response modeling results for the incidence of
peritoneal mesothelioma in male rats exposed to 1,4-dioxane vapors for 2-years
(Kasai et al., 2009)
X2 Residual of BMCio BMCLio
Polynomial Degree AIC p-value Interest (ppm) (ppm)
(1°) First3 155.433 0.8509 -0.204 82.21 64.38
(2°) Second 157.168 0.8053 -0.204 96.23 65.15
(3°) Third 157.168 0.8053 0 96.23 65.15
a Best-fitting model based on AIC.
Data from Kasai et al. (2009).
G-14
-------
Multistage Cancer Model with 0.95 Confidence Level
o
'•G
ro
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
BMDL BMD
200
400
600
dose
800
1000
1200
10:31 11/172010
Data points obtained from Kasai et al. (2009).
Figure G-5. Multistage model (First-degree (1°)) for male rat peritoneal mesotheliomas.
MS_COMBO. (Version: 1.4; Date: 10/20/2010)
Input Data File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.(d)
Gnuplot Plotting File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.pit
Wed Nov 17 10:57:55 2010
BMDS_Model_Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = 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 = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
G-15
-------
Default Initial Parameter Values
Background = 0.0172414
Beta(l) = 0.00135351
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.45
Beta(l) -0.45 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.033631 * * *
Beta(l) 0.00128167 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -75.553 4
Fitted model -75.7165 2 0.326905 2 0.8492
Reduced model -123.008 1 94.9105 3 <.0001
AIC: 155.433
Log-likelihood Constant 68.666413125908832
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0336 1.682 2 50 0.250
50.0000 0.0936 4.681 4 50 -0.331
250.0000 0.2986 14.928 14 50 -0.287
1,250.0000 0.8053 40.265 41 50 0.263
Chi^2 = 0.32 d.f. = 2 P-value = 0.8509
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 82.2057
BMDL = 64.3808
BMDU = 107.497
Taken together, (64.3808, 107.497) is a 90% two-sided confidence interval for the BMD
G.2.5. Mammary Gland Fibroadenoma
The incidence data for mammary gland fibroadenomas were monotonic non-decreasing functions
of dose; therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
BMDS modeling for the multistage cancer model for first (1°)-, second (2°), and third (3°)-degree
polynomials are shown in Table G-7. Since quadratic and cubic terms of the multistage models evaluated
resulted in the estimates on the boundary, i.e., equal to 0, the first (l°)-degree polynomial was selected
G-16
-------
based on model parsimony. The plot (Figure G-6) and model output for the first (l°)-degree model are
shown below.
Table G-7 BMDS Multistage cancer dose-response modeling results for the incidence of
mammary gland fibroadenoma in male rats exposed to 1,4-dioxane vapors for
2-years (Kasai et al., 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
86.29
86.29
86.29
p-value
0.7904
0.7904
0.7904
X2 Residual of
Interest
-0.149
-0.149
-0.149
BMC10
(ppm)
1,635.46
1,635.46
1,635.46
BMCL10
(ppm)
703.03
703.03
703.03
aAII model fits were equivalent based on AIC. Selected 1st-degree model based on parsimony.
Source: Kasai et al. (2009).
Multistage Cancer Model with 0.95 Confidence Level
I
C
o
'
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMID
0 200
10:34 11/172010
Data points obtained from Kasai et al. (2009).
400
600
800
dose
1000 1200 1400 1600
Figure G-6. Multistage model (First-degree (1°)) for male rat mammary gland
fibroadenoma.
G-17
-------
MS_COMBO. (Version: 1.4; Date: 10/20/2010)
Input Data File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.(d)
Gnuplot Plotting File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.pit
Wed Nov 17 10:57:55 2010
BMDS_Model_Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = 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 = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0335609
Beta(l) = 5.91694e-005
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.61
Beta(l) -0.61 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0315836 * * *
Beta(l) 6.44224e-005 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -40.9017 4
Fitted model -41.145 2 0.486662 2 0.784
Reduced model -42.5964 1 3.3895 3 0.3354
AIC: 86.29
Log-likelihood Constant 35.472345543489602
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0316 1.579 1 50 -0.468
50.0000 0.0347 1.735 2 50 0.205
250.0000 0.0471 2.353 3 50 0.432
1,250.0000 0.1065 5.326 5 50 -0.149
G-18
-------
= 0.47 d.f. = 2 P-value = 0.7904
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1,635.46
BMDL = 703.034
BMDU = 1.9523e+009
Taken together, (703.034, 1.9523e+009) is a 90% two-sided confidence interval for the
BMD
G-19
-------
G.2.6. Subcutis Fibroma
The incidence data for subcutis fibroma were monotonic non-decreasing functions of dose for the
control (0 ppm), low (50 ppm), and mid-dose (250 ppm); however, the incidence rate at the high dose
(1,250 ppm) was lower than observed at the mid-dose. No BMDS model had reasonable fit to the data
without dropping the high dose. The results of the BMDS modeling for the multistage cancer model for
first (1°)-, second (2°), and third (3°)-degree polynomials with the high dose dropped are shown in
Table G-8. Since quadratic and cubic terms of multistage models evaluated resulted in the estimates on
the boundary, i.e., equal to 0,, the first (l°)-degree polynomial was selected based on model parsimony.
The plot (Figure G-7) and model output for the first (l°)-degree model are shown below.
Table G-8 BMDS Multistage cancer dose-response modeling results for the incidence of
subcutis fibromas in male rats exposed to 1,4-dioxane vapors for 2-years (Kasai
et al., 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
89.2094
89.2094
89.2094
p-value
0.5245
0.5245
0.5245
X2 Residual of
Interest
0.537
0.537
0.537
BMCio
(ppm)
141.76
141.76
141.76
BMCLio
(ppm)
81.92
81.92
81.92
aAII model fits were equivalent based on AIC. Selected 1st-degree model based on parsimony.
Data from Kasai et al. (2009).
G-20
-------
Multistage Cancer Model with 0.95 Confidence Level
o
'•G
ro
0.35
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
0 -
100 150
dose
200
250
10:56 11/172010
Data points obtained from Kasai et al. (2009).
Figure G-7. Multistage model (First-degree (1°)) for male rat subcutis fibroma (high
dose dropped).
MS_COMBO. (Version: 1.4; Date: 10/20/2010)
Input Data File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.(d)
Gnuplot Plotting File: C:\Documents and
Settings\emclanah\Desktop\BMD_14D_Cancer\Data\New.pit
Wed Nov 17 10:57:55 2010
BMDS_Model_Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl) ]
The parameter betas are restricted to be positive
Dependent variable = 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
G-21
-------
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 = 0.0327631
Beta(l) = 0.000673665
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.68
Beta(l) -0.68 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0262054 * * *
Beta(l) 0.00074322 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -42.4101 3
Fitted model -42.6047 2 0.389155 1 0.5327
Reduced model -46.5274 1 8.23466 2 0.01629
AIC: 89.2094
Log-likelihood Constant 37.900888781466982
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0262 1.310 1 50 -0.275
50.0000 0.0617 3.086 4 50 0.537
250.0000 0.1913 9.566 9 50 -0.204
Chi^2 = 0.41 d.f. = 1 P-value = 0.5245
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 141.762
BMDL = 81.9117
BMDU = 364.364
Taken together, (81.9117, 364.364) is a 90% two-sided confidence interval for the BMD
G-22
-------
G.3. Multitumor Analysis Using BMDS MS_Combo
The combined tumor analysis was also performed with beta version of the MS_Combo model in
BMDS (Version 2.2beta). The model resulted in similar results to the Bayesian method and model output
is shown below for the combined calculation.
**** Start of combined BMD and BMDL Calculations.****
Combined Log-Likelihood -277.79874987953076
Combined Log-likelihood Constant 246.62591390071873
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 40.4937
BMDL = 32.331
G.4. Multitumor analysis using Bayesian Methods
Given the multiplicity of tumor sites, basing the IUR on one tumor site will likely underestimate
the carcinogenic potential of 1,4-dioxane. Simply pooling the counts of animals with one or more tumors
(i.e., counts of tumor bearing animals) would tend to underestimate the overall risk when tumors are
independent across sites and ignores potential differences in the dose-response relationships across the
sites (NRC, 1994; Bogen. 1990). NRC (1994) also noted that the assumption of independence across
tumor types is not likely to produce substantial error in the risk estimates unless tumors are known to be
biologically dependent.
Kopylev et al. (2009) describe a Markov Chain Monte Caro (MCMC) computational approach to
calculating the dose associated with a specified composite risk under assumption of independence of
tumors. The current Guidelines for Carcinogen Risk Assessment recommend calculation of an upper
bound to account for uncertainty in the estimate (U.S. EPA. 2005a). For uncertainty characterization,
MCMC methods have the advantage of providing information about the full distribution of risk and/or
benchmark dose, which can be used in generating a confidence bound. This MCMC approach building on
the re-sampling approach recommended by Bogen (1990). and also provides a distribution of the
combined potency across sites.
For individual tumor data modeled using the multistage model:
P(d | q) = 1 - exp[-(q0 + qid + q2d2 + ... + qkdk)], q; > 0
the model for the combined tumor risk is still multistage, with a functional form that has the sum of
stage-specific multistage coefficients as the corresponding multistage coefficient;
Pc(d | q) = 1 - exp[-(qsoi + qsnd + qS2id2 + ... + qskidk)],
G-23
-------
the resulting equation for fixed extra risk (BMR) is polynomial in dose (when logarithms of both sides are
taken) and can be straightforwardly solved for a combined BMC. Computation of the confidence bound
on combined risk BMC can be accomplished via likelihood methods (BMDS-MS_Combo), re-sampling
(bootstrap) or Bayesian methods.
The MCMC computations were conducted using WinBUGS (Spiegelhalter et al.. 2003) (freeware
developed by the MRC Biostatistical Unit, Cambridge, United Kingdom, available at
http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml). The model code was checked and
compiled within, and the data read into, WinBUGS. Three chains were used for the analysis. Initial values
for each variable were generated using a Uniform (0, 1) distribution and read into WinBUGS. The
WinBUGS code calculates the BMC directly (U.S. EPA. 2013d).
In a Bayesian analysis, the choice of an appropriate prior probability is important. In the
examples developed by Kopylev et al. (2009). a diffuse (i.e., high variance or low tolerance) Gaussian
prior restricted to be nonnegative was used; such diffuse priors performed reasonably well.
The mean and the 5th percentile of the posterior distribution of combined BMC provide estimates
of the mean BMC and the lower bound on the BMC (BMCL), respectively, for the combined tumor risk.
The values calculated using this method were: mean BMCio 39.2 ppm, and BMCLio 31.4 ppm.
G-24
-------
APPENDIX H. ASSESSMENTS BY OTHER
NATIONAL AND INTERNATIONAL HEALTH
AGENCIES
Table H-l Health assessments, guideline levels, and regulatory limits by other national
and international agencies
Organization
Toxicity Value or Determination
Noncancer: oral values
ATSDR (2012)
An acute oral minimum risk level (MRL) of 5 mg/kg-day was derived from a no-
observed-adverse-effect level (NOAEL) of 516 mg/kg-day for developmental and
maternal effects in rats from Giavini et al. (1985) and using an uncertainty factor of 100.
An intermediate oral MRL of 0.5 mg/kg-day was derived from a NOAEL of 52 mg/kg-
day for liver effects in rats from Kano et al. (2008) and using an uncertainty factor of
100.
A chronic oral MRL of 0.1 mg/kg-day was derived from a NOAEL of 9.6 mg/kg-day for
liver effects in rats from Kociba et al. (1974) and using an uncertainty factor of 100.
Noncancer: inhalation values
ATSDR (2012)
ACGIH (2011)
NIOSH (2010)
OSHA (2004a. b, c)
CalEPA (2008)
CalEPA (2000)
An acute inhalation MRL of 2 ppm was derived from a NOAEL of 20 ppm for eye and
respiratory irritation and pulmonary function effects in humans from Ernstgard et al.
(2006) using an uncertainty factor of 10.
An intermediate inhalation MRL of 0.2 ppm was derived from a Benchmark
Concentration (BMCL-io) of 27.99 ppm (subsequently adjusted for duration) for
increased incidence of nasal lesions in rats from Kasai et al. (2008) using an
uncertainty factor of 30.
A chronic inhalation MRL of 0.03 ppm was derived from a lowest-observed-adverse-
effect level (LOAEL) of 50 ppm (subsequently adjusted for duration) for increased
incidence of nasal lesions in rats from Kasai et al. (2009) using an uncertainty factor of
300.
Threshold limit value (TLV) time weighted average (TWA) of 20 ppm
Reference exposure level (REL) (30-minute ceiling TWA) 1 ppm
Immediately dangerous to life and health (IDLH) 500 ppm
Permissible exposure limit (PEL) (8-hour TWA) for general industry 100 ppm
PEL (8-hour TWA) for construction industry 100 ppm
PEL (8-hour TWA) for shipyard industry 100 ppm
Acute REL = 3,000 ug/m3 (0.8 ppm) based on respiratory and eye irritation in humans
(Younqetal., 1977)
Chronic REL = 3,000 mg/m3 (0.8 ppm), based on liver, kidney, and hematologic
chanqes in rats (Torkelson et al., 1974).
H-l
-------
Table H-l (Continued): Health assessments, guideline levels, and regulatory limits by other
national and international agencies
Organization
Toxicity Value or Determination
Cancer characterization
IARC (1999)
NIOSH (2004)
NTP (2011)
CalEPA (2013)
ACGIH (2011)
Possibly carcinogenic to humans (Group 2B) (based on inadequate evidence in
humans and sufficient evidence in experimental animals)
Potential occupational carcinogen
Reasonably anticipated to be a human carcinogen
Listed on Proposition 65 as a carcinogen
Confirmed animal carcinogen with unknown relevance to humans (Group A3)
Regulatory limits and guideline levels
MAS (2003)
FDA (2006)
California (2011)
Connecticut (2012)
Maine (2012)
Massachusetts (2012)
New Hampshire (2011)
WHO (2005)
Established a maximum specification of 10 ppm for 1,4-dioxane in the ingredient
polysorbate, a food additive.
Limited 1,4-dioxane to 10 mg/kg in approving glycerides and polyclycerides for use as
excipients in products such as dietary supplements. Regulation located in 21 CFR
172.736.
Drinking water notification level of 1 ug/L. Drinking water response level of 35 ug/L.
Drinking water action level of 3 ug/L, bathing/showering action level of 50 ug/L
Drinking water maximum exposure guideline of 4 ug/L
Drinking water guideline of 0.3 ug/L
Ambient groundwater quality standard of 3 ug/L
Drinking water guideline of 50 ug/L
ATSDR = Agency for Toxic Substances Disease Registry; ACGIH = American Conference of Industrial Hygienists; CalEPA =
California Environmental Protection Agency; IARC = International Agency for Research on Cancer; NIOSH = National Institute for
Occupational Safety and Health; OSHA = Occupational Safety and Health Administration; NAS = National Academy of Sciences;
NTP = National Toxicology Program; WHO = World Health Organization.
H-2
-------
APPENDIX I. DOCUMENTATION OF
IMPLEMENTATION OF THE 2011 NATIONAL
RESEARCH COUNCIL RECOMMENDATIONS
Background: On December 23, 2011, The Consolidated Appropriations Act, 2012, was signed
into law (U.S. Congress. 2011). The report language included direction to EPA for the Integrated Risk
Information System (IRIS) Program related to recommendations provided by the National Research
Council (NRC) in their review of EPA's draft IRIS assessment of formaldehyde (NRC. 2011). The report
language included the following:
The Agency shall incorporate, as appropriate, based on chemical-specific datasets
and biological effects, the recommendations of Chapter 7 of the National Research
Council's Review of the Environmental Protection Agency's Draft IRIS Assessment
of Formaldehyde into the IRIS process.. .For draft assessments released in fiscal year
2012, the Agency shall include documentation describing how the Chapter 7
recommendations of the National Academy of Sciences (NAS) have been
implemented or addressed, including an explanation for why certain
recommendations were not incorporated.
The NRC's recommendations, provided in Chapter 7 of their review report, offered suggestions
to EPA for improving the development of IRIS assessments. Consistent with the direction provided by
Congress, documentation of how the recommendations from Chapter 7 of the NRC report have been
implemented in this assessment is provided in the tables below. Where necessary, the documentation
includes an explanation for why certain recommendations were not incorporated.
The IRIS Program's implementation of the NRC recommendations is following a phased
approach that is consistent with the NRC's "Roadmap for Revision" as described in Chapter 7 of the
formaldehyde review report. The NRC stated that "the committee recognizes that the changes suggested
would involve a multi-year process and extensive effort by the staff at the National Center for
Environmental Assessment and input and review by the EPA Science Advisory Board and others."
Phase 1 of implementation has focused on a subset of the short-term recommendations, such as
editing and streamlining documents, increasing transparency and clarity, and using more tables, figures,
and appendices to present information and data in assessments. Phase 1 also focused on assessments near
the end of the development process and close to final posting. The 1,4-dioxane (with inhalation update)
IRIS assessment is in Phase 1 of implementation. The 2010 IRIS Toxicological Review of 1,4-Dioxane
was completed prior to the release of NRC's 2011 recommendations and, as such, does not incorporate
the recommendations. To the extent possible, the 2013 reassessment of the inhalation exposure
information has followed the Phase 1 changes. Chemical assessments in Phase 2 of the implementation
will address all of the short-term recommendations from Table 1-1. The IRIS Program is implementing all
of these recommendations but recognizes that achieving full and robust implementation of certain
recommendations will be an evolving process with input and feedback from the public, stakeholders, and
external peer review committees. Chemical assessments in Phase 3 of implementation will incorporate the
longer-term recommendations made by the NRC as outlined below in Table 1-2. including the
1-1
-------
development of a standardized approach to describe the strength of evidence for noncancer effects. On
May 16, 2012, EPA announced (U.S. EPA. 2012c) that as a part of a review of the IRIS Program's
assessment development process, the NRC will also review current methods for weight-of-evidence
analyses and recommend approaches for weighing scientific evidence for chemical hazard identification.
This effort is included in Phase 3 of EPA's implementation plan.
Table 1-1. National Research Council recommendations that EPA is implementing in the
short-term
NRC recommendations that EPA is
implementing in the short-term
Implementation in the 1,4-dioxane assessment
General recommendations for completing the IRIS formaldehyde assessment that EPA will adopt for all IRIS
assessments (see p. 152 of the NRC Report)
1. To enhance the clarity of the document, the
draft IRIS assessment needs rigorous editing to
reduce the volume of text substantially and
address redundancies and inconsistencies. Long
descriptions of particular studies should be
replaced with informative evidence tables. When
study details are appropriate, they could be
provided in appendices.
Partially Implemented. Since the inhalation assessment was
an addition to a recently peer-reviewed and finalized oral
assessment (U.S. EPA, 2010), rigorous editing and streamlining
of the original document was not performed. In order to
maintain consistency within this assessment, the new text in
support of the inhalation assessment was added in a manner
consistent with the scope, appearance, and format of the oral
assessment. However, the new inhalation information was
described and analyzed in a manner to provide transparency
without redundancy in an effort to limit the volume of text.
For example, in the new inhalation cancer assessment,
supporting evidence from the oral database was referenced
rather than repeated.
2. Chapter 1 needs to be expanded to describe
more fully the methods of the assessment,
including a description of search strategies used
to identify studies with the exclusion and inclusion
criteria articulated and a better description of the
outcomes of the searches and clear descriptions
of the weight-of-evidence approaches used for
the various noncancer outcomes. The committee
emphasizes that it is not recommending the
addition of long descriptions of EPA guidelines to
the introduction, but rather clear concise
statements of criteria used to exclude, include,
and advance studies for derivation of the RfCs and
unit risk estimates.
Partially Implemented. Additional text on the literature
search strategy used to identify health effect studies has been
added to Section 1. A link to EPA's Health and Environmental
Research Online (HERO) database (www.epa.gov/hero) that
contains the references that were cited in the document is
also provided in Section L There were a limited number of
new inhalation studies and they were all incorporated into the
assessment. Inclusion/exclusion criteria for the oral
assessment were not added as that assessment was
previously finalized as indicated above. Statements of criteria
used to exclude, include, and advance studies for derivation
of toxicity values are being developed as part of Phase 2.
3. Standardized evidence tables for all health
outcomes need to be developed. If there were
appropriates tables, long text descriptions of
studies could be moved to an appendix of
deleted.
Not Implemented. The inhalation assessment was largely
finalized before the release of the NRC recommendations,
thus development of evidence tables was not implemented as
part of Phase 1. Evidence tables will be prepared for
assessments that are part of Phase 2 of the implementation
process.
1-2
-------
NRC recommendations that EPA is
implementing in the short-term
Implementation in the 1,4-dioxane assessment
4. All critical studies need to be thoroughly
evaluated with standardized approaches that are
clearly formulated and based on the type of
research, for example, observational
epidemiologic or animal bioassays. The findings of
the reviews might be presented in tables to
ensure transparency.
Partially implemented. Standardized approaches were used
to thoroughly evaluate each potential inhalation critical study
by use of EPA guidelines. EPA guidance documents that were
used to guide the evaluation of human and animals study
were identified in Section 1. Standardized approaches for
evaluating studies are under development as part of Phases 2
and 3.
5. The rationales for the selection of the studies
that are advanced for consideration in calculating
the RfCs and unit risks need to be expanded. All
candidate RfCs should be evaluated together with
the aid of graphic displays that incorporate
selected information on attributes relevant to the
database.
Partially implemented. Section 5, the dose-response analysis
section of the document provides a clear explanation of the
rationale used to select and advance studies that were
considered for calculating toxicity values. Rationales for the
selection of studies advanced for reference value derivation
are supported by streamlined and concise text. In support of
the RfC derivations potential points of departures and
candidate RfCs are depicted in Figure 5-5.
6. Strengthened, more integrative and more
transparent discussions of weight-of-evidence are
needed. The discussions would benefit from more
rigorous and systematic coverage of the various
determinants of weight-of-evidence, such as
consistency.
Partially implemented. Weight-of-evidence tables (Table 4-27
and Table 4-28) for the temporal sequence and dose-response
relationship for possible key events for nasal and liver tumors
in rats and mice were included in the oral assessment and
updated with the data from the inhalation studies. A more
rigorous and formalized approach for developing weight of
evidence tables and characterizing the weight-of-evidence will
be completed as a part of Phase 2 and 3 of the
implementation process
General Guidance for the Overall Process (p. 164 of the NRC Report)
7. Elaborate an overall, documented, and quality-
controlled process for IRIS assessments.
8. Ensure standardization of review and
evaluation approaches among contributors and
teams of contributors; for example, include
standard approaches for reviews of various types
of studies to ensure uniformity.
9. Assess disciplinary structure of teams needed
to conduct the assessments.
Partially implemented. EPA has created Chemical Assessment
Support Teams in response to the NRC recommendations to
formalize an internal process to provide additional overall
quality control for the development of IRIS assessments. This
initiative uses a team approach to making timely, consistent
decisions about the development of IRIS assessments across
the Program. This team approach has been utilized in revising
the 1,4-dioxane assessment in response to external peer
review comments and preparing a final Toxicological Review.
Additional objectives of the teams are to help ensure that the
necessary disciplinary expertise is available for assessment
development and review, to provide a forum for identifying
and addressing key issues raised during assessment
development and review, and to monitor progress in
implementing the NRC recommendations. Further
standardization of document development and review among
contributors is ongoing as a part of Phase 2 of the
implementation process.
1-3
-------
NRC recommendations that EPA is
implementing in the short-term
Implementation in the 1,4-dioxane assessment
Evidence Identification: Literature Collection and Collation Phase (p. 164 of the NRC Report)
10. Select outcomes on the basis of available
evidence and understanding of mode of action.
11. Establish standard protocols for evidence
identification.
12. Develop a template for description of the
search approach.
13. Use a database, such as the Health and
Environmental Research Online (HERO) database,
to capture study information and relevant
quantitative data.
Partially implemented. The hazards associated with
1,4-dioxane exposure by the oral and inhalation pathways are
based on a synthesis of the available evidence; the synthesis is
presented in Sections 4.6.1 and 4.6.2 for noncancer effects
and Sections 4.7.1 and 4.7.2 for cancer. Current
understanding of the cancer mode of action is presented in
Section 4.7.3. As discussed in Section 4.7.3.7, the available
evidence in support of any hypothesized mode of action by
which 1,4-dioxane (or a transient or terminal metabolite)
induces tumors in rats and mice is not conclusive.
Each study that is cited in this document is included in the
HERO database (www.epa.gov/hero). Each citation in the
Toxicological Review is linked to HERO such that the public
can access the references and abstracts to the scientific
studies used in the assessment.
Standard protocols for evidence identification and templates
for describing the search approach are being implemented as
a part of Phase 2.
Evidence Evaluation: Hazard Identification and Dose-Response Modeling (p. 165 of the NRC Report)
14. Standardize the presentation of reviewed
studies in tabular or graphic form to capture the
key dimensions of study characteristics, weight-
of- evidence, and utility as a basis for deriving
reference values and unit risks.
15. Develop templates for evidence tables, forest
plots, or other displays.
16. Establish protocols for review of major types
of studies, such as epidemiologic and bioassay.
Partially Implemented. The use of standardized tables and
graphics will be included in assessments that are part of
Phase 2 of the implementation process. The addition of these
standardized tables and graphics was not implemented as
part of Phase 1. The Toxicological Review does provide a
graphical representation of candidate points of departure
(i.e., NOAEL, LOAEL, BMDL values) for various effects resulting
from exposure to 1,4-dioxane (Figure 5-1 through Figure 5-5).
These graphical arrays inform the identification of doses
associated with specific effects, the weight of evidence for
those effects, and the relative specie sensitivity of the effects.
Not Implemented. Evidence table templates will be included
in assessments that are part of Phase 2 of the implementation
process. The application of templates for evidence tables and
exposure-response arrays was not implemented as part of
Phase 1.
Partially implemented. Citations to EPA guidance documents
that were used to guide the review of epidemiology and
animal bioassays were included in the Toxicological Review
(e.g., in Section ].). More formalized protocols for review of
studies will be developed as a part of Phase 2.
1-4
-------
NRC recommendations that EPA is
implementing in the short-term
Implementation in the 1,4-dioxane assessment
Selection of Studies for Derivation of Reference Values and Unit Risks (p. 165 of the NRC Report)
17. Establish clear guidelines for study selection.
a. Balance strengths and weaknesses.
b. Weigh human vs. experimental evidence
c. Determine whether combining estimates
among studies is warranted.
Partially implemented. As discussed above, citations to EPA
guidance documents that were used to guide study selection,
including consideration of the strengths and weaknesses of
individual studies considered for reference value derivation,
were included in the Toxicological Review (e.g., in Section 1).
In future assessments, combining estimates across studies will
be routinely considered.
Calculation of Reference Values and Unit Risks (pp. 165-166 of the NRC Report)
18. Describe and justify assumptions and models
used. This step includes review of dosimetry
models and the implications of the models for
uncertainty factors; determination of appropriate
points of departure (such as benchmark dose, no-
observed-adverse-effect level, and lowest
observed-adverse-effect level), and assessment of
the analyses that underlie the points of departure.
Implemented as applicable.The rationale for the selection of
the point of departure for the derivation of the oral Rf D and
inhalation RfCfor 1,4-dioxane and each of the uncertainty
factors is transparently described in Sections 5.1 (RfD) and 5.2
(RfC).
19. Provide explanation of the risk-estimation
modeling processes (for example, a statistical or
biologic model fit to the data) that are used to
develop a unit risk estimate.
Implemented as applicable. The rationale for derivation of an
oral cancer slope factor based on mouse liver tumors,
including selection of the statistical model fit to the data, is
transparently described in Section 5.4. The rationale for
derivation of an inhalation unit risk based on combined tumor
analysis, including the modeling approach, is transparently
described in Section 5.4 and APPENDIX G.
20. Provide adequate documentation for
conclusions and estimation of reference values
and unit risks. As noted by the committee
throughout the present report, sufficient support
for conclusions in the formaldehyde draft IRIS
assessment is often lacking. Given that the
development of specific IRIS assessments and
their conclusions are of interest to many
stakeholders, it is important that they provide
sufficient references and supporting
documentation for their conclusions. Detailed
appendixes, which might be made available only
electronically, should be provided when
appropriate.
Implemented. The Toxicological Review provides a clear
explanation of the literature and methods used to develop
the 1,4-dioxane reference values. The document provides a
clear description of the decisions applied in developing the
hazard identification and dose-response analysis, including
documentation of the information to support conclusion and
reference to relevant EPA guidelines that guided decision
making. As recommended, supplementary information
(including PBPK model evaluation and detailed
documentation of BMD modeling) is provided in appendices.
1-5
-------
Table 1-2. National Research Council recommendations that the EPA is generally
implementing in the long-term
NRC recommendations that the EPA is
generally implementing in the
long-term
Implementation in the 1,4-dioxane assessment
Weight-of-Evidence Evaluation: Synthesis of
Evidence for Hazard Identification (p. 165 of the
NRC Report)
1. Review use of existing weight-of-evidence
guidelines.
2. Standardize approach to using weight-of-
evidence guidelines.
3. Conduct agency workshops on approaches to
implementing weight-of-evidence guidelines.
4. Develop uniform language to describe strength
of evidence on noncancer effects.
5. Expand and harmonize the approach for
characterizing uncertainty and variability.
6. To the extent possible, unify consideration of
outcomes around common modes of action
rather than considering multiple outcomes
separately.
Not implemented. As indicated above, Phase 3 of EPA's
implementation plan will incorporate the longer-term
recommendations made by the NRC, including the
development of a standardized approach to describe the
strength of evidence for noncancer effects. On May 16, 2012,
EPA announced (U.S. EPA. 2012C) that as a part of a review
of the IRIS Program's assessment development process, the
NRC will also review current methods for weight-of-evidence
analyses and recommend approaches for weighing scientific
evidence for chemical hazard identification. In addition, EPA
held a workshop on August 26, 2013, on issues related to
weight-of-evidence to inform future assessments.
Calculation of Reference Values and Unit Risks
(pp. 165-166 of the NRC Report)
7. Assess the sensitivity of derived estimates to
model assumptions and end points selected. This
step should include appropriate tabular and
graphic displays to illustrate the range of the
estimates and the effect of uncertainty factors on
the estimates.
Partially implemented. As indicated above, Phase 3 of EPA's
implementation plan will incorporate the longer-term
recommendations made by the NRC, including assessment of
the sensitivity of derived estimates to model assumptions and
endpoint selection. As discussed in Sections 4.6.1 and 4.6.2,
the primary targets of toxicity of 1,4-dioxane are the kidney,
liver, and respiratory tract. Candidate RfDs are evaluated in
Figure 5-1 through Figure 5-4 and candidate RfCs are
evaluated in Figure 5-5.
1-6
------- |