DRAFT - DO NOT CITE OR QUOTE
EPA/635/R-11/003A
— ^_ ^_ _ www. ep a. gov/iris
oEPA
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 2011
NOTICE
This document is an External Review draft. This information is distributed solely for the purpose of
pre-dissemination peer review under applicable information quality guidelines. It has not been formally
disseminated by EPA. It does not represent and should not be construed to represent any Agency
determination or policy. It is being circulated for review of its technical accuracy and science policy
implications.
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIME R
This document is a preliminary draft for review purposes only. This
information is distributed solely for the purpose of pre-dissemination peer
review under applicable information quality guidelines. It has not been
formally disseminated by EPA. It does not represent and should not be
construed to represent any Agency determination or policy. Mention of trade
names or commercial products does not constitute endorsement or
recommendation for use.
DRAFT - DO NOT CITE OR QUOTE
-------
CONTENTS: TOX 1C 0 L 0 G 1C A L REVIEW OF
1,4-DIOXANE (CAS NO. 123-91-1)
LIST OF ABBREVIATIONS AND ACRONYMS XIV
FOREWORD XVI
AUTHORS, CONTRIBUTORS, AND REVIEWERS XVII
1.1 INTRODUCTION 1
2 CHEMICAL AND PHYSICAL INFORMATION 3
Figure 2-1 1,4-Dioxane chemical structure. 3
Table 2-1 Physical properties and chemical identity of 1,4-dioxane 3
TOXICOKINETICS
3.1
3.2
3.3
3.4
3.5
3,
3,
3,
Absorption
Distribution
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
Elimination
Physiologically Based Pharmacokinetic Models
Figure 3-3 General PBPK model structure.
.5.1 Available Pharmacokinetic Data
.5.2 Published PBPK Models for 1 ,4-Dioxane
3.5.2.1 Leung and Paustenbach
3.5.2.2 Reitzetal.
3.5.2.3 Fisher et al.
3.5.2.4 Sweeney et al.
.5.3 Implementation of Published PBPK Models for 1 ,4-Dioxane
5
6
7
8
9
10
11
12
12
14
14
15
16
16
17
3.6 Rat Nasal Exposure via Drinking Water 20
HAZAR D IDE NTIFICATION 21
4.1 Studies in Humans - Epidemiology, Case Reports, Clinical Controls 21
4.1.1 Thiessetal. 22
4.1.2 Buffleretal. 23
4.2 Subchronic and Chronic Studies and Cancer Bioassays in Animals - Oral and
Inhalation 24
4.2.1 Oral Toxicity 25
4.2.1.1 Subchronic Oral Toxicity 25
Table 4-1 Incidence of histopathological lesions in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 13 weeks 28
Table 4-2 Incidence of histopathological lesions in Crj:BDF1 mice exposed to
1,4-dioxane in drinking water for 13 weeks 29
4.2.1.2 Chronic Oral Toxicity and Carcinogenicity 30
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 32
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 34
Table 4-5 Incidence of nonneoplastic lesions in Osborne-Mendel rats exposed
to 1,4-dioxane in drinking water 36
Table 4-6 Incidence of nasal cavity sguamous cell carcinoma and liver
hepatocellular adenoma in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water 37
Table 4-7 Incidence of hepatocellular adenoma or carcinoma in B6C3F-: mice
exposed to 1,4-dioxane in drinking water 38
in
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8 Incidence of histopathological lesions in male F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years 42
Table 4-9 Incidence of histopathological lesions in female F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years 42
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 44
Table 4-11 Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane in
drinking water for 2 years 44
Table 4-12 Incidence of histopathological lesions in male Crj:BDF1 mice exposed
to 1,4-dioxane in drinking water for 2 years 46
Table 4-13 Incidence of histopathological lesions in female Crj:BDF1 mice
exposed to 1,4-dioxane in drinking water for 2 years 46
Table 4-14 Incidence of tumors in Crj:BDF1 mice exposed to 1,4-dioxane in
drinking water for 2 years 47
4.2.2 Inhalation Toxicity 48
4.2.2.1 Subchronic Inhalation Toxicity 48
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 50
Table 4-16 Hematology and clinical chemistry of F344/DuCrj rats exposed to
1,4-dioxane vapor by whole-body inhalation for 13 weeks 51
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 51
4.2.2.2 Chronic Inhalation Toxicity and Carcinogenicity 52
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 56
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 56
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 57
Table 4-21 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 2 years 58
4.2.3 Initiation/Promotion Studies 58
4.2.3.1 Bulletal. 58
4.2.3.2 Kingetal. 58
4.2.3.3 Lundberg et al. 59
4.3 Reproductive/Developmental Studies—Oral and Inhalation 60
4.3.1 Giavini et al. 60
4.4 Other Duration or Endpoint Specific Studies 61
4.4.1 Acute and Short-term Toxicity 61
4.4.1.1 Oral Toxicity 61
4.4.1.2 Inhalation Toxicity 61
Table 4-22 Acute and short-term toxicity studies of 1,4-dioxane 62
4.4.2 Neurotoxicity 63
4.4.2.1 Frantiketal. 64
4.4.2.2 Goldberg et al. 64
4.4.2.3 Kanada et al. 65
4.4.2.4 Knoefel 65
4.5 Mechanistic Data and Other Studies in Support of the Mode of Action 65
4.5.1 Genotoxicity 65
Table 4-23 Genotoxicity studies of 1,4-dioxane; in vitro 69
Table 4-24 Genotoxicity studies of 1,4-dioxane; mammalian in vivo 71
4.5.2 Mechanistic Studies 72
4.5.2.1 Free Radical Generation 72
4.5.2.2 Induction of Metabolism 73
4.5.2.3 Mechanisms of Tumor Induction 73
4.6 Synthesis of Major Noncancer Effects 75
4.6.1 Oral 75
Table 4-25 Oral toxicity studies (noncancer effects) for 1,4-dioxane 77
4.6.2 Inhalation 79
Table 4-26 Inhalation toxicity studies (noncancer effects) for 1,4-dioxane 81
4.6.2.1 Mode of Action Information 81
4.7 Evaluation of Carcinogenicity 83
4.7.1 Summary of Overall Weight of Evidence 83
4.7.2 Synthesis of Human, Animal, and Other Supporting Evidence 84
4.7.3 Mode of Action Information 86
4.7.3.1 Identification of Key Events for Carcinogenicity 87
iv
DRAFT - DO NOT CITE OR QUOTE
-------
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_
4.7.3.2 Strength, Consistency, Specificity of Association 89
4.7.3.3 Dose-Response Relationship 90
Table 4-27 Temporal sequence and dose-response relationship for possible key
events and liver tumors in rats and mice 91
Table 4-28 Temporal sequence and dose-response relationship for possible key
events and nasal tumors in rats and mice
4.7.3.4 Temporal Relationship
4.7.3.5 Bioloqical Plausibility and Coherence
4.7.3.6 Other Possible Modes of Action
4.7.3.7 Conclusions About the Hypothesized Mode of Action
4.7.3.8 Relevance of the Mode of Action to Humans
Susceptible Populations and Life Stages
93
94
95
96
97
97
98
4.8
DOS E-RESPONSEASSESSMENTS 99
5.1 Oral Reference Dose (RfD) 99
5.1.1 Choice of Principal Studies and Critical Effect with Rationale and Justification 99
5.1.2 Methods of Analysis—Including Models (PBPK, BMD, etc.) 100
Table 5-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years 102
Table 5-2 BMD and BMDL values derived from BMD modeling of cortical tubule
degeneration in male and female Osborne-Mendel rats exposed to
1,4-dioxane in drinking water for 2 years 102
Table 5-3 Incidence of liver hyperplasia in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years 102
Table 5-4 BMD and BMDL values derived from BMD modeling of liver
hyperplasia in male and female F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 2 years 103
5.1.3 RfD Derivation - Including Application of Uncertainty Factors (UFs) 103
5.1.4 RfD Comparison Information 104
Figure 5-1 Potential points of departure (POD) for liver toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following
oral exposure to 1,4-dioxane. 105
Figure 5-2 Potential points of departure (POD) for kidney toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following
oral exposure to 1,4-dioxane. 106
Figure 5-3 Potential points of departure (POD) for nasal inflammation with
corresponding applied uncertainty factors and derived sample RfDs
following oral exposure to 1,4-dioxane. 107
Figure 5-4 Potential points of departure (POD) for organ specific toxicity
endpoints with corresponding applied uncertainty factors and derived
sample RfDs following oral exposure to 1,4-dioxane. 108
5.1.5 Previous RfD Assessment 108
5.2 Inhalation Reference Concentration (RfC) 108
5.2.1 Choice of Principal Study and Candidate Critical Effect(s) with Rationale and
Justification 108
Table 5-5 Incidences of nonneoplastic lesions resulting from chronic exposure
(ppm) to 1,4-dioxane considered for identification of a critical effect. 111
5.2.2 Methods of Analysis 111
5.2.3 Exposure Duration and Dosimetric Adjustments 111
Table 5-6 Duration adjusted POD estimates for BMDLs (from best fitting BMDS
models) or NOAELs/LOAELs from chronic exposure to 1,4-dioxane 112
5.2.4 RfC Derivation- Including Application of Uncertainty Factors (UFs) 115
5.2.5 RfC Comparison Information 115
Figure 5-5 Potential points of departure (POD) for candidate endpoints with
corresponding applied uncertainty factors and derived sample RfCs
following inhalation exposure to 1,4-dioxane. 116
5.2.6 Previous RfC Assessment 116
5.3 Uncertainties in the Oral Reference Dose and Inhalation Reference Concentration 116
5.4 Cancer Assessment 118
5.4.1 Choice of Study/Data - with Rationale and Justification 118
5.4.1.1 Oral Study/Data 118
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-7 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) 120
5.4.1.2 Inhalation Study/Data 120
Table 5-8 Incidence of liver, nasal cavity, kidney, peritoneal, and mammary
gland, Zymbal gland, and subcutis tumors in rats exposed to
1,4-dioxane vapors for 2 years. 122
5.4.2 Dose-Response Data 122
5.4.2.1 Oral Data 122
Table 5-9 Incidence of hepatocellular adenoma or carcinoma in rats and mice
exposed to 1,4-dioxane in drinking water for2 years 123
5.4.2.2 Inhalation Data 123
Table 5-10 Incidence of tumors in F344 male rats exposed to 1,4-dioxane for 104
weeks (6 hours/day, 5 days/week) 124
5.4.3 Dose Adjustments and Extrapolation Method(s) 124
5.4.3.1 Oral 124
Table 5-11 Calculated HEDs for the tumor incidence data used for
dose-response modeling 125
5.4.3.2 Inhalation 126
5.4.4 Oral Slope Factor and Inhalation Unit Risk 128
5.4.4.1 Oral Slope Factor 128
Table 5-12 BMD HED and BMDI_HED values from models fit to tumor incidence data
for rats and mice exposed to 1,4-dioxane in drinking water for 2 years
and corresponding oral CSFs 129
5.4.4.2 Inhalation Unit Risk 130
Table 5-13 Dose-response modeling summary results for male rat tumors
associated with inhalation exposure to 1,4-dioxane for 2 years 131
5.4.5 Previous Cancer Assessment 132
5.5 Uncertainties in Cancer Risk Values 133
5.5.1 Sources of Uncertainty 133
5.5.1.1 Choice of Low-Dose Extrapolation Approach 133
5.5.1.2 Dose Metric 134
5.5.1.3 Cross-Species Scaling 134
5.5.1.4 Statistical Uncertainty at the POD 135
5.5.1.5 Bioassay Selection 135
5.5.1.6 Choice of Species/Gender 135
5.5.1.7 Relevance to Humans 136
5.5.1.8 Human Population Variability 136
Table 5-14 Summary of uncertainty in the 1,4-dioxane cancer risk estimation 137
6 MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE 1 38
6.1 Human Hazard Potential 138
6.2 DOSE RESPONSE 139
6.2.1 Noncancer/Oral 139
6.2.2 Noncancer/lnhalation 140
6.2.3 Cancer 140
6.2.3.1 Oral 140
6.2.3.2 Inhalation 140
6.2.3.3 Choice of Low-Dose Extrapolation Approach 141
6.2.3.4 Dose Metric 142
6.2.3.5 Cross-Species Scaling 142
6.2.3.6 Statistical Uncertainty at the POD 142
6.2.3.7 Bioassay Selection 143
6.2.3.8 Choice of Species/Gender 143
6.2.3.9 Relevance to Humans 143
6.2.3.10 Human Population Variability 144
REFERENCES 145
APPENDIX A. SUMMARY OF EXTERNAL PEER RE VIEW AND PUBLIC COMMENTS
AND DIS P OS ITION A-1
A.1 External Peer Review Panel Comments A-1
A. 1.1 General Charge Questions A-1
vi
DRAFT - DO NOT CITE OR QUOTE
-------
A. 1.2 Oral reference dose (RfD) for 1,4-dioxane A-5
A.1.3 Carcinogenicity of 1,4-dioxane A-10
A.2 Public Comments A-15
A.2.1 Oral reference dose (RfD) for 1,4-dioxane A-15
A.2.2 Carcinogenicity of 1,4-dioxane A-15
A.2.3 PBPK Modeling A-18
A.2.4 Other Comments A-18
APPENDIX B. EVALUATION OF EXISTING PBPK MODELS FOR 1,4-DIOXANE B-1
B.1 Background B-1
B.2 Scope B-1
B.3 Implementation of the Empirical Models in aclsXtreme B-2
B.3.1 Model Descriptions B-2
Figure B-1 Schematic representation of empirical model for 1,4-dioxane in rats. B-3
Figure B-2 Schematic representation of empirical model for 1,4-dioxane in
humans. B-3
B.3.2 Modifications to the Empirical Models B-3
B.3.3 Results B-4
Figure B-3 Output of 1,4-dioxane blood level data from the acslXtreme
implementation (left) and published (right) empirical rat model
simulations of i.v. administration experiments. B-5
Figure B-4 Output of HEAA urine level data from acslXtreme implementation
(left) and published (right) empirical rat model simulations of i.v.
administration experiments. B-5
Figure B-5 acslXtreme predictions of blood 1,4-dioxane and urine HEAA levels
from the empirical rat model simulations of a 6-hour, 50-ppm
inhalation exposure. B-6
Figure B-6 Output of 1,4-dioxane blood level data from the acslXtreme
implementation (left) and published (right) empirical human model
simulations of a 6-hour, 50-ppm inhalation exposure. B-7
Figure B-7 Observations and acslXtreme predictions of cumulative HEAA in
human urine following a 6-hour, 50-ppm inhalation exposure. B-7
Figure B-8 EPA-modified Young et al. empirical model prediction (line) of plasma
1,4-dioxane levels in rats following exposure to 1,4-dioxane for 13
weeks compared to data from Kasai et al. (2008). B-8
B.3.4 Conclusions for Empirical Model Implementation B-8
B.4 Initial Recalibration of the PBPK Model B-9
B.4.1 Sources of Values for Flow Rates B-9
Table B-1 Human PBPK model parameter values for 1,4-dioxane B-10
B.4.2 Sources of Values for Partition Coefficients B-10
B.4.3 Calibration Method B-10
B.4.4 Results B-11
Table B-2 PBPK metabolic and elimination parameter values resulting from
re-calibration of the human model using alternative values for
physiological flow rates3 and tissue:air partition coefficients B-11
Figure B-9 Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following re-calibration of the human
PBPK model with tissue:air partition coefficient values. B-12
Figure B-10 Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following re-calibration of the human
PBPK model with tissue:air partition coefficient values. B-12
Figure B-11 Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) B-13
using EPA estimated biologically plausible parameters (Table B-1). B-13
B.4.5 Conclusions for PBPK Model Implementation B-13
B.4.6 Sensitivity Analysis B-14
B.4.7 Method B-14
B.4.8 Results B-15
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-15
B.5 PBPK Model Exercises Using Biologically Plausible Paramter Boundaries B-15
B.5.1 Observations Regarding the Volume of Distribution B-16
B.5.2 Defining Boundaries for Parameter Values B-16
B.5.3 Results B-16
vii
DRAFT - DO NOT CITE OR QUOTE
-------
Figure B-13 Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients with empirical model
predictions and experimental observations for blood 1,4-dioxane
concentrations (left) and urinary HEAA levels (right) from a 6-hour,
50-ppm inhalation exposure.
Figure B-14 Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients with empirical model
predictions and experimental observations for blood 1,4-dioxane
concentrations (left) and urinary HEAA levels (right) from a 6-hour,
50-ppm inhalation exposure.
Table B-3
B.5.4
Figure B-17 Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a zero-order metabolism rate constant, ki_c, and slowly
perfused tissue:air partition coefficient to the experimental data.
B.6 Conclusions
B.7 aclsXtreme Code for the Young et al. Empricial Model for 1,4-Dioxane in Rats
B.8 aclsXtreme Code for the Young et al. Empricial Model for 1,4-Dioxane in Humans_
B.9 aclsXtreme Code for the Reitz et al. PBPK Model For 1,4-Dioxane
B-17
B-18
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-18
Alternative Model Parameterization B-18
Figure B-15 Predictions of blood 1,4-dioxane concentration following calibration of
a zero-order metabolism rate constant, ki_c, to the experimental data. B-19
Figure B-16 Predictions of blood 1,4-dioxane concentration following calibration of
a zero-order metabolism rate constant, ki_c, to only the exposure
phase of the experimental data._ B-20
B-21
B-21
"B-22
B-24
"B-25
APPENDIX C.
DETAILS OF BMD ANALYSIS FOR ORAL RFD FOR 1,4-DIOXANE C-1
C.1 Cortical Tubule Degeneration_
Table C-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years
Table C-2 Goodness-of-fit statistics and BMDio 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
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 to
Figure C-2 BMD Weibull model of cortical tubule degeneration incidence data for
female rats exposed to 1,4-dioxane in drinking water for 2 years to
C.2
Table C-4
Figure C-3
Figure C-4
Table C-4.
Figure C-5
Table C-4.
Figure C-6
Table C-4.
Figure C-7
Benchmark dose modeling results based on the incidence of liver
hyperplasias in male and female F344 rats exposed to 1,4-dioxane in
drinking water for 2 years_
C-1
C-1
C-2
C-3
Liver hyperplasi;
Table C-3
support the results in Table C-2
a
Incidence of
1,4-dioxane
liver hyperplasia in
in drinking water3
F344/DuCrj
rats exposed
to
C-5
C-6
C-7
C-8
BMD gamma model of liver hyperplasia incidence data for F344 male
rats exposed to 1,4-dioxane in drinking water for 2 years to support
results Table C-4. C-9
BMD multistage (2 degree) model of liver hyperplasia incidence data
for F344 male rats exposed to 1,4-dioxane in drinking water for 2
years to support results C-11
C-12
BMD Weibull model of liver hyperplasia incidence data for F344 male
rats exposed to 1,4-dioxane in drinking water for 2 years to support
the results in C-13
C-14
BMD quantal-linear model of liver hyperplasia incidence data for F344
male rats exposed to 1,4-dioxane in drinking water for 2 years to
support the results in C-15
C-16
BMD Multistage model (third (3°)of liver hyperplasia incidence data
for F344 female rats exposed to 1,4-dioxane in drinking water for 2
years to support the results in C-17
vm
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4. C-18
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.1.1 Combining Data on Adenomas and Carcinomas_
D.1.2 Model Selection Criteria
D.1.3 Summary
Table D-1
Recommended models for rodents exposed to 1,4-dioxane in drinking
water (Kano et al., 2009)
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).
D.3
Figure D-1 Multistage BMD model (2 degree) for the combined incidence of
hepatic adenomas and carcinomas in female F344 rats.
Male F344 Rats: Hepatic Carcinomas and Adenomas_
Table D-5
Figure D-2
Figure D-3
D.4
Multistage BMD model (3 degree) for the combined incidence of
hepatic adenomas and carcinomas in male F344 rats.
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)
LogLogistic 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 etal., 2009).
Table D-12
Table D-13
Figure D-10
D.6
LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 10%.
Figure D-11 LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 30%.
Figure D-12 LogLogistic 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.
Male BDF1 Mice: Hepatic Carcinomas and Adenomas
D-2
D-2
"D-3
-D-4
D-4
-D-4
D-5
_D-6
D-6
"D-8
BMDS dose-response modeling results for the combined incidence of
adenomas and carcinomas in livers of male F344 rats (Kano et al.,
2009) D-9
Probit BMD model for the combined incidence of hepatic adenomas
and carcinomas in male F344 rats. D-9
D-11
.D-13
.D-13
.D-14
.D-14
.D-17
.D-17
.D-20
.D-20
.D-22
D-24
~_D-24
D-26
~D-27
D-28
BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in female BDF1 mice (Kano et al.,
2009)
BMDS LogLogistic 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).
D-29
.D-29
.D-30
.D-32
.D-34
D-36
D-37
IX
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-14
TableD-15
Figure D-14
Figure D-15
Data for hepatic adenomas and carcinomas in male BDF1 mice
(Kanoetal., 2009).
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)
TableD-17
TableD-18
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 _
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_
TableD-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_
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.
D.7.2 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 the 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 the drinking water for 2 years_
Table D-22
Figure D-21 LogLogistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in female Osborne-Mendel rats exposed to
1,4-dioxane in drinking water.
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 LogLogistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in male Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water.
D-38
BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.,
2009)
LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice.
.D-38
.D-39
D-41
D-42
D-43
D-43
D-44
D-44
Figure D-16 Probit BMD model for the incidence of hepatocellular carcinoma in
male and female Sherman rats exposed to 1,4-dioxane in drinking
water. D-45
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-47
D-49
D-49
D-51
D-52
D-53
Figure D-19 LogLogistic BMD model for the incidence of hepatocellular adenoma
in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water. D-53
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-55
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-57
D-57
D-59
.D-61
D-61
"D-61
DRAFT - DO NOT CITE OR QUOTE
-------
Figure D-24 Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel ratsexposed to
1,4-dioxane in drinking water.
D.7.3 Hepatocellular Adenoma or Carcinoma in B6C3F-I Mice (NCI, 1978)
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
D-63
"D-64
D-66
Table D-25
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-66
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.
D-67
Table D-26
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_
D-69
Figure D-26 Gamma BMD model for the incidence of hepatocellular adenoma or
carcinoma in male B6C3F-: mice exposed to 1,4-dioxane in drinking
water. D-69
Figure D-27 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-71
APPENDIX E. COMPARISON OF SEVERAL DATA REPORTS FOR THE JBRC 2-YEAR
1,4-DIOXANE DRINKING WATER STUDY E-1
Table E-1
Table E-2
Table E-3
Table E-4
Table E-5
Table E-6
Table E-7
Table E-8
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinking water study in male F344 rats
E-2
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinking water study in female F344 rats_
E-3
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in male F344 rats_
E-4
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in female F344 rats
E-5
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinking water study in male Crj:BDF1 mice E-6
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinking water study in female Crj:BDF1 mice E-7
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in male Crj:BDF1 mice_ E-8
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in female Crj:BDF1 mice
E-9
APPENDIX F.
DETAILS OF BMD ANALYSIS FOR INHALATION RFC FOR 1,4-DIOXANE F-1
F.1 Centrilobular Necrosis of the Liver
F-1
Table F-1
Table F-2
Incidence of centrilobular necrosis of the liver in F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years
F-1
F.2
Goodness-of-fit statistics and BMD-io and BMDL-io 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)
Figure F-1 BMD Dichotomous Hill model of centrilobular necrosis incidence data
for male rats exposed to 1,4-dioxane vapors for 2 years to support the
results in Table F-2.
Spongiosis Hepatis
F-2
Table F-3
Table F-4
Figure F-2
Figure F-3
Incidence of spongiosis hepatis of the liver in F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years
F-2
.F-4
F-4
Goodness-of-fit statistics and BMD-io and BMDL-io values from models
fit to incidence data for spongiosis hepatis of the liver in male
F344/DuCrj rats (NCI, 1978) exposed to 1,4-dioxane vapors
F-5
BMD Dichotomous-Hill model of spongiosis hepatis incidence data for
male rats exposed to 1,4-dioxane vapors for 2 years to support the
results in Table F-4.
F-5
BMD Log-Logistic model of spongiosis hepatis incidence data for
male rats exposed to 1,4-dioxane vapors for 2 years to support the
results in Table F-4.
F-7
XI
DRAFT - DO NOT CITE OR QUOTE
-------
F.3 Squamous Cell Metaplasia_
F-8
Table F-5
Table F-6
Incidence of squamous cell metaplasia of the respiratory epithelium in
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years F-9
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
(Kasaietal.,2009)_ F-10
F.4
Figure F-4 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 to support the results in Table F-6.
Squamous Cell Hyperplasia_
Table F-7
Table F-8
F.5
Figure F-5 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 to support the results in Table F-8.
Respiratory Metaplasia
Table F-9
Table F-10
Table F-11
Figure F-6
F.6 Atrophy_
Table F-12
TableF-13
Figure F-7
F.7
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 respiratory metaplasia of the olfactory epithelium in
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 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 to
support the results in Table F-13.
Hydropic Change
Table F-14 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
TableF-15
Figure F-8
F.8 Sclerosis
Goodness-of-fit statistics and BMD-io and BMDL-io 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)
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 to support the results in Table F-16.
F-10
Incidence of squamous cell hyperplasia of the respiratory epithelium
in F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years F-12
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) F-13
F-13
Incidence of respiratory metaplasia of the olfactory epithelium in
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years F-15
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 F-16
Goodness-of-fit statistics and BMDio 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 F-16
F-17
.F-18
F-19
F-20
F-20
.F-22
F-22
F-23
F-23
.F-25
F-25
Table F-16 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-17 Goodness-of-fit statistics and BMDio and BMDL-io 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) F-26
Figure F-9 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 to support the results in Table F-18. F-28
Xll
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIXG. RFC DERIVATION: ALTERNATIVE APPROACH IN THE APPLICATION OF
THE DOSIMETRIC ADJUSTMENT FACTOR G-1
G.1 Application of DAF for Category 1 Gases G-2
G.2 Application of Uncertainty Factors G-2
APPENDIX H. DETAILS OF BMD ANALYSIS FOR INHALATION UNIT RISK FOR
1,4-DIOXANE H-1
H.1 General Issues and Approaches to BMDS and Multitumor Modeling H-1
H.1.1 Combining Data tumor types H-1
H.1.2 Summary H-1
Table H-1 Summary of BMC-io and BMCL-io model results for individual tumor
types and combined tumor analysis for male rats exposed to
1,4-dioxane vapors (Kasai etal., 2009) H-2
H.2 BMDS Model Output for Multistage Cancer Models for Inidividual Tumor Types H-2
Table H-2 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 2 years H-3
H.2.1 Nasal Squamous Cell Carcinoma H-3
Table H-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 for2-years (Kasai etal., 2009) H-4
Figure H-1 Multistage model (First (1°)-degree) for male rat nasal squamous cell
carcinomas. H-4
H.2.2 Hepatocellular Adenoma and Carcinoma H-6
Table H-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 for2-years (Kasai et al., 2009) H-7
Figure H-2 Multistage model (First-degree (1°)) for male rat hepatocellular
adenomas and carcinomas. H-7
H.2.3 Renal Cell Carcinoma and Zymbal Gland Adenoma H-9
Table H-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 for2-years (Kasai et al.,
2009) H-10
Figure H-3 Multistage model (Second-degree (2°)) for male rat renal cell
carcinomas and Zymbal gland adenomas. H-10
Figure H-4 Multistage model (Third-degree (3°)) for male rat renal cell
carcinomas. H-12
H.2.4 Peritoneal Mesothelioma H-14
Table H-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) H-15
Figure H-5 Multistage model (First-degree (1°)) for male rat peritoneal
mesotheliomas. H-15
H.2.5 Mammary Gland Fibroadenoma H-17
Table H-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 etal., 2009) H-18
Figure H-6 Multistage model (First-degree (1°)) for male rat mammary gland
fibroadenoma. H-18
H.2.6 Subcutis Fibroma H-20
Table H-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) H-21
Figure H-7 Multistage model (First-degree (1°)) for male rat subcutis fibroma
(high dose dropped). H-21
H.2.7 Multitumor analysis using Bayesian Methods H-23
H.3 Multitumor Analysis Using BMDS MSCOMBO (BETA) H-24
xm
DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS
AIC Akaike's Information Criterion
ALP alkaline phosphatase
ALT alanine aminotransferase
AST aspartate aminotransferase
ATSDR Agency for Toxic Substances and Disease Registry
BMC benchmark concentration
BMCL benchmark concentration, lower 95% confidence limit
BMCLio benchmark concentration, lower 95% confidence limit at 10% extra risk
BMD benchmark dose
BMD-io benchmark dose at 10% extra risk
BMDso benchmark dose at 30% extra risk
BMDso benchmark dose at 50% extra risk
BMDL benchmark dose, lower 95% confidence limit
BMDL-io benchmark dose, lower 95% confidence limit at 10% extra risk
BMDLso benchmark dose, lower 95% confidence limit at 30% extra risk
BMDLso 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
CHO Chinese hamster ovary (cells)
Cl confidence interval(s)
CNS central nervous system
CPK creatinine phosphokinase
CREST antikinetochore
CSF cancer slope factor
CV concentration in venous blood
CYP450 cytochrome P450
DEN diethylnitrosamine
FISH fluorescence in situ hybridization
G-6-Pase glucose-6-phosphatase
GC gas chromatography
GGT Y-g|utamyl transpeptidase
GST-P qlutathione 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
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
kiNH 1 st order 1,4-dioxane inhalation rate constant
ki_c 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 1st order elimination rate of HEAA (1,4-dioxane metabolite)
koc soil organic carbon-water portioning coefficient
LAP leucine aminopeptidase
LDso median lethal dose
LDH lactate dehydrogenase
LOAEL lowest-observed-adverse-effect-level
MCH mean corpuscular hemoglobin
MCV
MOA
MS
mean corpuscular volume
mode of action
mass spectrometry, multi-stage
xiv
DRAFT - DO NOT CITE OR QUOTE
-------
MTD maximum tolerated dose
MVK Moolgavkar-Venzon-Knudsen (model)
NCE normochromatic erythrocyte
NCI National Cancer Institute
ND no data, not detected
NE not estimated
NOAEL no-observed-adverse-effect-level
NRC National Research Council
NTP National Toxicology Program
OCT ornithine carbamyl transferase
ODC ornithine decarboxylase
OECD Organization for Economic Co-operation and Development
PB blood:air partition coefficient
PBPK physiologically based pharmacokinetic
PC partition coefficient
PCS polychlorinated biphenyl
PCE polychromatic erythrocyte
PFA fat:air partition coefficient
PLA liverair partition coefficient
POD point of departure
ppm parts per million
PRA rapidly perfused tissue:air partition coefficient
PSA 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
SCE sister chromatid exchange
SDH sorbitol dehydrogenase
SMR standardized mortality ratio
SRC Syracuse Research Corporation
TPA 12-O-tetradecanoylphorbol-13-acetate
TWA time-weighted average
UF uncertainty factor
UNEP United Nations Environment Programme
U.S. 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
xv
DRAFT - DO NOT CITE OR QUOTE
-------
FOR EWOR D
1 The purpose of this Toxicological Review is to provide scientific support and rationale for the
2 hazard and dose-response assessment in IRIS pertaining to chronic exposure to 1,4-dioxane. It is not
3 intended to be a comprehensive treatise on the chemical or toxicological nature of 1,4-dioxane.
4 The intent of Section 6, Major Conclusions in the Characterization of Hazard and Dose
5 Response, is to present the major conclusions reached in the derivation of the reference dose, reference
6 concentration and cancer assessment, where applicable, and to characterize the overall confidence in the
7 quantitative and qualitative aspects of hazard and dose response by addressing the quality of data and
8 related uncertainties. The discussion is intended to convey the limitations of the assessment and to aid
9 and guide the risk assessor in the ensuing steps of the risk assessment process.
10 For other general information about this assessment or other questions relating to IRIS, the reader
11 is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
12 hotline.iris@epa.gov (email address).
13 NOTE: New studies (Kasai et al., 2009; Kasai et al., 2008) regarding the toxicity of 1.4-dioxane through
14 the inhalation route of exposure are available that were not included in the 1.4-dioxane assessment that
15 was posted on the IRIS database in 2010 (U.S. EPA. 2010).
16 These studies have been incorporated into the previously posted assessment (U.S. EPA. 2010) for
17 review. Sections including new information can be identified by the red underlined text in the document.
18 The entire document is provided for completeness.
DRAFT-DO NOT CITE OR QUOTE xvi
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGERS/AUTHORS
Patricia Gillespie. Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Eva D. McLanahan, Ph.D.
Lieutenant Commander, U.S. Public Health Service
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
ReederSams II, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
CO-AUTHORS AND CONTRIBUTORS
J. Allen Davis, MSPH
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Hisham EI-Masri, Ph.D.
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
JeffS. Gift, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Karen Hogan
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
Leonid Kopvlev. Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
William Lefew. Ph.D.
National Health and Environmental Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Fernando Llados
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Michael Lumpkin, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
DRAFT-DO NOT CITE OR QUOTE xvii
-------
Allan Marcus, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Marc Odin, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Susan Rieth
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC
Andrew Rooney, Ph.D.*
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
'Currently at National Toxicology Program , National Institute of Environmental Health Sciences; Research
Triangle Park, NC
Paul Schlosser, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
John Stanek. Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Julie Stickney, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
John Vandenberg, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
Debra Walsh, M.S.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC
TECHNICAL SUPPORT
Ellen Lorang, M.S.
National Center for Environmental Assessment
Research Triangle Park, NC
J. Sawyer Lucy, B.A.
Student Services Contractor
National Center for Environmental Assessment
Research Triangle Park, NC
Deborah Wales
National Center for Environmental Assessment
Research Triangle Park, NC
DRAFT-DO NOT CITE OR QUOTE xvii
i
-------
R EVIE WE RS
1 This document has been provided for review to EPA scientists, interagency reviewers from other
2 federal agencies and White House offices. The comments and responses in Appendix A were in regards to
3 the oral assessment previously reviewed. A summary of external peer review and public comments and
4 disposition following review of the inhalation assessment for 1.4-dioxane will be included when they
5 become available.
INTERNAL EPA REVIEWERS (ORAL AS S E S S ME NT)
Anthony DeAngelo, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Nagu Keshava, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
Jason Lambert, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
Connie Meacham, M.S.
National Center for Environmental Assessment
Research Triangle Park, NC
Douglas Wolf, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
DRAFT-DO NOT CITE OR QUOTE xix
-------
EXTERNAL PEER REVIEWERS
(ORAL AS S ES S ME NT)
George V. Alexeeff, Ph.D., DABT
Office of Environmental Health Hazard Assessment (OEHHA)
California EPA
Bruce C. Allen, M.S.
Bruce Allen Consulting
James V. Bruckner, Ph.D.
Department of Pharmaceutical and Biomedical Sciences
College of Pharmacy
The University of Georgia
Harvey J. Clewell III, Ph.D., DABT
Center for Human Health Assessment
The Hamner Institutes for Health Sciences
Lena Ernstgard, Ph.D.
Institute of Environmental Medicine
Karolinska Institutet
Frederick J. Kaskel, M.D., Ph.D.
Children's Hospital at Montefiore
Albert Einstein College of Medicine of Yeshiva University
Kannan Krishnan, Ph.D., DABT
Inter-University Toxicology Research Center (CIRTOX)
Universite de Montreal
Raghubir P. Sharma, DVM, Ph.D.
Department of Physiology and Pharmacology
College of Veterinary Medicine (retired)
The University of Georgia
DRAFT-DO NOT CITE OR QUOTE xx
-------
1.1 INTRODUCTION
1 This document presents background information and justification for the Integrated Risk
2 Information System (IRIS) Summary of the hazard and dose-response assessment of 1,4-Dioxane.
3 IRIS Summaries may include oral reference dose (RfD) and inhalation reference concentration (RfC)
4 values for chronic and other exposure durations, and a carcinogenicity assessment.
5 The RfD and RfC, if derived, provide quantitative information for use in risk assessments for
6 health effects known or assumed to be produced through a nonlinear (presumed threshold) mode of
7 action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with uncertainty spanning
8 perhaps an order of magnitude) of a daily exposure to the human population (including sensitive
9 subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime. The
10 inhalation RfC (expressed in units of mg/m3) is analogous to the oral RfD, but provides a continuous
11 inhalation exposure estimate. The inhalation RfC considers toxic effects for both the respiratory system
12 (portal-of-entry) and for effects peripheral to the respiratory system (extrarespiratory or systemic effects).
13 Reference values are generally derived for chronic exposures (up to a lifetime), but may also be derived
14 for acute (< 24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
15 lifetime) exposure durations, all of which are derived based on an assumption of continuous exposure
16 throughout the duration specified. Unless specified otherwise, the RfD and RfC are derived for chronic
17 exposure duration.
18 The carcinogenicity assessment provides information on the carcinogenic hazard potential of the
19 substance in question and quantitative estimates of risk from oral and inhalation exposure may be derived.
20 The information includes a weight-of-evidence judgment of the likelihood that the agent is a human
21 carcinogen and the conditions under which the carcinogenic effects may be expressed. Quantitative risk
22 estimates may be derived from the application of a low-dose extrapolation procedure. If derived, the oral
23 slope factor is a plausible upper bound on the estimate of risk per mg/kg-day of oral exposure. Similarly,
24 an inhalation unit risk is a plausible upper bound on the estimate of risk per ug/m3 air breathed.
25 Development of these hazard identification and dose-response assessments for 1,4-dioxane has
26 followed the general guidelines for risk assessment as set forth by the National Research Council (NRC,
27 1983). U.S. Environmental Protection Agency (U.S. EPA) Guidelines and Risk Assessment Forum
28 technical panel reports that may have been used in the development of this assessment include the
29 following Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S. EPA. 1986b).
30 Guidelines for Mutagenicity Risk Assessment (U.S. EPA. 1986a). Recommendations for and
31 Documentation of Biological Values for Use in Risk Assessment (U.S. EPA. 1988), Guidelines for
32 Developmental Toxicity Risk Assessment (U.S. EPA. 1991). Interim Policy for Particle Size and Limit
33 Concentration Issues in Inhalation Toxicity (U.S. EPA. 1994a). Methods for Derivation of Inhalation
34 Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA. 1994b). Use of the
35 Benchmark Dose Approach in Health Risk Assessment (U.S. EPA. 1995). Guidelines for Reproductive
36 Toxicity Risk Assessment (U.S. EPA. 1996). Guidelines for Neurotoxicity Risk Assessment (U.S. EPA.
1
DRAFT - DO NOT CITE OR QUOTE
-------
1 1998). Science Policy Council Handbook: Risk Characterization (U.S. EPA. 2000a). Benchmark Dose
1 Technical Guidance Document (External Review Draft) (U.S. EPA. 2000c). Supplementary Guidance for
3 Conducting Health Risk Assessment of Chemical Mixtures (U.S. EPA. 2000b). A Review of the Reference
4 Dose and Reference Concentration Processes (U.S. EPA. 2002b). Guidelines for Carcinogen Risk
5 Assessment (U.S. EPA, 2005b), Supplemental Guidance for Assessing Susceptibility from Early-Life
6 Exposure to Carcinogens (U.S. EPA. 2005a). Science Policy Council Handbook: Peer Review (U.S. EPA.
7 2006b). and A Framework for Assessing Health Risks of Environmental Exposures to Children (U.S.
8 EPA. 2006a).
9 In 2010. an updated health assessment for oral exposures to 1.4-dioxane was released (U.S. EPA.
10 2010). During the development of the 2010 health assessment, new studies (Kasai et al.. 2009; Kasai et
11 al.. 2008) regarding the toxicity of 1.4-dioxane through the inhalation route of exposure became available
12 that were not included in the 1.4-dioxane assessment that was posted on the IRIS database in 2010. These
13 new inhalation studies have been incorporated into the previously posted assessment for this review.
14 Sections including new information can be identified in this draft assessment by underlined red text.
15 Tables containing new information can be identified by red text, but for improved legibility the new
16 information presented in the tables has not been underlined. The entire document is provided for
17 completeness.
18 The literature search strategy employed for 1,4-dioxane was based on the chemical name,
19 Chemical Abstracts Service Registry Number (CASRN), and multiple common synonyms. Any pertinent
20 scientific information submitted by the public to the IRIS Submission Desk was also considered in the
21 development of this document. Primary, peer-reviewed-literature was reviewed through September 2009
22 for the oral assessment and through July 2011 for the inhalation assessment and was included where that
23 literature was determined to be critical to the assessment. The relevant literature included publications on
24 1,4-dioxane which were identified through Toxicology Literature Online (TOXLINE), PubMed, the Toxic
25 Substance Control Act Test Submission Database (TSCATS), the Registry of Toxic Effects of Chemical
26 Substances (RTECS), the Chemical Carcinogenesis Research Information System (CCRIS), the
27 Developmental and Reproductive Toxicology/Environmental Teratology Information Center
28 (DART/ETIC), the Environmental Mutagens Information Center (EMIC) and Environmental Mutagen
29 Information Center Backfile (EMICBACK) databases, the Hazardous Substances Data Bank (HSDB), the
30 Genetic Toxicology Data Bank (GENE-TOX), Chemical abstracts, and Current Contents. Other peer-
31 reviewed information, including health assessments developed by other organizations, review articles, and
32 independent analyses of the health effects data were retrieved and may be included in the assessment
33 where appropriate.
34
DRAFT - DO NOT CITE OR QUOTE
-------
2 CHEMICAL AND PHYSICAL INFORMATION
1 1,4-Dioxane, a volatile organic compound (VOC), is a colorless liquid with a pleasant odor
2 (Hawley and Lewis. 2001; Lewis. 2000). Synonyms include diethylene ether, 1,4-diethylene dioxide,
3 diethylene oxide, dioxyethylene ether, and dioxane (Hawley and Lewis. 2001). The chemical structure of
4 1,4-dioxane is shown in Figure 2-1. Selected chemical and physical properties of this substance are in
5 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
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 q/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 (Hansch et al., 1995)
Henry's Law constant: 4.80 x 10'6 atm-m3/molecule at25°C (Parket al.. 1987)
OH reaction rate constant: 1.09 x 10"11 cm3/molecule sec at 25°C (Atkinson. 1989)
Koc: 17 (estimated using log Kow) (ACS Handbook (Lvman et al.. 1990))
Bioconcentration factor: 0.4 (estimated using log Kow) (Meylan et al., 1999)
1 ppm = 3.6 mg/m ; 1 mg/m = 0.278 ppm
Conversion factors (in air):
v ' (25 C and 1 atm) (HSDB, 2007)
6 1,4-Dioxane is produced commercially through the dehydration and ring closure of diethylene
7 glycol (Surprenant. 2002). Concentrated sulfuric acid is used as a catalyst (Surprenant. 2002). This is a
8 continuous distillation process with operating temperatures and pressures of 130-200°C and 188-
9 825 mmHg, respectively (Surprenant 2002). During the years 1986 and 1990, the U.S. production of
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-dioxane reported by manufacturers was within the range of 10-50 million pounds (U.S. EPA. 2002b).
2 The production volume reported during the years 1994, 1998, and 2002 was within the range of 1-
3 10 million pounds (U.S. EPA. 2002b).
4 Historically, 1,4-dioxane has been used as a stabilizer for the solvent 1,1,1-trichloro-ethane
5 (Surprenant 2002). However, this use is no longer expected to be important due to the 1990 Amendments
6 to the Clean Air Act and the Montreal Protocol, which mandate the eventual phase-out of
7 1,1,1 -trichloroethane production in the U.S. (ATSDR. 2007; U.N. Environment Programme. 2000;
8 "Amendments to the Clean Air Act. Sec. 604. Phase-out of production and consumption of class I
9 substances." 1990). 1,4-Dioxane is a contaminant of some ingredients used in the manufacture of personal
10 care products and cosmetics. 1,4-Dioxane is also used as a solvent for cellulosics, organic products,
11 lacquers, paints, varnishes, paint and varnish removers, resins, oils, waxes, dyes, cements, fumigants,
12 emulsions, and polishing compositions (Hawley and Lewis, 2001; Merck Index. 2001; IARC. 1999).
13 1,4-Dioxane has been used as a solvent in the formulation of inks, coatings, and adhesives and in the
14 extraction of animal and vegetable oil (Surprenant 2002). Reaction products of 1,4-dioxane are used in
15 the manufacture of insecticides, herbicides, plasticizers, and monomers (Surprenant. 2002).
16 When 1,4-dioxane enters the air, it will exist as a vapor, as indicated by its vapor pressure
17 (HSDB. 2007). It is expected to be degraded in the atmosphere through photooxidation with hydroxyl
18 radicals (HSDB. 2007; Surprenant 2002). The estimated half-life for this reaction is 6.7 hours (HSDB.
19 2007). It may also be broken down by reaction with nitrate radicals, although this removal process is not
20 expected to compete with hydroxyl radical photooxidation (Grosjean. 1990). 1,4-Dioxane is not expected
21 to undergo direct photolysis (Wolfe and Jeffers. 2000). 1,4-Dioxane is primarily photooxidized to
22 2-oxodioxane and through reactions with nitrogen oxides (NOX) results in the formation of ethylene
23 glycol diformate (Platz et al.. 1997). 1,4-Dioxane is expected to be highly mobile in soil based on its
24 estimated Koc and is expected to leach to lower soil horizons and groundwater (ATSDR, 2007; ACS
25 Handbook (LymanetaL 1990). This substance may volatilize from dry soil surfaces based on its vapor
26 pressure (HSDB. 2007). The estimated bioconcentration factor value indicates that 1,4-dioxane will not
27 bioconcentrate in aquatic or marine organisms (Tvleylan et al.. 1999; Franke etal.. 1994). 1,4-Dioxane is
28 not expected to undergo hydrolysis or to biodegrade readily in the environment (ATSDR, 2007; HSDB,
29 2007). Therefore, volatilization is expected to be the dominant removal process for moist soil and surface
30 water. Based on a Henry's Law constant of 4.8x 10"6 atm-m3/mole, the half-life for volatilization of
31 1,4-dioxane from a model river is 5 days and that from a model lake is 56 days (HSDB. 2007; Lyman et
32 al.. 1990; Park etal.. 1987). 1,4-Dioxane may be more persistent in groundwater where volatilization is
33 hindered.
34 Recent environmental monitoring data for 1,4-dioxane are lacking. Existing data indicate that
35 1,4-dioxane may leach from hazardous waste sites into drinking water sources located nearby (Yasuhara
36 etal.. 2003; Yasuhara et al.. 1997; Lesage et al.. 1990). 1,4-Dioxane has been detected in contaminated
37 surface and groundwater samples collected near hazardous waste sites and industrial facilities (Derosa et
38 al.. 1996).
DRAFT - DO NOT CITE OR QUOTE
-------
3 TOXICOKINETICS
1 Data for the toxicokinetics of 1,4-dioxane in humans are very limited. However, absorption,
2 distribution, metabolism, and elimination of 1,4-dioxane are well described in rats exposed via the oral,
3 inhalation, or intravenous (i.v.) routes. 1,4-Dioxane is extensively absorbed and metabolized in humans
4 and rats. The metabolite most often measured and reported is (3-hydroxyethoxy acetic acid (HEAA),
5 which is predominantly excreted in the urine; however, other metabolites have also been identified.
6 Saturation of 1,4-dioxane metabolism has been observed in rats and would be expected in humans;
7 however, human exposure levels associated with nonlinear toxicokinetics are not known.
8 Important data elements that have contributed to our current understanding of the toxicokinetics
9 of 1,4-dioxane are summarized in the following sections.
3.1 Absorption
10 Absorption of 1,4-dioxane following inhalation exposure has been qualitatively demonstrated in
11 workers and volunteers. Workers exposed to a time-weighted average (TWA) of 1.6 parts per
12 million (ppm) of 1,4-dioxane in air for 7.5 hours showed a HEAA/1,4-dioxane ratio of 118:1 in urine
13 (Young et al. 1976). The authors assumed lung absorption to be 100% and calculated an average
14 absorbed dose of 0.37 mg/kg, although no exhaled breath measurements were taken. In a study with four
15 healthy male volunteers, Young et al. (1977) reported 6-hour inhalation exposures of adult volunteers to
16 50 ppm of 1,4-dioxane in a chamber, followed by blood and urine analysis for 1,4-dioxane and HEAA.
17 The study protocol was approved by a seven-member Human Research Review Committee of the Dow
18 Chemical Company, and written informed consent of study participants was obtained. At a concentration
19 of 50 ppm, uptake of 1,4-dioxane into plasma was rapid and approached steady-state conditions by
20 6 hours. The authors reported a calculated absorbed dose of 5.4 mg/kg. However, the exposure chamber
21 atmosphere was kept at a constant concentration of 50 ppm and exhaled breath was not analyzed.
22 Accordingly, gas uptake could not be measured. As a result, the absorbed fraction of inhaled 1,4-dioxane
23 could not be accurately determined in humans. Rats inhaling 50 ppm for 6 hours exhibited 1,4-dioxane
24 and HEAA in urine with an HEAA to 1,4-dioxane ratio of over 3,100:1 (Young etal., 1978a; 1978b).
25 Plasma concentrations at the end of the 6-hour exposure period averaged 7.3 ug/mL. The authors
26 calculated an absorbed 1,4-dioxane dose of 71.9 mg/kg; however, the lack of exhaled breath data and
27 dynamic exposure chamber precluded the accurate determination of the absorbed fraction of inhaled
28 1,4-dioxane.
29 No human data are available to evaluate the oral absorption of 1,4-dioxane. Gastrointestinal
30 absorption was nearly complete in male Sprague Dawley rats orally dosed with 10-1,000 mg/kg of
31 [14C]-l,4-dioxane given as a single dose or as 17 consecutive daily doses (Young et al., 1978a; 1978b).
32 Cumulative recovery of radiolabel in the feces was
-------
1 No human data are available to evaluate the dermal absorption of 1,4-dioxane; however,
2 Bronaugh (1982) reported an in vitro study in which 1,4-dioxane penetrated excised human skin 10 times
3 more under occluded conditions (3.2% of applied dose) than unoccluded conditions (0.3% of applied
4 dose). [14C]-l,4-Dioxane was dissolved in lotion, applied to the excised skin in occluded and unoccluded
5 diffusion cells, and absorption of the dose was recorded 205 minutes after application. Bronaugh (1982)
6 also reported observing rapid evaporation, which further decreased the small amount available for skin
7 absorption.
8 Dermal absorption data in animals are also limited. Dermal absorption in animals was reported to
9 be low following exposure of forearm skin of monkeys (Marzulli et al.. 1981). In this study, Rhesus
10 monkeys were exposed to [14C]-l,4-dioxane in methanol or skin lotion vehicle for 24 hours (skin was
11 uncovered/unoccluded). Only 2-3% of the original radiolabel was cumulatively recovered in urine over a
12 5-day period.
3.2 Distribution
13 No data are available for the distribution of 1,4-dioxane in human tissues. No data are available
14 for the distribution of 1,4-dioxane in animals following oral or inhalation exposures.
15 Mikheev et al. (1990) studied the distribution of [14C]-l,4-dioxane in the blood, liver, kidney,
16 brain, and testes of rats (strain not reported) for up to 6 hours following intraperitoneal (i.p.) injection of
17 approximately one-tenth the median lethal dose (LD50) (actual dose not reported). While actual tissue
18 concentrations were not reported, tissue:blood ratios were given for each tissue at six time points ranging
19 from 5 minutes to 6 hours. The time to reach maximum accumulation of radiolabel was shorter for liver
20 and kidney than for blood or the other tissues, which the authors suggested was indicative of selective
21 membrane transport. Tissue:blood ratios were less than one for all tissues except testes, which had a ratio
22 greater than one at the 6-hour time point. The significance of these findings is questionable since the
23 contribution of residual blood in the tissues was unknown (though saline perfusion may serve to clear
24 tissues of highly water-soluble 1,4-dioxane), the tissue concentrations of radiolabel were not reported, and
25 data were collected from so few time points.
26 Woo et al. (1977a) administered i.p. doses of [3H]-1,4-dioxane (5 mCi/kg body weight [BW]) to
27 male Sprague Dawley rats with and without pretreatment using mixed-function oxidase inducers
28 (phenobarbital, 3-methylcholanthrene, or polychlorinated biphenyls [PCBs]). Liver, kidney, spleen, lung,
29 colon, and skeletal muscle tissues were collected from 1,2,6, and 12 hours after dosing. Distribution was
30 generally uniform across tissues, with blood concentrations higher than tissues at all times except for
31 1 hour post dosing, when kidney levels were approximately 20% higher than blood. Since tissues were
32 not perfused prior to analysis, the contribution of residual blood to radiolabel measurements is unknown,
33 though loss of 1,4-dioxane from tissues would be unknown had saline perfusion been performed.
34 Covalent binding reached peak percentages at 6 hours after dosing in liver (18.5%), spleen (22.6%), and
35 colon (19.5%). At 16 hours after dosing, peak covalent binding percentages were observed in whole blood
36 (3.1%), kidney (9.5%), lung (11.2%), and skeletal muscle (11.2%). Within hepatocytes, radiolabel
6
DRAFT - DO NOT CITE OR QUOTE
-------
1 distribution at 6 hours after dosing was greatest in the cytosolic fraction (43.8%) followed by the
2 microsomal (27.9%), mitochondrial (16.6%), and nuclear (11.7%) fractions. While little covalent binding
3 of radiolabel was measured in the hepatic cytosol (4.6%), greater binding was observed at 16 hours after
4 dosing in the nuclear (64.8%), mitochondrial (45.7%), and microsomal (33.4%) fractions. Pretreatment
5 with inducers of mixed-function oxidase activity did not significantly change the extent of covalent
6 binding in subcellular fractions.
3.3 Metabolism
7 The major product of 1,4-dioxane metabolism appears to be HEAA, although there is one report
8 that identified l,4-dioxane-2-one as a major metabolite (Woo et al.. 1977a). However, the presence of this
9 compound in the sample was believed to result from the acidic conditions (pH of 4.0-4.5) of the
10 analytical procedures. The reversible conversion of HEAA and p-l,4-dioxane-2-one is pH-dependent
11 (Braun and Young. 1977). Braun and Young (1977) identified HEAA (85%) as the major metabolite,
12 with most of the remaining dose excreted as unchanged 1,4-dioxane in the urine of Sprague Dawley rats
13 dosed with 1,000 mg/kg of uniformly labeled l,4-[14C]dioxane. In fact, toxicokinetic studies of
14 1,4-dioxane in humans and rats (Young et al. (1978b: 1978a: 1977)) employed an analytical technique
15 that converted HEAA to the more volatile l,4-dioxane-2-one prior to gas chromatography (GC); however,
16 it is still unclear as to whether HEAA or l,4-dioxane-2-one is the major metabolite of 1,4-dioxane.
17 A proposed metabolic scheme for 1,4-dioxane metabolism (Woo et al.. 1977a) in
18 Sprague Dawley rats is shown in Figure 3-1. Oxidation of 1,4-dioxane to diethylene glycol (pathway a),
19 l,4-dioxane-2-ol (pathway c), or directly to l,4-dioxane-2-one (pathway b) could result in the production
20 of HEAA. 1,4-Dioxane oxidation appears to be cytochrome P450 (CYP450)-mediated, as CYP450
21 induction with phenobarbital or Aroclor 1254 (a commercial PCB mixture) and suppression with
22 2,4-dichloro-6-phenylphenoxy ethylamine or cobaltous chloride were effective in significantly increasing
23 and decreasing, respectively, the appearance of HEAA in the urine of male Sprague Dawley rats
24 following 3 g/kg i.p. dose (Woo et al.. 1978. 1977c). 1,4-Dioxane itself induced CYP450-mediated
25 metabolism of several barbiturates in Hindustan mice given i.p. injections of 25 and 50 mg/kg
26 1,4-dioxane (Mungikar and Pawar. 1978). Of the three possible pathways proposed in this scheme,
27 oxidation to diethylene glycol and HEAA appears to be the most likely, because diethylene glycol was
28 found as a minor metabolite in Sprague Dawley rat urine following a single 1,000 mg/kg gavage dose of
29 1,4-dioxane (Braun and Young, 1977). Additionally, i.p. injection of 100-400 mg/kg diethylene glycol in
30 Sprague Dawley rats resulted in urinary elimination of HEAA (Woo etal.. 1977b).
DRAFT - DO NOT CITE OR QUOTE
-------
C> OH ^OH 0
°- (a) HOH2C CH2OH HOH2C COOH
O
[I] ""---....,
(b)"
+ hi,O
-hLO
~0
[IV]
Source: Adapted with permission of Elsevier Ltd., Woo et al. (1977a: 1977c).
Figure 3-1 Suggested metabolic pathways of 1,4-dioxane in the rat.
Legend: I = 1,4-dioxane; II = diethylene glycol; III = P-hydroxyethoxy acetic acid (HEAA);
IV = l,4-dioxane-2-one; V = l,4-dioxane-2-ol; VI = P-hydroxyethoxy acetaldehyde. Note:
Metabolite [V] is a likely intermediate in pathway b as well as pathway c. The proposed
pathways are based on the metabolites identified; the enzymes responsible for each reaction
have not been determined. The proposed pathways do not account for metabolite
degradation to the labeled carbon dioxide (CO2) identified in expired air after labeled
1,4-dioxane exposure.
1 Metabolism of 1,4-dioxane in humans is extensive. In a survey of 1,4-dioxane plant workers
2 exposed to a TWA of 1.6 ppm of 1,4-dioxane for 7.5 hours, Young et al. (1976) found HEAA and
3 1,4-dioxane in the worker's urine at a ratio of 118:1. Similarly, in adult male volunteers exposed to
4 50 ppm for 6 hours (Young et al.. 1977). over 99% of inhaled 1,4-dioxane (assuming negligible exhaled
5 excretion) appeared in the urine as HEAA. The linear elimination of 1,4-dioxane in both plasma and urine
6 indicated that 1,4-dioxane metabolism was a nonsaturated, first-order process at this exposure level.
7 Like humans, rats extensively metabolize inhaled 1,4-dioxane, as HEAA content in urine was
8 over 3,000-fold higher than that of 1,4-dioxane following exposure to 50 ppm for 6 hours (Young et al..
9 1978a: 1978b). 1,4-Dioxane metabolism in rats was a saturable process, as exhibited by oral and i.v.
10 exposures to various doses of [14C]-l,4-dioxane (Young et al.. 1978a; 1978b). Plasma data from
11 Sprague Dawley rats given single i.v. doses of 3, 10, 30, 100, 300, or 1,000 mg [14C]-l,4-dioxane/kg
12 demonstrated a dose-related shift from linear, first-order to nonlinear, saturable metabolism of
13 1,4-dioxane between plasma 1,4-dioxane levels of 30 and 100 ug/mL (Figure 3-2). Similarly, in rats
14 given, via gavage in distilled water, 10, 100, or 1,000 mg [14C]-l,4-dioxane/kg singly or 10 or 1,000 mg
15 [14C]-l,4-dioxane/kg in 17 daily doses, the percent urinary excretion of the radiolabel decreased
16 significantly with dose while radiolabel in expired air increased. Specifically, with single
17 [14C]-l,4-dioxane/kg doses, urinary radiolabel decreased from 99 to 76% and expired 1,4-dioxane
DRAFT - DO NOT CITE OR QUOTE
-------
1 increased from <1 to 25% as dose increased from 10 to 1,000 mg/kg. Likewise, with multiple daily doses
2 10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary radiolabel decreased from 99 to 82% and expired
3 1,4-dioxane increased from 1 to 9% as dose increased. The differences between single and multiple doses
4 in urinary and expired radiolabel support the notion that 1,4-dioxane may induce its own metabolism.
5 Induction of 1.4-dioxane metabolism was evaluated in a 13 week inhalation study by Kasai et al.
6 (2008). In this study, male and female F344 rats were exposed daily to concentrations of 0 (control). 100.
7 200. 400. 1.600. and 3.200 ppm. Plasma levels of 1.4-dioxane linearly increased with increasing
8 inhalation concentration, suggesting that metabolic saturation was not achieved during the course of the
9 experiments for plasma levels up to 730 and 1.054 ug/mL in male and female rats, respectively, at the
10 highest exposure concentration (3.200 ppm). In contrast. Young et al. (1978b) single dose experiments of
11 inhalation exposure to 50 ppm in male rats showed possible saturation of metabolism at plasma levels of
12 100 u,g/mL. Therefore, lack of the metabolic saturation of 1.4-dioxane found in the Kasai et al. (2008)
13 study is likely attributed to enhanced metabolism by the induction of P450 enzymes, including CYP2E1.
14 by 13 weeks of repeated inhalation exposure to 1.4-dioxane at concentrations up to 3.200 ppm (Kasai et
15 al.. 2008).
10,000
O 5 10 >5
25 3O 35 4O 45 5O 55 60 6S 7C
Source: Reprinted with permission of Taylor and Francis, Young et al. (1978b).
Figure 3-2 Plasma 1,4-dioxane levels in rats following i.v. doses of 3-5,600 mg/kg
[y-axis is plasma concentration of 1,4-dioxane (jig/mL) and x-axis is time (hr)]
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-Dioxane has been shown to induce several isoforms of CYP450 in various tissues following
2 acute oral administration by gavage or drinking water (Nannelli et al.. 2005). Male Sprague Dawley rats
3 were exposed to either 2,000 mg/kg 1,4-dioxane via gavage for 2 consecutive days or by ingestion of a
4 1.5% 1,4-dioxane drinking water solution for 10 days. Both exposures resulted in significantly increased
5 CYP2B1/2, CYP2C11, and CYP2E1 activities in hepatic microsomes. The gavage exposure alone
6 resulted in increased CYP3A activity. The increase in 2C11 activity was unexpected, as that isoform has
7 been observed to be under hormonal control and was typically suppressed in the presence of 2B1/2 and
8 2E1 induction. In the male rat, hepatic 2C11 induction is associated with masculine pulsatile plasma
9 profiles of growth hormone (compared to the constant plasma levels in the female), resulting in
10 masculinization of hepatocyte function ("Waxman et al.. 1991). The authors postulated that 1,4-dioxane
11 may alter plasma growth hormone levels, resulting in the observed 2C11 induction. However, growth
12 hormone induction of 2C11 is primarily dependent on the duration between growth hormone pulses and
13 secondarily on growth hormone plasma levels (Agrawal and Shapiro. 2000; Waxman etal., 1991). Thus,
14 the induction of 2C11 by 1,4-dioxane may be mediated by changes in the time interval between growth
15 hormone pulses rather than changes in growth hormone levels. This may be accomplished by 1,4-dioxane
16 temporarily influencing the presence of growth hormone cell surface binding sites (Agrawal and Shapiro.
17 2000). However, no studies are available to confirm the influence of 1,4-dioxane on either growth
18 hormone levels or changes in growth hormone pulse interval.
19 In nasal and renal mucosal cell microsomes, CYP2E1 activity, but not CYP2B1/2 activity, was
20 increased. Pulmonary mucosal CYP450 activity levels were not significantly altered. Observed increases
21 in 2E1 mRNA in rats exposed by gavage and i.p. injection suggest that 2E1 induction in kidney and nasal
22 mucosa is controlled by a transcriptional activation of 2E1 genes. The lack of increased mRNA in
23 hepatocytes suggests that induction is regulated via a post-transcriptional mechanism. Differences in 2E1
24 induction mechanisms in liver, kidney, and nasal mucosa suggest that induction is controlled in a
25 tissue-specific manner.
3.4 Elimination
26 In workers exposed to a TWA of 1.6 ppm for 7.5 hours, 99% of 1,4-dioxane eliminated in urine
27 was in the form of HEAA (Young et al., 1976). The elimination half-life was 59 minutes in adult male
28 volunteers exposed to 50 ppm 1,4-dioxane for 6 hours, with 90% of urinary 1,4-dioxane and 47% of
29 urinary HEAA excreted within 6 hours of onset of exposure (Young et al., 1977). There are no data for
30 1,4-dioxane elimination in humans from oral exposures.
31 Elimination of 1,4-dioxane in rats (Young et al.. 1978a: 1978b). was primarily via urine. As
32 comparably assessed in humans, the elimination half-life in rats exposed to 50 ppm 1,4-dioxane for
33 6 hours was calculated to be 1.01 hours. In Sprague Dawley rats given single daily doses of 10, 100, or
34 1,000 mg [14C]-l,4-dioxane/kg or multiple doses of 10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary
35 radiolabel ranged from 99% down to 76% of total radiolabel. Fecal elimination was less than 2% for all
36 doses. The effect of saturable metabolism on expired 1,4-dioxane was apparent, as expired 1,4-dioxane in
10
DRAFT - DO NOT CITE OR QUOTE
-------
1 singly dosed rats increased with dose from 0.4 to 25% while expired 14CO2 changed little (between 2 and
2 3%) across doses. The same relationship was seen in Sprague Dawley rats dosed i.v. with 10 or 1,000 mg
3 [14C]-l,4-dioxane/kg. Higher levels of 14CO2 relative to 1,4-dioxane were measured in expired air of the
4 10 mg/kg group, while higher levels of expired 1,4-dioxane relative to 14CO2 were measured in the
5 1,000 mg/kg group.
3.5 Physiologically Based Pharmacokinetic Models
6 Physiologically based pharmacokinetic models (PBPK) models have been developed for
7 1.4-dioxane in rats (Sweeney et al.. 2008; Leung and Paustenbach. 1990; Reitz et al.. 1990). mice (Reitz
8 etal.. 1990). humans (Sweeney et al. 2008; Leung and Paustenbach. 1990; Reitz etal.. 1990). and
9 lactating women (Fisher etal.. 1997). Each of the models simulates the body as a series of compartments
10 representing tissues or tissue groups that receive blood from the central vascular compartment
11 (Figure 3-3). Modeling was conducted under the premise that transfers of 1,4-dioxane between blood and
12 tissues occur sufficiently fast to be effectively blood flow-limited, which is consistent with the available
13 data (Ramsey and Andersen. 1984). Blood time course and metabolite production data in rats and humans
14 suggest that absorption and metabolism are accomplished through common mechanisms in both species
15 (Young et al. (1978b; 1978a; 1977)). allowing identical model structures to be used for both species (and
16 by extension, for mice as well). In all three models, physiologically relevant, species-specific parameter
17 values for tissue volume, blood flow, and metabolism and elimination are used. The models and
18 supporting data are reviewed below, from the perspective of assessing their utility for predicting internal
19 dosimetry and for cross-species extrapolation of exposure-response relationships for critical neoplastic
20 and nonneoplastic endpoints (also see Appendix B).
11
DRAFT - DO NOT CITE OR QUOTE
-------
IV
infusion
J
inhalation
4 t
Lung blood
-Q
O
CD
Z3
O
£=
CD
>
L J
f Rapidly 1
[^ perfused J
[Slowly 1
perfused J
4 F^t L
^ Liver L
"D
O
m
cu
t;
Gl
absorption
metabolism
Figure 3-3 General PBPK model structure.
Consisting of blood-flow limited tissue compartments connected via arterial and venous
blood flows. Note: Orally administered chemicals are absorbed directly into the liver while
inhaled and intravenously infused chemicals enter directly into the arterial and venous
blood pools, respectively.
3.5.1 Available Pharmacokinetic Data
1 Animal and human data sets available for model calibration derive from Young et al. (1978b:
2 1978a: 1977). Mikheev et al. (1990). and Woo et al. (1977a; 1977b). Young et al. Q978b; 1978a) studied
3 the disposition of radiolabeled [14C]-l,4-dioxane in adult male Sprague Dawley rats following i.v.,
4 inhalation, and single and multiple oral gavage exposures. Plasma concentration-time profiles were
5 reported for i.v. doses of 3, 10, 30, 100, and 1,000 mg/kg. In addition, exhaled 14CO2 and urinary
6 1,4-dioxane and HEAA profiles were reported following i.v. doses of 10 and 1,000 mg/kg. The plasma
7 1,4-dioxane concentration-time course, cumulative urinary 1,4-dioxane and cumulative urinary HEAA
8 concentrations were reported following a 6-hour inhalation exposure to 50 ppm. Following oral gavage
9 doses of 10-1,000 mg/kg, percentages of total orally administered radiolabel were measured in urine,
10 feces, expired air, and the whole body.
11 Oral absorption of 1,4-dioxane was extensive, as only approximately 1% of the administered dose
12 appeared in the feces within 72 hours of dosing (Young et al.. 1978a: 1978b). Although it may be
12
DRAFT - DO NOT CITE OR QUOTE
-------
1 concluded that the rate of oral absorption was high enough to ensure nearly complete absorption by
2 72 hours, a more quantitative estimate of the rate of oral absorption is not possible due to the absence of
3 plasma time course data by oral exposure.
4 Saturable metabolism of 1,4-dioxane was observed in rats exposed by either the i.v. or oral routes
5 (Young et al.. 1978a: 1978b). Elimination of 1,4-dioxane from plasma appeared to be linear following i.v.
6 doses of 3-30 mg/kg, but was nonlinear following doses of 100-1,000 mg/kg. Accordingly, 10 mg/kg i.v.
7 doses resulted in higher concentrations of 14CO2 (from metabolized 1,4-dioxane) in expired air relative to
8 unchanged 1,4-dioxane, while 1,000 mg/kg i.v. doses resulted in higher concentrations of expired
9 1,4-dioxane relative to 14CO2. Thus, at higher i.v. doses, a higher proportion of unmetabolized
10 1,4-dioxane is available for exhalation. Taken together, the i.v. plasma and expired air data from Young et
11 al. (1978b: 1978a) corroborate previous studies describing the saturable nature of 1,4-dioxane metabolism
12 in rats (1977a; Woo et al.. 1977b) and are useful for optimizing metabolic parameters (Vmax and Km) in a
13 PBPK model.
14 Similarly, increasing single or multiple oral doses of 10-1,000 mg/kg resulted in increasing
15 percentage of 1,4-dioxane in exhaled air and decreasing percentage of radiolabel (either as 1,4-dioxane or
16 a metabolite) in the urine, with significant differences in both metrics being observed between doses of 10
17 and 100 mg/kg (Young et al., 1978a; 1978b). These data identify the region (10-100 mg/kg) in which oral
18 exposures will result in nonlinear metabolism of 1,4-dioxane and can be used to test whether metabolic
19 parameter value estimates derived from i.v. dosing data are adequate for modeling oral exposures.
20 Post-exposure plasma data from a single 6-hour, 50 ppm inhalation exposure in rats were reported
21 (Young et al., 1978a; 1978b). The observed linear elimination of 1,4-dioxane after inhalation exposure
22 suggests that, via this route, metabolism is in the linear region at this exposure level.
23 The only human data adequate for use in PBPK model development (Young etal. 1977) come
24 from adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours. Plasma 1,4-dioxane and HEAA
25 concentrations were measured both during and after the exposure period, and urine concentrations were
26 measured following exposure. Plasma levels of 1,4-dioxane approached steady-state at 6 hours. HEAA
27 data were insufficient to describe the appearance or elimination of HEAA in plasma. Data on elimination
28 of 1,4-dioxane and HEAA in the urine up to 24 hours from the beginning of exposure were reported. At
29 6 hours from onset of exposure, approximately 90% and 47% of the cumulative (0-24 hours) urinary
30 1,4-dioxane and HEAA, respectively, were measured in the urine. The ratio of HEAA to 1,4-dioxane in
31 urine 24 hours after onset of exposure was 192:1 (similar to the ratio of 118:1 observed by Young et al.
32 (1976) in workers exposed to 1.6 ppm for 7.5 hours), indicating extensive metabolism of 1,4-dioxane As
33 with Sprague Dawley rats, the elimination of 1,4-dioxane from plasma was linear across all observations
34 (6 hours following end of exposure), suggesting that human metabolism of 1,4-dioxane is linear for a
35 50 ppm inhalation exposure to steady-state. Thus, estimation of human Vmax and Km from these data will
36 introduce uncertainty into internal dosimetry performed in the nonlinear region of metabolism.
37 Further data were reported for the tissue distribution of 1,4-dioxane in rats. Mikheev et al. (1990)
38 administered i.p. doses of [14C]-l,4-dioxane to white rats (strain not reported) and reported time-to-peak
39 blood, liver, kidney, and testes concentrations. They also reported ratios of tissue to blood concentrations
13
DRAFT - DO NOT CITE OR QUOTE
-------
1 at various time points after dosing. Woo et al. (1977a; 1977b) administered i.p. doses of [14C]-l,4-dioxane
2 to Sprague Dawley rats and measured radioactivity levels in urine. However, since i.p. dosing is not
3 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
4 Leung and Paustenbach (1990) developed a PBPK model for 1,4-dioxane and its primary
5 metabolite, HEAA, in rats and humans. The model, based on the structure of a PBPK model for styrene
6 (Ramsey and Andersen. 1984). consists of a central blood compartment and four tissue compartments:
7 liver, fat, slowly perfused tissues (mainly muscle and skin), and richly perfused tissues (brain, kidney, and
8 viscera other than the liver). Tissue volumes were calculated as percentages of total BW, and blood flow
9 rates to each compartment were calculated as percentages of cardiac output. Equivalent cardiac output
10 and alveolar ventilation rates were allometrically scaled to a power (0.74) of BW for each species. The
11 concentration of 1,4-dioxane in alveolar blood was assumed to be in equilibrium with alveolar air at a
12 ratio equal to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane
13 between blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
14 between blood and tissue, governed by tissue:blood equilibrium partition coefficients. The latter were
15 derived from the quotient of blood:air and tissue:air partition coefficients, which were measured in vitro
16 (Leung and Paustenbach. 1990) for blood, liver, fat, and skeletal muscle (slowly perfused tissue).
17 Blood:air partition coefficients were measured for both humans and rats. Rat tissue:air partition
18 coefficients were used as surrogate values for humans, with the exception of slowly perfused tissue:blood,
19 which was estimated by optimization to the plasma time-course data. Portals of entry included i.v.
20 infusion (over a period of 36 seconds) into the venous blood, inhalation by diffusion from the alveolar air
21 into the lung blood at the rate of alveolar ventilation, and oral administration via zero-order absorption
22 from the gastrointestinal tract to the liver. Elimination of 1,4-dioxane was accomplished through
23 pulmonary exhalation and saturable hepatic metabolism. Urinary excretion of HEAA was assumed to be
24 instantaneous with the generation of HEAA from the hepatic metabolism of 1,4-dioxane.
25 The parameter values for hepatic metabolism of 1,4-dioxane, Vmax and Km, were optimized and
26 validated against plasma and/or urine time course data for 1,4-dioxane and HEAA in rats following i.v.
27 and inhalation exposures and humans following inhalation exposure (Young et al. (1978b: 1978a: 1977)):
28 the exact data (i.e., i.v., inhalation, or both) used for the optimization and calibration were not reported.
29 Although the liver and fat were represented by tissue-specific compartments, no tissue-specific
30 concentration data were available for model development, raising uncertainty as the model's ability to
31 adequately predict exposure to these tissues. The human inhalation exposure of 50 ppm for 6 hours
32 (Young et al.. 1977) was reported to be in the linear range for metabolism; thus, uncertainty exists in the
33 ability of the allometrically-scaled value for the human metabolic Vmax to accurately describe 1,4-dioxane
34 metabolism from exposures resulting in metabolic saturation. Nevertheless, these values resulted in the
14
DRAFT - DO NOT CITE OR QUOTE
-------
1 model producing good fits to the data. For rats, the values for Vmax had to be adjusted upwards by a factor
2 of 1.8 to reasonably simulate exposures greater than 300 mg/kg. The model authors attributed this to
3 metabolic enzyme induction by high doses of 1,4-dioxane.
3.5.2.2 Reitzetal.
4 Reitz et al. (1990) developed a model for 1,4-dioxane and HEAA in the mouse, rat, and human.
5 This model, also based on the styrene model of Ramsey and Andersen (1984). included a central blood
6 compartment and compartments for liver, fat, and rapidly and slowly perfused tissues. Tissue volumes
7 and blood flow rates were defined as percentages of total BW and cardiac output, respectively.
8 Physiological parameter values were similar to those used by Andersen et al. (1987). except that flow
9 rates for cardiac output and alveolar ventilation were doubled in order to produce a better fit of the model
10 to human blood level data (Young et al.. 1977). Portals of entry included i.v. injection into the venous
11 blood, inhalation, oral bolus dosing, and oral dosing via drinking water. Oral absorption of 1,4-dioxane
12 was simulated, in all three species, as a first-order transfer to liver (halftime approximately 8 minutes).
13 Alveolar blood levels of 1,4-dioxane were assumed to be in equilibrium with alveolar air at a
14 ratio equal to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane
15 between blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
16 between blood and tissue, governed by tissue:blood equilibrium partition coefficients. These coefficients
17 were derived by dividing experimentally measured (Leung and Paustenbach. 1990) in vitro blood:air and
18 tissue:air partition coefficients for blood, liver, fat. Blood:air partition coefficients were measured for both
19 humans and rats. The mouse blood:air partition coefficient was different from rat or human values; the
20 source of the partition coefficient for blood in mice was not reported. Rat tissue:air partition coefficients
21 were used as surrogate values for humans. Rat tissue partition coefficient values were the same values as
22 used in the Leung and Paustenbach (1990) model (with the exception of slowly perfused tissues) and were
23 used in the models for all three species. The liver value was used for the rapidly perfused tissues, as well
24 as slowly perfused tissues. Although slowly perfused tissue:air partition coefficients for rats were
25 measured, the authors suggested that 1,4-dioxane in the muscle and air may not have reached equilibrium
26 in the highly gelatinous tissue homogenate (Reitzetal.. 1990). Substitution of the liver value provided
27 much closer agreement to the plasma data than when the muscle value was used. Further, doubling of the
28 measured human blood:air partition coefficient improved the fit of the model to the human blood level
29 data compared to the fit resulting from the measured value (Reitzetal.. 1990). The Reitz et al. (1990)
30 model simulated three routes of 1,4-dioxane elimination: pulmonary exhalation, hepatic metabolism to
31 HEAA, and urinary excretion of HEAA. The elimination of HEAA was modeled as a first-order transfer
32 of 1,4-dioxane metabolite to urine.
33 Values for the metabolic rate constants, Vmax and Km, were optimized to achieve agreement with
34 various observations. Reitz et al. (1990) optimized values for human Vmax and Km against the
35 experimental human 1,4-dioxane inhalation data (Young et al.. 1977). As noted previously, because the
36 human exposures were below the level needed to exhibit nonlinear kinetics, uncertainty exists in the
15
DRAFT - DO NOT CITE OR QUOTE
-------
1 ability of the optimized value of Vmax to simulate human 1,4-dioxane metabolism above the concentration
2 that would result in saturation of metabolism. Rat metabolic rate constants were obtained by optimization
3 to simulated data from a two compartment empirical pharmacokinetic model, which was fitted to i.v.
4 exposure data (Young etal.. 1978a: 1978b). As with the Leung and
5 The Leung and Paustenbach model (1990) and the Reitz et al. (1990) model included
6 compartments for the liver and fat, although no tissue-specific concentration data were available to
7 validate dosimetry for these organs. The derivations of human and rat HEAA elimination rate constants
8 were not reported. Since no pharmacokinetics data for 1,4-dioxane in mice were available, mouse
9 metabolic rate constants were allometrically scaled from rat and human values.
3.5.2.3 Fisher etal.
10 A PBPK model was developed by Fisher et al. (1997) to simulate a variety of volatile organic
11 compounds (VOCs, including 1,4-dioxane) in lactating humans. This model was similar in structure to
12 those of Leung and Paustenbach (1990) and Reitz et al. (1990) with the addition of elimination of
13 1,4-dioxane to breast milk. Experimental measurements were made for blood:air and milk:air partition
14 coefficients. Other partition coefficient values were taken from Reitz et al. (1990). The model was not
15 optimized, nor was performance tested against experimental exposure data. Thus, the ability of the model
16 to simulate 1,4-dioxane exposure data is unknown.
3.5.2.4 Sweeney et al.
17 The Sweeney et al. (2008) model consisted of fat liver, slowly perfused, and other well perfused
18 tissue compartments. Lung and stomach compartments were used to describe the route of exposure, and
19 an overall volume of distribution compartment was used for calculation of urinary excretion levels of
20 1.4-dioxane and HEAA. Blood, saline, and tissue to air partition coefficient values for 1.4-dioxane were
21 experimentally determined for rats and mice. Average values of the rat and mouse partition coefficients
22 were used for humans. Metabolic constants (VmaxC and Km) for the rat were derived by. optimization of
23 data from an i.v. exposure of 1.000 mg/kg (Young et al.. 1978b) for inducible metabolism. For uninduced
24 VmaxC estimation, data generated by_ i.v. exposures to 3^ 10. 30. and 100 mg/kg were used (Young et al..
25 1978b). Sweeney et al. (2008) determined best fit values for VmaxC by fitting to blood data in Young et
26 al. (1978b). The best fit VmaxC values were 7.5. 10.8. and 12.7 mg/hr-kg°75 for i.v. doses of 3 to 100.
27 300. and 1.000 mg/kg. suggesting a gradual dose dependent increase in metabolic rate over i.v. doses
28 ranging from 3 to 1.000 mg/kg. Although the Sweeney et al. (2008) model utilized two values for VmaxC
29 (induced and uninduced). the PBPK model does not include a dose-dependent function description of the
30 change of Vmax for i.v. doses between metabolic induced and uninduced exposures. Mouse VmaxC and
31 absorption constants were derived by optimizing fits to the blood 1.4-dioxane concentrations in mice
32 administered nominal doses of 200 and 2.000 mg/kg 1.4-dioxane via gavage in a water vehicle (Young et
33 al.. 1978b). The in vitro Vmax values for rats and mice were scaled to estimate in vivo rates. The scaled
16
DRAFT - DO NOT CITE OR QUOTE
-------
1 and optimized rat VmaxC values were similar. The discrepancy between the scaled and optimized mouse
2 values was larger, which was attributed to possible induction in mice at the lowest dose tested (200
3 mg/kg). The ratio of optimized/scaled values for the rat was used to adjust the scaled human VmaxC and
4 Km values to projected in vivo values.
5 The Sweeney et al. (2008) model outputs were compared, by visual inspection, with data not used
6 in fitting model parameters. The model predictions gave adequate match to the 1.4-dioxane exhalation
7 data in rats after a 1.000 mg/kg i.v. dose. 1.4-Dioxane exhalation was overpredicted by a factor of about 3
8 after a 10 mg/kg i.v. dose. Similarly, the simulations of exhaled 1.4-dioxane after oral dosing were
9 adequate at 1.000 mg/kg and 100 mg/kg (within 50%). but poor at 10 mg/kg (model over predicted by a
10 factor of 5). The model did not adequately fit the human data (Young et al., 1977). Using physiological
11 parameters of Brown et al. (1997) and measured partitioning parameters (Sweeney et al.. 2008; Leung and
12 Paustenbach. 1990) with no metabolism, measured blood 1.4-dioxane concentrations reported by Young
13 et al. (1977) could not be achieved unless the estimated exposure concentration was increased by 2-fold.
14 As expected, inclusion of any metabolism resulted in a decrease in predicted blood concentrations. If
15 estimated metabolism rates were used with the reported exposure concentration, urinary metabolite
16 excretion was also underpredicted (Sweeney et al.. 2008).
3.5.3 Implementation of Published PBPK Models for 1,4-Dioxane
17 As previously described, several pharmacokinetic models have been developed to predict the
18 absorption, distribution, metabolism, and elimination of 1,4-dioxane in rats and humans. Single
19 compartment, empirical models for rats (Young et al.. 1978a: 1978b) and humans (Young et al.. 1977)
20 were developed to predict blood levels of 1,4-dioxane and urine levels of the primary metabolite, HEAA.
21 PBPK models that describe the kinetics of 1,4-dioxane using biologically realistic flow rates, tissue
22 volumes, enzyme affinities, metabolic processes, and elimination behaviors were also developed
23 (Sweeney et al.. 2008: Fisher etal.. 1997: Leung and Paustenbach. 1990: Reitzetal.. 1990).
24 In developing updated toxicity values for 1,4-dioxane the available PBPK models were evaluated
25 for their ability to predict observations made in experimental studies of rat and human exposures to
26 1,4-dioxane (Appendix B). The Reitz et al. (1990) and Leung and Paustenbach (1990) PBPK models were
27 both developed from a PBPK model of styrene (Ramsey and Andersen. 1984). with the exception of
28 minor differences in the use of partition coefficients and biological parameters. The model code for Leung
29 and Paustenbach (1990) was unavailable in contrast to Reitz et al. (1990). The model of Reitz et al.
30 (1990) was identified for further consideration to assist in the derivation of toxicity values, and the
31 Sweeney et al. (2008) PBPK model was also evaluated.
32 The biological plausibility of parameter values in the Reitz et al. (1990) human model were
33 examined. The model published by Reitz et al. (1990) was able to predict the only available human
34 inhalation data (50 ppm 1,4-dioxane for 6 hours; Young et al., (1977)) by increasing (i.e., approximately
35 doubling) the parameter values for human alveolar ventilation (30 L/hour/kg°74), cardiac output (30
36 L/hour/kg°74), and the blood:air partition coefficient (3,650) above the measured values of
17
DRAFT - DO NOT CITE OR QUOTE
-------
1 13 L/minute/kg074 (Brown et al.. 1997). 14 L/hour/kg°74 (Brown etal.. 1997). and 1,825 (Leung and
2 Paustenbach. 1990). respectively. Furthermore, Reitz et al. (1990) replaced the measured value for the
3 slowly perfused tissue:air partition coefficient (i.e., muscle—value not reported in manuscript) with the
4 measured liver value (1,557) to improve the fit. Analysis of the Young et al. (1977) human data suggested
5 that the apparent volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold higher in rats
6 than humans, presumably due to species differences in tissue partitioning or other process not represented
7 in the model. Based upon these observations, several model parameters (e.g., metabolism/elimination
8 parameters) were re-calibrated using biologically plausible values for flow rates and tissue:air partition
9 coefficients.
10 Appendix B describes all activities that were conducted in the evaluation of the empirical models
11 and the re-calibration and evaluation of the Reitz et al. (1990) PBPK model to determine the adequacy
12 and preference for the potential use of the models.
13 The evaluation consisted of implementation of the Young et al. (1978b; 1978a; 1977) empirical
14 rat and human models using the acslXtreme simulation software, re-calibration of the Reitz et al. (1990)
15 human PBPK model, and evaluation of the model parameters published by Sweeney et al. (2008). Using
16 the model descriptions and equations given in Young et al. (1978b: 1978a: 1977). model code was
17 developed for the empirical models and executed, simulating the reported experimental conditions. The
18 model output was then compared with the model output reported in Young et al. (1978b; 1978a; 1977).
19 The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for cardiac and
20 alveolar flow rates and tissue:air partition coefficients. The predictions of blood and urine levels of
21 1,4-dioxane and HEAA, respectively, from the re-calibrated model were compared with the empirical
22 model predictions of the same dosimeters to determine whether the re-calibrated PBPK model could
23 perform similarly to the empirical model. As part of the PBPK model evaluation, EPA performed a
24 sensitivity analysis to identify the model parameters having the greatest influence on the primary
25 dosimeter of interest, the blood level of 1,4-dioxane. Variability data for the experimental measurements
26 of the tissue: air partition coefficients were incorporated to determine a range of model outputs bounded
27 by biologically plausible values for these parameters. Model parameters from Sweeney et al. (2008) were
28 also tested to evaluate the ability of the PBPK model to predict human data following exposure to
29 1,4-dioxane.
30 The rat and human empirical models of Young et al. (1978b; 1978a; 1977) were successfully
31 implemented in acslXtreme and perform identically to the models reported in the published papers
32 (Figure B-3 through Figure B-7), with the exception of the lower predicted HEAA concentrations and
33 early appearance of the peak HEAA levels in rat urine. The early appearance of peak HEAA levels cannot
34 presently be explained, but may result from manipulations of kme or other parameters by Young et al.
35 (1978b: 1978a) that were not reported. The lower predictions of HEAA levels are likely due to reliance on
36 a standard urine volume production rate in the absence of measured (but unreported) urine volumes.
37 While the human urinary HEAA predictions were lower than observations, this is due to parameter fitting
38 of Young et al. (1977). No model output was published in Young et al. (1977) for comparison. The
39 empirical models were modified to allow for user-defined inhalation exposure levels. However, no
18
DRAFT - DO NOT CITE OR QUOTE
-------
1 modifications were made to model oral exposures as adequate data to parameterize such modifications do
2 not exist for rats or humans. Further evaluations of the Young et aL (1977) modified model were
3 conducted against data from the Kasai et al. (2008) subchronic inhalation study. The results of this
4 evaluation are shown in Appendix B (Figure B-8). It shows that the Young et al. (1977) inhalation
5 empirical model failed to provide an adequate simulation of the 13 week inhalation exposure blood data
6 of Kasai et al. (2008). Since the Young et al. (1977) model consistently overpredicted the Kasai et al.
7 (2008) data, the lack of model fit is most likely due to the lack of inclusion of other metabolic processes
8 or parameters.
9 Several procedures were applied to the Reitz et al. (1990) human PBPK model to determine if an
10 adequate fit of the model to the empirical model output or experimental observations could be attained
11 using biologically plausible values for the model parameters. The re-calibrated model predictions for
12 blood 1,4-dioxane levels do not come within 10-fold of the experimental values using measured tissue:air
13 partition coefficients from Leung and Paustenbach (1990) or Sweeney et al. (2008) (Figure B-9 and
14 Figure B-10). The utilization of a slowly perfused tissue:air partition coefficient 10-fold lower than
15 measured values produces exposure-phase predictions that are much closer to observations, but does not
16 replicate the elimination kinetics (Figure B-l 1). Recalibration of the model with upper bounds on the
17 tissue:air partition coefficients results in predictions that are still six- to sevenfold lower than empirical
18 model prediction or observations (Figure B-13 and Figure B-14). Exploration of the model space using an
19 assumption of zero-order metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate
20 fit to the exposure and elimination data can be achieved only when unrealistically low values are assumed
21 for the slowly perfused tissue:air partition coefficient (Figure B-17). Artificially low values for the other
22 tissue:air partition coefficients are not expected to improve the model fit, as these parameters are shown
23 in the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km. In the absence
24 of actual measurements for the human slowly perfused tissue:air partition coefficient, high uncertainty
25 exists for this model parameter value. Differences in the ability of rat and human blood to bind
26 1,4-dioxane may contribute to the difference in Vd. However, this is expected to be evident in very
27 different values for rat and human blood:air partition coefficients, which is not the case (Table B-l).
28 Therefore, some other, as yet unknown, modification to model structure may be necessary.
29 Similarly, Sweeney et al. (2008) also evaluated the available PBPK models (Leung and
30 Paustenbach. 1990; Reitz etal., 1990) for 1,4-dioxane. To address uncertainties and deficiencies in these
31 models, the investigators conducted studies to fill data gaps and reduce uncertainties pertaining to the
32 pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. The following studies were
33 performed:
34 • Partition coefficients, including measurements for mouse blood and tissues (liver, kidney, fat, and
35 muscle) and confirmatory measurements for human blood and rat blood and muscle.
36 • Blood time course measurements in mice conducted for gavage administration of nominal single
37 doses (20, 200, or 2,000 mg/kg) of 1,4-dioxane administered in water.
38 • Metabolic rate constants for rat, mouse, and human liver based on incubations of 1,4-dioxane
39 with rat, mouse, and human hepatocytes and measurement of HEAA.
19
DRAFT - DO NOT CITE OR QUOTE
-------
1 The studies conducted by Sweeney et al. (2008) resulted in partition coefficients that were
2 consistent with previously measured values and those used in the Leung and Paustenbach (1990) model.
3 Of noteworthy significance, the laboratory results of Sweeney et al. (2008) did not confirm the human
4 blood:air partition coefficient Reitz et al. (1990) reported. Furthermore, Sweeney et al. (2008) estimated
5 metabolic rate constants (VmaxC and Km) within the range used in the previous models (Leung and
6 Paustenbach. 1990; Reitz etal.. 1990). Overall, the Sweeney et al. (2008) model utilized more rodent in
7 vivo and in vitro data in model parameterization and refinement; however, the model was still unable to
8 adequately predict the human blood data from Young et al. (1977).
9 Updated PBPK models were developed based on these new data and data from previous kinetic
10 studies in rats, workers, and human volunteers reported by Young et al. (1978b; 1978a; 1977; 1976). The
11 optimized rate of metabolism for the mouse was significantly higher than the value previously estimated.
12 The optimized rat kinetic parameters were similar to those in the 1990 models. Of the two available
13 human studies (Young (1977; 1976). model predictions were consistent with one study, but did not fit the
14 second as well.
3.6 Rat Nasal Exposure via Drinking Water
15 Sweeney et al. (2008) conducted a rat nasal exposure study to explore the potential for direct
16 contact of nasal tissues with 1,4-dioxane-containing drinking water under bioassay conditions. Two
17 groups of male Sprague Dawley rats (5/group) received drinking water in 45-mL drinking water bottles
18 containing a fluorescent dye mixture (Cell Tracker Red/FluoSpheres). The drinking water for one of these
19 two groups also contained 0.5% 1,4-dioxane, a concentration within the range used in chronic toxicity
20 studies. A third group of five rats received tap water alone (controls). Water was provided to the rats
21 overnight. The next morning, the water bottles were weighed to estimate the amounts of water consumed.
22 Rats were sacrificed and heads were split along the midline for evaluation by fluorescence microscopy.
23 One additional rat was dosed twice by gavage with 2 mL of drinking water containing fluorescent dye
24 (the second dose was 30 minutes after the first dose; total of 4 mL administered) and sacrificed 5 hours
25 later to evaluate the potential for systemic delivery of fluorescent dye to the nasal tissues.
26 The presence of the fluorescent dye mixture had no measurable impact on water consumption;
27 however, 0.5% 1,4-dioxane reduced water consumption by an average of 62% of controls following a
28 single, overnight exposure. Fluorescent dye was detected in the oral cavity and nasal airways of each
29 animal exposed to the Cell Tracker Red/FluoSpheres mixture in their drinking water, including numerous
30 areas of the anterior third of the nose along the nasal vestibule, maxillary turbinates, and dorsal
31 nasoturbinates. Fluorescent dye was occasionally detected in the ethmoid turbinate region and
32 nasopharynx. 1,4-Dioxane had no effect on the detection of the dye. Little or no fluorescence at the
33 wavelength associated with the dye mixture was detected in control animals or in the single animal that
34 received the dye mixture by oral gavage. The investigators concluded that the findings indicate rat nasal
35 tissues are exposed by direct contact with drinking water under bioassay conditions.
20
DRAFT - DO NOT CITE OR QUOTE
-------
4 HAZAR D IDE NOTIFICATION
4.1 Studies in Humans - Epidemiology, Case Reports, Clinical
Controls
1 Case reports of acute occupational poisoning with 1,4-dioxane indicated that exposure to high
2 concentrations resulted in liver, kidney, and central nervous system (CNS) toxicity (Johnstone. 1959;
3 Barber. 1934). Barber (1934) described four fatal cases of hemorrhagic nephritis and centrilobular
4 necrosis of the liver attributed to acute inhalation exposure to high (unspecified) concentrations of
5 1,4-dioxane. Death occurred within 5-8 days of the onset of illness. Autopsy findings suggested that the
6 kidney toxicity may have been responsible for lethality, while the liver effects may have been compatible
7 with recovery. Jaundice was not observed in subjects and fatty change was not apparent in the liver.
8 Johnstone (1959) presented the fatal case of one worker exposed to high concentrations of 1,4-dioxane
9 through both inhalation and dermal exposure for a 1 week exposure duration. Measured air concentrations
10 in the work environment of this subject were 208-650 ppm, with a mean value of 470 ppm. Clinical signs
11 that were observed following hospital admission included severe epigastric pain, renal failure, headache,
12 elevation in blood pressure, agitation and restlessness, and coma. Autopsy findings revealed significant
13 changes in the liver, kidney, and brain. These included centrilobular necrosis of the liver and hemorrhagic
14 necrosis of the kidney cortex. Perivascular widening was observed in the brain with small foci of
15 demyelination in several regions (e.g., cortex, basal nuclei). It was suggested that these neurological
16 changes may have been secondary to anoxia and cerebral edema.
17 Several studies examined the effects of acute inhalation exposure in volunteers. In a study
18 performed at the Pittsburgh Experimental Station of the U.S. Bureau of Mines, eye irritation and a
19 burning sensation in the nose and throat were reported in five men exposed to 5,500 ppm of 1,4-dioxane
20 vapor for 1 minute (Yant et al.. 1930). Slight vertigo was also reported by three of these men. Exposure to
21 1,600 ppm of 1,4-dioxane vapor for 10 minutes resulted in similar symptoms with a reduced intensity of
22 effect. In a study conducted by the Government Experimental Establishment at Proton, England (Fairley
23 etal.. 1934). four men were exposed to 1,000 ppm of 1,4-dioxane for 5 minutes. Odor was detected
24 immediately and one volunteer noted a constriction in the throat. Exposure of six volunteers to 2,000 ppm
25 for 3 minutes resulted in no symptoms of discomfort. Wirth and Klimmer (1936). of the Institute of
26 Pharmacology, University of Wurzburg, reported slight mucous membrane irritation in the nose and
27 throat of several human subjects exposed to concentrations greater than 280 ppm for several minutes.
28 Exposure to approximately 1,400 ppm for several minutes caused a prickling sensation in the nose and a
29 dry and scratchy throat. Silverman et al. (1946) exposed 12 male and 12 female subjects to varying air
30 concentrations of 1,4-dioxane for 15 minutes. A 200 ppm concentration was reported to be tolerable,
31 while a concentration of 300 ppm caused irritation to the eyes, nose, and throat. The study conducted by
32 Silverman et al. (1946) was conducted by the Department of Industrial Hygiene, Harvard School of
33 Public Health, and was sponsored and supported by a grant from the Shell Development Company. These
34 volunteer studies published in the 1930s and 1940s (Silverman et al.. 1946; Wirth and Klimmer. 1936;
35 Fairley et al.. 1934; Yantetal.. 1930) did not provide information on the human subjects research ethics
21
DRAFT - DO NOT CITE OR QUOTE
-------
1 procedures undertaken in these studies; however, there is no evidence that the conduct of the research was
2 fundamentally unethical or significantly deficient relative to the ethical standards prevailing at the time
3 the research was conducted.
4 Young et al. (1977) exposed four healthy adult male volunteers to a 50-ppm concentration of
5 1,4-dioxane for 6 hours. The investigators reported that the protocol of this study was approved by a
6 seven-member Human Research Review Committee of the Dow Chemical Company and was followed
7 rigorously. Perception of the odor of 1,4-dioxane appeared to diminish over time, with two of the four
8 subjects reporting inability to detect the odor at the end of the exposure period. Eye irritation was the only
9 clinical sign reported in this study. The pharmacokinetics and metabolism of 1,4-dioxane in humans were
10 also evaluated in this study (see Section 3.3). Clinical findings were not reported in four workers exposed
11 in the workplace to a TWA concentration of 1.6 ppm for 7.5 hours (Young et al.. 1976).
12 Ernstgard et al. (2006) examined the acute effects of 1,4-dioxane vapor in male and female
13 volunteers. The study protocol was approved by the Regional Ethics Review Board in Stockholm, and
14 performed following informed consent and according to the Helsinki declaration. In a screening study by
15 these investigators, no self-reported symptoms (based on a visual analogue scale (VAS) that included
16 ratings for discomfort in eyes, nose, and throat, breathing difficulty, headache, fatigue, nausea, dizziness,
17 or feeling of intoxication) were observed at concentrations up to 20 ppm; this concentration was selected
18 as a tentative no-observed-adverse-effect-level (NOAEL) in the main study. In the main study, six male
19 and six female healthy volunteers were exposed to 0 or 20 ppm 1,4-dioxane, at rest, for 2 hours. This
20 exposure did not significantly affect symptom VAS ratings, blink frequency, pulmonary function or nasal
21 swelling (measured before and at 0 and 3 hours after exposure), or inflammatory markers in the plasma
22 (C-reactive protein and interleukin-6) of the volunteers. Only ratings for "solvent smell" were
23 significantly increased during exposure.
24 Only two well documented epidemiology studies were available for occupational workers
25 exposed to 1,4-dioxane (Buffler et al.. 1978; Thiess et al., 1976). These studies did not provide evidence
26 of effects in humans; however, the cohort size and number of reported cases were small.
4.1.1 Thiess etal.
27 A cross-sectional survey was conducted by Thiess et al. (1976) in German workers exposed to
28 1,4-dioxane. The study evaluated health effects in 74 workers, including 24 who were still actively
29 employed in 1,4-dioxane production at the time of the investigation, 23 previously exposed workers who
30 were still employed by the manufacturer, and 27 retired or deceased workers. The actively employed
31 workers were between 32 and 62 years of age and had been employed in 1,4-dioxane production for 5-
32 41 years. Former workers (age range not given) had been exposed to 1,4-dioxane for 3-38 years and
33 retirees (age range not given) had been exposed for 12-41 years. Air concentrations in the plant at the
34 time of the study were 0.06-0.69 ppm. A simulation of previous exposure conditions (prior to 1969)
35 resulted in air measurements between 0.06 and 7.2 ppm.
22
DRAFT - DO NOT CITE OR QUOTE
-------
1 Active and previously employed workers underwent a thorough clinical examination and X-ray,
2 and hematological and serum biochemistry parameters were evaluated. The examination did not indicate
3 pathological findings for any of the workers and no indication of malignant disease was noted.
4 Hematology results were generally normal. Serum transaminase levels were elevated in 16 of the
5 47 workers studied; however, this finding was consistent with chronic consumption of more than
6 80 grams of alcohol per day, as reported for these workers. No liver enlargement or jaundice was found.
7 Renal function tests and urinalysis were normal in exposed workers. Medical records of the 27 retired
8 workers (15 living at the time of the study) were reviewed. No symptoms of liver or kidney disease were
9 reported and no cancer was detected. Medical reasons for retirement did not appear related to 1,4-dioxane
10 exposure (e.g., emphysema, arthritis).
11 Chromosome analysis was performed on six actively employed workers and six control persons
12 (not characterized). Lymphocyte cultures were prepared and chromosomal aberrations were evaluated. No
13 differences were noted in the percent of cells with gaps or other chromosome aberrations. Mortality
14 statistics were calculated for 74 workers of different ages and varying exposure periods. The proportional
15 contribution of each of the exposed workers to the total time of observation was calculated as the sum of
16 man-years per 10-year age group. Each person contributed one man-year per calendar year to the specific
17 age group in which he was included at the time. The expected number of deaths for this population was
18 calculated from the age-specific mortality statistics for the German Federal Republic for the years 1970-
19 1973. From the total of 1,840.5 person-years, 14.5 deaths were expected; however, only 12 deaths were
20 observed in exposed workers between 1964 and 1974. Two cases of cancer were reported, including one
21 case of lamellar epithelial carcinoma and one case of myelofibrosis leukemia. These cancers were not
22 considered to be the cause of death in these cases and other severe illnesses were present. Standardized
23 mortality ratios (SMRs) for cancer did not significantly differ from the control population (SMR for
24 overall population = 0.83; SMR for 65-75-year-old men = 1.61; confidence intervals (CIs) were not
25 provided).
4.1.2 Buffleretal.
26 Buffler et al. (1978) conducted a mortality study on workers exposed to 1,4-dioxane at a chemical
27 manufacturing facility in Texas. 1,4-Dioxane exposure was known to occur in a manufacturing area and
28 in a processing unit located 5 miles from the manufacturing plant. Employees who worked between April
29 1, 1954, and June 30, 1975, were separated into two cohorts based on at least 1 month of exposure in
30 eitherthe manufacturing plant (100 workers) or the processing area (65 workers). Company records and
31 follow-up techniques were used to compile information on name, date of birth, gender, ethnicity, job
32 assignment and duration, and employment status at the time of the study. Date and cause of death were
33 obtained from copies of death certificates and autopsy reports (if available). Exposure levels for each job
34 category were estimated using the 1974 Threshold Limit Value for 1,4-dioxane (i.e., 50 ppm) and
35 information from area and personal monitoring. Exposure levels were classified as low (<25 ppm),
36 intermediate (50-75 ppm), and high (>75 ppm). Monitoring was not conducted prior to 1968 in the
37 manufacturing areas or prior to 1974 in the processing area; however, the study authors assumed that
23
DRAFT - DO NOT CITE OR QUOTE
-------
1 exposures would be comparable, considering that little change had been made to the physical plant or the
2 manufacturing process during that time. Exposure to 1,4-dioxane was estimated to be below 25 ppm for
3 all individuals in both cohorts. Manufacturing area workers were exposed to several other additional
4 chemicals and processing area workers were exposed to vinyl chloride.
5 Seven deaths were identified in the manufacturing cohort and five deaths were noted for the
6 processing cohort. The average exposure duration was not greater for those workers who died, as
7 compared to those still living at the time of the study. Cancer was the underlying cause of death for two
8 cases from the manufacturing area (carcinoma of the stomach, alveolar cell carcinoma) and one case from
9 the processing area (malignant mediastinal tumor). The workers from the manufacturing area were
10 exposed for 28 or 38 months and both had a positive smoking history (>1 pack/day). Smoking history was
11 not available for processing area workers. The single case of cancer in this area occurred in a 21-year-old
12 worker exposed to 1,4-dioxane for 1 year. The mortality data for both industrial cohorts were compared to
13 age-race-sex specific death rates for Texas (1960-1969). Person-years of observation contributed by
14 workers were determined over five age ranges with each worker contributing one person-year for each
15 year of observation in a specific age group. The expected number of deaths was determined by applying
16 the Texas 1960-1969 death rate statistics to the number of person years calculated for each cohort. The
17 observed and expected number of deaths for overall mortality (i.e., all causes) was comparable for both
18 the manufacturing area (7 observed versus 4.9 expected) and the processing area (5 observed versus
19 4.9 expected). No significant excess in cancer-related deaths was identified for both areas of the facility
20 combined (3 observed versus 1.7 expected). A separate analysis was performed to evaluate mortality in
21 manufacturing area workers exposed to 1,4-dioxane for more than 2 years. Six deaths occurred in this
22 group as compared to 4.1 expected deaths. The use of a conditional Poisson distribution indicated no
23 apparent excess in mortality or death due to malignant neoplasms in this study. It is important to note that
24 the cohorts evaluated were limited in size. In addition, the mean exposure duration was less than 5 years
25 (<2 years for 43% of workers) and the latency period for evaluation was less than 10 years for 59% of
26 workers. The study authors recommended a follow-up investigation to allow for a longer latency period;
27 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
28 The majority of the subchronic and chronic studies conducted for 1.4-dioxane were drinking
29 water studies. To date, there are only two subchronic inhalation studies (Kasai et al., 2008; Fairley et al.
30 1934) and two chronic inhalation studies (Kasai et al.. 2009; Torkelson et al.. 1974). The effects
31 following oral and inhalation exposures are described in detail below.
24
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.1 OralToxicity
4.2.1.1 S ubchronic Oral Toxicity
1 Six rats and six mice (unspecified strains) were given drinking water containing 1.25%
2 1,4-dioxane for up to 67 days (Fairley et al.. 1934). Using reference BWs and drinking water ingestion
3 rates for rats and mice (U.S. EPA. 1988). it can be estimated that these rats and mice received doses of
4 approximately 1,900 and 3,300 mg/kg-day, respectively. Gross pathology and histopathology were
5 evaluated in all animals. Five of the six rats in the study died or were killed in extremis prior to day 34 of
6 the study. Mortality was lower in mice, with five of six mice surviving up to 60 days. Kidney enlargement
7 was noted in 5/6 rats and 2/5 mice. Renal cortical degeneration was observed in all rats and 3/6 mice.
8 Large areas of necrosis were observed in the cortex, while cell degeneration in the medulla was slight or
9 absent. Tubular casts were observed and vascular congestion and hemorrhage were present throughout the
10 kidney. Hepatocellular degeneration with vascular congestion was also noted in five rats and three mice.
11 For this assessment, EPA identified the tested doses of 1,900 mg/kg-day in rats and 3,300 mg/kg-day in
12 mice as the lowest-observed-adverse-effect-levels (LOAELs) for liver and kidney degeneration in this
13 study.
14 4.2.1.1.1 S toner et a I. 1,4-Dioxane was evaluated by Stoner et al. (1986) for its ability to
15 induce lung adenoma formation in A/J mice. Six- to 8-week-old male and female A/J mice (16/sex/group)
16 were given 1,4-dioxane by gavage or i.p. injection, 3 times/week for 8 weeks. Total cumulative dose
17 levels were given as 24,000 mg/kg (oral), and 4,800, 12,000, or 24,000 mg/kg (i.p.). Average daily dose
18 estimates were calculated to be 430 mg/kg-day (oral), and 86, 210, or 430 mg/kg-day (i.p.) by assuming
19 an exposure duration of 56 days. The authors indicated that i.p. doses represent the maximum tolerated
20 dose (MTD), 0.5 times the MTD, and 0.2 times the MTD. Mice were killed 24 weeks after initiation of
21 the bioassay, and lungs, liver, kidney, spleen, intestines, stomach, thymus, salivary, and endocrine glands
22 were examined for gross lesions. Histopathology examination was performed if gross lesions were
23 detected. 1,4-Dioxane did not induce lung tumors in male or female A/J mice in this study.
14 4.2.1.1.2 Stottetal. In the Stott et al. (1981) study, male Sprague Dawley rats
15 (4-6/group) were given average doses of 0, 10, or 1,000 mg/kg-day 1,4-dioxane (>99% pure) in their
16 drinking water, 7 days/week for 11 weeks. It should be noted that the methods description in this report
17 stated that the high dose was 100 mg/kg-day, while the abstract, results, and discussion sections indicated
18 that the high dose was 1,000 mg/kg-day. Rats were implanted with a [6"3H]thymidine loaded osmotic
19 pump 7 days prior to sacrifice. Animals were sacrificed by cervical dislocation and livers were removed,
20 weighed, and prepared for histopathology evaluation. [3F£]-Thymidine incorporation was measured by
21 liquid scintillation spectroscopy.
14 An increase in the liver to BW ratio was observed in rats from the high dose group (assumed to
15 be 1,000 mg/kg-day). Histopathological alterations, characterized as minimal centrilobular swelling, were
16 also seen in rats from this dose group (incidence values were not reported). Hepatic DNA synthesis,
17 measured by [3H]-thymidine incorporation, was increased 1.5-fold in high-dose rats. No changes relative
25
DRAFT - DO NOT CITE OR QUOTE
-------
1 to control were observed for rats exposed to 10 mg/kg-day. EPA found a NOAEL value of 10 mg/kg-day
2 and a LOAEL value of 1,000 mg/kg-day for this study based on histopathological changes in the liver.
3 Stott et al. (1981) also performed several acute experiments designed to evaluate potential
4 mechanisms for the carcinogenicity of 1,4-dioxane. These experiments are discussed separately in Section
5 4.5.2 (Mechanistic Studies).
6 4.2.1.1.3 Kanoetal. In the Kano et al. (2008) study, groups of 6-week-old F344/DuCrj
7 rats (10/sex/group) and Crj:BDFl mice (10/sex/group) were administered 1,4-dioxane (>99% pure) in the
8 drinking water for 13 weeks. The animals were observed daily for clinical signs of toxicity. Food
9 consumption and BWs were measured once per week and water consumption was measured twice
10 weekly. Food and water were available ad libitum. The concentrations of 1,4-dioxane in the water for rats
11 and mice were 0, 640, 1,600, 4,000, 10,000, or 25,000 ppm. The investigators used data from water
12 consumption and BW changes to calculate a daily intake of 1,4-dioxane by the male and female animals.
13 Thus, male rats received doses of approximately 0, 52, 126, 274, 657, and 1,554 mg 1,4-dioxane/kg-day
14 and female rats received 0, 83, 185, 427, 756, and 1,614 mg/kg-day. Male mice received 0, 86, 231, 585,
15 882, or 1,570 mg/kg-day and female mice received 0, 170, 387, 898, 1,620, or 2,669 mg/kg-day.
6 No information was provided as to when the blood and urine samples were collected.
7 Hematology analysis included red blood cell (RBC) count, hemoglobin, hematocrit, mean corpuscular
8 volume (MCV), platelet count, white blood cell (WBC) count, and differential WBCs. Serum
9 biochemistry included total protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only),
10 alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), leucine
11 aminopeptidase (LAP), alkaline phosphatase (ALP), creatinine phosphokinase (CPK) (rat only), urea
12 nitrogen, creatinine (rat only), sodium, potassium, chloride, calcium (rat only), and inorganic phosphorous
13 (rat only). Urinalysis parameters were pH, protein, glucose, ketone body, bilirubin (rat only), occult
14 blood, and urobilinogen. Organ weights (brain, lung, liver, spleen, heart, adrenal, testis, ovary, and
15 thymus) were measured, and gross necropsy and histopathologic examination of tissues and organs were
16 performed on all animals (skin, nasal cavity, trachea, lungs, bone marrow, lymph nodes, thymus, spleen,
17 heart, tongue, salivary glands, esophagus, stomach, small and large intestine, liver, pancreas, kidney,
18 urinary bladder, pituitary thyroid adrenal, testes, epididymis, seminal vesicle, prostate, ovary, uterus,
19 vagina, mammary gland, brain, spinal cord, sciatic nerve, eye, Harderian gland, muscle, bone, and
20 parathyroid). Dunnett's test and %2 test were used to assess the statistical significance of changes in
21 continuous and discrete variables, respectively.
22 Clinical signs of toxicity in rats were not discussed in the study report. One female rat in the high
23 dose group (1,614 mg/kg-day) group died, but cause and time of death were not specified. Final BWs
24 were reduced at the two highest dose levels in females (12 and 21%) and males (7 and 21%), respectively.
25 Food consumption was reduced 13% in females at 1,614 mg/kg-day and 8% in 1,554 mg/kg-day males. A
26 dose-related decrease in water consumption was observed in male rats starting at 52 mg/kg-day (15%)
27 and in females starting at 185 mg/kg-day (12%). Increases in RBCs, hemoglobin, hematocrit, and
28 neutrophils, and a decrease in lymphocytes were observed in males at 1,554 mg/kg-day. In females, MCV
29 was decreased at doses > 756 mg/kg and platelets were decreased at 1,614 mg/kg-day. With the exception
26
DRAFT - DO NOT CITE OR QUOTE
-------
1 of the 30% increase in neutrophils in high-dose male rats, hematological changes were within 2-15% of
2 control values. Total serum protein and albumin were significantly decreased in males at doses >
3 274 mg/kg-day and in females at doses > 427 mg/kg-day. Additional changes in high-dose male and
4 female rats included decreases in glucose, total cholesterol, triglycerides, and sodium (and calcium in
5 females), and increases in ALT (males only), AST, ALP, and LAP. Serum biochemistry parameters in
6 treated rats did not differ more than twofold from control values. Urine pH was decreased in males at >
7 274 mg/kg-day and in females at > 756 mg/kg-day.
8 Kidney weights were increased in females at >185 mg/kg-day with a maximum increase of 15%
9 and 44% at 1,614 mg/kg-day for absolute and relative kidney weight, respectively. No organ weight
10 changes were noted in male rats. Histopathology findings in rats that were related to exposure included
11 nuclear enlargement of the respiratory epithelium, nuclear enlargement of the olfactory epithelium,
12 nuclear enlargement of the tracheal epithelium, hepatocyte swelling of the centrilobular area of the liver,
13 vacuolar changes in the liver, granular changes in the liver, single cell necrosis in the liver, nuclear
14 enlargement of the proximal tubule of the kidneys, hydropic changes in the proximal tubule of the
15 kidneys, and vacuolar changes in the brain. The incidence data for histopathological lesions in rats are
16 presented in Table 4-1. The effects that occurred at the lowest doses were nuclear enlargement of the
17 respiratory epithelium in the nasal cavity and hepatocyte swelling in the central area of the liver in male
18 rats. Based on these histopathological findings the study authors identified the LOAEL as 126 mg/kg-day
19 and the NOAEL as 52 mg/kg-day.
27
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1 Incidence of histopathological lesions in F344/DuCrj rats exposed to 1,4-dioxane in
drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; renal proximal tubule
Hydropic change; renal proximal tubule
Vacuolic change; brain
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
52
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
126
9/10"
0/10
0/10
9/10"
0/10
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
0
0/10
0/10
0/10
0/10
0/10
2/10
2/10
0/10
0/10
0/10
83
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
185
5/1 Oc
0/10
0/10
0/10
0/10
1/10
1/10
0/10
0/10
0/10
274
10/10"
10/10"
10/10"
10/10"
0/10
5/1 Oc
5/1 Oc
1/10
0/10
0/10
657
9/10"
9/10"
10/10"
10/10"
10/10"
2/10
2/10
5/1 Oc
0/10
0/10
1,554
10/10"
10/10"
10/10"
10/10"
10/10"
10/10"
10/10"
9/10"
7/10"
10/10"
(mg/kg-day)a
427
10/10"
9/10"
9/10"
0/10
0/10
5/1 Oc
5/10
0/10
0/10
0/10
756
10/10"
10/10"
10/10"
9/10"
0/10
5/1 Oc
5/10
8/10"
0/10
0/10
1,614
8/9"
8/9"
9/9"
9/9"
9/9"
8/9"
8/9"
9/9"
5/9c
9/9"
aData are presented for sacrificed animals.
bp < 0.01 by x2 test.
°p < 0.05.
Source: Kano et al. (2008)
1 Clinical signs of toxicity in mice were not discussed in the study report One male mouse in the
2 high-dose group (1,570 mg/kg-day) died, but no information was provided regarding cause or time of
3 death. Final BWs were decreased 29% in male mice at 1,570 mg/kg-day, but changed less than 10%
4 relative to controls in the other male dose groups and in female mice. Food consumption was not
5 significantly reduced in any exposure group. Water consumption was reduced 14-18% in male mice
6 exposed to 86, 231, or 585 mg/kg-day. Water consumption was further decreased by 48 and 70% in male
7 mice exposed to 882 and 1,570 mg/kg-day, respectively. Water consumption was also decreased 31 and
8 57% in female mice treated with 1,620 and 2,669 mg/kg-day, respectively. An increase in MCV was
9 observed in the two highest dose groups in both male (882 and 1,570 mg/kg-day) and female mice (1,620
10 and 2,669 mg/kg-day). Increases in RBCs, hemoglobin, and hematocrit were also observed in high dose
11 males (1,570 mg/kg-day). Hematological changes were within 2-15% of control values. Serum
12 biochemistry changes in exposed mice included decreased total protein (at 1,570 mg/kg-day in males,
13 >1,620 mg/kg-day in females), decreased glucose (at 1,570 mg/kg-day in males, >1,620 mg/kg-day in
14 females), decreased albumin (at 1,570 mg/kg-day in males, 2,669 mg/ kg-day in females), decreased total
15 cholesterol (> 585 mg/kg-day in males, >1,620 mg/kg-day in females), increased serum ALT (at
16 1,570 mg/kg-day in males, > 620 mg/kg-day in females), increased AST (at 1,570 mg/kg-day in males,
17 2,669 mg/kg-day in females), increased ALP (> 585 mg/kg-day in males, 2,669 mg/kg-day in females),
18 and increased LDH (in females only at doses > 1,620 mg/kg-day). With the exception of a threefold
28
DRAFT - DO NOT CITE OR QUOTE
-------
1 increase in ALT in male and female mice, serum biochemistry parameters in treated rats did not differ
2 more than twofold from control values. Urinary pH was decreased in males at > 882 mg/kg-day and in
3 females at > 1,620 mg/kg-day.
4 Absolute and relative lung weights were increased in males at 1,570 mg/kg-day and in females at
5 1,620 and 2,669 mg/kg-day. Absolute kidney weights were also increased in females at 1,620 and
6 2,669 mg/kg-day and relative kidney weight was elevated at 2,669 mg/kg-day. Histopathology findings in
7 mice that were related to exposure included nuclear enlargement of the respiratory epithelium, nuclear
8 enlargement of the olfactory epithelium, eosinophilic change in the olfactory epithelium, vacuolic change
9 in the olfactory nerve, nuclear enlargement of the tracheal epithelium, accumulation of foamy cells in the
10 lung and bronchi, nuclear enlargement and degeneration of the bronchial epithelium, hepatocyte swelling
11 of the centrilobular area of the liver, and single cell necrosis in the liver. The incidence data for
12 histopathological lesions in mice are presented in Table 4-2. Based on the changes in the bronchial
13 epithelium in female mice, the authors identified the dose level of 387 mg/kg-day as the LOAEL for
14 mice; the NOAEL was 170 mg/kg-day (Kano et al., 2008).
15
Table 4-2 Incidence of histopathological lesions in
drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Crj:BDF1 mice exposed to
1,4-dioxane in
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0
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
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
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
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
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
DRAFT - DO NOT CITE OR QUOTE
29
-------
Single cell necrosis; liver
0/10
0/10
0/10
7/1 Oc
10/10C
9/1 Oc
aData are presented for sacrificed animals.
bp < 0.01 by x2 test.
°p < 0.05.
Source: Kano et al (2008).
1 4.2.1.1.4 Yamamoto eta I. Studies (Yamamoto et al., 1998a; Yamamoto et al., 1998b) in
2 rasH2 transgenic mice carrying the human prototype c-Ha-ras gene have been investigated as a bioassay
3 model for rapid carcinogenicity testing. As part of validation studies of this model, 1,4-dioxane was one
4 of many chemicals that were evaluated. RasH2 transgenic mice were Fl offspring of transgenic male
5 C57BLr6J and normal female BALB/cByJ mice. CB6F] mice were used as a nontransgenic control.
6 Seven- to nine-week-old mice (10-15/group) were exposed to 0, 0.5, or 1% 1,4-dioxane in drinking water
7 for 26 weeks. An increase in lung adenomas was observed in treated transgenic mice, as compared to
8 treated nontransgenic mice. The tumor incidence in transgenic animals, however, was not greater than
9 that observed in vehicle-treated transgenic mouse controls. Further study details were not provided.
4.2.1.2 Chronic Oral Toxicity and Carcinogenicity
1 4.2.1.2.1 A rg us et a I. Twenty-six adult male Wistar rats (Argus et al., 1965) weighing
2 between 150 and 200 g were exposed to 1,4-dioxane (purity not reported) in the drinking water at a
3 concentration of 1% for 64.5 weeks. A group of nine untreated rats served as control. Food and water
4 were available ad libitum. The drinking water intake for treated animals was reported to be 30 mL/day,
5 resulting in a dose/rat of 300 mg/day. Using a reference BW of 0.462 kg for chronic exposure to male
6 Wistar rats (U.S. EPA. 1988). it can be estimated that these rats received daily doses of approximately
7 640 mg/kg-day. All animals that died or were killed during the study underwent a complete necropsy. A
8 list of specific tissues examined microscopically was not provided; however, it is apparent that the liver,
9 kidneys, lungs, lymphatic tissue, and spleen were examined. No statistical analysis of the results was
10 conducted.
1 Six of the 26 treated rats developed hepatocellular carcinomas, and these rats had been treated for
2 an average of 452 days (range, 448-455 days). No liver tumors were observed in control rats. In two rats
3 that died after 21.5 weeks of treatment, histological changes appeared to involve the entire liver. Groups
4 of cells were found that had enlarged hyperchromic nuclei. Rats that died or were killed at longer
5 intervals showed similar changes, in addition to large cells with reduced cytoplasmic basophilia. Animals
6 killed after 60 weeks of treatment showed small neoplastic nodules or multifocal hepatocellular
7 carcinomas. No cirrhosis was observed in this study. Many rats had extensive changes in the kidneys
8 often resembling glomerulonephritis, however, incidence data was not reported for these findings. This
9 effect progressed from increased cellularity to thickening of the glomerular capsule followed by
10 obliteration of the glomeruli. One treated rat had an early transitional cell carcinoma in the kidney's
11 pelvis; this rat also had a large tumor in the liver. The lungs from many treated and control rats (incidence
12 not reported) showed severe bronchitis with epithelial hyperplasia and marked peribronchial infiltration,
13 as well as multiple abscesses. One rat treated with 1,4-dioxane developed leukemia with infiltration of all
30
DRAFT - DO NOT CITE OR QUOTE
-------
1 organs, particularly the liver and spleen, with large, round, isolated neoplastic cells. In the liver, the
2 distribution of cells in the sinusoids was suggestive of myeloid leukemia. The dose of 640 mg/kg-day
3 tested in this study was a free-standing LOAEL, identified by EPA, for glomerulonephritis in the kidney
4 and histological changes in the liver (hepatocytes with enlarged hyperchromic nuclei, large cells with
5 reduced cytoplasmic basophilia).
6 4.2.1.2.2 Argus et al.; Hoch-Ligeti et al. Five groups (28-3 2/dose group) of male
7 Sprague Dawley rats (2-3 months of age) weighing 110-230 g at the beginning of the experiment were
8 administered 1,4-dioxane (purity not reported) in the drinking water for up to 13 months at concentrations
9 of 0, 0.75, 1.0, 1.4, or 1.8% (Argus etal., 1973; Hoch-Ligeti et al.. 1970). The drinking water intake was
10 determined for each group over a 3-day measurement period conducted at the beginning of the study and
11 twice during the study (weeks were not specified). The rats were killed with ether at 16 months or earlier
12 if nasal tumors were clearly observable. Complete autopsies were apparently performed on all animals,
13 but only data from the nasal cavity and liver were presented and discussed. The nasal cavity was studied
14 histologically only from rats in which gross tumors in these locations were present; therefore, early
15 tumors may have been missed and pre-neoplastic changes were not studied. No statistical analysis of the
16 results was conducted. Assuming a BW of 0.523 kg for an adult male Sprague Dawley rat (U.S. EPA.
17 1988) and a drinking water intake of 30 mL/day as reported by the study authors, dose estimates were 0,
18 430, 574, 803, and 1,032 mg/kg-day. The progression of liver tumorigenesis was evaluated by an
19 additional group of 10 male rats administered 1% 1,4-dioxane in the drinking water (574 mg/kg-day), 5 of
20 which were sacrificed after 8 months of treatment and 5 were sacrificed after 13 months of treatment.
21 Liver tissue from these rats and control rats was processed for electron microscopy examination.
6 Nasal cavity tumors were observed upon gross examination in six rats (1/30 in the 0.75% group,
7 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
8 the tumors visible either at the tip of the nose, bulging out of the nasal cavity, or on the back of the nose
9 covered by intact or later ulcerated skin. As the tumors obstructed the nasal passages, the rats had
10 difficulty breathing and lost weight rapidly. No neurological signs or compression of the brain were
11 observed. In all cases, the tumors were squamous cell carcinomas with marked keratinization and
12 formation of keratin pearls. Bony structure was extensively destroyed in some animals with tumors, but
13 there was no invasion into the brain. In addition to the squamous carcinoma, two adenocarcinomatous
14 areas were present. One control rat had a small, firm, well-circumscribed tumor on the back of the nose,
15 which proved to be subcutaneous fibroma. The latency period for tumor onset was 329-487 days.
16 Evaluation of the latent periods and doses received did not suggest an inverse relationship between these
17 two parameters.
18 Argus et al. (1973) studied the progression of liver tumorigenesis by electron microscopy of liver
19 tissues obtained following interim sacrifice at 8 and 13 months of exposure (5 rats/group,
20 574 mg/kg-day). The first change observed in the liver was an increase in the size of the nucleus of the
21 hepatocytes, mostly in the periportal area. Precancerous changes were characterized by disorganization of
22 the rough endoplasmic reticulum, an increase in smooth endoplasmic reticulum, and a decrease in
23 glycogen and increase in lipid droplets in hepatocytes. These changes increased in severity in the
24 hepatocellular carcinomas in rats exposed to 1,4-dioxane for 13 months.
31
DRAFT - DO NOT CITE OR QUOTE
-------
1 Three types of liver nodules were observed in exposed rats at 13-16 months. The first consisted
2 of groups of cells with reduced cytoplasmic basophilia and a slightly nodular appearance as viewed by
3 light microscopy. The second type of circumscribed nodule was described consisting of large cells,
4 apparently filled and distended with fat. The third type of nodule was described as finger-like strands, 2-
5 3 cells thick, of smaller hepatocytes with large hyperchromic nuclei and dense cytoplasm. This third type
6 of nodule was designated as an incipient hepatoma, since it showed all the histological characteristics of a
7 fully developed hepatoma. All three types of nodules were generally present in the same liver. Cirrhosis
8 of the liver was not observed. The numbers of incipient liver tumors and hepatomas in rats from this study
9 (treated for 13 months and observed at 13-16 months) are presented in Table 4-3.
Table 4-3 Number of incipient liver tumors and hepatomas in male Sprague-Dawley rats
exposed to 1,4-dioxane in drinking water for 13 months
Dose (mg/kg-day) Incipient tumors Hepatomas Total
430 4 0 4
574 9 0 9
803 13 3 16
1,032 11 12 23
aPrecise incidences cannot be calculated since the number of rats per group was reported as 28-32; incidence in control rats was
not reported; no statistical analysis of the results was conducted in the study.
Source: Argus et al. (1973).
10 Treatment with all dose levels of 1,4-dioxane induced marked kidney alterations, but quantitative
11 incidence data were not provided. Qualitatively, the changes indicated glomerulonephritis and
12 pyelonephritis, with characteristic epithelial proliferation of Bowman's capsule, periglomerular fibrosis,
13 and distension of tubules. No kidney tumors were found. No tumors were found in the lungs. One rat at
14 the 1.4% treatment level showed early peripheral adenomatous change of the alveolar epithelium and
15 another rat in the same group showed papillary hyperplasia of the bronchial epithelium. The lowest dose
16 tested (430 mg/kg-day) was considered a LOAEL by EPA for hepatic and renal effects in this study.
17 4.2.1.2.3 Hoch-Ligeti and Argus. Hoch-Ligeti and Argus (1970) provided a brief
18 account of the results of exposure of guinea pigs to 1,4-dioxane. A group of 22 male guinea pigs (neither
19 strain nor age provided) was administered 1,4-dioxane (purity not provided) in the drinking water for at
20 least 23 months and possibly up to 28 months. The authors stated that the concentration of 1,4-dioxane
21 was regulated so that normal growth of the guinea pigs was maintained, and varied 0.5-2% (no further
22 information provided). The investigators further stated that the amount of 1,4-dioxane received by the
23 guinea pigs over a 23-month period was 588-635 g. Using a reference BW of 0.89 kg for male guinea
24 pigs in a chronic study (U.S. EPA. 1988) and assuming an exposure period of 700 days (23 months), the
25 guinea pigs received doses between 944 and 1,019 mg 1,4-dioxane/kg-day. A group often untreated
26 guinea pigs served as controls. All animals were sacrificed within 28 months, but the scope of the
27 postmortem examination was not provided.
17 Nine treated guinea pigs showed peri- or intrabronchial epithelial hyperplasia and nodular
18 mononuclear infiltration in the lungs. Also, two guinea pigs had carcinoma of the gallbladder, three had
32
DRAFT - DO NOT CITE OR QUOTE
-------
1 early hepatomas, and one had an adenoma of the kidney. Among the controls, four guinea pigs had
2 peripheral mononuclear cell accumulation in the lungs, and only one had hyperplasia of the bronchial
3 epithelium. One control had formation of bone in the bronchus. No further information was presented in
4 the brief narrative of this study. Given the limited reporting of the results, a NOAEL or LOAEL value
5 was not provided for this study.
6 4.2.1.2.4 Kociba et al. Groups of 6-8-week-old Sherman rats (60/sex/dose level) were
7 administered 1,4-dioxane (purity not reported) in the drinking water at levels of 0 (controls), 0.01, 0.1, or
8 1.0% for up to 716 days (Kociba et al.. 1974). The drinking water was prepared twice weekly during the
9 first year of the study and weekly during the second year of the study. Water samples were collected
10 periodically and analyzed for 1,4-dioxane content by routine gas liquid chromatography. Food and water
11 were available ad libitum. Rats were observed daily for clinical signs of toxicity, and BWs were measured
12 twice weekly during the first month, weekly during months 2-7, and biweekly thereafter. Water
13 consumption was recorded at three different time periods during the study: days 1-113, 114-198, and
14 446-460. Blood samples were collected from a minimum of five male and five female control and
15 high-dose rats during the 4th, 6th, 12th, and 18th months of the study and at termination. Each sample
16 was analyzed for packed cell volume, total erythrocyte count, hemoglobin, and total and differential WBC
17 counts. Additional endpoints evaluated included organ weights (brain, liver, kidney, testes, spleen, and
18 heart) and gross and microscopic examination of major tissues and organs (brain, bone and bone marrow,
19 ovaries, pituitary, uterus, mesenteric lymph nodes, heart, liver, pancreas, spleen, stomach, prostate, colon,
20 trachea, duodenum, kidneys, esophagus, jejunum, testes, lungs, spinal cord, adrenals, thyroid,
21 parathyroid, nasal turbinates, and urinary bladder). The number of rats with tumors, hepatic tumors,
22 hepatocellular carcinomas, and nasal carcinomas were analyzed for statistical significance with Fisher's
23 Exact test (one-tailed), comparing each treatment group against the respective control group. Survival
24 rates were compared using %2 Contingency Tables and Fisher's Exact test. Student's test was used to
25 compare hematological parameters, body and organ weights, and water consumption of each treatment
26 group with the respective control group.
6 Male and female rats in the high-dose group (1% in drinking water) consumed slightly less water
7 than controls. BW gain was depressed in the high-dose groups relative to the other groups almost from
8 the beginning of the study (food consumption data were not provided). Based on water consumption and
9 BW data for specific exposure groups, Kociba et al. (1974) calculated mean daily doses of 9.6, 94, and
10 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
11 the 0.01, 0.1, and 1.0% concentration levels, respectively. Treatment with 1,4-dioxane significantly
12 increased mortality among high-dose males and females beginning at about 2-4 months of treatment.
13 These rats showed degenerative changes in both the liver and kidneys. From the 5th month on, mortality
14 rates of control and treated groups were not different. There were no treatment-related alterations in
15 hematological parameters. At termination, the only alteration in organ weights noted by the authors was a
16 significant increase in absolute and relative liver weights in male and female high-dose rats (data not
17 shown). Fiistopathological lesions were restricted to the liver and kidney from the mid- and high-dose
18 groups and consisted of variable degrees of renal tubular epithelial and hepatocellular degeneration and
19 necrosis (no quantitative incidence data were provided). Rats from these groups also showed evidence of
33
DRAFT - DO NOT CITE OR QUOTE
-------
1 hepatic regeneration, as indicated by hepatocellular hyperplastic nodule formation and evidence of renal
2 tubular epithelial regenerative activity (observed after 2 years of exposure). These changes were not seen
3 in controls or in low-dose rats. The authors determined a LOAEL of 94 mg/kg-day based on the liver and
4 kidney effects in male rats. The corresponding NOAEL value was 9.6 mg/kg-day.
5 Histopathological examination of all the rats in the study revealed a total of 132 tumors in
6 114 rats. Treatment with 1% 1,4-dioxane in the drinking water resulted in a significant increase in the
7 incidence of hepatic tumors (hepatocellular carcinomas in six males and four females). In addition, nasal
8 carcinomas (squamous cell carcinoma of the nasal turbinates) occurred in one high-dose male and two
9 high-dose females. Since 128 out of 132 tumors occurred in rats from the 12th to the 24th month, Kociba
10 et al. (1974) assumed that the effective number of rats was the number surviving at 12 months, which was
11 also when the first hepatic tumor was noticed. The incidences of liver and nasal tumors from Kociba et al.
12 (1974) are presented in Table 4-4. Tumors in other organs were not elevated when compared to control
13 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
1307
Effective
number of
animals3
106
110
106
66
Number of
tumor-bearing
animals
31
34
28
21
Number of animals
Hepatic tumors
(all types)
2
0
1
12°
Hepatocellular
carcinomas
1
0
1
10C
Nasal
carcinomas
0
0
0
3a
"Rats 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).
14 The high-dose level was the only dose that increased the formation of liver tumors over control
15 (males 1,015 mg/kg-day; females 1,599 mg/kg-day) and also caused significant liver and kidney toxicity
16 in these animals. The mid-dose group (males 94 mg/kg-day; females 148 mg/kg-day) experienced hepatic
17 and renal degeneration and necrosis, as well as regenerative proliferation in hepatocytes and renal tubule
18 epithelial cells. No increase in tumor formation was seen in the mid-dose group. No toxicity or tumor
19 formation was observed in either sex in the low-dose (males 9.6 mg/kg-day; females 19 mg/kg-day) group
20 of rats.
34
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.2.1.2.5 National Cancer Institute (NCI). Groups of Osborne-Mendel rats
2 (35/sex/dose) and B6C3F] mice (50/sex/dose) were administered 1,4-dioxane (> 99.95% pure) in the
3 drinking water for 110 or 90 weeks, respectively, at levels of 0 (matched controls), 0.5, or 1% (NCI.
4 1978). Solutions of 1,4-dioxane were prepared with tap water. The report indicated that at 105 weeks
5 from the earliest starting date, a new necropsy protocol was instituted. This affected the male controls and
6 high-dose rats, which were started a year later than the original groups of rats and mice. Food and water
7 were available ad libitum. Endpoints monitored in this bioassay included clinical signs (twice daily), BWs
8 (once every 2 weeks for the first 12 weeks and every month during the rest of the study), food and water
9 consumption (once per month in 20% of the animals in each group during the second year of the study),
10 and gross and microscopic appearance of all major organs and tissues (mammary gland, trachea, lungs
11 and bronchi, heart, bone marrow, liver, bile duct, spleen, thymus, lymph nodes, salivary gland, pancreas,
12 kidney, esophagus, thyroid, parathyroid, adrenal, gonads, brain, spinal cord, sciatic nerve, skeletal muscle,
13 stomach, duodenum, colon, urinary bladder, nasal septum, and skin). Based on the measurements of water
14 consumption and BWs, the investigators calculated average daily intakes of 1,4-dioxane of 0, 240, and
15 530 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
16 mice, and 0, 380, and 860 mg/kg-day in female mice. According to the report, the doses of 1,4-dioxane in
17 high-dose male mice were only slightly higher than those of the low-dose group due to decreased fluid
18 consumption in high-dose male mice.
1 During the second year of the study, the BWs of high-dose rats were lower than controls, those of
2 low-dose males were higher than controls, and those of low-dose females were comparable to controls.
3 The fluctuations in the growth curves were attributed to mortality by the investigators; quantitative
4 analysis of BW changes was not done. Mortality was significantly increased in treated rats, beginning at
5 approximately 1 year of study. Analysis of Kaplan-Meier curves (plots of the statistical estimates of the
6 survival probability function) revealed significant positive dose-related trends (p < 0.001, Tarone test). In
7 male rats, 33/35 (94%) in the control group, 26/35 (74%) in the mid-dose group, and 33/35 (94%) in the
8 high-dose group were alive on week 52 of the study. The corresponding numbers for females were 35/35
9 (100%), 30/35 (86%), and 29/35 (83%). Nonneoplastic lesions associated with treatment with 1,4-dioxane
10 were seen in the kidneys (males and females), liver (females only), and stomach (males only). Kidney
11 lesions consisted of vacuolar degeneration and/or focal tubular epithelial regeneration in the proximal
12 cortical tubules and occasional hyaline casts. Elevated incidence of hepatocytomegaly also occurred in
13 treated female rats. Gastric ulcers occurred in treated males, but none were seen in controls. The
14 incidence of pneumonia was increased above controls in high-dose female rats. The incidence of
15 nonneoplastic lesions in rats following drinking water exposure to 1,4-dioxane is presented in Table 4-5.
16 EPA identified the LOAEL in rats from this study as 240 mg/kg-day for increased incidence of gastric
17 ulcer and cortical tubular degeneration in the kidney in males; a NOAEL was not established.
35
DRAFT - DO NOT CITE OR QUOTE
-------
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
7/3 1a
(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).
1 Neoplasms associated with 1,4-dioxane treatment were limited to the nasal cavity (squamous cell
2 carcinomas, adenocarcinomas, and one rhabdomyoma) in both sexes, liver (hepatocellular adenomas) in
3 females, and testis/epididymis (mesotheliomas) in males. The first tumors were seen at week 52 in males
4 and week 66 in females. The incidence of squamous cell carcinomas in the nasal turbinates in male and
5 female rats is presented in Table 4-6. Squamous cell carcinomas were first seen on week 66 of the study.
6 Morphologically, these tumors varied from minimal foci of locally invasive squamous cell proliferation to
7 advanced growths consisting of extensive columns of epithelial cells projecting either into free spaces of
8 the nasal cavity and/or infiltrating into the submucosa. Adenocarcinomas of the nasal cavity were
9 observed in 3 of 34 high-dose male rats, 1 of 35 low-dose female rats, and 1 of 35 high-dose female rats.
10 The single rhabdomyoma (benign skeletal muscle tumor) was observed in the nasal cavity of a male rat
11 from the low-dose group. A subsequent re-examination of the nasal tissue sections by Goldsworthy et al.
12 (1991) concluded that the location of the tumors in the nasal apparatus was consistent with the possibility
13 that the nasal tumors resulted from inhalation of water droplets by the rats (see Section 4.5.2 for more
14 discussion of Goldsworthy et al. (1991)).
36
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-6 Incidence of nasal cavity squamous cell carcinoma and liver hepatocellular adenoma
in Osborne-Mendel rats exposed to 1,4-dioxane in drinking water
Males (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/33 (0%)
2/31 (6%)
240°
12/33(36%)
2/32 (6%)
530
16/34(47%)c
1/33 (3%)
Females (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/34 (0%)°
0/31 (0%)T
350
10/35(29%)e
10/33(30%)e
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 Cochran-Armitage test.
Source: NCI (1978).
1 The incidence of hepatocellular adenomas in male and female rats is presented in Table 4-6.
2 Hepatocellular adenomas were first observed in high-dose females in week 70 of the study. These tumors
3 consisted of proliferating hepatic cells oriented as concentric cords. Hepatic cell size was variable;
4 mitoses and necrosis were rare. Mesothelioma of the vaginal tunics of the testis/epididymis was seen in
5 male rats (2/33, 4/33, and 5/34 in controls, low-, and high-dose animals, respectively). The difference
6 between the treated groups and controls was not statistically significant. These tumors were characterized
7 as rounded and papillary projections of mesothelial cells, each supported by a core of fibrous tissue. Other
8 reported neoplasms were considered spontaneous lesions not related to treatment with 1,4-dioxane.
9 In mice, mean BWs of high-dose female mice were lower than controls during the second year of
10 the study, while those of low-dose females were higher than controls. In males, mean BWs of high-dose
11 animals were higher than controls during the second year of the study. According to the investigators,
12 these fluctuations could have been due to mortality; no quantitative analysis of BWs was done. No other
13 clinical signs were reported. Mortality was significantly increased in female mice (p < 0.001, Tarone test),
14 beginning at approximately 80 weeks on study. The numbers of female mice that survived to 91 weeks
15 were 45/50 (90%) in the control group, 39/50 (78%) in the low-dose group, and 28/50 (56%) in the
16 high-dose group. In males, at least 90% of the mice in each group were still alive at week 91.
17 Nonneoplastic lesions that increased significantly due to treatment with 1,4-dioxane were pneumonia in
18 males and females and rhinitis in females. The incidences of pneumonia were 1/49 (2%), 9/50 (18%), and
19 17/47 (36%) in control, low-dose, and high-dose males, respectively; the corresponding incidences in
20 females were 2/50 (4%), 33/47 (70%), and 32/36 (89%). The incidences of rhinitis in female mice were
21 0/50, 7/48 (14%), and 8/39 (21%) in control, low-dose, and high-dose groups, respectively. Pair-wise
22 comparisons of low-dose and high-dose incidences with controls for incidences of pneumonia and rhinitis
23 in females using Fisher's Exact test (done for this review) yielded p-values < 0.001 in all cases.
24 Incidences of other lesions were considered to be similar to those seen in aging mice. The authors stated
25 that hepatocytomegaly was commonly found in dosed mice, but the incidences were not significantly
26 different from controls and showed no dose-response trend. EPA concluded the LOAEL for 1,4-dioxane
37
DRAFT - DO NOT CITE OR QUOTE
-------
1 in mice was 380 mg/kg-day based on the increased incidence of pneumonia and rhinitis in female mice; a
2 NOAEL was not established in this study.
3 As shown in Table 4-7, treatment with 1,4-dioxane significantly increased the incidence of
4 hepatocellular carcinomas or adenomas in male and female mice in a dose-related manner. Tumors were
5 first observed on week 81 in high-dose females and in week 58 in high-dose males. Tumors were
6 characterized by parenchymal cells of irregular size and arrangement, and were often hypertrophic with
7 hyperchromatic nuclei. Mitoses were seldom seen. Neoplasms were locally invasive within the liver, but
8 metastasis to the lungs was rarely observed.
Table 4-7 Incidence of hepatocellular adenoma or carcinoma in B6C3F1 mice exposed to
1,4-dioxane in drinking water
Males (mg/kg-day)a
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
0
2/49 (4%)c
8/49 (16%)D
720
18/50(36%)c
19/50(38%)°
830
24/47 (51 %)c
28/47 (60%)c
Females (mg/kg-day)a
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
0
0/50 (0%)c
0/50 (0%)D
380
12/48(25%)c
21/48 (44%)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).
°p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (1978).
9 In addition to liver tumors, a variety of other benign and malignant neoplasms occurred.
10 However, the report (NCI. 1978) indicated that each type had been encountered previously as a
11 spontaneous lesion in the B6C3FJ mouse. The report further stated that the incidences of these neoplasms
12 were unrelated by type, site, group, or sex of the animal, and hence, not attributable to exposure to
13 1,4-dioxane. There were a few nasal adenocarcinomas (1/48 in low-dose females and 1/49 in high-dose
14 males) that arose from proliferating respiratory epithelium lining of the nasal turbinates. These growths
15 extended into the nasal cavity, but there was minimal local tissue infiltration. Nasal mucosal polyps were
16 rarely observed. The polyps were derived from mucus-secreting epithelium and were otherwise
17 unremarkable. There was a significant negative trend for alveolar/bronchiolar adenomas or carcinomas of
18 the lung in male mice, such that the incidence in the matched controls was higher than in the dosed
19 groups. The report (NCI. 1978) indicated that the probable reason for this occurrence was that the dosed
20 animals did not live as long as the controls, thus diminishing the possibility of the development of tumors
21 in the dosed groups.
38
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.2.1.2.6 Kano et al.; Japan Bioassay Research Center; Yamazaki et al. The
2 Japan Bioassay Research Center (JBRC) conducted a 2-year drinking water study determining the effects
3 of 1,4-dioxane on both sexes of rats and mice. The study results have been reported several times: once as
4 conference proceedings (Yamazaki et al.. 1994). once as a laboratory report (JBRC. 1998). and most
5 recently as a peer-reviewed manuscript (Kano et al.. 2009). Dr. Yamazaki also provided some detailed
6 information (Yamazaki. 2006). Variations in the data between these three reports were noted and
7 included: (1) the level of detail on dose information reported; (2) categories for incidence data reported
8 (e.g., all animals or sacrificed animals); and (3) analysis of non- and neoplastic lesions.
1 The 1,4-dioxane dose information provided in the reports varied. Specifically, Yamazaki et al.
2 (1994) only included drinking water concentrations for each dose group. In contrast, JBRC (1998)
3 included drinking water concentrations (ppm), in addition using body weights and water consumption
4 measurements to calculate daily chemical intake (mg/kg-day). JBRC (1998) reported daily chemical
5 intake for each dose group as a range. Thus, for the External Peer Review draft of this Toxicological
6 Review ofl,4-Dioxane (U.S. EPA. 2009b). the midpoint of the range was used. Kano et al. (2009) also
7 reported a calculation of daily chemical intake based on body weight and water consumption
8 measurements; however, for each dose group they reported a mean and standard deviation estimate.
9 Therefore, because the mean more accurately represents the delivered dose than the midpoint of a range,
10 the Kano et al. (2009) calculated mean chemical intake (mg/kg-day) is used for quantitative analysis of
11 this data.
12 The categories for which incidence rates were described also varied among the reports. Yamazaki
13 et al. (1994) and Kano et al. (2009) reported histopathological results for all animals, including dead and
14 moribund animals; however, the detailed JBRC laboratory findings (1998) included separate incidence
15 reports for dead and moribund animals, sacrificed animals, and all animals.
16 Finally, the criteria used to evaluate some of the data were updated when JBRC published the
17 most recent manuscript by Kano et al. (2009). The manuscript by Kano et al. (2009) stated that the lesions
18 diagnosed in the earlier reports (JBRC. 1998; Yamazaki et al.. 1994) were re-examined and recategorized
19 as appropriate according to current pathological diagnostic criteria (see references in Kano et al. (2009)).
20 Groups of F344/DuCrj rats (50/sex/dose level) were exposed to 1,4-dioxane (>99% pure) in the
21 drinking water at levels of 0, 200, 1,000, or 5,000 ppm for 2 years. Groups of Crj:BDFl mice
22 (50/sex/dose level) were similarly exposed in the drinking water to 0, 500, 2,000, or 8,000 ppm of
23 1,4-dioxane. The high doses were selected based on results from the Kano et al. (2008) 13-week drinking
24 water study so as not to exceed the maximum tolerated dose (MTD) in that study. Both rats and mice
25 were 6 weeks old at the beginning of the study. Food and water were available ad libitum. The animals
26 were observed daily for clinical signs of toxicity; and BWs were measured once per week for 14 weeks
27 and once every 2 weeks until the end of the study. Food consumption was measured once a week for
28 14 weeks and once every 4 weeks for the remainder of the study. The investigators used data from water
29 consumption and BW to calculate an estimate of the daily intake of 1,4-dioxane (mg/kg-day) by male and
30 female rats and mice. Kano et al. (2009) reported a calculated mean ± standard deviation for the daily
31 doses of 1,4-dioxane for the duration of the study. Male rats received doses of approximately 0, 11±1,
39
DRAFT - DO NOT CITE OR QUOTE
-------
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
2 received doses of 0, 49±5, 191±21, or 677±74 mg/kg-day and female mice received 0, 66±10, 278±40, or
3 964±88 mg/kg-day. For the remainder of this document, including the dose-response analysis, the mean
4 calculated intake values are used to identify dose groups. The Kano et al. (2009) study was conducted in
5 accordance with the Organization for Economic Co-operation and Development (OECD) Principles for
6 Good Laboratory Practice (GLP).
7 No information was provided as to when urine samples were collected. Blood samples were
8 collected only at the end of the 2-year study (Yamazaki. 2006). Hematology analysis included RBCs,
9 hemoglobin, hematocrit, MCV, platelets, WBCs and differential WBCs. Serum biochemistry included
10 total protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only), phospholipid, ALT, AST,
11 LDH, LAP, ALP, y-glutamyl transpeptidase (GGT), CPK, urea nitrogen, creatinine (rat only), sodium,
12 potassium, chloride, calcium, and inorganic phosphorous. Urinalysis parameters were pH, protein,
13 glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ weights (brain, lung,
14 liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and gross necropsy and
15 histopathologic examination of tissues and organs were performed on all animals (skin, nasal cavity,
16 trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart, tongue, salivary glands, esophagus,
17 stomach, small and large intestine, liver, pancreas, kidney, urinary bladder, pituitary, thyroid, adrenal,
18 testes, epididymis, seminal vesicle, prostate, ovary, uterus, vagina, mammary gland, brain, spinal cord,
19 sciatic nerve, eye, Harderian gland, muscle, bone, and parathyroid). Dunnett's test and %2 test were used to
20 assess the statistical significance of changes in continuous and discrete variables, respectively.
21 For rats, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
22 study. Both male and female rats in the high dose groups (274 and 429 mg/kg-day, respectively) exhibited
23 slower growth rates and terminal body weights that were significantly different (p < 0.05) compared to
24 controls. A statistically significant reduction in terminal BWs was observed in high-dose male rats (5%, p
25 < 0.01) and in high-dose female rats (18%, p < 0.01) (Kanoetal.. 2009). Food consumption was not
26 significantly affected by treatment in male or female rats; however, water consumption in female rats
27 administered 18 mg/kg-day was significantly greater (p < 0.05) .
28 All control and exposed rats lived at least 12 months following study initiation (Yamazaki. 2006);
29 however, survival at the end of the 2-year study in the high dose group of male and female rats (274 and
30 429 mg/kg-day, respectively) was approximately 50%, which was significantly different compared to
31 controls. The investigators attributed these early deaths to the increased incidence in nasal tumors and
32 peritoneal mesotheliomas in male rats and nasal and hepatic tumors in female rats. (Yamazaki. 2006).
33 Several hematological changes were noted in the JBRC report (1998): Decreases in RBC (male
34 rats only), hemoglobin, hematocrit, and MCV; and increases in platelets in high-dose groups were
35 observed (JBRC. 1998). These changes (except for MCV) also occurred in mid-dose males. With the
36 exception of a 23% decrease in hemoglobin in high-dose male rats and a 27% increase in platelets in
37 high-dose female rats, hematological changes were within 15% of control values. Significant changes in
38 serum chemistry parameters occurred only in high-dose rats (males: increased phospholipids, AST, ALT,
39 LDH, ALP, GGT, CPK, potassium, and inorganic phosphorus and decreased total protein, albumin, and
40
DRAFT - DO NOT CITE OR QUOTE
-------
1 glucose; females: increased total bilirubin, cholesterol, phospholipids, AST, ALT, LDH, GOT, ALP,
2 CPK, and potassium, and decreased blood glucose) (JBRC. 1998). Increases in serum enzyme activities
3 ranged from <2- to 17-fold above control values, with the largest increases seen for ALT, AST, and GGT.
4 Urine pH was significantly decreased at 274 mg/kg-day in male rats (not tested at other dose levels) and
5 at 83 and 429 mg/kg-day in female rats (JBRC. 1998). Also, blood in the urine was seen in female rats at
6 83 and 429 mg/kg-day (JBRC. 1998). In male rats, relative liver weights were increased at 55 and
7 274 mg/kg-day (Kano et al., 2009). In female rats, relative liver weight was increased at 429 mg/kg-day
8 (Kano etal.. 2009).
9 Microscopic examination of the tissues showed nonneoplastic alterations in the nasal cavity, liver,
10 and kidneys mainly in high-dose rats and, in a few cases, in mid-dose rats (Table 4-8 and Table 4-9).
11 Alterations in high-dose (274 mg/kg-day) male rats consisted of nuclear enlargement and metaplasia of
12 the olfactory and respiratory epithelia, atrophy of the olfactory epithelium, hydropic changes and sclerosis
13 of the lamina propria, adhesion, and inflammation. In female rats, nuclear enlargement of the olfactory
14 epithelium occurred at doses > 83 mg/kg-day, and nuclear enlargement and metaplasia of the respiratory
15 epithelium, squamous cell hyperplasia, respiratory metaplasia of the olfactory epithelium, hydropic
16 changes and sclerosis of the lamina propria, adhesion, inflammation, and proliferation of the nasal gland
17 occurred at 429 mg/kg-day. Alterations were seen in the liver at > 55 mg/kg-day in male rats (spongiosis
18 hepatis, hyperplasia, and clear and mixed cell foci) and at 429 mg/kg-day in female rats (hyperplasia,
19 spongiosis hepatis, cyst formation, and mixed cell foci). Nuclear enlargement of the renal proximal tubule
20 occurred in males at 274 mg/kg-day and in females at > 83 mg/kg-day (JBRC. 1998).
41
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8 Incidence of histopathological lesions in male F344/DuCrj rats exposed to 1,4-dioxane
in drinking water for 2 years
Dose (mg/kg-day)a'D
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 epithelium0
Atrophy; nasal olfactory epithelium0
Hydropic change; lamina propria0
Sclerosis; lamina propria0
Adhesion; nasal cavity0
Inflammation; nasal cavity0
Hyperplasia; liver0
Spongiosis hepatis; liver0
Clear cell foci; liver0
Acidophilic cell foci; liver0
Basophilic cell foci; liver0
Mixed-cell foci; liver0
Nuclear enlargement; kidney proximal tubule0
0
0/50
0/50
0/50
0/50
12/50
0/50
0/50
0/50
0/50
0/50
3/50
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
2/50
20/50
3/50
8/50
11/50
8/50
0/50
55
0/50
0/50
0/50
5/50'
20/50
0/50
0/50
1/50
0/50
0/50
10/50
25/50'
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
24/50
40/50
8/50
5/50
16/50'
13/508
50/50
"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 Y2 test.
fp < 0.05 by x test.
Sources: Kano et al. (2009) and JBRC (1998).
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'D
Nuclear enlargement; nasal respiratory epithelium0
Squamous cell metaplasia; nasal respiratory epithelium0
Squamous cell hyperplasia; nasal cavity0
Nuclear enlargement; nasal olfactory epithelium °
Respiratory metaplasia; nasal olfactory epithelium0
Atrophy; nasal olfactory epithelium0
Hydropic change; lamina propria0
Sclerosis; lamina propria0
Adhesion; nasal cavity0
Inflammation; nasal cavity0
Proliferation; nasal gland0
Hyperplasia; liver0
Spongiosis hepatis; liver0
Cyst formation; liver0
Acidophilic cell foci; liver0
Basophilic cell foci; liver0
Clear cell foci; liver0
Mixed-cell foci; liver0
Nuclear enlargement; kidney proximal tubule0
0
0/50
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
0/50
0/50
3/50
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
2/50
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
11 /508
1/50
1/50
1/50
31/50
5/50
3/50
6/50
429
13/508
35/50 8
5/50
39/50
42/50
40/50
46/50
48/50
46/50
15/50
11/50
47/50
20/50
8/50
1/50
8/508
4/50
11/50'
39/50
DRAFT - DO NOT CITE OR QUOTE
42
-------
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 Y2 test.
fp < 0.05 by x test.
Sources: Kano et al. (2009) and JBRC (1998).
1 NOAEL and LOAEL values for rats in this study were identified by EPA as 55 and
2 274 mg/kg-day, respectively, based on toxicity observed in nasal tissue of male rats (i.e., atrophy of
3 olfactory epithelium, adhesion, and inflammation). Metaplasia and hyperplasia of the nasal epithelium
4 were also observed in high-dose male and female rats. These effects are likely to be associated with the
5 formation of nasal cavity tumors in these dose groups. Nuclear enlargement was observed in the nasal
6 olfactory epithelium and the kidney proximal tubule at a dose of 83 mg/kg-day in female rats; however, it
7 is unclear whether these alterations represent adverse toxicological effects. Hematological effects noted in
8 male rats given 55 and 274 mg/kg-day (decreased RBCs, hemoglobin, hematocrit, increased platelets)
9 were within 20% of control values. In female rats decreases in hematological effects were observed in the
10 high dose group (429 mg/kg-day). A reference range database for hematological effects in laboratory
11 animals (Wolford et al.. 1986) indicates that a 20% change in these parameters may fall within a normal
12 range (10th-90th percentile values) and may not represent a treatment-related effect of concern. Liver
13 lesions were also seen at a dose of 55 mg/kg-day in male rats; these changes are likely to be associated
14 with liver tumorigenesis. Clear and mixed-cell foci are commonly considered preneoplastic changes and
15 would not be considered evidence of noncancer toxicity. The nature of spongiosis hepatis as a
16 preneoplastic change is less well understood (Bannasch. 2003; Karbe and Kerlin, 2002; Stroebel et al.,
17 1995). Spongiosis hepatis is a cyst-like lesion that arises from the perisinusoidal (Ito) cells (PSC) of the
18 liver. It is commonly seen in aging rats, but has been shown to increase in incidence following exposure
19 to hepatocarcinogens. Spongiosis hepatis can be seen in combination with preneoplastic foci in the liver
20 or with hepatocellular adenoma or carcinoma and has been considered a preneoplastic lesion (Bannasch.
21 2003; Stroebel etal.. 1995). This change can also be associated with hepatocellular hypertrophy and liver
22 toxicity and has been regarded as a secondary effect of some liver carcinogens (Karbe and Kerlin. 2002).
23 In the case of the JBRC (1998) study, spongiosis hepatis was associated with other preneoplastic changes
24 in the liver (clear and mixed-cell foci). No other lesions indicative of liver toxicity were seen in this
25 study; therefore, spongiosis hepatis was not considered indicative of noncancer effects. Serum chemistry
26 changes (increases in total protein, albumin, and glucose; decreases in AST, ALT, LDH, and ALP,
27 potassium, and inorganic phosphorous) were observed in both male and female rats (JBRC. 1998) in the
28 high dose groups, 274 and 429 mg/kg-day, respectively. These serum chemistry changes seen in terminal
29 blood samples from high-dose male and female rats are likely related to tumor formation in these dose
30 groups.
31 Significantly increased incidences of liver tumors (adenomas and carcinomas) and tumors of the
32 nasal cavity occurred in high-dose male and female rats (Table 4-10 and Table 4-11) treated with
33 1,4-dioxane for 2 years (Kano et al.. 2009). The first liver tumor was seen at 85 weeks in high-dose male
34 rats and 73 weeks in high-dose female rats (vs. 101-104 weeks in lower dose groups and controls)
35 (Yamazaki. 2006). In addition, a significant increase (p < 0.01, Fisher's Exact test) in mesotheliomas of
43
DRAFT - DO NOT CITE OR QUOTE
-------
1 the peritoneum was seen in high-dose males (28/50 versus 2/50 in controls). Mesotheliomas were the
2 single largest cause of death among high-dose male rats, accounting for 12 of 28 pretermination deaths
3 (Yamazaki. 2006). Also, in males, there were increasing trends in mammary gland fibroadenoma and
4 fibroma of the subcutis, both statistically significant (p < 0.01) by the Peto test of dose-response trend.
5 Females showed a significant increasing trend in mammary gland adenomas (p < 0.01 by Peto's test). The
6 tumor incidence values presented in Table 4-10 and Table 4-11 were not adjusted for survival.
Table 4-10 Incidence of nasal cavity, peritoneum, and mammary gland tumors in F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years
Males
Dose (mg/kg-day)
0
11
55
274
0
Females
18
83
429
Nasal cavity
Squamous cell carcinoma
Sarcoma
Rhabdomyosarcoma
Esthesioneuroepithelioma
0/50
0/50
0/50
0/50
0/50
0/50
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'D
0/50
0/50
1/50
Peritoneum
Mesothelioma
2/50
2/50
5/50
28/50a'c
1/50
0/50
0/50
0/50
Mammary gland
Fibroadenoma
Adenoma
Either adenoma or fibroadenoma
1/50
0/50
1/50
1/50
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
aStatistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
bSignificantly different from control by Fisher's exact test (p < 0.01).
cSignificantly 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
Males
Dose (mg/kg-day)
Hepatocellular adenoma
Hepatocellular carcinoma
Either adenoma or carcinoma
0
3/50
0/50
3/50
11
4/50
0/50
4/50
55
7/50
0/50
7/50
274
32/50a'D
14/50a'°
39/50a'c
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'°
10/50a'°
48/50a'D
"Significantly 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).
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
7 For mice, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
8 study. Similar to rats, the growth rates of male and female mice were slower than controls and terminal
9 body weights were lower for the mid (p < 0.01 for males administered 191 mg/kg-day and p < 0.05 for
10 females administered 278 mg/kg-day) and high doses (p < 0.05 for males and females administered 677
11 and 964 mg/kg-day, respectively). There were no differences in survival rates between control and treated
12 male mice; however, survival rates were significantly decreased compared to controls for female mice in
13 the mid (278 mg/kg-day, approximately 40% survival) and high (964 mg/kg-day, approximately 20%
14 survival) dose groups. The study authors attributed these early female mouse deaths to the significant
44
DRAFT - DO NOT CITE OR QUOTE
-------
1 incidence of hepatic tumors, and Kano et al. (2009) reported tumor incidence for all animals in the study
2 (N=50), including animals that became moribund or died before the end of the study. Additional data on
3 survival rates of mice were provided in a personal communication from Dr. Yamazaki (2006). who
4 reported that the survival of mice was low in all male groups (31/50, 33/50, 25/50 and 26/50 in control,
5 low-, mid-, and high-dose groups, respectively) and particularly low in high-dose females (29/50, 29/50,
6 17/50, and 5/50 in control, low-, mid-, and high-dose groups, respectively). These deaths occurred
7 primarily during the second year of the study. Survival at 12 months in male mice was 50/50, 48/50,
8 50/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively. Female mouse survival at
9 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively
10 (Yamazaki. 2006). Furthermore, these deaths were primarily tumor related. Liver tumors were listed as
11 the cause of death for 31 of the 45 pretermination deaths in high-dose female Crj:BDFl mice (Yamazaki.
12 2006). For mice, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
13 study. Similar to rats, the growth rates of male and female mice were slower than controls and terminal
14 body weights were lower for the mid (p < 0.01 for males administered 191 mg/kg-day and p < 0.05 for
15 females administered 278 mg/kg-day) and high doses (p < 0.05 for males and females administered 677
16 and 964 mg/kg-day, respectively).
17 Food consumption was not significantly affected, but water consumption was reduced 26% in
18 high-dose male mice and 28% in high-dose female mice. Final BWs were reduced 43% in high-dose male
19 mice and 15 and 45% in mid- and high-dose female mice, respectively. Male mice showed increases in
20 RBC counts, hemoglobin, and hematocrit, whereas in female mice, there was a decrease in platelets in
21 mid- and high-dose rats. With the exception of a 60% decrease in platelets in high-dose female mice,
22 hematological changes were within 15% of control values. Serum AST, ALT, LDH, and ALP activities
23 were significantly increased in mid- and high-dose male mice, whereas LAP and CPK were increased
24 only in high-dose male mice. AST, ALT, LDH, and ALP activities were increased in mid- and high-dose
25 female mice, but CPK activity was increased only in high-dose female mice. Increases in serum enzyme
26 activities ranged from less than two- to sevenfold above control values. Glucose and triglycerides were
27 decreased in high-dose males and in mid- and high-dose females. High-dose female mice also showed
28 decreases in serum phospholipid and albumin concentrations (not reported in males). Blood calcium was
29 lower in high-dose females and was not reported in males. Urinary pH was decreased in high-dose males,
30 whereas urinary protein, glucose, and occult blood were increased in mid- and high-dose female mice.
31 Relative and absolute lung weights were increased in high-dose males and in mid- and high-dose females
32 (JBRC. 1998). Microscopic examination of the tissues for nonneoplastic lesions showed significant
33 alterations in the epithelium of the respiratory tract, mainly in high-dose animals, although some changes
34 occurred in mid-dose mice (
35 Table 4-12 and Table 4-13). Commonly seen alterations included nuclear enlargement, atrophy,
36 and inflammation of the epithelium. Other notable changes observed included nuclear enlargement of the
37 proximal tubule of the kidney and angiectasis in the liver in high-dose male mice.
45
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-12 Incidence of histopathological lesions in male Crj:BDF1 mice exposed to 1,4-dioxane
in drinking water for 2 years
Dose (mg/kg-day)
49
191
677
Nuclear enlargement; nasal respiratory epithelium0
0/50
0/50
0/50
31/508
Nuclear enlargement; nasal olfactory epithelium
0/50
0/50
9/50e
49/508
Atrophy; nasal olfactory epithelium
0/50
0/50
1/50
48/50
Inflammation; nasal cavity
1/50
2/50
1/50
25/50
Atrophy; tracheal epithelium
0/50
0/50
0/50
42/50
Nuclear enlargement; tracheal epithelium
0/50
0/50
0/50
17/50
Nuclear enlargement; bronchial epithelium
0/50
0/50
0/50
41/50
Atrophy; lung/bronchial epithelium
0/50
0/50
0/50
43/50
Accumulation of foamy cells; lung
1/50
0/50
0/50
27/50
Angiectasis; liver
2/50
3/50
4/50
16/50
Nuclear enlargement; kidney proximal tubule
0/50
0/50
0/50
"Data from JBRC (1j
ep < 0.01 by x2 test.
Sources: Kano et al. (2009) and JBRC (1998).
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).
. JBRC did not report statistical significance for the "All animals" comparison.
Table 4-13 Incidence of histopathological lesions in female Crj:BDF1 mice exposed to
1,4-dioxane in drinking water for 2 years
Dose (mg/kg-day)a'D
Nuclear enlargement; nasal respiratory epithelium0
Nuclear enlargement; nasal olfactory epithelium0
Atrophy; nasal olfactory epithelium0
Inflammation; nasal cavity0
Atrophy; tracheal epithelium0
Nuclear enlargement; bronchial epithelium0
Atrophy; lung/bronchial epithelium0
Accumulation of foamy cells; lung0
0
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
66
0/50
0/50
0/50
0/50
0/50
1/50
0/50
1/50
278
0/50
41/508
1/50
7/50
2/50
22/50
7/50
4/50
964
41/508
33/508
42/50
42/50
49/50
48/50
50/50
45/50
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).
1 NOAEL and LOAEL values for mice in this study were identified by EPA as 66 and
2 278 mg/kg-day, respectively, based on nasal inflammation observed in female mice. Nuclear enlargement
3 of the nasal olfactory epithelium and bronchial epithelium was also observed at a dose of 278 mg/kg-day
4 in female mice; however, it is unclear whether these alterations represent adverse toxicological effects.
5 The serum chemistry changes seen in terminal blood samples from male and female mice (mid- and
6 high-dose groups) are likely related to tumor formation in these animals. Liver angiectasis, an abnormal
46
DRAFT - DO NOT CITE OR QUOTE
-------
1 dilatation and/or lengthening of a blood or lymphatic vessel, was seen in male mice given 1,4-dioxane at a
2 dose of 677 mg/kg-day.
3 Treatment with 1,4-dioxane resulted in an increase in the formation of liver tumors (adenomas
4 and carcinomas) in male and female mice. The incidence of hepatocellular adenoma was statistically
5 increased in male mice in the mid-dose group only. The incidence of male mice with hepatocellular
6 carcinoma or either tumor type (adenoma or carcinoma) was increased in the low, mid, and high-dose
7 groups. The appearance of the first liver tumor occurred in male mice at 64, 74, 63, and 59 weeks in the
8 control, low- mid-, and high-dose groups, respectively (Yamazaki, 2006). In female mice, increased
9 incidence was observed for hepatocellular carcinoma in all treatment groups, while an increase in
10 hepatocellular adenoma incidence was only seen in the 66 and 278 mg/kg-day dose groups (Table 4-14).
11 The appearance of the first liver tumor in female mice occurred at 95, 79, 71, and 56 weeks in the control,
12 low-, mid-, and high-dose groups, respectively (Yamazaki, 2006). The tumor incidence data presented for
13 male and female mice in Table 4-14 are based on reanalyzed sample data presented in Kano et al. (2009)
14 that included lesions in animals that became moribund or died prior to the completion of the 2-year study.
15 Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and carcinomas in
16 control male and female BDF1 mice from ten 2-year bioassays at the JBRC. For female mice, out of 499
17 control mice, the incidence rates were 4.4% for hepatocellular adenomas and 2.0% for hepatocellular
18 carcinomas. Kano et al. (2009) reported a 10% incidence rate for hepatocellular adenomas and a 0%
19 incidence rate for hepatocellular carcinomas in control female BDF1. The background incidence rates for
20 male BDF1 mice were 15% and 22.8% for hepatocellular adenomas and carcinomas, respectively, out of
21 500 control mice in ten 2-year bioassays (Katagiri et al.. 1998). Background rates for B6C3Fi mice
22 evaluated by the National Toxicology Program are similar (10.3% and 21.3% for hepatocellular
23 adenomas and carcinomas in male mice, respectively; 4.0% and 4.1% for hepatocellular adenomas and
24 carcinomas in female mice, respectively) to the BDF1 mice background rates observed by JBRC
25 (Haseman et al.. 1984). Thus, the BDF1 mouse is not particularly sensitive compared to the commonly
26 used B6C3FJ strain and indicates that the results obtained by JBRC are reasonable.
Table 4-14 Incidence of tumors in Crj:BDF1 mice exposed to 1,4-dioxane in drinking water for
2 years
Males
Dose (mg/kg-day)
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'c
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'c
46/50a'b
"Significantly 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).
27 A weight of evidence evaluation of the carcinogenicity studies presented in Section 4.2.1.2is
28 located in Section 4.7 and Table 4-19.
47
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.2 Inhalation Toxicity
4.2.2.1 Subchronic Inhalation Toxicity
1 4.2.2.1.1 Fairleyetal. Rabbits, guinea pigs, rats, and mice (3-6/species/group) were
2 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 (3 hours/day) for 5 days/week and 1.5 hours on the 6th day (16.5 hours/week) (Fairleyetal.. 1934).
4 Animals were exposed until death occurred or were sacrificed at varying time periods. At the 10,000 ppm
5 concentration, only one animal (rat) survived a 7-day exposure. The rest of the animals (six guinea pigs,
6 three mice, and two rats) died within the first five exposures. Severe liver and kidney damage and acute
7 vascular congestion of the lungs were observed in these animals. Kidney damage was described as patchy
8 degeneration of cortical tubules with vascular congestion and hemorrhage. Liver lesions varied from
9 cloudy hepatocyte swelling to large areas of necrosis. At 5,000 ppm, mortality was observed in two mice
10 and one guinea pig following 15-34 exposures. The remaining animals were sacrificed following
11 49.5 hours (3 weeks) of exposure (three rabbits) or 94.5 hours (5 weeks) of exposure (three guinea pigs).
12 Liver and kidney damage in both dead and surviving animals was similar to that described for the
13 10,000 ppm concentration. Animals (four rabbits, four guinea pigs, six rats, and five mice) were exposed
14 to 2,000 ppm for 45-102 total exposure hours (approximately 2-6 weeks). Kidney and liver damage was
15 still apparent in animals exposed to this concentration. Animals exposed to 1,000 ppm were sacrificed at
16 intervals with the total exposure duration ranging between 78 and 202.5 hours (approximately 4-
17 12 weeks). Cortical kidney degeneration and hepatocyte degeneration and liver necrosis were observed in
18 these animals (two rabbits, three guinea pigs, three rats, and four mice). The low concentration of
19 1,000 ppm was identified by EPA as a LOAEL for liver and kidney degeneration in rats, mice, rabbits,
20 and guinea pigs in this study.
48
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.2.2.1.2 Kasai et a I. Male and female 6-week-old F344/DuCrj rats (10/sex/group) were
2 exposed to nominal concentrations of 0 (clean air). 100. 200. 400. 800. 1.600. 3.200. or 6.400 ppm (0.
3 360. 720. 1.400. 2.900. 5.800. 1.2000. and 23.000 mg/m3. respectively) of vaporized 1.4-dioxane (>99%
4 pure) for 6 hours/day. 5 days/week, for 13 weeks in whole body inhalation chambers (Kasai et al.. 2008).
5 Each inhalation chamber housed 20 individual cages for 10 males and 10 females. During exposure, the
6 concentration of 1.4-dioxane vapor was determined every 15 minutes by gas chromatography. In addition.
7 during exposure, animals received food and water ad libitum and the following data were collected: 1)
8 clinical signs and mortality (daily): 2) BW and food intake (weekly): 3) urinary parameters using Ames
9 reagent strips (measured during week 13 of the exposure); and 4) 1.4-dioxane content in plasma from
10 three rats of both sexes (measured on the third day of exposure during weeks 12 and 13 at 1 hour
11 postmortem). At the end of the 13-week exposure period or at the time of an animal's death during
12 exposure, all organs were collected, weighed, and evaluated for macroscopic lesions. Histopathological
13 evaluations of organs and tissues were conducted in accordance with the OECD test guidelines, including
14 all tissues of the respiratory tract. Liver sections from male and female rats exposed to 800. 1.600 and
15 3.200 ppm of 1.4-dioxane were also analyzed for foci (in the absence of tumor formation) by
16 immunohistochemical expression of glutathione S-transferase placental form (GST-P). Hematological and
17 clinical chemistry parameters were measured using blood collected from the abdominal aorta of rats
18 following an overnight fasting at the end of the 13-week exposure period. The measured hematological
19 and clinical chemistry parameters included: red blood cell count, hemoglobin, hematocrit. MCV. AST.
20 ALT, glucose, andtriglyceride. Statistically significant differences (p-value of 0.05) between 1.4-dioxane
21 and clean air exposed groups were determined by study authors using Dunnett's test or y2test.
1 All rats exposed to 6.400 ppm of 1.4-dioxane died by the end of the first week of exposure: the
2 determined cause of death was renal failure and diagnosed as necrosis of the renal tubules. At
3 concentrations lower than 6.400 ppm. mortality was not observed and all exposed rats were absent of
4 clinical signs. Exposure-related effects on final BWs. organ weights, and hematological and clinical
5 chemistry parameters were reported as compared to controls and these changes are outlined in Table 4-15
6 and Table 4-16. Briefly, terminal BWs were significantly decreased in both sexes at 200 ppm; and
7 additionally in females at 800 and 1.600 ppm. Statistically significant increases in several organ weights
8 were observed, including lung (> 1.600 ppm. males; >200 ppm. females); liver (>800 ppm. both sexes).
9 and kidneys (3.200 ppm. males: >800 ppm. females). Statistically significant changes in hematological
10 parameters and clinical chemistry were observed in both sexes at 3.200 ppm including increased levels of
11 hemoglobin ALT. RBC. AST .and MCV. In females only, at 3.200 ppm. increased levels of hematocrit
12 was noted; and in males at this exposure concentration decreased levels of glucose and triglyceride were
13 observed, in addition to slightly decreased urinary protein. However, the urinary protein data were not
14 shown in this study. At 200 ppm. an increased AST level in females was noted. Blood plasma levels of
15 1.4-dioxane were also evaluated and in both sexes, a linear increase in 1.4-dioxane levels was detected at
16 exposure concentrations of 400 ppm and above. The highest blood levels of 1.4-dioxane were detected in
17 females.
18 Exposure and/or sex-related histopathology findings also reported by the study authors included
19 nuclear enlargement of the nasal respiratory, nasal olfactory, tracheal. and bronchial epithelium; vacuolic
49
DRAFT - DO NOT CITE OR QUOTE
-------
1 change in the olfactory and bronchial epithelium; atrophy of the nasal epithelium; hydropic change in the
2 proximal tubules of the kidney: and single-cell necrosis and centrilobular swelling in the liver. Table 4-17
3 presents a summary of these histopathological lesions, including incidence and severity data. Further
4 microscopic evaluation of liver tissue revealed GST-P positive liver foci in both sexes at 3.200 ppm (3/10
5 males. 2/10 females) and in females at 1.600 ppm (4/10).
6 The study authors determined nuclear enlargement in the respiratory epithelium as the most
7 sensitive lesion and a LOAEL value of 100 ppm was identified by the study authors based on the
8 incidence data of this lesion in both male and female rats.
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
Males
Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
Females
Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
Males3
1,4-dioxane vapor concentration (ppm)
lean air)
323 ± 14
0.310±0.011
2.610 ±0.069
0.589± 0.016
100
323 ± 14
0.312 ±
0.007
2.697 ±
0.092
0.596 ±
0.021
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
Females3
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
1,4-dioxane vapor concentration (ppm)
0 (clean air)
187 ±5
0.402 ±0.013
2.353 ±0.081
0.647± 0.014
100
195 ±8
0.402 ±
0.015
2.338 ±
0.092
0.631 ±
0.019
200
174+ 10"
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 ±
0.1 44b
0.749 ±
0.024b
aData are presented for 10 sacrificed animals.
bp < 0.01 by Dunnett's test.
cp < 0.05 by Dunnett's test.
Source: Kasai et al. (20081
DRAFT - DO NOT CITE OR QUOTE
50
-------
Table 4-16 Hematology and clinical chemistry of F344/DuCrj rats exposed to 1,4-dioxane vapor
by whole-body inhalation for 13 weeks
Males
Red blood cell (10b/Ml)
Hemoglobin (g/dl)
Hematocrit (%)
MCV (fl)
AST(IU/I)
ALT(IU/I)
Glucose (mg/dl)
Triglyceride (mg/dl)
Females
Red blood cell (10b/Ml)
Hemoglobin (g/dl)a
Hematocrit (%)a
MCV(fl)a
AST(IU/l)a
ALT(IU/l)a
Glucose (mg/dl) a
Triglyceride (mg/dl)
Males3
1,4-dioxane vapor concentration (ppm)
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
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
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
Females3
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 ±0.5°
73 ±4
30 ±2
170± 11°
87 ± 22C
1,4-dioxane vapor concentration (ppm)
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
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
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±0.12C
16.6±0.2C
47.5 ± 0.6 c
52.0 ±0.7
76 ±5°
30 ±3"
139± 18
42 ±7
"Data are presented for 10 sacrificed animals.
bp < 0.01 by Dunnett's test.
cp < 0.05 by Dunnett's test.
dData were reported for 9/10 female rats.
Source: Kasai et al. (20081
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
Males
Effect0
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
Males3
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+)
5/10°
(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 Oa
(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+)
6/10°
(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+)
6/10°
(6, 1+)
Atrophy; olfactory epithelium8 - ...
Hepatocyte centrilobular swelling
Hepatocyte single-cell necrosis
0/10
0/10
0/10
0/10
0/10
0/10
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
Females
Females3
51
DRAFT - DO NOT CITE OR QUOTE
-------
1,4-dioxane vapor concentration (ppm)
Effect0
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
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 Oa
(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+)
5/10°
(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+)
5/10°
(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 Oa
(6, 1+)
3Data 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 for
a given grade of lesion severity. Severity key: 1+, slight and , 2+, moderate.
°p < 0.01 by x2 test.
dp < 0.05 by x2 test.
eData were not reported for male rats.
Source: Kasai et al. (20081
4.2.2.2 Chronic Inhalation Toxicity and Carcinogenicity
1 4.2.2.2.1 Torkelson et al. Whole body exposures of male and female Wistar rats
2 (288/sex) to 1,4-dioxane vapors (99.9% pure) at a concentration of 0.4 mg/L (111 ppm), were carried out
3 7 hours/day, 5 days/week for 2 years (Torkelson et al.. 1974). The age of the animals at the beginning of
4 the study was not provided. The concentration of 1,4-dioxane vapor during exposures was determined
5 with infrared analyzers. Food and water were available ad libitum except during exposures. Endpoints
6 examined included clinical signs, eye and nasal irritation, skin condition, respiratory distress, and tumor
7 formation. BWs were determined weekly. Standard hematological parameters were determined on all
8 surviving animals after 16 and 23 months of exposure. Blood collected at termination was used also for
9 determination of clinical chemistry parameters (serum AST and ALP activities, blood urea nitrogen
10 [BUN], and total protein). Liver, kidneys, and spleen were weighed and the major tissues and organs were
11 processed for microscopic examination (lungs, trachea, thoracic lymph nodes, heart, liver, pancreas,
12 stomach, intestine, spleen, thyroid, mesenteric lymph nodes, kidneys, urinary bladder, pituitary, adrenals,
13 testes, ovaries, oviduct, uterus, mammary gland, lacrimal gland, lymph nodes, brain, vagina, and bone
14 marrow, and any abnormal growths). Nasal tissues were not obtained for histopathological evaluation.
15 Control and experimental groups were compared statistically using Student's t test, Yates corrected
16 %2 test, or Fisher's Exact test.
52
DRAFT - DO NOT CITE OR QUOTE
-------
1 Exposure to 1,4-dioxane vapors had no significant effect on mortality or BW gain and induced no
2 signs of eye or nasal irritation or respiratory distress. Slight, but statistically significant, changes in
3 hematological and clinical chemistry parameters were within the normal physiological limits and were
4 considered to be of no toxicological importance by the investigators. Altered hematological parameters
5 included decreases in packed cell volume, RBC count, and hemoglobin, and an increase in WBC count in
6 male rats. Clinical chemistry changes consisted of a slight decrease in both BUN (control—23 ± 9.9;
7 111-ppm 1,4-dioxane—19.8 ±8.8) and ALP activity (control—34.4 ± 12.1; 111-ppm 1,4-dioxane—29.9
8 ± 9.2) and a small increase in total protein (control—7.5 ± 0.37; 111-ppm 1,4-dioxane—7.9 ± 0.53) in
9 male rats (values are mean ± standard deviation). Organ weights were not significantly affected.
10 Microscopic examination of organs and tissues did not reveal any treatment-related effects. Based on the
11 lack of significant effects on several endpoints, EPA identified the exposure concentration of 0.4 mg/L
12 (111 ppm) as a free standing NOAEL. The true NOAEL was likely to be higher.
13 Tumors, observed in all groups including controls, were characteristic of the rat strain used and
14 were considered unrelated to 1,4-dioxane inhalation. The most common tumors were reticulum cell
15 sarcomas and mammary tumors. Using Fisher's Exact test and a significance level of p < 0.05, no one
16 type of tumor occurred more frequently in treated rats than in controls. No hepatic or nasal cavity tumors
17 were seen in any rat.
53
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.2.2.2.2 Kasai et al. Groups of male 6-week-old F344/DuCrj rats (50/group) weighing
2 120 ± 5g (mean ± SD) at the beginning of the study were exposed via inhalation to nominal
3 concentrations of 0 (clean air). 50. 250. and 1.250 ppm (0. 180. 900. and 4.500 mg/m3. respectively) of
4 vaporized 1.4-dioxane (>99% pure) for 6 hours/day. 5 days/week, for 104 weeks (2 years') in whole body
5 inhalation chambers (Kasai et al., 2009). Each inhalation chamber housed male rats individually in
6 stainless-steel wire hanging cages. The authors stated female counterparts were not exposed given data
7 illustrating the absence of induced mesotheliomas following exposure to 1.4-dioxane in drinking water
8 (Yamazaki et al.. 1994). During exposure, the concentration of 1.4-dioxane vapor was determined every
9 15 minutes by gas chromatography and animals received food and water ad libitum. In addition, during
10 the 2-year exposure period, clinical signs and mortality were recorded daily. BW and food intake were
11 measured once weekly for the first 14 weeks of exposure, and thereafter, every 4 weeks. At the end of the
12 2-year exposure period or at the time of an animal's death during exposure, all organs were collected.
13 weighed, and evaluated for macroscopic lesions. Additional examinations were completed on rats
14 sacrificed at the end of the 2-year exposure period. Endpoints examined included: 1) measurement of
15 hematological and clinical chemistry parameters using blood collected from the abdominal aorta of rats
16 following an overnight fasting at the end of the 2-year exposure period: 2) measurement of urinary
17 parameters using Ames reagent strips during the last week of the exposure period; and 3)
18 histopathological evaluations of organs and tissues outlined in the OECD test guideline which included
19 all tissues of the respiratory tract. For measured hematological and clinical chemistry parameters.
20 analyses included: red blood cell count, hemoglobin, hematocrit. MCV. mean corpuscular hemoglobin
21 (MCH). AST. ALT. ALP, and y-GTP. Organs and tissues collected for histopathological examination
22 were fixed in 10% neutral buffered formalin with the exception of nasal cavity samples. Nasal tissue was
23 trimmed transversely at three levels after decalcification and Fixation in a formic acid-formalin solution.
24 The levels were demarcated at the following points: at the posterior edge of the upper incisor teeth (level
25 1). at the incisive papilla (level 2). and at the anterior edge of the upper molar teeth (level 3). All tissue
26 samples were embedded in paraffin, and then sectioned (at 5_ [im thickness) and stained with hematoxylin
27 and eosin (H&E). Dunnett's test. y2test. and Fisher's exact test were used by study authors to determine
28 statistical differences (p-value of 0.05) between 1.4-dioxane exposed and clean air exposed group data.
1 Deformity in the nose was the only clinical sign reported in this study. This deformity was seen at
2 exposure weeks 74 and 79 in one rat each, exposed to 250 ppm and 1.250 ppm of 1.4-dioxane.
3 respectively. Both of these rats did not survive the 2-year exposure with deaths caused by malignant nasal
4 tumors.
5 Growth rates and survival rates were analyzed. Growth rates were not significantly affected by
6 1.4-dioxane exposures, but a decreasing trend in growth was observed during the latter half of the 2-year
7 exposure period for all exposure doses (i.e.. 50. 250. and 1.250 ppm). Survival rates were significantly
8 decreased following 91 weeks of exposure to 1.250 ppm of 1.4-dioxane. The authors attributed these
9 deaths to increased incidences of peritoneal mesotheliomas. but also noted that nasal tumors could have
10 been a contributing factor. Terminal survival rates were 37/50. 37/50. 29/50. and 25/50 for 0. 50. 250. and
11 1.250 ppm exposed groups, respectively.
54
DRAFT - DO NOT CITE OR QUOTE
-------
1 Exposure-related effects on final BWs. organ weights, and hematological and clinical chemistry
2 parameters were reported. Changes in these effects, as compared to control are outlined in Table 4-18 and
3 Table 4-19. Briefly, at 1.250 ppm terminal BWs were significantly decreased and relative liver and lung
4 weights were significantly increased. It is of note that the observed change in terminal body weight was
5 not an effect of food consumption, which was determined to be unaltered by the study authors. Altered
6 hematological and clinical chemistry parameters were also observed with significant changes at
7 1.250 ppm. Altered endpoints included decreased hemoglobin. MCV. and MCH. and increased AST.
8 ALT. ALP, and y-GTP (p < 0.01) levels. In addition, urine pH was significantly decreased in 1.250 ppm
9 exposed rats.
10 Histopathology Findings of pre- and nonneoplastic lesions associated with 1.4-dioxane treatment
11 were seen in the nasal cavity, liver, and kidneys (Table 4-20). At the highest concentration of 1.250 ppm.
12 all pre- and nonneoplastic lesions were significantly increased, as compared to controls, with the
13 exception of clear and mixed cell foci in the liver. At the lowest concentration of 50 ppm. nuclear
14 enlargement of the respiratory epithelium was the most sensitive lesion observed in the nasal cavity.
15 Based on this finding, the study authors identified a LOAEL of 50 ppm in male rats.
16 Tumor development was observed in the nasal cavity (squamous cell carcinoma), liver
17 (hepatocellular adenoma and carcinoma), peritoneum (peritoneal mesothelioma). kidney (renal cell
18 carcinoma), mammary gland (fibroadenoma and adenoma). Zvmbal gland (adenoma), and subcutaneous
19 tissue (subcutis fibroma). Tumor incidences with a dose-dependent, statistically significant positive trend
20 (Peto's test) included nasal squamous cell carcinoma, hepatocellular adenoma, peritoneal mesothelioma.
21 mammary gland fibroadenoma. and Zymbal gland adenoma. Renal cell carcinoma was also identified as
22 statistically significant with a positive dose-dependent trend; however, no tumor incidences were reported
23 at 50 and 250 ppm. At 1.250 ppm. significant increases in nasal squamous cell carcinoma, hepatocellular
24 adenoma, and peritoneal mesothelioma were observed. At 250 ppm. significant increases in peritoneum
25 mesothelioma and subcutis fibroma were observed. Table 4-21 presents a summary of tumor incidences
26 found in this study. Further characterizations of neoplasms revealed nasal squamous cell carcinoma
27 occurred at the dorsal area of the nose (levels 1-3) marked by keratinization and the progression of growth
28 into surrounding tissue. Peritoneal mesotheliomas were characterized by complex branching structures
29 originating from the mesothelium of the scrotal sac. Invasive growth into surrounding tissues was
30 occasionally observed for peritoneal mesotheliomas.
55
DRAFT - DO NOT CITE OR QUOTE
-------
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
29
376 ± 38
0.45 ±0.1 8
3.58 ±0.52
0.81 ±0.13
25
359 ± 129"
0.46 ± 0.07a
4.53 ±0.71"
0.86 ±0.12
ap < 0.01 by Dunnett's test.
bp < 0.05 by Dunnett's test.
Source: Kasai et al. (20081
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 (10B/ul)
Hemoglobin (g/dl)
Hematocrit (%)
MCV (fl)
MCH (pg)
AST(IU/I)
ALT(IU/I)
ALP (IU/I)
Y-GTP (IU/I)
Urinary pH
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
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
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 ± 171
10±8
7.1 ±0.6
25
7.0 ± 1.8
10.9 ±2.8"
34.3 ±7.6
49.4 ±4.0"
15.5± 1.3a
98 ± 52a
72± 36a
212 ± 109a
40 ± 26a
6.6 ±0.4°
ap < 0.01 by Dunnett's test.
bp < 0.05 by Dunnett's test.
Source: Kasai et al. (20081
56
DRAFT - DO NOT CITE OR QUOTE
-------
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
Effect
1,4-dioxane vapor concentration (ppm)
3 (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
6/50"
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 x" test.
bp < 0.05 by x2 test.
Source: Kasai et al. (2009).
DRAFT - DO NOT CITE OR QUOTE
57
-------
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
1/50
3/50
1/50
0/50
14/50a
3/50
0/50
0/50
9/50a
1,250
6/50D'c
21/50a'c
2/50
4/50c
41/50a'c
5/50a
1/50
4/50c
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
cp < 0.01 by Peto's test for dose-related trend.
dp < 0.05 by Peto's test for dose-related trend.
Source: Kasai et al. (2009).
4.2.3 Initiation/Promotion Studies
4.2.3.1 Bulletal.
1 Bull et al. (1986) tested 1,4-dioxane as a cancer initiator in mice using oral, subcutaneous, and
2 topical routes of exposure. A group of 40 female SENCAR mice (6-8 weeks old) was administered a
3 single dose of 1,000 mg/kg 1,4-dioxane (purity >99%) by gavage, subcutaneous injection, or topical
4 administration (vehicle was not specified). A group of rats was used as a vehicle control (number of
5 animals not specified). Food and water were provided ad libitum. Two weeks after administration of
6 1,4-dioxane, 12-O-tetradecanoylphorbol-13-acetate (TPA) (1.0 (ig in 0.2 mL of acetone) was applied to
7 the shaved back of mice 3 times/week for a period of 20 weeks. The yield of papillomas at 24 weeks was
8 selected as a potential predictor of carcinoma yields at 52 weeks following the start of the promotion
9 schedule. Acetone was used instead of TPA in an additional group of 20 mice in order to determine
10 whether a single dose of 1,4-dioxane could induce tumors in the absence of TPA promotion.
11 1,4-Dioxane did not increase the formation of papillomas compared to mice initiated with vehicle
12 and promoted with TPA, indicating lack of initiating activity under the conditions of the study. Negative
13 results were obtained for all three exposure routes. A single dose of 1,4-dioxane did not induce tumors in
14 the absence of TPA promotion.
4.2.3.2 Kingetal.
15 1,4-Dioxane was evaluated for complete carcinogenicity and tumor promotion activity in mouse
16 skin (King etal.. 1973). In the complete carcinogenicity study, 0.2 mL of a solution of 1,4-dioxane (purity
58
DRAFT - DO NOT CITE OR QUOTE
-------
1 not specified) in acetone was applied to the shaved skin of the back of Swiss Webster mice (30/sex)
2 3 times/week for 78 weeks. Acetone was applied to the backs of control mice (30/sex) for the same time
3 period. In the promotion study, each animal was treated with 50 ug of dimethylbenzanthracene 1 week
4 prior to the topical application of the 1,4-dioxane solution described above (0.2 mL, 3 times/week,
5 78 weeks) (30 mice/sex). Acetone vehicle was used in negative control mice (30/sex). Croton oil was
6 used as a positive control in the promotion study (30/sex). Weekly counts of papillomas and suspect
7 carcinomas were made by gross examination. 1,4-Dioxane was also administered in the drinking water
8 (0.5 and 1%) to groups of Osborne-Mendel rats (35/sex/group) and B6C3Fi mice for 42 weeks (control
9 findings were only reported for 34 weeks).
10 1,4-Dioxane was negative in the complete skin carcinogenicity test using dermal exposure. One
11 treated female mouse had malignant lymphoma; however, no papillomas were observed in male or female
12 mice by 60 weeks. Neoplastic lesions of the skin, lungs, and kidney were observed in mice given the
13 promotional treatment with 1,4-dioxane. In addition, the percentage of mice with skin tumors increased
14 sharply after approximately 10 weeks of promotion treatment. Significant mortality was observed when
15 1,4-dioxane was administered as a promoter (only 4 male and 5 female mice survived for 60 weeks), but
16 not as a complete carcinogen (22 male and 25 female mice survived until 60 weeks). The survival of
17 acetone-treated control mice in the promotion study was not affected (29 male and 26 female mice
18 survived until 60 weeks); however, the mice treated with croton oil as a positive control experienced
19 significant mortality (0 male and 1 female mouse survived for 60 weeks). The incidence of mice with
20 papillomas was similar for croton oil and 1,4-dioxane; however, the tumor multiplicity (i.e., number of
21 tumors/mouse) was higher for the croton oil treatment.
22 Oral administration of 1,4-dioxane in drinking water caused appreciable mortality in rats, but not
23 mice, and increased weight gain in surviving rats and male mice. Histopathological lesions (i.e.,
24 unspecified liver and kidney effects) were also reported in exposed male and female rats; however, no
25 histopathological changes were indicated for mice.
26 1,4-Dioxane was demonstrated to be a tumor promoter, but not a complete carcinogen in mouse
27 skin, in this study. Topical administration for 78 weeks following initiation with dimethylbenzanthracene
28 caused an increase in the incidence and multiplicity of skin tumors in mice. Tumors were also observed at
29 remote sites (i.e., kidney and lung), and survival was affected. Topical application of 1,4-dioxane for
30 60 weeks in the absence of the initiating treatment produced no effects on skin tumor formation or
31 mortality in mice.
4.2.3.3 Lundberg et al.
32 Lundberg et al. (1987) evaluated the tumor promoting activity of 1,4-dioxane in rat liver. Male
33 Sprague Dawley rats (8/dose group, 19 for control group) weighing 200 g underwent a partial
34 hepatectomy followed 24 hours later by an i.p. injection of 30 mg/kg diethylnitrosamine (DEN) (initiation
35 treatment). 1,4-Dioxane (99.5% pure with 25 ppm butylated hydroxytoluene as a stabilizer) was then
36 administered daily by gavage (in saline vehicle) at doses of 0, 100, or 1,000 mg/kg-day, 5 days/week for
59
DRAFT - DO NOT CITE OR QUOTE
-------
1 7 weeks. Control rats were administered saline daily by gavage, following DEN initiation. 1,4-Dioxane
2 was also administered to groups of rats that were not given the DEN initiating treatment (saline used
3 instead of DEN). Ten days after the last dose, animals were sacrificed and liver sections were stained for
4 GOT. The number and total volume of GGT-positive foci were determined.
5 1,4-Dioxane did not increase the number or volume of GGT-foci in rats that were not given the
6 DEN initiation treatment. The high dose of 1,4-dioxane (1,000 mg/kg-day) given as a promoting
7 treatment (i.e., following DEN injection) produced an increase in the number of GGT-positive foci and
8 the total foci volume. Histopathological changes were noted in the livers of high-dose rats. Enlarged,
9 foamy hepatocytes were observed in the midzonal region of the liver, with the foamy appearance due to
10 the presence of numerous fat-containing cytoplasmic vacuoles. These results suggest that cytotoxic doses
11 of 1,4-dioxane may be associated with tumor promotion of 1,4-dioxane in rat liver.
4.3 Reproductive/Developmental Studies—Oral and Inhalation
4.3.1 Giavini et al.
12 Pregnant female Sprague Dawley rats (18-20 per dose group) were given 1,4-dioxane (99% pure,
13 0.7% acetal) by gavage in water at concentrations of 0, 0.25, 0.5, or 1 mL/kg-day, corresponding to dose
14 estimates of 0, 250, 500, or 1,000 mg/kg-day (density of 1,4-dioxane is approximately 1.03 g/mL)
15 (Giavini et al., 1985). The chemical was administered at a constant volume of 3 mL/kg on days 6-15 of
16 gestation. Food consumption was determined daily and BWs were measured every 3 days. The dams were
17 sacrificed with chloroform on gestation day 21 and the numbers of corpora lutea, implantations,
18 resorptions, and live fetuses were recorded. Fetuses were weighed and examined for external
19 malformations prior to the evaluation of visceral and skeletal malformations (Wilson's free-hand section
20 method and staining with Alizarin red) and a determination of the degree of ossification.
21 Maternal weight gain was reduced by 10% in the high-dose group (1,000 mg/kg-day). Food
22 consumption for this group was 5% lower during the dosing period, but exceeded control levels for the
23 remainder of the study. No change from control was observed in the number of implantations, live
24 fetuses, or resorptions; however, fetal birth weight was 5% lower in the highest dose group (p < 0.01).
25 1,4-Dioxane exposure did not increase the frequency of major malformations or minor anomalies and
26 variants. Ossification of the sternebrae was reduced in the 1,000 mg/kg-day dose group (p < 0.05). The
27 study authors suggested that the observed delay in sternebrae ossification combined with the decrease in
28 fetal birth weight indicated a developmental delay related to 1,4-dioxane treatment. NOAEL and LOAEL
29 values of 500 and 1,000 mg/kg-day were identified from this study by EPA and based on delayed
30 ossification of the sternebrae and reduced fetal BWs.
60
DRAFT - DO NOT CITE OR QUOTE
-------
4.4 Other Duration or E ndpoint Specific Studies
4.4.1 Acute and Short-term Toxicity
1 The acute (< 24 hours) and short-term toxicity studies (<30 days) of 1,4-dioxane in laboratory
2 animals are summarized in Table 4-22. Several exposure routes were employed in these studies, including
3 dermal application, drinking water exposure, gavage, vapor inhalation, and i.v. or i.p. injection.
4.4.1.1 Oral Toxicity
4 Mortality was observed in many acute high-dose studies, and LD50 values for 1,4-dioxane were
5 calculated for rats, mice, and guinea pigs (Pozzani etal., 1959; HF Jr et al., 1941; Laug et al., 1939).
6 Clinical signs of CNS depression were observed, including staggered gait, narcosis, paralysis, coma, and
7 death (Nelson. 1951; Laug et al.. 1939; Schrenk and Yant 1936; de Navasquez. 1935). Severe liver and
8 kidney degeneration and necrosis were often seen in acute studies (JBRC. 1998; David. 1964; Kesten et
9 al. 1939; Laugetal.. 1939; Schrenk and Yant. 1936; de Navasquez. 1935). JBRC (1998) additionally
10 reported histopathological lesions in the nasal cavity and the brain of rats following 2 weeks of exposure
11 to 1,4-dioxane in the drinking water.
4.4.1.2 Inhalation Toxicity
12 Acute and short-term toxicity studies (all routes) are summarized in Table 4-18. Mortality
13 occurred in many high-concentration studies (Pozzani et al., 1959; Nelson. 1951; Wirth and Klimmer.
14 1936). Inhalation of 1,4-dioxane caused eye and nasal irritation, altered respiration, and pulmonary edema
15 and congestion (Yantet al.. 1930). Clinical signs of CNS depression were observed, including staggered
16 gait, narcosis, paralysis, coma, and death (Nelson. 1951; Wirth and Klimmer. 1936). Liver and kidney
17 degeneration and necrosis were also seen in acute and short-term inhalation studies (Drew et al.. 1978;
18 Fairlev et al.. 1934).
61
DRAFT - DO NOT CITE OR QUOTE
-------
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
Crj:BDF1 mouse
Dog
Oral via
drinking
water
Oral via
drinking
water
Oral via
drinking
water
Gavage
Gavage
Gavage
Gavage
Gavage
Gavage
Oral via
drinking
water
Drinking
water
ingestion
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 LDso
Single dose,
LDso 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
14-day exposure
3-1 0 days of exposure
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
Mortality, decreased
BWs, histopathological
lesions in the nasal
cavity, liver, kidney, and
brain
Clinical signs of CMS
depression, and liver
and kidney
degeneration
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)
LDso =6, 400 mg/kg
(14,200ppm)
LDso (mg/kg):
rat = 7, 120
guinea pig - 3,150
LDso (mg/kg):
mouse = 5,900
rat = 5,400
1,034 mg/kg-day
3,160 mg/kg
10,800 mg/kg-day;
hepatocellular swelling
11,000 mg/kg-day
(5%)
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)
JBRC
(1998)
Schrenk
and Yant
(1936)
Inhalation studies
Male CD1 rat
Vapor
inhalation
Serum enzymes
measured before and
after a single 4 hour
Increase in ALT, AST,
and OCT; no change in
G-6-Pase
1,000 ppm
Drew et al.
(1978)
exposure
DRAFT - DO NOT CITE OR QUOTE
62
-------
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
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
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
6,000 ppm
LCso = 51.3 mg/L
8,400 ppm
0.5% by volume
10,000 ppm
Nelson
(1951)
Pozzani et
al. (1959)
Wirth and
Klimmer
(1936)
Yant et al.
(193Q)
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
"Lowest effective dose for positive
ND = no data; OCT =
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 CNS
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):
24 hours = 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)
results/ highest dose tested for negative results.
ornithine carbamyl transferase; ODC =
ornithine decarboxylase; SDH
= sorbitol dehydrogenase
4.4.2 Neurotoxicity
1 Clinical signs of CNS depression have been reported in humans and laboratory animals following
2 high dose exposure to 1,4-dioxane (see Sections 4.1 and 4.2.1.1). Neurological symptoms were reported
3 in the fatal case of a worker exposed to high concentrations of 1,4-dioxane through both inhalation and
4 dermal exposure (Johnstone. 1959). These symptoms included headache, elevation in blood pressure,
5 agitation and restlessness, and coma. Autopsy findings demonstrated perivascular widening in the brain,
6 with small foci of demyelination in several regions (e.g., cortex, basal nuclei). It was suggested that these
7 neurological changes may have been secondary to anoxia and cerebral edema. In laboratory animals, the
8 neurological effects of acute high-dose exposure included staggered gait, narcosis, paralysis, coma, and
9 death (Nelson. 1951: Laugetal. 1939: Schrenk and Yant. 1936: de Navasquez. 1935: Yantetal.. 1930).
10 The neurotoxicity of 1,4-dioxane was further investigated in several studies described below (Frantik et
11 al.. 1994: Kanadaetal.. 1994: Goldberg et al.. 1964: Knoefel. 1935V
63
DRAFT - DO NOT CITE OR QUOTE
-------
4.4.2.1 Frantiketal.
1 The acute neurotoxicity of 1,4-dioxane was evaluated following a 4-hour inhalation exposure to
2 male Wistar rats (four per dose group) and a 2-hour inhalation exposure to female H-strain mice (eight
3 per dose group) (Frantiketal., 1994). Three exposure groups and a control group were used in this study.
4 Exposure concentrations were not specified, but apparently were chosen from the linear portion of the
5 concentration-effect curve. The neurotoxicity endpoint measured in this study was the inhibition of the
6 propagation and maintenance of an electrically-evoked seizure discharge. This endpoint has been
7 correlated with the behavioral effects and narcosis that occur following acute exposure to higher
8 concentrations of organic solvents. Immediately following 1,4-dioxane exposure, a short electrical
9 impulse was applied through ear electrodes (0.2 seconds, 50 hertz (Hz), 180 volts (V) in rats, 90 V in
10 mice). Several time characteristics of the response were recorded; the most sensitive and reproducible
11 measures of chemically-induced effects were determined to be the duration of tonic hind limb extension
12 in rats and the velocity of tonic extension in mice.
13 Linear regression analysis of the concentration-effect data was used to calculate an isoeffective
14 air concentration that corresponds to the concentration producing a 30% decrease in the maximal response
15 to an electrically-evoked seizure. The isoeffective air concentrations for 1,4-dioxane were 1,860 ±
16 200 ppm in rats and 2,400 ± 420 ppm in mice. A NOAEL value was not identified from this study.
4.4.2.2 Goldberg etal.
17 Goldberg et al. (1964) evaluated the effect of solvent inhalation on pole climb performance in
18 rats. Female rats (Carworth Farms Elias strain) (eight per dose group) were exposed to 0, 1,500, 3,000, or
19 6,000 ppm of 1,4-dioxane in air for 4 hours/day, 5 days/weeks, for 10 exposure days. Conditioned
20 avoidance and escape behaviors were evaluated using a pole climb methodology. Prior to exposure, rats
21 were trained to respond to a buzzer or shock stimulus by using avoidance/escape behavior within
22 2 seconds. Behavioral criteria were the abolishment or significant deferment (>6 seconds) of the
23 avoidance response (conditioned or buzzer response) or the escape response (buzzer plus shock response).
24 Behavioral tests were administered on day 1, 2, 3, 4, 5, and 10 of the exposure period. Rat BWs were also
25 measured on test days.
26 1,4-Dioxane exposure produced a dose-related effect on conditioned avoidance behavior in
27 female rats, while escape behavior was generally not affected. In the 1,500 ppm group, only one of eight
28 rats had a decreased avoidance response, and this only occurred on days 2 and 5 of exposure. A larger
29 number of rats exposed to 3,000 ppm (two or three of eight) experienced a decrease in the avoidance
30 response, and this response was observed on each day of the exposure period. The maximal decrease in
31 the avoidance response was observed in the 6,000 ppm group during the first 2 days of exposure
32 (75-100% of the animals were inhibited in this response). For exposure days 3-10, the percent of rats in
33 the 6,000 ppm group with significant inhibition of the avoidance response ranged from 37-62%. At the
34 end of the exposure period (day 10), the BWs for rats in the high exposure group were lower than
35 controls.
64
DRAFT - DO NOT CITE OR QUOTE
-------
4.4.2.3 Kanadaetal.
1 Kanada et al. evaluated the effect of oral exposure to 1,4-dioxane on the regional neurochemistry
2 of the rat brain (Kanadaetal.. 1994). 1,4-Dioxane was administered by gavage to male Sprague Dawley
3 rats (5/group) at a dose of 1,050 mg/kg, approximately equal to one-fourth the oral LD50. Rats were
4 sacrificed by microwave irradiation to the head 2 hours after dosing, and brains were dissected into small
5 brain areas. Each brain region was analyzed for the content of biogenic amine neurotransmitters and their
6 metabolites using high-performance liquid chromatography (HPLC) or GC methods. 1,4-Dioxane
7 exposure was shown to reduce the dopamine and serotonin content of the hypothalamus. The
8 neurochemical profile of all other brain regions in exposed rats was similar to control rats.
4.4.2.4 Knoefel
9 The narcotic potency of 1,4-dioxane was evaluated following i.p. injection in rats and gavage
10 administration in rabbits (Knoefel. 1935). Rats were given i.p. doses of 20, 30, or 50 mmol/kg. No
11 narcotic effect was seen at the lowest dose; however, rats given 30 mmol/kg were observed to sleep
12 approximately 8-10 minutes. Rats given the high dose of 50 mmol/kg died during the study. Rabbits were
13 given 1,4-dioxane at oral doses of 10, 20, 50, 75, or 100 mmol/kg. No effect on the normal erect animal
14 posture was observed in rabbits treated with less than 50 mmol/kg. At 50 and 75 mmol/kg, a semi-erect or
15 staggering posture was observed; lethality occurred at both the 75 and 100 mmol/kg doses.
4.5 Mechanistic Data and Other Studies in Support of the Mode of
Action
4.5.1 Genotoxicity
16 The genotoxicity data for 1,4-dioxane are presented in Table 4-23 and Table 4-24 for in vitro and
17 in vivo tests, respectively. 1,4-Dioxane has been tested for genotoxic potential using in vitro assay
18 systems with prokaryotic organisms, non-mammalian eukaryotic organisms, and mammalian cells, and in
19 vivo assay systems using several strains of rats and mice. In the large majority of in vitro systems,
20 1,4-dioxane was not genotoxic. Where a positive genotoxic response was observed, it was generally
21 observed in the presence of toxicity. Similarly, 1,4-dioxane was not genotoxic in the majority of available
22 in vivo studies. 1,4-Dioxane did not bind covalently to DNA in a single study with calf thymus DNA.
23 Several investigators have reported that 1,4-dioxane caused increased DNA synthesis indicative of cell
24 proliferation. Overall, the available literature indicates that 1,4-dioxane is nongenotoxic or weakly
25 genotoxic.
26 Negative findings were reported for mutagenicity in in vitro assays with the prokaryotic
27 organisms Salmonella typhimurium, Escherichia coli, and Photobacterium phosphoreum (Mutatox assay)
28 (Morita and Havashi. 1998; Hellmer and Bolcsfoldi. 1992; Kwanetal.. 1990; Khudolev et al.. 1987;
65
DRAFT - DO NOT CITE OR QUOTE
-------
1 Nestmann et al., 1984; Haworth et al., 1983; Stottet al., 1981). In in vitro assays with nonmammalian
2 eukaryotic organisms, negative results were obtained for the induction of aneuploidy in yeast
3 (Saccharomyces cerevisiae) and in the sex-linked recessive lethal test in Drosophila melanogaster (Yoon
4 et al.. 1985; Zimmermann et al.. 1985). In the presence of toxicity, positive results were reported for
5 meiotic nondisj unction in Drosophila (Munoz and Barnett. 2002).
6 The ability of 1,4-dioxane to induce genotoxic effects in mammalian cells in vitro has been
7 examined in model test systems with and without exogenous metabolic activation and in hepatocytes that
8 retain their xenobiotic-metabolizing capabilities. 1,4-Dioxane was reported as negative in the mouse
9 lymphoma cell forward mutation assay (Morita and Hayashi. 1998; McGregor et al.. 1991). 1,4-Dioxane
10 did not produce chromosomal aberrations or micronucleus formation in Chinese hamster ovary (CHO)
11 cells (Morita and Hayashi. 1998; Galloway et al.. 1987). Results were negative in one assay for sister
12 chromatid exchange (SCE) in CHO (Morita and Hayashi. 1998) and were weakly positive in the absence
13 of metabolic activation in another (Galloway et al.. 1987). In rat hepatocytes, 1,4-dioxane exposure in
14 vitro caused single-strand breaks in DNA at concentrations also toxic to the hepatocytes (Sina et al..
15 1983) and produced a positive genotoxic response in a cell transformation assay with BALB/3T3 cells
16 also in the presence of toxicity (Sheu et al.. 1988).
17 1,4-Dioxane was not genotoxic in the majority of available in vivo mammalian assays. Studies of
18 micronucleus formation following in vivo exposure to 1,4-dioxane produced mostly negative results,
19 including studies of bone marrow micronucleus formation in B6C3Fi, BALB/c, CBA, and C57BL6 mice
20 (McFee etal.. 1994; Mirkova. 1994; Tinwell and Ashby. 1994) and micronucleus formation in peripheral
21 blood of GDI mice (Morita and Hayashi. 1998; Morita. 1994). Mirkova (1994) reported a dose-related
22 increase in the incidence of bone marrow micronuclei in male and female C57BL6 mice 24 or 48 hours
23 after administration of 1,4-dioxane. At a sampling time of 24 hours, a dose of 450 mg/kg produced no
24 change relative to control, while doses of 900, 1,800, and 3,600 mg/kg increased the incidence of bone
25 marrow micronuclei by approximately two-, three-, and fourfold, respectively. A dose of 5,000 mg/kg
26 also increased the incidence of micronuclei by approximately fourfold at 48 hours. This compares with
27 the negative results for BALB/c male mice tested in the same study at a dose of 5,000 mg/kg and
28 sampling time of 24 hours. Tinwell and Ashby (1994) could not explain the difference in response in the
29 mouse bone marrow micronucleus assay with C57BL6 mice obtained in their laboratory (i.e.,
30 non-significant 1.6-fold increase over control) with the dose-related positive findings reported by
31 Mirkova (Mirkova. 1994) using the same mouse strain, 1,4-dioxane dose (3,600 mg/kg) and sampling
32 time (24 hours). Morita and Hayashi (1998) demonstrated an increase in micronucleus formation in
33 hepatocytes following 1,4-dioxane dosing and partial hepatectomy to induce cellular mitosis. DNA
34 single-strand breaks were demonstrated in hepatocytes following gavage exposure to female rats (Kitchin
35 and Brown. 1990).
36 Roy et al. (2005) examined micronucleus formation in male CD1 mice exposed to 1,4-dioxane to
37 confirm the mixed findings from earlier mouse micronucleus studies and to identify the origin of the
38 induced micronuclei. Mice were administered 1,4-dioxane by gavage at doses of 0, 1,500, 2,500, and
39 3,500 mg/kg-day for 5 days. The mice were also implanted with 5-bromo-2-deoxyuridine
40 (BrdU)-releasing osmotic pumps to measure cell proliferation in the liver and to increase the sensitivity of
66
DRAFT - DO NOT CITE OR QUOTE
-------
1 the hepatocyte assay. The frequency of micronuclei in the bone marrow erythrocytes and in the
2 proliferating BrdU-labeled hepatocytes was determined 24 hours after the final dose. Significant
3 dose-related increases in micronuclei were seen in the bone-marrow at all the tested doses (>
4 1,500 mg/kg-day). In the high-dose (3,500-mg/kg) mice, the frequency of bone marrow erythrocyte
5 micronuclei was about 10-fold greater than the control frequency. Significant dose-related increases in
6 micronuclei were also observed at the two highest doses (> 2,500 mg/kg-day) in the liver.
7 Antikinetochore (CREST) staining or pancentromeric fluorescence in situ hybridization (FISH) was used
8 to determine the origin of the induced micronuclei. The investigators determined that 80-90% of the
9 micronuclei in both tissues originated from chromosomal breakage; small increase in micronuclei
10 originating from chromosome loss was seen in hepatocytes. Dose-related statistically significant
11 decreases in the ratio of bone marrow polychromatic erythrocytes (PCE):normochromatic erythrocytes
12 (NCE), an indirect measure of bone marrow toxicity, were observed. Decreases in hepatocyte
13 proliferation were also observed. Based on these results, the authors concluded that at high doses
14 1,4-dioxane exerts genotoxic effects in both the mouse bone marrow and liver; the induced micronuclei
15 are formed primarily from chromosomal breakage; and 1,4-dioxane can interfere with cell proliferation in
16 both the liver and bone marrow. The authors noted that reasons for the discrepant micronucleus assay
17 results among various investigators was unclear, but could be related to the inherent variability present
18 when detecting moderate to weak responses using small numbers of animals, as well as differences in
19 strain, dosing regimen, or scoring criteria.
20 1,4-Dioxane did not affect in vitro or in vivo DNA repair in hepatocytes or in vivo DNA repair in
21 the nasal cavity (Goldsworthy et al.. 1991; Stott etal.. 1981). but increased hepatocyte DNA synthesis
22 indicative of cell proliferation in several in vivo studies (Miyagawa et al.. 1999; Uno et al.. 1994;
23 Goldsworthy et al., 1991; Stott et al., 1981). 1,4-Dioxane caused a transient inhibition of RNA
24 polymerase A and B in the rat liver (Kurl et al.. 1981). indicating a negative impact on the synthesis of
25 ribosomal and messenger RNA (DNA transcription). Intravenous administration of 1,4-dioxane at doses
26 of 10 or 100 mg/rat produced inhibition of both polymerase enzymes, with a quicker and more complete
27 recovery of activity for RNA polymerase A, the polymerase for ribosomal RNA synthesis.
28 1,4-Dioxane did not covalently bind to DNA under in vitro study conditions (Woo et al.. 1977b).
29 DNA alkylation was also not detected in the liver 4 hours following a single gavage exposure
30 (1,000 mg/kg) in male Sprague Dawley rats (Stott etal.. 1981).
31 Rosenkranz and Klopman (1992) analyzed 1,4-dioxane using the computer automated structure
32 evaluator (CASE) structure activity method to predict its potential genotoxicity and carcinogenicity. The
33 CASE analysis is based on information contained in the structures of approximately 3,000 chemicals
34 tested for endpoints related to mutagenic/genotoxic and carcinogenic potential. CASE selects descriptors
35 (activating [biophore] or inactivating [biophobe] structural fragments) from a learning set of active and
36 inactive molecules. Using the CASE methodology, Rosenkranz and Klopman (1992) predicted that
37 1,4-dioxane would be inactive for mutagenicity in several in vitro systems, including Salmonella,
38 induction of chromosomal aberrations in CHO cells, and unscheduled DNA synthesis in rat hepatocytes.
39 1,4-Dioxane was predicted to induce SCE in cultured CHO cells, micronuclei formation in rat bone
40 marrow, and carcinogenicity in rodents.
67
DRAFT - DO NOT CITE OR QUOTE
-------
1 Gene expression profiling in cultured human hepatoma HepG2 cells was performed using DNA
2 microarrays to discriminate between genotoxic and other carcinogens (van Delft et al.. 2004). Van Delft
3 et al. (2004) examined this method using a training set of 16 treatments (nine genotoxins and seven
4 nongenotoxins) and a validation set (three and three), with discrimination models based on Pearson
5 correlation analyses for the 20 most discriminating genes. As reported by the authors (van Delft et al..
6 2004). the gene expression profile for 1,4-dioxane indicated a classification of this chemical as a
7 "nongenotoxic" carcinogen, and thus, 1,4-dioxane was included in the training set as a "nongenotoxic"
8 carcinogen. The accuracy for carcinogen classification using this method ranged from 33 to 100%,
9 depending on which chemical data sets and gene expression signals were included in the analysis.
68
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-23 Genotoxicity studies of 1,4-dioxane; in vitro
Test system
Endpoint
Results3
Test conditions Without With
activation activation
Doseb
Source
Prokaryotic organisms in vitro
S. typhimurium
strains TA98,
TA100, TA1535,
TA1537
S. typhimurium
strains TA98,
TA100, TA1530,
TA1535,
TA1537
S. typhimurium
strains TA98,
TA100, TA1535,
TA1537
S. typhimurium
strains TA100,
TA1535
S. typhimurium
strains TA98,
TA100, TA1535,
TA1537,
TA1538
£. CO//K-12
uvrB/recA
£. co//
WP2/WP2uvrA
P. phosphoreum
M169
Nonmammalian
S. cerevisiae
D61.M
D. melanogaster
D. melanogaster
Reverse
mutation
Reverse
mutation
Reverse
mutation
Reverse
mutation
Reverse
mutation
DMA repair
Reverse
mutation
Mutagenicity,
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
1,150mmol/L
5,000 ug/plate
ND
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)
eukaryotic organisms in vitro
Aneuploidy
Meiotic
nondisjunction
Sex-linked
recessive lethal
test
Standard 16-hour
incubation or -T ND
cold-interruption regimen
Oocytes were obtained for
evaluation 24 and +TC NDd
48 hours after mating
Exposure by feeding and Nr.d
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)
Mammalian cells in vitro
Rat hepatocytes
Primary
hepatocyte
culture from
male F344 rats
L5178Y mouse
lymphoma cells
L5178Y mouse
lymphoma cells
DMA damage;
single-strand
breaks
measured by
alkaline elution
DMA repair
Forward
mutation assay
Forward
mutation assay
3-Hour exposure to
isolated primary +Te NDd
hepatocytes
Autoradiography - NDd
Thymidine kinase
mutagenicity assay
(trifluorothymidine
resistance)
Thymidine kinase
mutagenicity assay T
(trifluorothymidine
resistance)
0.3 mM
1 mM
5,000 ug/mL
5,000 ug/mL
Sina et al.
(1983)
Goldsworthy
et al. (1991)
McGregor et
al. (1991)
Morita and
Hayashi
(1998)
DRAFT - DO NOT CITE OR QUOTE
69
-------
BALB/3T3 cells
Cell
transformation
48-Hour exposure
followed by 4 weeks
incubation; 13 day
exposure followed by
2.5 weeks incubation
+Tf
NDa
0.5 mg/mL
Sheu et al.
(1988)
BrdU was added 2 hours
CHO cells
CHO cells
CHO cells
SCE
Chromosomal
aberration
SCE
auer i,t-uioxane auumon,
chemical treatment was ±g
2 hours with S9 and
25 hours without S9
Cells were harvested 8-
12 hours or 18-26 hours
after treatment (time of
first mitosis)
3 hour pulse treatment;
followed by continuous
treatment of BrdU for
23 or 26 hours
1 0,520 ug/ml_
1 0,520 ug/ml_
5,000 ug/mL
Galloway et
al. (1987)
Galloway et
al. (1987)
Morita and
Hayashi
(1998)
CHO cells
Chromosomal
aberration
5 hour pulse treatment,
20 hour pulse and
continuous treatments, or
44 hour continuous
treatment; cells were
harvested 20 or 44 hours
following exposure
5,000 ug/mL
Morita and
Hayashi
(1998)
CHO cells
Micronucleus
formation
5 hour pulse treatment or
44 hour continuous
treatment; cells were
harvested 42 hours
following exposure
5,000 ug/mL
Morita and
Hayashi
(1998)
Calf thymus
DMA
Covalent
binding to DMA
Incubation with
microsomes from
3-methylcholanthrene
treated rats
0.04 pmol/mg
DMA (bound)
Woo et al.
(1977b)
a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity. 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.
°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 DMA 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.
DRAFT - DO NOT CITE OR QUOTE
70
-------
Table 4-24 Genotoxicity studies of 1,4-dioxane; mammalian in vivo
Test system
Female
Sprague Dawley
Rat
Male
Sprague Dawley
Rat
Male
B6C3F-,
Mouse
Male and female
C57BL6
Mouse;
male BALB/c
Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
CBA and
C57BL6 Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
Sprague Dawley
Rat
Test system
Male
F344
Rat
Male
F344
Rat
Endpoint
DMA damage;
single-strand breaks
measured by alkaline
elution
DMA alkylation in
hepatocytes
Micronucleus
formation in bone
marrow
Micronucleus
formation in bone
marrow
Micronucleus
formation in
peripheral blood
Micronucleus
formation in
hepatocytes
Micronucleus
formation in
peripheral blood
Micronucleus
formation in bone
marrow
Micronuclei formation
in bone marrow
Micronuclei formation
in hepatocytes
DMA repair in
hepatocytes
Endpoint
DMA repair in
hepatocytes
(autoradiography)
DMA repair in nasal
epithelial cells from
the nasoturbinate or
maxilloturbinate
Test Conditions Results3
Two gavage doses given 21
and 4 hours prior to +c
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 + /C57BL6xd
polychromatic erythrocytes /DAI R/ (
24 or 48 hours after dosing ~(bALb/c)
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 +e
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 +f
24 hours after dosing
Gavage; analysis for
micronuclei 24 hours after +9
dosing
Drinking water; thymidine
incorporation with
hydroxyurea to repress
normal DMA synthesis
Test Conditions Results3
Gavage and drinking water
exposure; thymidine
incorporation
Gavage and drinking water
exposure; thymidine
incorporation
Dose0
2,550 mg/kg
1,000 mg/kg
Single dose of
4,000 mg/kg;
3 daily doses of
2,000
900 mg/kg
(C57BL6);
5,000 mg/kg
(BALB/c)
3,200 mg/kg
2,000 mg/kg
3,000 mg/kg
3,600 mg/kg
1,500 mg/kg-day
for 5 days
2,500 mg/kg-day
for 5 days
1,000 mg/kg-day
for 1 1 weeks
Dose0
1,000 mg/kg for
2 or 12 hours;
1,500 mg/kg-day
for 2 weeks or
3,000 mg/kg-day
for 1 week
1,500 mg/kg-day
for 8 days +
1,000 mg/kg
gavage dose
12 hours prior to
sacrifice
Source
Kitchin and
Brown (1990)
Stott et al.
(1981)
McFee et al.
(1994)
Mirkova
(1994)
Morita (1994)
Morita and
Hayashi
(1998)
Morita and
Hayashi
(1998)
Tinwell and
Ashby (1994)
Roy et al.
(2005)
Roy et
al.(2005)
Stott et al.
(1981)
Source
Goldsworthy
et al. (1991)
Goldsworthy
et al. (1991)
DRAFT - DO NOT CITE OR QUOTE
71
-------
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
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
hepatocytes
Gavage and drinking water +h
exposure; thymidine (1-2-week
incorporation exposure)
Drinking water exposure;
thymidine incorporation
i.v. injection; activity
measured in isolated +'
hepatocytes
Gavage; thymidine and +j
BrdU incorporation
Thymidine incorporation ±k
Drinking water; thymidine i
incorporation
1,000 mg/kg for
24 or 48 hours;
1,500 mg/kg-day
for 1 or 2 weeks
1,500 mg/kg-day
for 2 weeks
10 mg/rat
1,000 mg/kg
2,000 mg/kg
1,000 mg/kg-day
for 1 1 weeks
Goldsworthy
et al. (1991)
Goldsworthy
et al. (1991)
Kurl et al.
(1981)
Miyagawa
(1999)
Uno et al.
(1994)
Stott et al.
(1981)
a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity. Endogenous metabolic activation is not
applicable for in vivo studies.
Yowest 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.
4.5.2 Mechanistic Studies
4.5.2.1 Free Radical Generation
1 Burmistrov et al. (2001) investigated the effect of 1,4-dioxane inhalation on free radical processes
2 in the rat ovary and brain. Female rats (6-9/group, unspecified strain) were exposed to 0, 10, or
3 100 mg/m3 of 1,4-dioxane vapor for 4 hours/day, 5 days/week, for 1 month. Rats were sacrificed during
72
DRAFT - DO NOT CITE OR QUOTE
-------
1 the morning or evening following exposure and the ovaries and brain cortex were removed and frozen.
2 Tissue preparations were analyzed for catalase activity, glutathione peroxidase activity, and protein
3 peroxidation. Inhalation of 100 mg/m3 of 1,4-dioxane resulted in a significant increase (p < 0.05) in
4 glutathione peroxidase activity, and activation of free radical processes were apparent in both the rat
5 ovary and brain cortex. No change in catalase activity or protein peroxidation was observed at either
6 concentration. A circadian rhythm for glutathione peroxidase activity was absent in control rats, but
7 occurred in rat brain and ovary following 1,4-dioxane exposure.
4.5.2.2 Induction of Metabolism
8 The metabolism of 1,4-dioxane is discussed in detail in Section 3.3. 1,4-Dioxane has been shown
9 to induce its own metabolism (Young et al., 1978a; 1978b). Nannelli et al. (2005) (study details provided
10 in Section 3.3) characterized the CYP450 isozymes that were induced by 1,4-dioxane in the liver, kidney,
11 and nasal mucosa of the rat. In the liver, the activities of several CYP450 isozymes were increased (i.e.,
12 CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was inducible in the kidney and nasal mucosa.
13 CYP2E1 mRNA was increased approximately two- to threefold in the kidney and nasal mucosa, but
14 mRNA levels were not increased in the liver, suggesting that regulation of CYP2E1 is organ-specific.
15 Induction of hepatic CYPB1/2 and CYP2E1 levels by phenobarbital or fasting did not increase the liver
16 toxicity of 1,4-dioxane, as measured by hepatic glutathione content or serum ALT activity. This result
17 suggested that highly reactive and toxic intermediates did not play a large role in the liver toxicity of
18 1,4-dioxane, even under conditions where metabolism was enhanced. This finding is similar to an earlier
19 conclusion by Kociba et al. (1975) who evaluated toxicity from a chronic drinking water study alongside
20 data providing a pharmacokinetic profile for 1,4-dioxane. Kociba et al. (1975) concluded that liver
21 toxicity and eventual tumor formation occurred only at doses where clearance pathways were saturated
22 and elimination of 1,4-dioxane from the blood was reduced. Nannelli et al. (2005) further suggested that a
23 sustained induction of CYP2E1 may lead to generation of reactive oxygen species contributing to target
24 organ toxicity and regenerative cell proliferation; however, no data were provided to support this
25 hypothesis.
4.5.2.3 Mechanisms of Tumor Induction
26 Several studies have been performed to evaluate potential mechanisms for the carcinogenicity of
27 1,4-dioxane (Goldsworthy et al.. 1991; Kitchin and Brown. 1990; Stottetal. 1981). Stott et al. (1981)
28 evaluated 1,4-dioxane in several test systems, including salmonella mutagenicity in vitro, rat hepatocyte
29 DNA repair activity in vitro, DNA synthesis determination in male Sprague Dawley rats following acute
30 gavage dosing or an 11-week drinking water exposure (described in Section 4.2.1), and hepatocyte DNA
31 alkylation and DNA repair following a single gavage dose. This study used doses of 0, 10, 100, or
32 1,000 mg/kg-day, with the highest dose considered to be a tumorigenic dose level. Liver histopathology
33 and liver to BW ratios were also evaluated in rats from acute gavage or repeated dose drinking water
34 experiments.
73
DRAFT - DO NOT CITE OR QUOTE
-------
1 The histopathology evaluation indicated that liver cytotoxicity (i.e., centrilobular hepatocyte
2 swelling) was present in rats from the 1,000 mg/kg-day dose group that received 1,4-dioxane in the
3 drinking water for 11 weeks (Stott et al., 1981). An increase in the liver to BW ratio accompanied by an
4 increase in hepatic DNA synthesis was also seen in this group of animals. No effect on histopathology,
5 liver weight, or DNA synthesis was observed in acutely exposed rats or rats that were exposed to a lower
6 dose of 10 mg/kg-day for 11 weeks. 1,4-Dioxane produced negative findings in the remaining
7 genotoxicity assays conducted as part of this study (i.e., Salmonella mutagenicity, in vitro and in vivo rat
8 hepatocyte DNA repair, and DNA alkylation in rat liver). The study authors suggested that the observed
9 lack of genotoxicity at tumorigenic and cytotoxic dose levels indicates an epigenetic mechanism for
10 1,4-dioxane hepatocellular carcinoma in rats.
11 Goldsworthy et al. (1991) evaluated potential mechanisms for the nasal and liver carcinogenicity
12 of 1,4-dioxane in the rat. DNA repair activity was evaluated as a measure of DNA reactivity and DNA
13 synthesis was measured as an indicator of cell proliferation or promotional activity. In vitro DNA repair
14 was evaluated in primary hepatocyte cultures from control and 1,4-dioxane-treated rats (1 or 2% in the
15 drinking water for 1 week). DNA repair and DNA synthesis were also measured in vivo following a
16 single gavage dose of 1,000 mg/kg, a drinking water exposure of 1% (1,500 mg/kg-day) for 1 week, or a
17 drinking water exposure of 2% (3,000 mg/kg-day) for 2 weeks. Liver to BW ratios and palmitoyl CoA
18 oxidase activity were measured in the rat liver to determine whether peroxisome proliferation played a
19 role in the liver carcinogenesis of 1,4-dioxane. In vivo DNA repair was evaluated in rat nasal epithelial
20 cells derived from either the nasoturbinate or the maxilloturbinate of 1,4-dioxane-treated rats. These rats
21 received 1% 1,4-dioxane (1,500 mg/kg-day) in the drinking water for 8 days, followed by a single gavage
22 dose of 10, 100, or 1,000 mg/kg 12 hours prior to sacrifice. Archived tissues from the NCI (1978)
23 bioassay were reexamined to determine the primary sites for tumor formation in the nasal cavity
24 following chronic exposure in rats. Histopathology and cell proliferation were determined for specific
25 sites in the nasal cavity that were related to tumor formation. This evaluation was performed in rats that
26 were exposed to drinking water containing 1% 1,4-dioxane (1,500 mg/kg-day) for 2 weeks.
27 1,4-Dioxane and its metabolite l,4-dioxane-2-one did not affect in vitro DNA repair in primary
28 hepatocyte cultures (Goldsworthy et al., 1991). In vivo DNA repair was also unaffected by acute gavage
29 exposure or ingestion of 1,4-dioxane in the drinking water for a 1- or 2-week period. Hepatocyte cell
30 proliferation was not affected by acute gavage exposure, but was increased approximately twofold
31 following a 1-2-week drinking water exposure. A 5-day drinking water exposure to 1% 1,4-dioxane
32 (1,500 mg/kg-day) did not increase the activity of palmitoyl coenzyme A or the liver to BW ratio,
33 suggesting that peroxisome proliferation did not play a role in the hepatocarcinogenesis of 1,4-dioxane.
34 Nannelli et al. (2005) also reported a lack of hepatic palmitoyl CoA induction following 10 days of
35 exposure to 1.5% 1,4-dioxane in the drinking water (2,100 mg/kg-day).
36 Treatment of rats with 1% (1,500 mg/kg-day) 1,4-dioxane for 8 days did not alter DNA repair in
37 nasal epithelial cells (Goldsworthy et al., 1991). The addition of a single gavage dose of up to
38 1,000 mg/kg 12 hours prior to sacrifice also did not induce DNA repair. Reexamination of tissue sections
39 from the NCI (1978) bioassay suggested that the majority of nasal tumors were located in the dorsal nasal
40 septum or the nasoturbinate of the anterior portion of the dorsal meatus (Goldsworthy et al.. 1991). No
74
DRAFT - DO NOT CITE OR QUOTE
-------
1 histopathological lesions were observed in nasal section of rats exposed to drinking water containing 1%
2 1,4-dioxane (1,500 mg/kg-day) for 2 weeks and no increase was observed in cell proliferation at the sites
3 of highest tumor formation in the nasal cavity.
4 Female Sprague Dawley rats (three to nine per group) were given 0, 168, 840, 2,550, or
5 4,200 mg/kg 1,4-dioxane (99% purity) by corn oil gavage in two doses at 21 and 4 hours prior to sacrifice
6 (Kitchin and Brown. 1990). DNA damage (single-strand breaks measured by alkaline elution), ODC
7 activity, reduced glutathione content, and CYP450 content were measured in the liver. Serum ALT
8 activity and liver histopathology were also evaluated. No changes were observed in hepatic reduced
9 glutathione content or ALT activity. Light microscopy revealed minimal to mild vacuolar degeneration in
10 the cytoplasm of hepatocytes from three of five rats from the 2,550 mg/kg dose group. No
11 histopathological lesions were seen in any other dose group, including rats given a higher dose of
12 4,200 mg/kg. 1,4-Dioxane caused 43 and 50% increases in DNA single-strand breaks at dose levels of
13 2,550 and 4,200 mg/kg, respectively. CYP450 content was also increased at the two highest dose levels
14 (25 and 66% respectively). ODC activity was increased approximately two-, five-, and eightfold above
15 control values at doses of 840, 2,550, and 4,200 mg/kg, respectively. The results of this study
16 demonstrated that hepatic DNA damage can occur in the absence of significant cytotoxicity. Parameters
17 associated with tumor promotion (i.e., ODC activity, CYP450 content) were also elevated, suggesting that
18 promotion may play a role in the carcinogenesis of 1,4-dioxane.
4.6 Synthesis of Major Noncancer Effects
19 Liver, kidney, and nasal toxicity were the primary noncancer health effects associated with
20 exposure to 1,4-dioxane. In humans, several fatal cases of hemorrhagic nephritis and centrilobular
21 necrosis of the liver were related to occupational exposure (i.e., inhalation and dermal contact) to
22 1,4-dioxane (Johnstone. 1959; Barber. 1934). Neurological changes were also reported in one case;
23 including, headache, elevation in blood pressure, agitation and restlessness, and coma (Johnstone. 1959).
24 Perivascular widening was observed in the brain of this worker, with small foci of demyelination in
25 several regions (e.g., cortex, basal nuclei). In laboratory animals, following oral and inhalation exposure
26 to 1.4-dioxane. liver and kidney degeneration and necrosis were observed(JBRC. 1998; Drew et al. 1978;
27 David. 1964; Kesten et al., 1939; Laugetal.. 1939; Schrenk and Yant. 1936; de Navasquez. 1935; Fairley
28 etal.. 1934). m addition to changes in the nasal epithelium (JBRC. 1998)(Kano et al.. 2008)(Kano et al..
29 2009)(Kasai et al.. 2008)(Kasai et al.. 2009). The results of subchronic and chronic studies are discussed
30 below.
4.6.1 Oral
31 Table 4-25 presents a summary of the noncancer results for the subchronic and chronic oral
32 studies of 1,4-dioxane toxicity in experimental animals. Liver and kidney toxicity were the primary
33 noncancer health effects of oral exposure to 1,4-dioxane in animals. Kidney damage at high doses was
75
DRAFT - DO NOT CITE OR QUOTE
-------
1 characterized by degeneration of the cortical tubule cells, necrosis with hemorrhage, and
2 glomerulonephritis (NCI. 1978; Kocibaetal. 1974; Argus etal.. 1965; Fairley et al.. 1934). Renal cell
3 degeneration generally began with cloudy swelling of cells in the cortex (Fairley et al.. 1934). Nuclear
4 enlargement of proximal tubule cells was observed at doses below those producing renal necrosis (Kano
5 et al.. 2008; JBRC. 1998). but is of uncertain toxicological significance. The lowest dose reported to
6 produce kidney damage was 94 mg/kg-day, which produced renal degeneration and necrosis of tubule
7 epithelial cells in male rats in the Kociba et al. (1974) study. Cortical tubule degeneration was seen at
8 higher doses in the NCI (1978) bioassay (240 mg/kg-day, male rats), and glomerulonephritis was reported
9 for rats given doses of > 430 mg/kg-day (Argus etal.. 1973; Argus et al.. 1965).
76
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-25 Oral toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration , ^ , , L°AEHL ,
(mg/kg-day) (mg/kg-day)
Effect
Reference
Subchronic studies
Rat and Mouse
(6/species);
unknown strain
Male
Sprague Dawley
Rat
(4-6/group)
F344/DuCrj Rat
(10/sex/group)
Crj:BDF1 Mouse
(10/sex/group)
Rats
0 or 1,900 mg/kg-day;
Mice NA J'lnn •
0 or 3,300 mg/kg-day 3,300 mice
for 67 days
Rats
0, 10, or 1,000 mg/kg-day 10 1,000
for 1 1 weeks
Rats
Males 0, 52, 126,274,
657, or 1,554 mg/kg-day; „ ,9R
Females 0, 83, 185, 427,
756, or 1,614 mg/kg-day
for 13 weeks
Mice
Males 0, 86,231, 585,
882, or 1,570 mg/kg-day;
Females 0, 170, 387, 170 387
898, 1,620, or
2,669 mg/kg-day
for 13 weeks
Renal cortical degeneration
and necrosis, hemorrhage;
hepatocellular degeneration
Minimal centrilobular
hepatocyte swelling;
increased DMA synthesis
Nuclear enlargement of
nasal respiratory
epithelium; hepatocyte
swelling
Nuclear enlargement of
bronchial epithelium
Fairley et al.
(1934)
Stott et al.
(1981)
Kano et al.
(2008)
Kano et al.
(2008)
Chronic studies
Male
Wistar
Rat (26 treated,
9 controls)
Male
Sprague Dawley
Rat (30/group)
Sherman Rat
(60/sex/dose
group)
Osborne-Mendel
Rat
(35/sex/dose
level)
B6C3F-I Mouse
(50/sex/dose
level)
F344/DuCrj Rat
(50/sex/dose
level)
Rats
0 or 640 mg/kg-day NA 640
for 63 weeks
Rats
0, 430, 574, 803, or
1,032 mg/kg-day NA ^JU
for 13 months
Rats
Males 0, 9.6, 94, or
1,01 5 mg/kg-day; qfi q.
Females 0, 19, 148, or
1,599 mg/kg-day
for 2 years
Rats
Males 0, 240, or
530 mg/kg-day;
Females 0, 350, or
640 mg/kg-day
for 110 weeks
Mice
Males 0, 720, or
830 mg/kg-day;
Females 0, 380, or NA dOU
860 mg/kg-day
for 90 weeks
Rats
Males 0, 11, 55, or
274 mg/kg-day; „ „.
Females 0, 18, 83, or
429 mg/kg-day
for 2 years
Hepatocytes with enlarged
hyperchromic nuclei;
glomerulonephritis
Hepatocytomegaly;
glomerulonephritis
Degeneration and necrosis
of renal tubular cells and
hepatocytes
Pneumonia, gastric ulcers,
and cortical tubular
degeneration in the kidney
Pneumonia and rhinitis
Atrophy of nasal olfactory
epithelium; nasal adhesion
and inflammation
Argus et al.
(1965)
Argus et al.
(1973)
Kociba et al.
(1974)
NCI (1978)
NCI (1978)
JBRC (1998);
Kano et al.
(2009)
77
DRAFT - DO NOT CITE OR QUOTE
-------
Rats
F344/DuCri Rat MaleS °' 1 1 ' 55' Or
(50/sx/doe P T'lr^P, H
LvPh Females 0, 18, 83, or
; 429 mg/kg-day
for 2 years
Rats
F344/DuCri Rat MaleS °' 1 1- 55' °r
/en, M 274 mg/kg-day;
50/sex/dose Fema|L g, 18y'83] or 55
' 429 mg/kg-day
for 2 years
Mice
Crj:BDF1 Mouse Males 049 191 or
(50/sex/dose f7 ™9/knlay;07a 66
LvPh Females 0, 66, 278, or
; 964 mg/kg-day
for 2 years
Mice
Crj:BDF1 Mouse Males 049 191 or
.'. . , 677 mg/kg-day; ....
50/sex/dose Fema|L g, 66y 278, or 49
ve ' 964 mg/kg-day
for 2 years
55 Liver hyperplasia
Increases in serum liver
274 enzymes (GOT, GPT, LDH,
and ALP)
278 Nasal inflammation
Increases in serum liver
191 enzymes (GOT, GPT, LDH,
and ALP)
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
Dpf c
SpragueDawley Pregnant dams 0] 250]
/HO on/ x 500, or 1,000 mg/kg-day
(18-20/group) on gestation dayy£ 6y_15Y
Delayed ossification of the
1,000 sternebrae and reduced
fetal BWs
Giavini et al.
(1985)
1 Liver effects included degeneration and necrosis, hepatocyte swelling, cells with hyperchromic
2 nuclei, spongiosis hepatis, hyperplasia, and clear and mixed cell foci of the liver (Kano et al., 2008; NCI.
3 1978: Kocibaetal.. 1974: Argus etal.. 1973: Argus etal.. 1965: Fairlev et al.. 1934). Hepatocellular
4 degeneration and necrosis were seen at high doses in a subchronic study (1,900 mg/kg-day in rats)
5 (Fairlev et al.. 1934) and at lower doses in a chronic study (94 mg/kg-day, male rats) (Kociba et al..
6 1974). Argus et al. (1973) described a progression of preneoplastic effects in the liver of rats exposed to a
7 dose of 575 mg/kg-day. Early changes (8 months exposure) were described as an increased nuclear size of
8 hepatocytes, disorganization of the rough endoplasmic reticulum, an increase in smooth endoplasmic
9 reticulum, a decrease in glycogen, an increase in lipid droplets in hepatocytes, and formation of liver
10 nodules. Spongiosis hepatis, hyperplasia, and clear and mixed-cell foci were also observed in the liver of
11 rats (doses >55 mg/kg-day in male rats) (Kano et al.. 2009: JBRC. 1998). Clear and mixed-cell foci are
12 commonly considered preneoplastic changes and would not be considered evidence of noncancer toxicity
13 when observed in conjunction with tumor formation. If exposure to 1,4-dioxane had not resulted in tumor
14 formation, these lesions could represent potential noncancer toxicity. The nature of spongiosis hepatis as a
15 preneoplastic change is less well understood (Bannasch. 2003: Karbe and Kerlin. 2002: Stroebel et al..
16 1995). Spongiosis hepatis is a cyst-like lesion that arises from the perisinusoidal Ito cells of the liver. This
17 change is sometimes associated with hepatocellular hypertrophy and liver toxicity (Karbe and Kerlin.
18 2002). but may also occur in combination with preneoplastic foci, or hepatocellular adenoma or
19 carcinoma (Bannasch. 2003: Stroebel et al.. 1995). In the case of the JBRC (1998) study, spongiosis
20 hepatis was associated with other preneoplastic changes in the liver (hyperplasia, clear and mixed-cell
21 foci). No other lesions indicative of liver toxicity were seen in this study; therefore, spongiosis hepatis
78
DRAFT - DO NOT CITE OR QUOTE
-------
1 was not considered indicative of noncancer effects. The activity of serum enzymes (i.e., AST, ALT,
2 LDH, and ALP) was increased in rats and mice exposed to 1,4-dioxane, although only in groups with
3 high incidence of liver tumors. Blood samples were collected only at the end of the 2-year study, so
4 altered serum chemistry may be associated with the tumorigenic changes in the liver.
5 Hematological changes were reported in the JBRC (1998) study only. Mean doses are reported
6 based on information provided in Kano et al. (2009). Observed increases in RBCs, hematocrit,
7 hemoglobin in high-dose male mice (677 mg/kg-day) may be related to lower drinking water
8 consumption (74% of control drinking water intake). Hematological effects noted in male rats given
9 55 mg/kg-day (decreased RBCs, hemoglobin, hematocrit, increased platelets) were within 20% of control
10 values. A reference range database for hematological effects in laboratory animals (Wolford et al., 1986)
11 indicates that a 20% change in these parameters may fall within a normal range (10th-90th percentile
12 values) and may not represent a treatment-related effect of concern.
13 Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
14 dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice). The
15 JBRC (1998) study also demonstrates atrophy of the nasal epithelium and adhesion in rats and mice.
16 Nasal inflammation may be a response to direct contact of the nasal mucosa with drinking water
17 containing 1,4-dioxane (Sweeney et al., 2008; Goldsworthy et al., 1991) or could result from systemic
18 exposure. Regardless, inflammation may indicate toxicity due to 1,4-dioxane exposure. A significant
19 increase in the incidence of pneumonia was reported in mice from the NCI (1978) study. The significance
20 of this effect is unclear, as it was not observed in other studies that evaluated lung histopathology (Kano
21 et al.. 2008; JBRC. 1998; Kociba et al.. 1974). No studies were available regarding the potential for
22 1,4-dioxane to cause immunological effects. Metaplasia and hyperplasia of the nasal epithelium were also
23 observed in high-dose male and female rats (JBRC. 1998): however, these effects are likely to be
24 associated with the formation of nasal cavity tumors in these dose groups. Nuclear enlargement of the
25 nasal olfactory epithelium was observed at a dose of 83 mg/kg-day in female rats (Kano et al.. 2009):
26 however, it is unclear whether this alteration represents an adverse toxicological effect. Nuclear
27 enlargement of the tracheal and bronchial epithelium and an accumulation of foamy cells in the lung were
28 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
29 Two subchronic (Kasai et al., 2008: Fairley et al., 1934) and two chronic inhalation studies (Kasai
30 et al.. 2009: Torkelson et al.. 1974) were identified. Nasal, liver, and kidney toxicity were the primary
31 noncancer health effects of inhalation exposure to 1.4-dioxane in animals. Table 4-26 presents a summary
32 of the noncancer results for the subchronic and chronic inhalation studies of 1.4-dioxane toxicity in
33 laboratory animals.
34 Of the inhalation studies, nasal tissue was only collected in rat studies conducted by Kasai et al.
35 (2009: 2008). Damage to nasal tissue was reported frequently in these studies and statistically significant
36 observations were noted as low as 50 ppm. Nasal effects included deformity of the nose and
79
DRAFT - DO NOT CITE OR QUOTE
-------
1 histopathological lesions characterized by enlarged epithelial nuclei (respiratory epithelium, olfactory
2 epithelium, trachea, and bronchus), atrophy (olfactory epithelium), vacuolic change (olfactory epithelium
3 and bronchial epithelium), squamous cell metaplasia and hyperplasia (respiratory epithelium), respiratory
4 metaplasia (olfactory epithelium), inflammation (respiratory and olfactory epithelium), hydropic change
5 (lamina propria). and sclerosis (lamina propria). In both studies, a concentration-dependent, statistically
6 significant change in enlarged nuclei of the respiratory epithelium was considered the most sensitive nasal
7 effect by the study authors; however, the toxicological significance of nuclear enlargement is uncertain.
8 At high doses, liver damage was characterized by cell degeneration which varied from swelling
9 (Kasai et al.. 2008: Fairlev et al.. 1934) to necrosis (Kasai et al.. 2009: Kasai et al.. 2008: Fairlev et al..
10 1934). spongiosis hepatis (Kasai et al.. 2009). nuclear enlargement of centrilobular cells (Kasai et al..
11 2009) and basophilic and acidophilic cell foci (Kasai et al.. 2009). Altered cell foci are commonly
12 considered preneoplastic changes and would not be considered evidence of noncancer toxicity when
13 observed in conjunction with tumor formation (Bannasch et al.. 1982). Since exposure to 1.4-dioxane
14 resulted in tumor formation in the liver, these lesions are not considered as potential noncancer toxicity.
15 At concentrations ranging from 200 ppm to 3.200 ppm. altered liver enzymes (i.e.. AST. ALT.
16 ALP, and y-GTP). increased liver weights, and induction of GST-P was also observed (Kasai et al.. 2009;
17 Kasai et al.. 2008). Changes in the activity of serum enzymes were mostly observed in exposed rat groups
18 of high 1.4-dioxane concentrations (Kasai et al.. 2009; Kasai et al.. 2008). Induction of GST-P positive
19 hepatocytes was observed in female rats at 1.600 ppm and male and female rats at 3.200 ppm following
20 13 weeks of exposure to 1.4-dioxane. GST-P is considered a good enzymatic marker for early detection of
21 chemical hepatocarcinogenesis (Sato. 1989). Although. GST-P positive liver foci were not observed in the
22 2 year bioassay. the focally and proliferating GST-P positive hepatocytes noted in the H week study
23 suggests eventual progression to hepatocellular tumors after 2 years of exposure and therefore would not
24 be a potential noncancer effect.
25 The lowest concentration reported to produce liver lesions was 1.250 ppm. characterized by
26 necrosis of centrilobular cells, spongiosis hepatis. and nuclear enlargement in the Kasai et al. (2009)
27 study. However, as previously stated, the toxicological significance of nuclear enlargement lesions is
28 uncertain.
29 Kidney effects were reported less frequently in these inhalation studies and were generally
30 observed at higher exposure concentrations than nasal and liver effects. Kidney damage was described as
31 patchy degeneration of cortical tubules with vascular congestion and hemorrhage (Fairlev et al.. 1934).
32 hydropic change of proximal tubules (Kasai et al.. 2009; Kasai et al.. 2008). and as nuclear enlargement
33 of proximal tubules cells (Kasai et al.. 2009). Changes in serum chemistry and urinalysis variables were
34 also noted as evidence of renal damage. In a 13 week inhalation study of male and female rats (Kasai et
35 al.. 2008) kidney toxicity was only observed in female rats exposed to 3.200 ppm of 1.4-dioxane (i.e.
36 hydropic change in the renal proximal tubules), which suggests a possible increased susceptibility of
37 female rats to renal damage following inhalation exposure to 1.4-dioxane.
38 Other noted noncancer effects in laboratory animals included acute vascular congestion of the
39 lungs (Fairlev et al.. 1934); changes in relative lung weights (Kasai et al.. 2008); and decrease in body
80
DRAFT - DO NOT CITE OR QUOTE
-------
1 weight gain (Kasai et al.. 2009; Kasai et al.. 2008). Following a 13-week exposure, higher 1.4-dioxane
2 plasma levels were found in female rats as compared to male rats (Kasai et al.. 2008). 1.4-Dioxane was
3 observed in plasma along with systemic effects following subchronic inhalation exposure to 1.4-dioxane
4 in rats.
Table 4-26 Inhalation toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration N^
™
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 ppmfor7
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
Renal cortical
degeneration and
1,000 hemorrhage;
hepatocellular
degeneration and necrosis
Respiratory epithelium:
1 00 nuclear enlargement of
epithelial cells
Fairley et al.
(1934)
Kasai et al.
(20081
Chronic studies
Wistar rat (288/sex)
F344/DuCrj male
rat
(50/group)
111 ppm for /hours/day, 111 (free
5days/wk, for 2 years standing)
0, 50, 250, or 1,250 ppm
for 6 hours/day, 5 N/A
days/wk for 2 years
No significant effects were
observed on BWs,
NA survival, organ weights,
hematology, clinical
chemistry, or
histopathology
Respiratory epithelium:
„ nuclear enlargement of
epithelial cells, atrophy,
and metaplasia
Torkelson et
al. (19741
Kasai et al.
(20091
4.6.2.1 Mode of Action Information
5 The metabolism of 1,4-dioxane in humans was extensive at low doses (<50 ppm). The linear
6 elimination of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism was a
7 nonsaturated, first-order process at this exposure level (Young et al.. 1977; 1976). Like humans, rats
8 extensively metabolized inhaled 1,4-dioxane; however, plasma data from rats given single i.v. doses of 3,
9 10, 30, 100, or 1,000 mg [14C]-l,4-dioxane/kg demonstrated a dose-related shift from linear, first-order to
10 nonlinear, saturable metabolism of 1,4-dioxane (Young et al.. 1978a; 1978b). Conversely, using the
11 Young et al. (1978b; 1978a) rat model, the metabolism of 1.4-dioxane in rats that were exposed to 400.
12 800. 1.600. and 3.200 ppm via inhalation for 13 weeks could not be accurately depicted due to a lack of
13 knowledge on needed model parameters and biological processes (See Section 3.5.3 and Appendix B). It
14 appears, following prolonged inhalation exposure to 1.4-dioxane at concentrations up to 3.200 ppm. that
15 metabolism is induced (Appendix B).
81
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-Dioxane oxidation appeared to be CYP450-mediated, as CYP450 induction with
2 phenobarbital or Aroclor 1254 and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine or
3 cobaltous chloride was effective in significantly increasing and decreasing, respectively, the appearance
4 of HEAA in the urine of rats (Wooetal.. 1978. 1977c). 1,4-Dioxane itself induced CYP450-mediated
5 metabolism of several barbiturates in Hindustan mice given i.p. injections of 25 and 50 mg/kg of
6 1,4-dioxane (Mungikar and Pawar. 1978). The differences between single and multiple doses in urinary
7 and expired radiolabel support the notion that 1,4-dioxane may induce its own metabolism. 1,4-Dioxane
8 has been shown to induce several isoforms of CYP450 in various tissues following acute oral
9 administration by gavage or drinking water (Nannelli et al.. 2005). In the liver, the activity of several
10 CYP450 isozymes was increased (i.e., CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was
11 inducible in the kidney and nasal mucosa. CYP2E1 mRNA was increased approximately two- to threefold
12 in the kidney and nasal mucosa, but mRNA levels were not increased in the liver, suggesting that
13 regulation of CYP2E1 was organ-specific.
14 Nannelli et al. (2005) investigated the role of CYP450 isozymes in the liver toxicity of
15 1,4-dioxane. Hepatic CYPB1/2 and CYP2E1 levels were induced by phenobarbital or fasting and liver
16 toxicity was measured as hepatic glutathione content or serum ALT activity. No increase in glutathione
17 content or ALT activity was observed, suggesting that highly reactive and toxic intermediates did not play
18 a large role in the liver toxicity of 1,4-dioxane, even under conditions where metabolism was enhanced.
19 Pretreatment with inducers of mixed-function oxidases also did not significantly change the extent of
20 covalent binding in subcellular fractions (Woo etal.. 1977b). Covalent binding was measured in liver,
21 kidney, spleen, lung, colon, and skeletal muscle 1-12 hours after i.p. dosing with 1,4-dioxane. Covalent
22 binding was highest in liver, spleen, and colon. Within hepatocytes, 1,4-dioxane distribution was greatest
23 in the cytosolic fraction, followed by the microsomal, mitochondrial, and nuclear fractions.
24 The absence of an increase in toxicity following an increase in metabolism suggests that
25 accumulation of the parent compound may be related to 1,4-dioxane toxicity. This hypothesis is supported
26 by a comparison of the pharmacokinetic profile of 1,4-dioxane with the toxicology data from a chronic
27 drinking water study (Kociba et al.. 1975). This analysis indicated that liver toxicity did not occur unless
28 clearance pathways were saturated and elimination of 1,4-dioxane from the blood was reduced. A
29 dose-dependent increase of 1.4-dioxane accumulation in the blood was seen, which correlated to the
30 observed dose-dependent increase in incidences of nasal liver, and kidney toxicities (Kasai et al., 2008).
31 Alternative metabolic pathways (i.e., not CYP450 mediated) may be present at high doses of 1,4-dioxane;
32 however, the available studies have not characterized these pathways or identified any possible reactive
33 intermediates. Thus, the mechanism by which 1,4-dioxane induces tissue damage is not known, nor is it
34 known whether the toxic moiety is 1,4-dioxane or a transient or terminal metabolite.
82
DRAFT - DO NOT CITE OR QUOTE
-------
4.7 Evaluation of Carcinogenicity
4.7.1 S ummary of Overall Weight of Evidence
1 Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is "likely
2 to be carcinogenic to humans" based on evidence of carcinogenicity in several 2-year bioassays
3 conducted in four strains of rats, two strains of mice, and in guinea pigs (Kano et al.. 2009; Kasai et al..
4 2009: JBRC. 1998: Yamazaki et al. 1994: NCI. 1978: Kocibaetal.. 1974: Argus etal.. 1973: Hoch-
5 Ligeti and Argus. 1970: Hoch-Ligeti et al., 1970: Argus et al., 1965). Tissue sites where tumors have been
6 observed in these laboratory animals due to exposure to 1.4-dioxane include, peritoneum (Kano et al..
7 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). mammary gland (Kano et al.. 2009: Kasai
8 et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). liver (Kano et al.. 2009: Kasai et al.. 2009). kidney
9 (Kasai et al.. 2009). Zymbal gland (Kasai et al.. 2009). subcutaneous (Kasai et al.. 2009). nasal tissue
10 (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kociba et al.. 1974:
11 Argus etal.. 1973: Hoch-Ligeti et al.. 1970). and lung (Hoch-Ligeti and Argus. 1970). Studies in humans
12 are inconclusive regarding evidence for a causal link between occupational exposure to 1,4-dioxane and
13 increased risk for cancer; however, only two studies were available and these were limited by small
14 cohort size and a small number of reported cancer cases (Buffler et al.. 1978: Thiess etal.. 1976).
15 The available evidence is inadequate to establish a mode of action (MOA) by which 1,4-dioxane
16 or a transient or terminal metabolite induces liver tumors in rats and mice. A MOA hypothesis involving
17 sustained proliferation of spontaneously transformed liver cells has some support from data indicating
18 that 1,4-dioxane acts as a tumor promoter in mouse skin and rat liver bioassays (Lundberg et al.. 1987:
19 King etal.. 1973). Dose-response and temporal data support the occurrence of cell proliferation and
20 hyperplasia prior to the development of liver tumors (JBRC. 1998: Kocibaetal.. 1974) in the rat model.
21 However, the dose-response relationship for induction of hepatic cell proliferation has not been
22 characterized, and it is unknown if it would reflect the dose-response relationship for liver tumors in the
23 2-year rat and mouse studies. Conflicting data from rat and mouse bioassays (JBRC. 1998: Kociba et al..
24 1974) suggest that cytotoxicity may not be a required precursor event for 1,4-dioxane-induced cell
25 proliferation. Data regarding a plausible dose response and temporal progression (see Table 4-21) from
26 cytotoxicity and cell proliferation to eventual liver tumor formation are not available.
27 For nasal tumors, there is no known MOA. There is a hypothesized MOA that includes metabolic
28 induction, cytotoxicity. and regenerative cell proliferation (Kasai et al.. 2009). The induction of CYP450
29 has some support from data illustrating that following acute oral administration of 1.4-dioxane by gavage
30 or drinking water. CYP2E1 was inducible in nasal mucosa (Nannelli et al.. 2005). CYP2E1 mRNA was
31 increased approximately two- to threefold in nasal mucosa (and in the kidney, see section 3.3) in the
32 Nannelli et al. (2005) study. While cell proliferation was observed following 1.4-dioxane exposure in
33 both a 2-year inhalation study in male rats (1.250 ppm) (Kasai et al.. 2009) and a 2-year drinking water
34 study in male (274 mg/kg-day) and female rats (429 mg/kg-day). no evidence of cvtotoxicity in the nasal
35 cavity was observed (Kasai et al.. 2009): therefore, cytotoxicity. as a key event, is not supported.
83
DRAFT - DO NOT CITE OR QUOTE
-------
1 Following a 13-week inhalation study in rats, a concentration-dependent accumulation of 1.4-dioxane in
2 the blood was observed (Kasai et al.. 2008). Studies have shown that water-soluble, gaseous irritants
3 cause nasal injuries such as squamous cell carcinomas (Morgan et al., 1986). Similarly. 1.4-dioxane.
4 which has been reported as a miscible compound (Hawley and Lewis. 2001). also caused nasal injuries
5 that were concentration-dependent, including nasal tumors (Kasai et al., 2009). Additionally, it has been
6 suggested that in vivo genotoxicity may contribute to the carcinogenic MOA for 1.4-Dioxane (Kasai et
7 al., 2009) (see Section 4.7.3.6 for further discussion). Collectively, these data are insufficient to support
8 the hypothesized MOAs.
9 The MOA by which 1,4-dioxane produces kidneyi lung, peritoneal (mesotheliomas), mammary
10 gland, Zymbal gland, and subcutis tumors is also unknown, and there are no available data regarding any
11 hypothesized carcinogenic MOA for 1,4-dioxane in these tissues.
12 U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) indicate that for
13 tumors occurring at a site other than the initial point of contact, the weight of evidence for carcinogenic
14 potential may apply to all routes of exposure that have not been adequately tested at sufficient doses. An
15 exception occurs when there is convincing information (e.g., toxicokinetic data) that absorption does not
16 occur by other routes. Information available on the carcinogenic effects of 1,4-dioxane via the oral route
17 demonstrates that tumors occur in tissues remote from the site of absorption. In addition, information on
18 the carcinogenic effects of 1.4-dioxane via the inhalation route in animals also demonstrates that tumors
19 occur at tissue sites distant from the portal of entry. Information on the carcinogenic effects of
20 1,4-dioxane via the inhalation and dermal routes in humans and via the dermal route in animals is absent.
21 Based on the observance of systemic tumors following oral and inhalation exposure, it is assumed that an
22 internal dose will be achieved regardless of the route of exposure. Therefore, 1,4-dioxane is "likely to be
23 carcinogenic to humans" by all routes of exposure.
4.7.2 Synthesis of Human, Animal, and Other Supporting Evidence
24 Human studies of occupational exposure to 1,4-dioxane were inconclusive; in each case, the
25 cohort size was limited and number of reported cases was small (Buffler et al.. 1978; Thiess et al.. 1976).
26 Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
27 pigs (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kociba et al..
28 1974; Torkelson et al.. 1974; Argus etal.. 1973; Hoch-Ligeti and Argus. 1970; Hoch-Ligeti et al.. 1970;
29 Argus etal.. 1965). Liver tumors have been observed following drinking water exposure in male Wistar
30 rats (Argus et al.. 1965). male guinea pigs (Hoch-Ligeti and Argus. 1970). male Sprague Dawley rats
31 (Argus etal.. 1973: Hoch-Ligeti et al.. 1970). male and female Sherman rats (Kociba et al.. 1974). female
32 Osborne-Mendel rats (NCI. 1978). male and female F344/DuCrj rats (Kano et al.. 2009: JBRC. 1998:
33 Yamazaki et al.. 1994). male and female B6C3Fi mice (NCI. 1978). and male and female Crj:BDFl mice
34 (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994); and following inhalation exposure in male F344
35 rats (Kasai et al.. 2009). In the earliest cancer bioassays, the liver tumors were described as hepatomas
36 (Argus etal.. 1973: Hoch-Ligeti and Argus. 1970: Hoch-Ligeti et al.. 1970: Argus etal.. 1965): however,
84
DRAFT - DO NOT CITE OR QUOTE
-------
1 later studies made a distinction between hepatocellular carcinoma and hepatocellular adenoma (Kano et
2 al. 2009: Kasai et al. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kocibaetal. 1974). Both
3 tumor types have been seen in rats and mice exposed to 1,4-dioxane via drinking water and inhalation.
4 Kociba et al. (1974) noted evidence of liver toxicity at or below the dose levels that produced
5 liver tumors but did not report incidence data for these effects. Hepatocellular degeneration and necrosis
6 were observed in the mid- and high-dose groups of male and female Sherman rats exposed to 1,4-dioxane,
7 while tumors were only observed at the highest dose. Hepatic regeneration was indicated in the mid- and
8 high-dose groups by the formation of hepatocellular hyperplastic nodules. Kano et al., (2009) also
9 provided evidence of liver hyperplasia in male F344/DuCrj rats at a dose level below the dose that
10 induced a statistically significant increase in tumor formation. Kasai et al (2009) noted evidence of liver
11 toxicity and tumor incidences (i.e. hepatocellular adenoma) in male F344/DuCrj rats following inhalation
12 exposures to 1.250 ppm. Increased liver toxicities included hepatocellular necrosis, spongiosis hepatis.
13 and acidophilic and basophilic cell foci.
14 Nasal cavity tumors were also observed in Sprague Dawley rats (Argus et al.. 1973: Hoch-Ligeti
15 etal.. 1970). Osborne-Mendel rats (NCI. 1978). Sherman rats (Kocibaetal.. 1974). and F344/DuCrj rats
16 (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). Most tumors were
17 characterized as squamous cell carcinomas. Nasal tumors were not elevated in B6C3F] or Crj:BDFl mice.
18 Kano et aL (2009) and Kasai et al (2009) were the only studies that evaluated nonneoplastic changes in
19 nasal cavity tissue following prolonged exposure to 1,4-dioxane via oral and inhalation routes.
20 respectively.
21 Histopathological lesions in female F344/DuCrj rats following oral exposure to 1,4-dioxane were
22 suggestive of toxicity and regeneration in nasal tissue (i.e., atrophy, adhesion, inflammation, nuclear
23 enlargement, and hyperplasia and metaplasia of respiratory and olfactory epithelium). Some of these
24 effects occurred at a lower dose (83 mg/kg-day) than that shown to produce nasal cavity tumors
25 (429 mg/kg-day) in female rats. Re-examination of tissue sections from the NCI (1978) bioassay
26 suggested that the majority of nasal tumors were located in the dorsal nasal septum or the nasoturbinate of
27 the anterior portion of the dorsal meatus.
28 Histopathological lesions in male F344/DuCrj rats following exposure to 1.4-dioxane via
29 inhalation were also suggestive of toxicity and regeneration in nasal tissue (i.e. atrophy, inflammation.
30 nuclear enlargement, hyperplasia and metaplasia of the respiratory and olfactory epithelium, and
31 inflammation). Some of these effects occurred at lower concentrations (50 ppm and 250 ppm) than those
32 shown to produce nasal cavity tumors (1.250 ppm) in male rats. Nasal squamous cell carcinomas were
33 observed in the dorsal area of levels 1-3 of the nasal cavity and were characterized as we 11-differentiated
34 and keratinized. In two cases, invasive growth into adjacent tissue was noted, marked by carcinoma
35 growth out of the nose and through a destroyed nasal bone.
36 In addition to the liver and nasal tumors observed in several studies, a statistically significant
37 increase in mesotheliomas of the peritoneum was seen in male rats from the Kano et al. (2009) study
38 (JBRC. 1998: Yamazaki et al.. 1994) and the Kasai et al (2009) study. Female rats dosed with
39 429 mg/kg-day in drinking water for 2 years also showed a statistically significant increase in mammary
85
DRAFT - DO NOT CITE OR QUOTE
-------
1 gland adenomas (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). In male rats, exposed via
2 inhalation, a statistically significant positive trend of mammary gland adenomas was observed by Kasai et
3 al. (2009). A statistically significant increase and/or trend of subcutis fibroma. Zvmbal gland adenoma.
4 and renal cell carcinoma incidences was also observed in male rats exposed for 2 years via inhalation
5 (Kasai etal.. 2009). A significant increase in the incidence of these tumors was not observed in other
6 chronic oral or inhalation bioassays of 1,4-dioxane (NCI. 1978; Kocibaet al.. 1974; Torkelson et al..
7 1974).
4.7.3 Mode of Action Information
8 The MOA by which 1,4-dioxane produces liver, nasal, kidney, peritoneal (mesotheliomas),
9 mammary glandi Zymbal gland, and subcutis tumors is unknown, and the available data do not support
10 any hypothesized mode of carcinogenic action for 1,4-dioxane. Available data also do not clearly identify
11 whether 1,4-dioxane or one of its metabolites is responsible for the observed effects. Furthermore, tumor
12 initiation and promotion studies in mouse skin and rat liver suggested that 1.4-dioxane exposure does not
13 initiate the carcinogenic process, but instead may act as a tumor promoter (Lundberg et al.. 1987; Bull et
14 al.. 1986; King et al.. 1973) (see Section 4.2.3).
15 The hypothesized MOAs for 1,4-dioxane carcinogenicity are discussed below within the context
16 of the modified Hill criteria of causality as recommended in the most recent Agency guidelines (U.S.
17 EPA. 2005a). MOA analyses were not conducted for kidney, peritoneal, mammary gland, Zymbal gland.
18 or subcutis tumors due to the absence of any chemical specific information for these tumor types.
86
DRAFT - DO NOT CITE OR QUOTE
-------
4.7.3.1 Identification of Key Events for Carcinogenicity
1 4.7.3.7.7 Liver. A key event in this MOA hypothesis is sustained proliferation of
2 spontaneously transformed liver cells, resulting in the eventual formation of liver tumors. Precursor
3 events in which 1,4-dioxane may promote proliferation of transformed liver cells are uncertain. One study
4 suggests that induced liver cytotoxicity may be a key precursor event to cell proliferation leading to the
5 formation of liver tumors (Kocibaet al.. 1974). however, this study did not report incidence data for these
6 effects. Other studies suggest that cell proliferation can occur in the absence of liver cytotoxicity. Liver
7 tumors were observed in female rats and female mice in the absence of lesions indicative of cytotoxicity
8 (Kano et al.. 2008; JBRC. 1998; NCI. 1978). Figure 4-1 presents a schematic representation of possible
9 key events in the MOA for 1,4-dioxane liver carcinogenicity. These include: (1) oxidation by CYP2E1
10 and CYP2B1/2 (i.e., detoxification pathway for 1,4-dioxane), (2) saturation of metabolism/clearance
11 leading to accumulation of the parent 1,4-dioxane, (3) liver damage followed by regenerative cell
12 proliferation, or (4) cell proliferation in the absence of cytotoxicity (i.e., mitogenesis), (5) hyperplasia,
13 and (6) tumor formation. It is suggested that liver toxicity is related to the accumulation of the parent
14 compound following metabolic saturation at high doses (Kociba et al.. 1975): however, since no in vivo
15 or in vitro assays have examined the toxic moiety resulting from 1.4-dioxane exposure, liver toxicity due
16 to metabolites cannot be ruled out. Therefore, this hypothesis is not supported. Nannelli et al. (2005)
17 demonstrated that an increase in the oxidative metabolism of 1,4-dioxane via CYP450 induction using
18 phenobarbital or fasting does not result in an increase in liver toxicity. This result suggested that highly
19 reactive and toxic intermediates did not play a large role in the liver toxicity of 1,4-dioxane, even under
20 conditions where metabolism was enhanced. Alternative metabolic pathways (e.g., not CYP450
21 mediated) may be present at high doses of 1,4-dioxane; although the available studies have not
22 characterized these pathways nor identified any possible reactive intermediates. Tumor promotion studies
23 in mouse skin and rat liver suggest that 1,4-dioxane may enhance the growth of previously initiated cells
24 (Lundberg et al.. 1987; King et al.. 1973). This is consistent with the increase in hepatocyte cell
25 proliferation observed in several studies (Mivagawa et al.. 1999; Uno etal. 1994; Goldsworthy et al..
26 1991; Stott etal.. 1981). These mechanistic studies provide evidence of cell proliferation but do not
27 indicate whether mitogenesis or cytotoxicity is responsible for increased cell turnover.
87
DRAFT - DO NOT CITE OR QUOTE
-------
Toxicokinetics*
Oral-absorption- of
1.4-dioxane'
1
Metabolism- by
CYP2El-and
CYP2B1/21
1
HEAA-elimination-
in-the-urine1
Hypothesized- MOA-for-Liver- Tumors'"
Metabolic1
saturation- and
accumulation- of
l=4-dioxane-in-the
blood
*
H epato cellular-
cvtotoxicitv1
I
Regenerative- cell-
proliferation'
1
Hvperplasia1
1
Tumor- formation1
t
Cell-proliferation- in-
absence- of
cvtotoxicitV
1
Hvperplasia'
1
Tumor- promotion*
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
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.1.2 Nasal cavity. A possible key event in the MOA hypothesis for nasal tumors is
2 sustained proliferation of spontaneously transformed nasal epithelial cells, resulting in the eventual
3 formation of nasal cavity tumors (Kasai et al.. 2009). Cell proliferation was observed following
4 1.4-dioxane exposure in both a 2-year inhalation study in male rats (1.250 ppm) (Kasai et al.. 2009) and a
5 2-year drinking water study in male (274 mg/kg-day) and female rats (429 mg/kg-day) (Kano. et al.
6 2009). However, neither study reported evidence of cytotoxicity in the nasal cavity (Kasai et al.. 2009)
7 therefore, cytotoxicity as a key event is not supported. Kasai et al. (2009; 2008) suggest that nasal
8 toxicity is related to the accumulation of the parent compound following metabolic induction at high
9 doses up to 3.200 ppm; however, since no in vivo or in vitro assays have examined the toxic moiety
10 resulting from 1.4-dioxane exposure, nasal toxicity due to metabolites cannot be ruled out. Nannelli et al.
11 (2005) demonstrated that CYP2E1 was inducible in nasal mucosa following acute oral administration of
12 1.4-dioxane by gavage and drinking water, which could potentially lead to an increase in the oxidative
13 metabolism of 1.4-dioxane and nasal toxicity. However. Nannelli et al. (2005) did not characterize this
14 pathway nor identify any possible reactive intermediates or nasal toxicities.
4.7.3.2 Strength, Consistency, Specificity of Association
1 4.7.3.2.1 Liver. The plausibility of a MOA that would include liver cytotoxicity, with
2 subsequent reparative cell proliferation, as precursor events to liver tumor formation is minimally
3 supported by findings that nonneoplastic liver lesions occurred at exposure levels lower than those
4 resulting in significantly increased incidences of hepatocellular tumors (Kociba et al.. 1974) and the
5 demonstration of nonneoplastic liver lesions in subchronic (Kano et al.. 2008) and acute and short-term
6 oral studies (see Table 4-18). Because the incidence of nonneoplastic lesions was not reported by Kociba
7 et al. (1974). it is difficult to know whether the incidence of liver lesions increased with increasing
8 1,4-dioxane concentration. Contradicting the observations by Kociba et al. (1974). liver tumors were
9 observed in female rats and female mice in the absence of lesions indicative of cytotoxicity (Kano et al..
10 2008; JBRC. 1998; NCI. 1978). This suggests that cytotoxicity may not be a requisite step in the MOA
11 for liver cancer. Mechanistic and tumor promotion studies suggest that enhanced cell proliferation without
12 cytotoxicity may be a key event; however, data showing a plausible dose response and temporal
13 progression from cell proliferation to eventual liver tumor formation are not available (see Sections
14 4.7.3.3 and 4.7.3.4). Mechanistic studies that demonstrated cell proliferation after short-term exposure did
15 not evaluate liver cytotoxicity (Miyagawa et al.. 1999; Uno et al.. 1994; Goldsworthy et al.. 1991).
16 Studies have not investigated possible precursor events that may lead to cell proliferation in the absence
17 of cytotoxicity (i.e., genetic regulation of mitogenesis).
89
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.2.2 Nasal cavity. Nasal cavity tumors have been demonstrated in several rat strains
2 (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al. 1994: NCI. 1978: Kociba et al..
3 1974). but were not elevated in two strains of mice (Kano et al.. 2009: JBRC. 1998: Yamazaki et al..
4 1994: NCI. 1978). Irritation of the nasal cavity of rats was indicated in studies by the observation of
5 inflammation (Kasai et al (2009: 2008) and in one study, also rhinitis (JBRC. 1998). The Kasai et al
6 (2009: 2008) studies also showed atrophy of the nasal epithelium in rats, and the JRBC (1998) study also
7 observed atrophy of the nasal epithelium as well as adhesion in rats. Regeneration of the nasal epithelium
8 is demonstrated by metaplasia and hyperplasia observed in rats exposed to 1,4-dioxane (Kano et al.. 2009:
9 Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). Oxidation of 1,4-dioxane metabolism by
10 CYP450s is not supported as a key event in the MOA hypothesis of nasal tumors. Although Nannelli et
11 al. (2005) demonstrated that CYP2E1 was inducible in nasal mucosa following acute oral administration
12 of 1.4-dioxane by gavage and drinking water, the study lacked details regarding the toxic moiety (e.g.
13 parent compound or reactive intermediate) and resulting nasal toxicity. Accumulation of 1.4-dioxane in
14 blood, as a precursor event of nasal tumor formation is also not supported because the parent compound
15 1.4-dioxane was only measured in one subchronic study (Kasai et al.. 2008) and in this study no evidence
16 of nasal cvtotoxicitv. cell proliferation, or incidence of nasal tumors were reported.
4.7.3.3 Dose-Response Relationship
1 4.7.3.3.1 Liver. Table 4-27 presents the temporal sequence and dose-response
2 relationship for possible key events in the liver carcinogenesis of 1,4-dioxane. Dose-response information
3 provides some support for enhanced cell proliferation as a key event in the liver tumorigenesis of
4 1,4-dioxane; however, the role of cytotoxicity as a required precursor event is not supported by data from
5 more than one study. Kociba et al. (1974) demonstrated that liver toxicity and hepatocellular regeneration
6 occurred at a lower dose level than tumor formation. Hepatocellular degeneration and necrosis were
7 observed in the mid- and high-dose groups of Sherman rats exposed to 1,4-dioxane, although it is not
8 possible to discern whether this effect was observed in both genders due to the lack of incidence data
9 (Kociba et al.. 1974). Hepatic tumors were only observed at the highest dose (Kociba et al.. 1974).
10 Hepatic regeneration was indicated in the mid- and high-dose group by the formation of hepatocellular
11 hyperplastic nodules. Liver hyperplasia was also seen in rats from the JBRC (1998) study, at or below the
12 dose level that resulted in tumor formation (Kano et al.. 2009): however, hepatocellular degeneration and
13 necrosis were not observed. These results suggest that hepatic cell proliferation and hyperplasia may
14 occur in the absence of significant cytotoxicity. Liver angiectasis (i.e., dilation of blood or lymphatic
15 vessels) was observed in male mice at the same dose that produced liver tumors; however, the
16 relationship between this vascular abnormality and tumor formation is unclear.
90
DRAFT - DO NOT CITE OR QUOTE
-------
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 ......
Exposure (ppm) Metabohsm
1,4-dioxane
Cell
Liver damage pro|iferation
Hyperplasia
Adenomas
and/or
carcinomas
Kociba et al., (1974) — Sherman rats (male and female combined)
0 mg/kg-day — a
14 mg/kg-day +b
121 mg/kg-day +b
1,307 mg/kg-day +b
a
a
+c
+c
a
a
a
a
a
a
+c
+c
a
a
a
+c
NCI, (1978)— female Osborne-Mendel rats
0 mg/kg-day — a
350 mg/kg-day +b
640 mg/kg-day +b
a
a
a
a
a
a
a
a
a
a
+c
+c
NCI, (1978)— male B6C3Fi mice
0 mg/kg-day — a
720 mg/kg-day +b
830 mg/kg-day +b
a
a
a
a
a
a
a
a
a
a
+c
+c
NCI. (1978)— female B6C3Fi mice
0 mg/kg-day — a
380 mg/kg-day +b
860 mg/kg-day +b
Kano et al., (2009); JBRC, (1998)— male
0 mg/kg-day — a
1 1 mg/kg-day +b
55 mg/kg-day +b
274 mg/kg-day +b
a
a
a
F344/DuCrj rats
a
a
a
+c
-------
250 ppm
1,250 ppm
a— No evidence demonstrating key event.
b+ 1,4-dioxane metabolism was not evaluated as part of the chronic bioassays. Data from pharmacokinetic studies suggest that
metabolism of 1,4-dioxane by CYP2E1 and CYP2B2 occurs immediately and continues throughout the duration of exposure at
all exposure levels.
CD E \idence demonstrating key event.
dD Single cellnecrosis was observed in a 13 weekbioassay formale 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; however, information from JBRC
(1998) on incidence of liver hyperplasia was used to create this table.
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.
9AII rats died during the first week of the 13-week bioassay (Kasai etal.. 2008).
hKasai et al. (2009) reported incidence rates for centrilobular necrosis and hepatocellular adenomas in male rats (1,250 ppm).
4.7.3.3.2 Nasal cavity.
1 Table 4-28 presents the temporal sequence and dose-response relationship for possible key events
2 in the nasal tissue carcinogenesis of 1.4-dioxane. Toxicity and regeneration in nasal epithelium (i.e.,
3 atrophy, adhesion, inflammation, and hyperplasia and metaplasia of respiratory and olfactory epithelium)
4 was evident in one study at the same dose levels that produced nasal cavity tumors (Kano et al.. 2009;
5 JBRC. 1998). In another study, dose-response information provided some support for nasal toxicity and
6 regeneration in nasal epithelium occurring before tumor development (Kasai et al.. 2009). However, the
7 role of cvtotoxicity as a required precursor event is not supported by data from any of the reviewed
8 studies. The accumulation of parent 1.4-dioxane as a key event has some support since
9 concentration-dependent increases were noted for 1.4-dioxane in plasma concurrent with toxicities
10 observed that are possible precursor events (i.e.. regeneration in nasal epithelium) (Kasai et al.. 2008). In
11 a subsequent study by Kasai et al. (2009) some of these same possible precursor events were observed at
12 50. 250. and 1.250 ppm with evidence of nasal tumors at the highest concentration (1.250 ppm).
92
DRAFT - DO NOT CITE OR QUOTE
-------
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) * J ' —
or Exposure Metabolism Nasal Cell Adenomas
(ppm) 1,4-dioxane cytotoxicity proliferation Hyperplasia and/or
CdrCI nOiTldS
Kociba et al.. (1974)—Sherman rats (male and female combined)
0 mg/kg-day —a_ ^ ^ ^ ^
14 mg/kg-day +h —a —a —a —a
121 mg/kg-day +b —a —a —a —a
1,307 mg/kg-day +b —a —a —a —a
NCI. (1978)—female Osborne-Mendel rats
0 mg/kg-day —a ^ ^ ^ ^
350 mg/kg-day +b —a —a —a —a
640 mg/kg-day +b —a —a —a —a
NCI. (1978)—male B6C3Fi mice
0 mg/kg-day —a_ ^ ^ ^ ^
720 mg/kg-day +" —a —a —a —a
830 mg/kg-day +b —a —a —a —a
NCI. (1978)—female B6C3Fi mice
0 mg/kg-day —a ^ ^ ^ —a
380 mg/kg-day £ —* —* —* —*
860 mg/kg-day +b —a —a —a —a
Kano et al., (2009); JBRC. (1998)—male F344/DuCrj rats
0 mg/kg-day —a -? -? -? —a
11 mg/kg-day +b —a —a —a —a
55 mg/kg-day +b —a —a —a —a
274 mg/kg-day +b —a —a +c'd +c'd
Kano et al., (2009): JBRC. (1998)—female F344/DuCrj rats
0 mg/kg-day —a -? -? -? —a
18 mg/kg-day +b —a —a —a —a
83 mg/kg-day +b —a —a —a —a
429 mg/kg-day +b —a —a +c'd +c'd
Kano et al.. (2009): JBRC. (1998)—male Crj:BDF1 mice
0 mg/kg-day —a ^ ^ ^ —a
49 mg/kg-day +b —a —a —a —a
191 mg/kg-day +h —a —a —a —a
677 mg/kg-day +b —a —a —a —a
Kano et al., (2009); JBRC. (1998)—female Crj:BDF1 mice
0 mg/kg-day —a ^ ^ ^ —a
66 mg/kg-day +" —a —a —a —a
278 mg/kg-day +b —a —a —a —a
964 mg/kg-day +b —a —a —a —a
Kasai et al. (2008)—F344 rats (male and female combined)
Oppm —a_ -^ -^ -^ —a
100 ppm +b —a —a —a —a
200 ppm +b —a —a —a —a
400 ppm +c —a —a —a —a
800 ppm +^ —^ —^ —^ —^
1,600 ppm +c —a —a —a —a
3,200 ppm +c —a —a —a —a
6,400 ppm +aAt -aj -aj -aj -aj
Kasai et al. (2009)—male F344 rats
Oppm —a_ ^ ^ ^ —a
50 ppm +b —a —a —a —a
250 ppm ? ^ ^ ^ ^
93
DRAFT - DO NOT CITE OR QUOTE
-------
1,250 ppm
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.
CD 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.
f+ All rats died during the first week of the 13 week bioassay (Kasai et al.. 2008).
4.7.3.4 Temporal Relationship
1 4.7.3.4.1 L iver. Available information regarding temporal relationships between the key
2 event (sustained proliferation of spontaneously transformed liver cells) and the eventual formation of liver
3 tumors is limited. A comparison of 13-week and 2-year studies conducted in F344/DuCrj rats and
4 Crj :BDF1 mice at the same laboratory revealed that tumorigenic doses of 1,4-dioxane produced liver
5 toxicity by 13 weeks of exposure (Kano etal.. 2009; Kano et al.. 2008; JBRC. 1998). Hepatocyte swelling
6 of the centrilobular area of the liver, vacuolar changes in the liver, granular changes in the liver, and
7 single cell necrosis in the liver were observed in mice and rats given 1,4-dioxane in the drinking water for
8 13 weeks. Sustained liver damage may lead to regenerative cell proliferation and tumor formation
9 following chronic exposure. As discussed above, histopathological evidence of regenerative cell
10 proliferation has been seen following long-term exposure to 1,4-dioxane (JBRC. 1998; Kociba et al..
11 1974). Tumors occurred earlier at high doses in both mice and rats from this study (Yamazaki. 2006);
12 however, temporal information regarding hyperplasia or other possible key events was not available (i.e.,
13 interim blood samples not collected, interim sacrifices were not performed). Argus et al. (1973) studied
14 the progression of tumorigenesis by electron microscopy of liver tissues obtained following interim
15 sacrifices at 8 and 13 months of exposure (five rats/group, 574 mg/kg-day). The first change observed
16 was an increase in the size of the nuclei of the hepatocytes, mostly in the periportal area. Precancerous
17 changes were characterized by disorganization of the rough endoplasmic reticulum, increase in smooth
18 endoplasmic reticulum, and decrease in glycogen and increase in lipid droplets in hepatocytes. These
19 changes increased in severity in the hepatocellular carcinomas in rats exposed to 1,4-dioxane for
20 13 months.
1 Three types of liver nodules were observed in exposed rats at 13-16 months. The first consisted
2 of groups of these cells with reduced cytoplasmic basophilia and a slightly nodular appearance as viewed
3 by light microscopy. The second type of nodule was described consisting of large cells, apparently filled
4 and distended with fat. The third type of nodule was described as finger-like strands, 2-3 cells thick, of
5 smaller hepatocytes with large hyperchromic nuclei and dense cytoplasm. This third type of nodule was
6 designated as an incipient hepatoma, since it showed all the histological characteristics of a fully
7 developed hepatoma. All three types of nodules were generally present in the same liver.
94
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.4.2 Nasal cavity. No information was available regarding the temporal relationship
2 between toxicity in the nasal epithelium and the formation of nasal cavity tumors. Sustained nasal damage
3 may lead to regenerative cell proliferation and tumor formation following chronic exposure. As discussed
4 above (Section 4.2.2.2.1). no evidence of cytotoxicity has been observed following exposure to
5 1.4-dioxane. despite histopathological evidence of regenerative cell proliferation and nasal tumors at the
6 highest exposure concentration (Kano et al.. 2009)(Kasai et al.. 2009) (See Table 4-28)^ Other incidences
7 of nasal damage may have occurred before tumor formation; however, temporal information regarding
8 these events was not available (i.e.. interim sacrifices were not performed).
4.7.3.5 Biological Plausibility and Coherence
1 4.7.3.5.1 Liver. The hypothesis that sustained proliferation of spontaneously transformed
2 liver cells is a key event within a MOA is possible based on supporting evidence indicating that
3 1,4-dioxane is a tumor promoter of mouse skin and rat liver tumors (Lundberg et al.. 1987; Bull et al..
4 1986; King etal.. 1973). Further support for this hypothesis is provided by studies demonstrating that
5 1,4-dioxane increased hepatocyte DNA synthesis, indicative of cell proliferation (Mivagawa et al.. 1999;
6 Uno et al.. 1994; Goldsworthy et al.. 1991; Stott etal.. 1981). In addition, the generally negative results
7 for 1,4-dioxane in a number of genotoxicity assays indicates the carcinogenicity of 1,4-dioxane may not
8 be mediated by a mutagenic MOA. The importance of cytotoxicity as a necessary precursor to sustained
9 cell proliferation is biologically plausible, but is not supported by the dose-response in the majority of
10 studies of 1,4-dioxane carcinogenicity.
1 4.7.3.5.2 Nasal cavity. Sustained cell proliferation in response to cell death from toxicity
2 may be related to the formation of nasal cavity tumors; however, this MOA is also not established. Nasal
3 carcinogens are generally characterized as potent genotoxins (Ashby. 1994); however, other MOAs have
4 been proposed for nasal carcinogens that induce effects through other mechanisms (Rasper et al.. 2007;
5 Green et al.. 2000).
1 The National Toxicological Program (NTP) database identified 12 chemicals from approximately
2 500 bioassays as nasal carcinogens and 1,4-dioxane was the only identified nasal carcinogen that showed
3 little evidence of genotoxicity (Haseman and Hailey. 1997). Nasal tumors were not observed in an
4 inhalation study in Wistar rats exposed to 111 ppm for 5 days/week for 2 years (Torkelson et al.. 1974).
5 but were observed in an inhalation study in F344 rats exposed to 1.250 ppm for 5 days/week for 2 years.
6 Two human studies of occupational exposure, ranging from 0.06 ppm to 75 ppm for Imonth up to 41
7 years, reported inconclusive Findings regarding increased tumor risk (Buffler et al.. 1978; Thiess et al..
8 1976). It is important to note, that nasal tumors were not evaluated in the human studies and genotoxicity
9 was not assessed in either the human or animal studies.
10 While there is no known MOA for 1.4-dioxane and the human studies are inconclusive regarding
11 tumor risk, the noted nasal tumors in rats are considered biologically plausible and relevant to humans.
12 since similar cell types considered to be at risk are prevalent throughout the respiratory tract of rats and
13 humans. Differences in the anatomy of the upper respiratory tract and resulting differences in absorption
95
DRAFT - DO NOT CITE OR QUOTE
-------
1 or in local respiratory system effects between humans and rats are acknowledged and considered sources
2 of uncertainty.
4.7.3.6 Other Possible Modes of Action
3 An alternate MOA could be hypothesized that 1,4-dioxane alters DNA, either directly or
4 indirectly (Kasai et al. 2009). which causes mutations in critical genes for tumor initiation, such as
5 oncogenes or tumor suppressor genes. Following these events, tumor growth may be promoted by a
6 number of molecular processes leading to enhanced cell proliferation or inhibition of programmed cell
7 death. The results from in vitro and in vivo assays do not provide overwhelming support for the
8 hypothesis of a genotoxic MOA for 1,4-dioxane carcinogenicity. The genotoxicity data for 1,4-dioxane
9 were reviewed in Section 4.5.1 and were summarized in Table 4-23. Negative findings were reported for
10 mutagenicity in Salmonella typhimurium, Escherichia coll, and Photobacterium phosphoreum (Mutatox
11 assay) (Morita and Havashi. 1998; Hellmer and Bolcsfoldi. 1992; Kwanetal.. 1990; Khudolev et al..
12 1987; Nestmann et al.. 1984; Haworth et al.. 1983; Stottetal.. 1981). Negative results were also indicated
13 for the induction of aneuploidy in yeast (Saccharomyces cerevisiae) and the sex-linked recessive lethal
14 test in Drosophila melanogaster (Zimmermann et al.. 1985). In contrast, positive results were reported in
15 assays for sister chromatid exchange (Galloway et al.. 1987). DNA damage (Kitchin and Brown. 1990).
16 and in in vivo micronucleus formation in bone marrow (Roy et al.. 2005; Mirkova. 1994). and liver (Roy
17 etal.. 2005; Morita and Havashi. 1998). Lastly, in the presence of toxicity, positive results were reported
18 for meiotic nondisjunction in drosophila (Munoz and Barnett. 2002). DNA damage (Sinaet al.. 1983).
19 and cell transformation (Sheu etal.. 1988).
20 Additionally, 1,4-dioxane metabolism did not produce reactive intermediates that covalently
21 bound to DNA (Stottetal.. 1981; Woo etal.. 1977b) and DNA repair assays were generally negative
22 (Goldsworthy et al.. 1991; Stottetal.. 1981). No studies were available to assess the ability of
23 1,4-dioxane or its metabolites to induce oxidative damage to DNA.
96
DRAFT - DO NOT CITE OR QUOTE
-------
4.7.3.7 Conclusions About the Hypothesized Mode of Action
1 4.7.3.7.1 Liver. The MOA by which 1,4-dioxane produces liver tumors is unknown, and
2 available evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
3 inconclusive. A MOA hypothesis involving 1,4-dioxane induced cell proliferation is possible but data are
4 not available to support this hypothesis. Pharmacokinetic data suggest that clearance pathways were
5 saturable and target organ toxicity occurs after metabolic saturation. Liver toxicity preceded tumor
6 formation in one study (Kociba et al.. 1974) and a regenerative response to tissue injury was demonstrated
7 by histopathology. Liver hyperplasia and tumor formation have also been observed in the absence of
8 cytotoxicity (Kano et al.. 2009; JBRC. 1998). Cell proliferation and tumor promotion have been shown to
9 occur after prolonged exposure to 1,4-dioxane (Mivagawa et al.. 1999; Uno et al.. 1994; Goldsworthy et
10 al.. 1991;Lundbergetal.. 1987; Bull et al.. 1986; Stottet al.. 1981; King et al.. 1973).
1 4.7.3.7.2 Nasal cavity. The MOA forthe formation of nasal cavity tumors is unknown,
2 and evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is inconclusive.
3 Nasal carcinogens are generally characterized as potent genotoxins (Ashby. 1994); however, other MOAs
4 have been proposed for nasal carcinogens that induce effects through other mechanisms (Kasper et al..
5 2007; Green et al.. 2000). Neither nasal tumors in the human studies nor genotoxicity in human or animal
6 studies following exposure to 1.4-dioxane was evaluated, so the role of genotoxicity cannot be ruled out.
7 A MOA hypothesis involving nasal damage, cell proliferation, and hvperplasia is possible, but data are
8 not available to support this hypothesis. In studies that examined nasal effects after exposure to
9 1.4-dioxane. at least one of these events is missing. More specifically, nasal cavity tumors have been
10 reported by Kasai et al. (2009) in the absence of cytotoxicity and in Kano et al. (2009) in the absence of
11 hvperplasia. Therefore, as per EPA's Cancer Guidelines (U.S. EPA. 2005a). there is insufficient
12 biological support for potential key events and to have reasonable confidence in the sequence of events
13 and how they relate to the development of nasal tumors following exposure to 1.4-dioxane. Using the
14 modified Hill criteria, exposure-response and temporal relationships have not been established in support
15 of any hypothetical mode of carcinogenic action for 1.4-dioxane. Thus, the MOA cannot be established.
4.7.3.8 Relevance of the Mode of Action to Humans
1 Several hypothesized MOAs for 1,4-dioxane induced tumors in laboratory animals have been
2 discussed along with the supporting evidence for each. As was stated, the MOA by which 1,4-dioxane
3 produces liver, nasal, peritoneal, and mammary gland tumors is unknown. Some mechanistic information
4 is available to inform the MOA of the liver and nasal tumors but no information exists to inform the
5 MOA of the observed peritoneal or mammary gland tumors (Kano et al.. 2009; JBRC. 1998; Yamazaki et
6 al.. 1994).
97
DRAFT - DO NOT CITE OR QUOTE
-------
4.8 Susceptible Populations and Life Stages
1 There is no direct evidence to establish that certain populations and lifestages may be susceptible
2 to 1,4-dioxane. Changes in susceptibility with lifestage as a function of the presence of microsomal
3 enzymes that metabolize and detoxify this compound (i.e., CYP2E1 present in liver, kidney, and nasal
4 mucosa can be hypothesized). Vieira et al. (1996) reported that large increases in hepatic CYP2E1 protein
5 occur postnatally between 1 and 3 months in humans. Adult hepatic concentrations of CYP2E1 are
6 achieved sometime between 1 and 10 years. To the extent that hepatic CYP2E1 levels are lower, children
7 may be more susceptible to liver toxicity from 1,4-dioxane than adults. CYP2E1 has been shown to be
8 inducible in the rat fetus. The level of CYP2E1 protein was increased by 1.4-fold in the maternal liver and
9 2.4-fold in the fetal liver following ethanol treatment, as compared to the untreated or pair-fed groups
10 (Carpenter et al.. 1996). Pre- and postnatal induction of microsomal enzymes resulting from exposure to
11 1,4-dioxane or other drugs or chemicals may reduce overall toxicity following sustained exposure to
12 1,4-dioxane.
13 Genetic polymorphisms have been identified for the human CYP2E1 gene CWatanabe et al..
14 1994; Havashi et al.. 1991) and were considered to be possible factors in the abnormal liver function seen
15 in workers exposed to vinyl chloride (Huang et al.. 1997). Individuals with a CYP2E1 genetic
16 polymorphism resulting in increased expression of this enzyme may be less susceptible to toxicity
17 following exposure to 1,4-dioxane.
18 Gender differences were noted in subchronic and chronic toxicity studies of 1,4-dioxane in mice
19 and rats (see Sections 4.6 and 4.7). No consistent pattern of gender sensitivity was identified across
20 studies. In a 13 week inhalation study of male and female rats (Kasai et al., 2008) kidney toxicity. as
21 evidenced by hydropic change in the renal proximal tubules, was observed in female rats exposed to
22 3.200 ppm of 1.4-dioxane. but not male rats. This suggests a possible increased susceptibility of female
23 rats to renal damage following inhalation exposure to 1.4-dioxane.
98
DRAFT - DO NOT CITE OR QUOTE
-------
5 DOSE-RESPONSE ASSESSMENTS
5.1 Oral Reference Dose (RfD)
5.1.1 Choice of Principal Studies and Critical Effect with Rationale and
J ustification
1 Liver and kidney toxicity were the primary noncancer health effects associated with exposure to
2 1,4-dioxane in humans and laboratory animals. Occupational exposure to 1,4-dioxane has resulted in
3 hemorrhagic nephritis and centrilobular necrosis of the liver (Johnstone. 1959; Barber. 1934). In animals,
4 liver and kidney degeneration and necrosis were observed frequently in acute oral and inhalation studies
5 (JBRC. 1998: Drewetal.. 1978: David. 1964: Kestenetal.. 1939: Laugetal.. 1939: Schrenk and Yant.
6 1936: de Navasquez. 1935: Fairley et al., 1934). Liver and kidney effects were also observed following
7 chronic oral exposure to 1,4-dioxane in animals (Kano et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994:
8 NCI. 1978: Kociba et al.. 1974: Argus etal.. 1973: Argus etal.. 1965) (see Table 4-25).
9 Liver toxicity in the available chronic studies was characterized by necrosis, spongiosis hepatic,
10 hyperplasia, cyst formation, clear foci, and mixed cell foci. Kociba et al. (1974) demonstrated
11 hepatocellular degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats) or greater. The
12 NOAEL for liver toxicity was 9.6 mg/kg-day and 19 mg/kg-day in male and female rats, respectively. No
13 quantitative incidence data were provided in this study. Argus et al. (1973) described early preneoplastic
14 changes in the liver and JBRC (1998) demonstrated liver lesions that are primarily associated with the
15 carcinogenic process. Clear and mixed-cell foci in the liver are commonly considered preneoplastic
16 changes and would not be considered evidence of noncancer toxicity. In the JBRC (1998) study,
17 spongiosis hepatis was associated with other preneoplastic changes in the liver (clear and mixed-cell foci)
18 and no other lesions indicative of liver toxicity were seen. Spongiosis hepatis was therefore not
19 considered indicative of noncancer effects in this study. The activity of serum enzymes (i.e., AST, ALT,
20 LDH, and ALP) was increased in mice and rats chronically exposed to 1,4-dioxane (JBRC. 1998):
21 however, these increases were seen only at tumorigenic dose levels. Blood samples were collected at
22 study termination and elevated serum enzymes may reflect changes associated with tumor formation.
23 Histopathological evidence of liver toxicity was not seen in rats from the JBRC (1998) study. The highest
24 non-tumorigenic dose levels for this study approximated the LOAEL derived from the Kociba et al.
25 (1974) study (94 and 148 mg/kg-day for male and female rats, respectively).
26 Kidney damage in chronic toxicity studies was characterized by degeneration of the cortical
27 tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCL 1978: Kociba etal.. 1974: Argus et
28 al.. 1973: Argus etal.. 1965: Fairley et al.. 1934). Kociba et al. (1974) described renal tubule epithelial
29 cell degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats) or greater, with a NOAEL
30 of 9.6 mg/kg-day. No quantitative incidence data were provided in this study (Kociba et al.. 1974). Doses
31 of > 430 mg/kg-day 1,4-dioxane induced marked kidney alterations (Argus etal., 1973). The observed
99
DRAFT - DO NOT CITE OR QUOTE
-------
1 changes included glomerulonephritis and pyelonephritis, with characteristic epithelial proliferation of
2 Bowman's capsule, periglomerular fibrosis, and distension of tubules. Quantitative incidence data were
3 not provided in this study. In the NCI (1978) study, kidney lesions in rats consisted of vacuolar
4 degeneration and/or focal tubular epithelial regeneration in the proximal cortical tubules and occasional
5 hyaline casts. Kidney toxicity was not seen in rats from the JBRC (1998) study at any dose level (highest
6 dose was 274 mg/kg-day in male rats and 429 mg/kg-day in female rats).
7 Kociba et al. (1974) was chosen as the principal study for derivation of the RfD because the liver
8 and kidney effects in this study are considered adverse and represent the most sensitive effects identified
9 in the database (NOAEL 9.6 mg/kg-day, LOAEL 94 mg/kg-day in male rats). Kociba et al. (1974)
10 reported degenerative effects in the liver, while liver lesions reported in other studies (JBRC. 1998; Argus
11 et al.. 1973) appeared to be related to the carcinogenic process. Kociba et al. (1974) also reported
12 degenerative changes in the kidney. NCI (1978) and Argus et al. (1973) provided supporting data for this
13 endpoint; however, kidney toxicity was observed in these studies at higher doses. JBRC (1998) reported
14 nasal inflammation in rats (NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and mice (NOAEL
15 66 mg/kg-day, LOAEL 278 mg/kg-day).
16 Even though the study reported by Kociba et al. (1974) had one noteworthy weakness, it had
17 several noted strengths, including: (1) two-year study duration; (2) use of both male and female rats and
18 three dose levels, 10-fold apart, plus a control group; (3) a sufficient number of animals per dose group
19 (60 animals/sex/dose group; and (4) the authors conducted a comprehensive evaluation of the animals
20 including body weights and clinical observations, blood samples, organ weights of all the major tissues,
21 and a complete histopathological examination of all rats. The authors did not report individual incidence
22 data that would have allowed for a BMD analysis of this robust dataset.
5.1.2 Methods of Analysis—Including Models (PBPK, BMD, etc.)
23 Several procedures were applied to the human PBPK model to determine if an adequate fit of the
24 model to the empirical model output or experimental observations could be attained using biologically
25 plausible values for the model parameters. The re-calibrated model predictions for blood 1,4-dioxane
26 levels did not come within 10-fold of the experimental values using measured tissue: air partition
27 coefficients of Leung and Paustenbach (1990) or Sweeney et al. (2008) (Figure B-8 and Figure B-9). The
28 utilization of a slowly perfused tissue:air partition coefficient 10-fold lower than measured values
29 produces exposure-phase predictions that are much closer to observations, but does not replicate the
30 elimination kinetics (Figure B-10). Re-calibration of the model with upper bounds on the tissue:air
31 partition coefficients results in predictions that are still six- to sevenfold lower than empirical model
32 prediction or observations (Figure B-12 and Figure B-13). Exploration of the model space using an
33 assumption of zero-order metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate
34 fit to the exposure and elimination data can be achieved only when unrealistically low values are assumed
35 for the slowly perfused tissue:air partition coefficient (Figure B-16). Artificially low values for the other
36 tissue:air partition coefficients are not expected to improve the model fit, as these parameters are shown
100
DRAFT - DO NOT CITE OR QUOTE
-------
1 in the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km. This suggests
2 that the model structure is insufficient to capture the apparent 10-fold species difference in the blood
3 1,4-dioxane between rats and humans. In the absence of actual measurements for the human slowly
4 perfused tissue:air partition coefficient, high uncertainty exists for this model parameter value.
5 Differences in the ability of rat and human blood to bind 1,4-dioxane may contribute to the difference in
6 Vd. However, this is expected to be evident in very different values for rat and human blood:air partition
7 coefficients, which is not the case (Table B-l). Therefore, some other, as yet unknown, modification to
8 model structure may be necessary.
9 Kociba et al. (1974) did not provide quantitative incidence or severity data for liver and kidney
10 degeneration and necrosis. Benchmark dose (BMD) modeling could not be performed for this study and
11 the NOAEL for liver and kidney degeneration (9.6 mg/kg-day in male rats) was used as the point of
12 departure (POD) in deriving the RfD for 1,4-dioxane.
13 Alternative PODs were calculated using incidence data reported for cortical tubule degeneration
14 in male and female rats (NCI. 1978) and liver hyperplasia (JBRC. 1998). The incidence data for cortical
15 tubule cell degeneration in male and female rats exposed to 1,4-dioxane in the drinking water for 2 years
16 are presented in Table 5-1. Details of the BMD analysis of these data are presented in Appendix C. Male
17 rats were more sensitive to the kidney effects of 1,4-dioxane than females and the male rat data provided
18 the lowest POD for cortical tubule degeneration in the NCI (1978) study (BMDL10 of 22.3 mg/kg-day)
19 (Table 5-2). Incidence data (Kano et al.. 2009; JBRC. 1998) for liver hyperplasia in male and female rats
20 exposed to 1,4-dioxane in the drinking water for 2 years are presented in Table 5-3. Details of the BMD
21 analysis of these data are presented in Appendix C. Male rats were more sensitive to developing liver
22 hyperplasia due to exposure to 1,4-dioxane than females and the male rat data provided the lowest POD
23 for hyperplasia in the JBRC (1998) study (BMDL10 of 23.8 mg/kg-day) (Table 5-4). The BMDL10 values
24 of 22.3 mg/kg-day and 23.8 mg/kg-day from the NCI (1978) and JBRC (1998) studies, respectively, are
25 about double the NOAEL (9.6 mg/kg-day) observed by Kociba et al. (1974).
101
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water for 2 years
0
0/3 1a
Males (mg/kg-day)
240
20/3 1b
Females (mg/kg-day)
530
27/33b
0
0/3 1a
350
0/34
640
10/32b
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 < 0.001) performed for this review.
Source: NCI (1978).
Table 5-2 BMD and BMDL values derived from BMD modeling of cortical tubule degeneration in
male and female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water for 2
years
BMDio (mg/kg-day) BMDL10 (mg/kg-day)
Male rats 28.8 22.3
Female rats 596.4 452.4
Source: NCI (1978).
Table 5-3 Incidence of liver hyperplasia in F344/DuCrj rats exposed to 1,4-dioxane in drinking
water for 2 years
Males (mg/kg-day)a
0
3/40
11
2/45
55
9/35b
274
12/22C
0
0/38b
Females (mg/kg-day)a
18
0/37
83
1/38
429
14/24C
Dose information from Kano et al. (2009) and incidence data for sacrificed animals from JBRC (1998).
"Statistically significant compared to controls by the Dunnett's test (p < .05).
""Incidence significantly elevated compared to control by x2 test (p < 0.01).
Sources: Kano et al. (2009): JBRC (1998).
102
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-4 BMD and BMDL values derived from BMD modeling of liver hyperplasia in male and
female F344/DuCrj rats exposed to 1,4-dioxane in drinking water for 2 years
BMDio (mg/kg-day) BMDLio (mg/kg-day)
Male rats 35.9 23.8
Female rats 137.3 88.5
Source: Kano et al. (2009): JBRC (1998).
5.1.3 RfD Derivation - Including Application of Uncertainty Factors (UFs)
1 The RfD of 3 x 1CT2 mg/kg-day is based on liver and kidney toxicity in rats exposed to
2 1,4-dioxane in the drinking water for 2 years (Kociba et al., 1974). The Kociba et al. (1974) study was
3 chosen as the principal study because it provides the most sensitive measure of adverse effects by
4 1,4-dioxane. The incidence of liver and kidney lesions was not reported for each dose group. Therefore,
5 BMD modeling could not be used to derive a POD. The RfD for 1,4-dioxane is derived by dividing the
6 NOAEL of 9.6 mg/kg-day (Kociba etal., 1974) by a composite UF of 300, as follows:
7 RfD = NOAEL/UF
8 =9.6 mg/kg-day / 300
9 = 0.03 or 3 x 10~2 mg/kg-day
10 The composite UF of 300 includes factors of 10 for animal-to-human extrapolation and for
11 interindividual variability, and an UF of 3 for database deficiencies.
12 A default interspecies UF of 10 was used to account for pharmacokinetic and pharmacodynamic
13 differences across species. Existing PBPK models could not be used to derive an oral RfD for 1,4-dioxane
14 (Appendix B).
15 A default interindividual variability UF of 10 was used to account for variation in sensitivity
16 within human populations because there is limited information on the degree to which humans of varying
17 gender, age, health status, or genetic makeup might vary in the disposition of, or response to, 1,4-dioxane.
18 An UF of 3 for database deficiencies was applied due to the lack of a multigeneration
19 reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
20 1,4-dioxane (Giavini et al., 1985). This developmental study indicates that the developing fetus may be a
21 target of toxicity.
22 An UF to extrapolate from a subchronic to a chronic exposure duration was not necessary
23 because the RfD was derived from a study using a chronic exposure protocol.
24 An UF to extrapolate from a LOAEL to a NOAEL was not necessary because the RfD was based
25 on a NOAEL. Kociba et al. (1974) was a well-conducted, chronic drinking water study with an adequate
26 number of animals. Histopathological examination was performed for many organs and tissues, but
103
DRAFT - DO NOT CITE OR QUOTE
-------
1 clinical chemistry analysis was not performed. NOAEL and LOAEL values were derived by the study
2 authors based on liver and kidney toxicity; however quantitative incidence data was not reported. Several
3 additional oral studies (acute/short-term, subchronic, and chronic durations) were available that support
4 liver and kidney toxicity as the critical effect (Kano etal.. 2008: JBRC. 1998: NCI. 1978: Argus et al..
5 1973) (Table 4-15 and Table 4-17). Although degenerative liver and kidney toxicity was not observed in
6 rats from the JBRC (1998) study at doses at or below the LOAEL in the Kociba et al. (1974) study, other
7 endpoints such as metaplasia and hyperplasia of the nasal epithelium, nuclear enlargement, and
8 hematological effects, were noted.
5.1.4 RfD Comparison Information
9 PODs and sample oral RfDs based on selected studies included in Table 4-18 are arrayed in
10 Figure 5-1 to Figure 5-3, and provide perspective on the RfD supported by Kociba et al. (1974). These
11 figures should be interpreted with caution because the PODs across studies are not necessarily
12 comparable, nor is the confidence in the data sets from which the PODs were derived the same. PODs in
13 these figures may be based on a NOAEL, LOAEL, or BMDL (as indicated), and the nature, severity, and
14 incidence of effects occurring at a LOAEL are likely to vary. To some extent, the confidence associated
15 with the resulting sample RfD is reflected in the magnitude of the total UF applied to the POD (i.e., the
16 size of the bar); however, the text of Sections 5.1.1 and 5.1.2 should be consulted for a more complete
17 understanding of the issues associated with each data set and the rationale for the selection of the critical
18 effect and principal study used to derive the RfD.
19 The predominant noncancer effect of chronic oral exposure to 1,4-dioxane is degenerative effects
20 in the liver and kidney. Figure 5-1 provides a graphical display of effects that were observed in the liver
21 following chronic oral exposure to 1,4-dioxane. Information presented includes the PODs and UFs that
22 could be considered in deriving the oral RfD. As discussed in Sections 5.1.1 and 5.1.2, among those
23 studies that demonstrated liver toxicity, the study by Kociba et al. (1974) provided the data set most
24 appropriate for deriving the RfD. For degenerative liver effects resulting from 1,4-dioxane exposure, the
25 Kociba et al. (1974) study represents the most sensitive effect and dataset observed in a chronic bioassay
26 (Figure 5-1).
27 Kidney toxicity as evidenced by glomerulonephritis (Argus etal.. 1973: Argus etal.. 1965) and
28 degeneration of the cortical tubule (NCI. 1978: Kociba etal.. 1974) has also been observed in response to
29 chronic exposure to 1,4-dioxane. As was discussed in Sections 5.1 and 5.2, degenerative effects were
30 observed in the kidney at the same dose level as effects in the liver (Kociba et al.. 1974). A comparison of
31 the available datasets from which an RfD could potentially be derived is presented in Figure 5-2.
32 Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
33 dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice).
34 JBRC (1998) reported nasal inflammation in rats (NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and
35 mice (NOAEL 66 mg/kg-day, LOAEL 278 mg/kg-day). A comparison of the available datasets from
36 which an RfD could potentially be derived is presented in Figure 5-3.
104
DRAFT - DO NOT CITE OR QUOTE
-------
1 Figure 5-4 displays PODs for the major targets of toxicity associated with oral exposure to
2 1,4-dioxane. Studies in experimental animals have also found that relatively high doses of 1,4-dioxane
3 (1,000 mg/kg-day) during gestation can produce delayed ossification of the sternebrae and reduced fetal
4 BWs (Giavini et al.. 1985). This graphical display (Figure 5-4) compares organ specific toxicity for
5 1,4-dioxane, including a single developmental study. The most sensitive measures of degenerative liver
6 are and kidney effects. The sample RfDs for degenerative liver and kidney effects are identical since they
7 were derived from the same study and dataset (Kocibaetal.. 1974) and are presented for completeness.
100
Rat
Rat
Mouse
Rat
Rat
10
o
Q
0.1
0.01
ffi
I
• POD
[HAnimal-to-human
nHuman variation
0LOAELto NOAEL
DSubchronic to Chronic
BDatabase deficiencies
ORfD
Liver hyperplasia; Hepatocellular Increase in serum liver Increase in serum liver Liver hyperplasia;
NOAEL; 2 yr rat degeneration and enzymes; NOAEL; 2 yr enzymes; NOAEL; 2 yr BMDL10; 2 yr rat
drinking water study necrosis; NOAEL; 2 yr mouse drinking water rat drinking water study drinking water study
rat drinking water study study
Figure 5-1 Potential points of departure (POD) for liver toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following
oral exposure to 1,4-dioxane.
105
DRAFT - DO NOT CITE OR QUOTE
-------
Rat
Rat
Rat
1000
100
10
0.1 -
0.01
• POD
[HAnimal-to-human
QHuman variation
E2LOAELIONOAEL
dSubchronic to Chronic
• Database deficiencies
ORfD
Glomerulonephritis; LOAEL; 13 month Degeneration and necrosis of tubular Cortical tubule degeneration; BMDL10;
rat drinking water study epithelium; NOAEL; 2 yr rat drinking 2 yr rat drinking water study
water study
Figure 5-2 Potential points of departure (POD) for kidney toxicity endpoints with
corresponding applied uncertainty factors and derived RfDs following
oral exposure to 1,4-dioxane.
106
DRAFT - DO NOT CITE OR QUOTE
-------
Mouse
Rat
100
10
o
Q
0.1
I
O
I
O
• POD
[jAnimal-to-human
QHuman variation
0LOAELto NOAEL
Dsubchronic to Chronic
B Database deficiencies
ORfD
Nasal inflammation; NOAEL; 2 yr mouse drinking water study Nasal inflammation; NOAEL; 2 yr rat drinking water study
Figure 5-3 Potential points of departure (POD) for nasal inflammation with
corresponding applied uncertainty factors and derived sample RfDs
following oral exposure to 1,4-dioxane.
107
DRAFT - DO NOT CITE OR QUOTE
-------
1000
100 -
Rat
Rat
Rat
Mouse
10
m
8
Q
0.01
POD
OAnimal-to-human
QHuman variation
^LOAEL to NOAEL
QSubchronic to Chronic
^Database deficiencies
ORfD
Degeneration and necrosis Hepatocellular Delayed ossification of Nasal inflammation;
of tubular epithelium; degeneration and necrosis; sternebrae and reduced NOAEL; 2 yr mouse
NOAEL; 2 yr rat drinking NOAEL; 2 yr rat drinking fetal body weight; NOAEL; drinking water study
water study water sduy rat study gestation days 6-
15
Figure 5-4 Potential points of departure (POD) for organ specific toxicity
endpoints with corresponding applied uncertainty factors and derived
sample RfDs following oral exposure to 1,4-dioxane.
5.1.5 Previous RfD Assessment
An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
oral RfD was not developed as part of the 1988 assessment.
5.2 Inhalation Reference Concentration (RfC)
5.2.1 Choice of Principal Study and Candidate Critical Effect(s) with
Rationale and J ustification
3 Two human studies of occupational exposure to 1.4-dioxane have been published (Buffler et al.
4 1978; Thiess et al.. 1976); however, neither study provides sufficient information and data to quantify
5 subchronic or chronic noncancer effects. In each study. Findings were inconclusive and the cohort size
6 and number of reported cases were limited (Buffler et al.. 1978; Thiess et al.. 1976).
108
DRAFT - DO NOT CITE OR QUOTE
-------
1 Four inhalation studies in animals were identified in the literature; two, 13-week subchronic
2 studies in laboratory animals (Kasai et al.. 2008; Fairley et al.. 1934) and two. 2-year chronic studies in
3 rats (Kasai et al.. 2009; Torkelson et al.. 1974).
4 In the subchronic study by Fairley et al. (1934) rabbits, guinea pigs, rats, and mice
5 (3-6/species/group) were exposed to 1.000. 2.000. 5.000. or 10.000 ppm of 1.4-dioxane vapor for
6 1.5 hours two times a day for 5 days. 1.5 hours for one day, and no exposure on the seventh day. Animals
7 were exposed until death occurred or were sacrificed after various durations of exposure (3-202.5 hours).
8 Detailed dose-response information was not provided; however, severe kidney and liver damage and
9 acute vascular congestion of the lungs were observed at concentrations > 1.000 ppm. Kidney damage was
10 described as patchy degeneration of cortical tubules with vascular congestion and hemorrhage. Liver
11 lesions varied from cloudy hepatocyte swelling to large areas of necrosis. In this study, a LOAEL of
12 1.000 ppm for liver and kidney degeneration in rats, mice, rabbits, and guinea pigs was identified by EPA.
13 In the subchronic study by Kasai et al. (2008) male and female rats (10/group/sex) were exposed
14 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
15 weeks. This study observed a range of 1.4-dioxane induced nonneoplastic effects across several organ
16 systems including the liver and respiratory tract (from the nose to the bronchus region) in both sexes and
17 the kidney in females. Detailed dose-response information was provided, illustrating a
18 concentration-dependent increase of nuclear enlargement of nasal (respiratory and olfactory), trachea, and
19 bronchus epithelial cells (both sexes): vacuolic change of nasal and bronchial epithelial cells (both sexes).
20 necrosis and centrilobular swelling of hepatocytes (both sexes); and hydropic change in the proximal
21 tubules of the kidney (females). The study authors determined nuclear enlargement of the nasal
22 respiratory epithelium as the most sensitive lesion and a LOAEL of 100 ppm was identified based on this
23 effect.
24 Torkelson et al (1974) performed a chronic inhalation study in which male and female Wistar
25 rats (288/sex) were exposed to 111 ppm 1.4-dioxane vapor for 7 hours/day. 5 days/week for 2 years.
26 Control rats (192/sex) were exposed to filtered air. No significant effects were observed on BWs.
27 survival organ weights, hematology. clinical chemistry, or histopathology. A free standing NOAEL of
28 111 ppm was identified in this study by EPA.
29 Kasai et al. (2009) reported data for groups of male F344 rats (50/group) exposed to 0. 50. 250.
30 and 1.250 ppm of 1.4-dioxane for 6 hours/day. 5 days/week, for 2 years. In contrast to the subchronic
31 Kasai et al. (2008) study, this 2-year bioassay reported more nonneoplastic effects in multiple organ
32 systems. Additional noted incidences included: (1) inflammation of nasal respiratory and olfactory
33 epithelium. (2) squamous cell metaplasia and hyperplasia of nasal respiratory epithelium. (3) atrophy and
34 respiratory metaplasia of olfactory epithelium. (4) hydropic change and sclerosis in the lamina propria of
35 nasal cavity. (5) nuclear enlargement in proximal tubules of the kidney and in the centrilobular region of
36 the liver. (6) centrilobular necrosis in the liver, and (7) spongiosis hepatis. Some of these
37 histopathological lesions were significantly increased compared to controls at the lowest exposure level
38 (50 ppm). including nuclear enlargement of respiratory and olfactory epithelium; and atrophy and
109
DRAFT - DO NOT CITE OR QUOTE
-------
1 respiratory metaplasia of olfactory epithelium. Many of these histopathological lesions were increased in
2 a concentration-dependent manner.
3 The Fairley et al. (1934) study was insufficient to characterize the inhalation risks of 1.4-dioxane
4 because control animals were not used, thus limiting the ability to perform statistical analysis;
5 additionally, no data for low dose exposure were reported. Because Torkelson et al. (1974) identified a
6 free-standing NOAEL only, this study was also insufficient to characterize the inhalation risks of
7 1.4-dioxane. A route extrapolation from oral toxicity data was not performed because 1,4-dioxane
8 inhalation causes direct effects on the respiratory tract (i.e., respiratory irritation in humans, pulmonary
9 congestion in animals) (Wirth and Klimmer. 1936; Fairley et al.. 1934; Yant et al.. 1930). which would
10 not be accounted for in a cross-route extrapolation. In addition, available kinetic models are not
11 suitable for this purpose (Appendix B).
12 The chronic Kasai et al. (2009) study was selected as the principal study for the derivation of the
13 RfC. The Kasai et al. (2009) 2-year bioassay utilized 50 animals per exposure group, a range of exposure
14 concentrations which were based on the results of the subchronic study (Kasai et al.. 2008) and
15 thoroughly examined toxicity of l-4.dioxane in multiple organ systems. Based on the noncancer database
16 for 1.4-dioxane. this study demonstrated exposure concentration-related effects for histopathological
17 lesions at a lower concentration (50 ppm) compared to the subchronic Kasai et al. study (2008). The
18 2-year bioassay (Kasai et al.. 2009) did not observe effects in both sexes, but the use of only male rats
19 was proposed by the study authors as justified by data illustrating the absence of induced mesotheliomas
20 in female rats following exposure to 1.4-dioxane in drinking water (Yamazaki et al.. 1994). Additionally.
21 a similar pattern of effects was observed after oral exposure to 1.4-dioxane (Kano et al.. 2009; JBRC.
22 1998) as observed in the Kasai et al (2009) 2-year inhalation study.
23 Nonneoplastic lesions from the Kasai et al. (2009) study that were statistically increased as
24 compared to control were considered candidates for the critical effect. The candidate endpoints included
25 centrilobular necrosis of the liver, spongiosis hepatis. squamous cell metaplasia of nasal respiratory
26 epithelium, squamous cell hyperplasia of nasal respiratory epithelium, respiratory metaplasia of nasal
27 olfactory epithelium, sclerosis in lamina propria of nasal cavity, and two degenerative nasal lesions, that
28 is. atrophy of nasal olfactory epithelium and hydropic change in the lamina propria (Table 5-5). Despite
29 statistical increases at the low- and mid exposure concentrations (50 and 250 ppm. respectively).
30 incidences of nuclear enlargement of respiratory epithelium (nasal cavity), olfactory epithelium (nasal
31 cavity), and proximal tubule (kidney) were not considered candidates for the critical effect given that the
32 toxicological significance of nuclear enlargement is uncertain (See Section 4.6.2 and Table 4-22).
110
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-5 Incidences of nonneoplastic lesions resulting from chronic exposure (ppm) to
1,4-dioxane considered for identification of a critical effect.
Species/Strain
Rat/ F344 (male)
Centrilobular necrosis
Spongiosis hepatis
Squamous cell metaplasia;
respiratory epithelium
Squamous cell hyperplasia;
respiratory epithelium
Respiratory metaplasia;
asa olfactory epithelium
Atrophy; olfactory epithelium
Hydropic change;
lamina propria
Sclerosis; lamina propria
Concentration (ppm)
0
1/50
7/50
0/50
0/50
11/50
0/50
0/50
0/50
50
3/50
6/50
0/50
0/50
34/50a
40/50a
2/50
0/50
250
6/50
13/50
7/50b
1/50
49/50a
47/50a
36/50a
22/50a
1,250
12/50a
19/50a
44/50a
10/50a
48/50a
48/50a
49/50a
40/50a
ap<0.01 by x^ test.
bp < 0.05 by x2 test.
Source: Kasai et al. (2009).
5.2.2 Methods of Analysis
1 Benchmark dose (BMP) modeling methodology (U.S. EPA. 2000a) was used to analyze the
2 candidate endpoints identified for 1.4-dioxane. Use of BMP methods involves fitting mathematical
3 models to the observed dose-response data and provides a BMP and its 95% lower confidence limit
4 (BMDL) associated with a predetermined benchmark response (BMR). For 1.4-dioxane. the selected
5 datasets in Table 5-5 were considered as candidate critical effects and analyzed using BMP modeling to
6 determine potential POPs. Information regarding the degree of change in the selected endpoints that is
7 considered biologically significant was not available. Therefore, a BMR of 10% extra risk was selected
8 under the assumption that it represents a minimally biologically significant response level (U.S. EPA.
9 2000a).
10 The BMDs and BMDLs for centrilobular necrosis, spongiosis hepatis. squamous cell metaplasia
11 of the respiratory epithelium, and hydropic change of lamina propria are presented in Table 5-6. Due to
12 poor fit or substantial model uncertainty. BMP model results were inadequate for the following nasal
13 lesions: atrophy (olfactory epithelium), respiratory metaplasia (olfactory epithelium), and sclerosis
14 (lamina propria). Consequently, the NOAEL/LOAEL approach was used to determine potential POPs for
15 these endpoints. The detailed results of the BMP analysis are provided in Appendix F.
5.2.3 Exposure Duration and Dosimetric Adjustments
16 Because an RFC is a measure that assumes continuous human exposure over a lifetime, data
17 derived from animal studies need to be adjusted to account for the noncontinuous exposure protocols used
18 in animal studies. In the Kasai et al. (2009) study, rats were exposed to 1.4-dioxane for 6 hours/day. 5
111
DRAFT - DO NOT CITE OR QUOTE
-------
1 days/week for 2 years. Therefore, the duration-adjusted POPs for liver and nasal lesions in rats were
2 calculated as follows:
PODADJ(ppm) = POD(ppm):
hours exposed per day days exposed per week
24hours
Vdays
4 RfCs are typically expressed in units of mg/m3; so POD^m (ppm) values were converted using
5 the chemical specific conversion factor of 1 ppm = 3.6 mg/m3 for 1.4-dioxane (Table 2-1). The following
6 calculation was used:
POD ADJ (mg/m3) =
3.6 mg/m3
Ippm
8 The calculated POD/\DJ (mg/m3) values for all considered endpoints are presented in the last
9 column of Table 5 -6^
10
Table 5-6 Duration adjusted POD estimates for BMDLs (from best fitting BMDS models) or
NOAELs/LOAELs from chronic exposure to 1,4-dioxane
11
12
13
Endpoint . .
L,OAEITD Model
(ppm)
BMR
BMD
(ppm)
BMDL
(ppm)
PODADJ
(mg/m3)
Liver Effects
Centrilobular necrosis;
Liver
Spongiosis hepatis; Liver
Dichotomous-Hill
Log-logistic0
10
10
220
314
60
172
38.6
111
Nasal Effects
Squamous cell metaplasia;
respiratory epithelium
Squamous cell
hyperplasia; respiratory
epithelium
Respiratory metaplasia;
olfactory epithelium
Atrophy; olfactory
epithelium
Hydropic change;
lamina propria
Sclerosis; lamina propria 50
Log-probit
Log-probit
50
50
Log-logistic
-c
10
10
-
-
10
218
756
-
-
69
160
561
-
-
47
103
361
32.2
32.2
30.2
32.2e
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.
CBMDS model results are not adequate for use to derive a POD. Therefore, the NOAEL/LOAEL approach is used to determine a POD
for these endpoints. BMDS analysis for these endpoints is included in Appendix F.
dDichotomous Hill model had lowest BMDL, but model output warned that the BMDL estimate was "imprecise at best".
eBased on the NOAEL of 50 ppm.
Based on analysis of data in Table 5-6a the liver effects (i.e.. centrilobular necrosis and spongiosis
hepatis) were shown to be less sensitive than the nasal effects and were not considered further as
112
DRAFT - DO NOT CITE OR QUOTE
-------
1 candidate critical effects. Similarly, the squamous cell metaplasia and hyperplasia of the respiratory
2 epithelium yielded potential POPs that were 3-fold or greater than the remaining nasal effects: thus, these
3 effects were not considered further as candidate critical effects. The POPs adjusted for continuous
4 exposure for sclerosis of the lamina propria. atrophy of the olfactory epithelium, and respiratory
5 metaplasia of the olfactory epithelium were identical (32.2 mg/m3) and similar to the POD^m for hydropic
6 change of the lamina propria (30.2 mg/m3). Although the POD ADJ estimates were either identical or
7 similar, the responses (i.e.. increased incidence of effect) associated with the POD Am for these effects.
8 (i.e.. 0% for sclerosis. 10% for hydropic change. 59% for respiratory metaplasia. 80% for atrophy) varied.
9 As shown in Table 5-5. atrophy and respiratory metaplasia of the olfactory epithelium were the
10 most sensitive effects based on the responses of 80 and 59%. respectively, observed in animals exposed at
11 the lowest concentration (50 ppm). Increased incidences of the other nasal effects and liver effects were
12 observed at either 50 ppm or greater; however, not to the extent that was observed for atrophy and
13 respiratory metaplasia of the olfactory epithelium. Typically, chemical-induced nasal effects include
14 atrophy and/or necrosis, cell proliferation/hvperplasia. and metaplasia depending on the nature of the
15 tissue damage and exposure (Harkema et al.. 2006; Boorman et al.. 1990; Gaskell. 1990). However the
16 pathological progression of these events is uncertain and often accompanied by an inflammatory
17 response. Since the data do not support a continuum of pathological events associated with respiratory
18 tract effects, both atrophy and respiratory metaplasia of the olfactory epithelium are selected as co-critical
19 effects in this assessment.
20 For the derivation of a RfC based upon an animal study, the selected POD must be adjusted to
21 reflect the human equivalent concentration (HEC). The HEC was calculated by the application of a
22 dosimetric adjustment factor (DAF). in accordance with the U.S. EPA Methods for Derivation of
23 Inhalation Reference Concentrations and Application of Inhalation Dosimetry (hereafter referred to as the
24 RfC methodology) (U.S. EPA. 1994). DAFs are ratios of animal and human physiologic parameters, and
25 are dependent on the nature of the contaminant (particle or gas) and the target site (e.g.. respiratory tract
26 or remote to the portal-of-entry) (U.S. EPA. 1994).
27 1.4-Dioxane is miscible with water and has a high blood:air partition coefficient. Typically.
28 highly water-soluble and directly reactive chemicals (i.e. Category 1 gases) partition greatly into the
29 upper respiratory tract, induce portal-of-entry effects, and do not accumulate significantly in the blood.
30 1.4-Dioxane induces effects throughout the respiratory tract, liver, and kidneys, and it has been measured
31 in the blood after inhalation exposure (Kasai et al.. 2008). The observations of systemic (i.e..
32 nonrespiratory) effects and measured blood levels resulting from 1.4-dioxane exposure indicate that this
33 compound is absorbed into the bloodstream and distributed throughout the body. Furthermore, the lack of
34 an anterior to posterior gradient for the nasal effects induced by 1.4-dioxane is not typical of chemicals
35 which are predominantly directly reactive. Thus. 1.4-dioxane might be best described as a water-soluble
36 and non-directly reactive gas. Gases such as these are readily taken up into respiratory tract tissues and
37 can also diffuse into the blood capillaries (Medinsky and Bond. 2001). The effects in the olfactory
38 epithelium may be the result of the metabolism of 1.4-dioxane to an acid metabolite: however, for the
39 reasons stated above it is unclear whether or not these effects are solely the result of portal-of-entry or
113
DRAFT - DO NOT CITE OR QUOTE
-------
1 systemic delivery. A similar pattern of effects was observed after oral exposure to 1.4-dioxane (JBRC.
2 1998: KanoetaL 2009).
3 In consideration of the evidence described above, the human equivalent concentration (FffiC) for
4 1.4-dioxane was calculated by the application of the appropriate dosimetric adjustment factor (DAF) for
5 systemic acting gases (i.e.. Category 3 gases), in accordance with the U.S. EPA RfC methodology (U.S.
6 EPA. 1994). However, since 1.4-dioxane is water soluble and might induce portal-of-entry effects, an
7 alternative calculation of the FffiC for 1.4-dioxane. based on the application of the corresponding DAF for
8 portal-of-entry acting gases (i.e.. Category 1) is provided in Appendix G.
9 The calculation of the HEC used in this assessment is as follows:
10 DAF = (Hb/g)A/(Hb/g)H
11 DAF= 1.861/1.666
12 DAF= 1.12
13 where:
14 (FIb/g)A = the animal blood:air partition coefficient =1.861 (Sweeney et al.. 2008)
15 (FIb/g)H = the human blood:air partition coefficient =1.666 (Sweeney et al.. 2008)
16 Given that the animal blood:air partition coefficient is higher than the human value resulting in a
17 DAF>1. a default value of 1 is substituted in accordance with the U.S. EPA RfC methodology (U.S. EPA.
18 1994). Analysis of the existing inhalation dosimetry modeling database supports the application of a DAF
19 of 1 (U.S. EPA. 2009c). Application of these models to gases that have similar physicochemical
20 properties and induce similar nasal effects as 1.4-dioxane estimate DAFs > j^
21 Utilizing a DAF of 1. the FffiC for atrophy and respiratory metaplasia of the olfactory epithelium
22 in male F344/DuCrj rats is calculated as follows:
23 POD^ (mg/m3) = POD^m (mg/m3) x DAF
24 = PODAm (mg/m3) x. LO
25 =32. 2 mg/m3 x l.Q
26 = 32.2 mg/m3
27 Therefore, the P()DHEC of 32.2 mg/m3 for the co-critical effects of atrophy and respiratory
28 metaplasia of the olfactory epithelium is used for the derivation of a RfC for 1.4-dioxane.
114
DRAFT - DO NOT CITE OR QUOTE
-------
5.2.4 RfC Derivation- Including Application of Uncertainty Factors (UFs)
1 The RfC of 3 x ICT2 mg/m3 is based on atrophy and respiratory metaplasia of the olfactory
2 epithelium in male rats exposed to 1.4-dioxane via inhalation for 2 years (Kasai et al.. 2009). The RfC for
3 1.4-dioxane is derived by dividing the POPART by a composite UF of 1.000.
4 RfQ = PODHEc/UF
5 =32.2 mg/m3/1.000
6 = 0.0322 or 3 x 10~2 mg/m3 (rounded to 1 significant figure)
7 An UF of 10 was used to extrapolate from a LOAEL to a NOAEL because a LOAEL was used as
8 the POD for critical effects. A NOAEL for atrophy and respiratory metaplasia of the olfactory epithelium
9 was not identified in the study by Kasai et al. (2009).
10 A default interindividual variability UF of 10 was used to account for variation in sensitivity
11 within human populations because there is limited information on the degree to which humans of varying
12 gender, age, health status, or genetic makeup might vary in the disposition of. or response to. 1.4-dioxane.
13 An interspecies UF of 3 was used for animal-to-human extrapolation to account for
14 pharmacodynamic differences between species. This uncertainty factor is comprised of two separate areas
15 of uncertainty to account for differences in the toxicokinetics and toxicodynamics of animals and humans.
16 In this assessment, the toxicokinetic uncertainty was accounted for by the calculation of a HEC and
17 application of a dosimetric adjustment factor as outlined in the RfC methodology (U.S. EPA. 1994). As
18 the toxicokinetic differences are thus accounted for, only the toxicodynamic uncertainties remain, and an
19 UF of 3 is retained to account for this uncertainty.
20 An UF of 3 for database deficiencies was applied due to the lack of a multigeneration
21 reproductive toxicity study. The oral toxicity database included a single prenatal developmental study that
22 indicated the developing fetus may be a target of toxicity (Giavini et al.. 1985).
23 An UF of 1 was used to extrapolate from a subchronic to a chronic exposure duration because the
24 RfC was derived from a study using a chronic exposure protocol.
5.2.5 RfC Comparison Information
25 Figure 5-5 presents POPs, applied UFs. and derived sample RfCs for possible endpoints from the
26 chronic inhalation Kasai et al (2009) in male rats. The POPs are based on the BMDLjn. NOAEL. or
27 LOAEL and appropriate unit conversion, duration, and dosimetric adjustments were applied before
28 applications of UFs. The predominant noncancer effects of chronic inhalation exposure to 1.4-dioxane
29 include nasal and liver effects. Figure 5-5 provides a graphical display of effects that were observed in
30 the Kasai et al. (2009) study. Information presented includes the POPs and UFs that could be considered
31 in deriving the inhalation RfC. As discussed in Sections 5.2.1 and 5.2.3. the Kasai et al. (2009) study
115
DRAFT - DO NOT CITE OR QUOTE
-------
1 provided the data set for deriving the RfC. The nasal effects of the olfactory epithelium represent the
2 most sensitive effects.
I
s.
'•••-
o
c
o
u
£
I
Squarnous cell Squamous cell Respiratory Atrophy in the Hydropic
metaplasia in hyperplasia in metaplasia in nasal olfactory change in
the respiratory the respiratory the olfactory epithelium, lamina propria;
epithelium; epithelium; epithelium; LOAEL BMDLIO
BMDL10 BMDL1D LOAEL
• POD
UJ Animal- to-human
^]Human variation
LOAEL to NOAEL
dSubchromc to Chronic
• Database deficiencies
oRfC
Sclerosis of Centrilobular Spongiosis
the lamina necrosis in the hepatismthe
propria; liver; BMDL10 liver; BMDL10
NOAEL
Figure 5-5 Potential points of departure (POD) for candidate endpoints with
corresponding applied uncertainty factors and derived sample RfCs
following inhalation exposure to 1,4-dioxane.
Source: Kasai et al. (2009)
5.2.6 Previous RfC Assessment
An RfC for 1.4-dioxane was not previously available on the IRIS database.
5.3 Uncertainties in the Oral Reference Dose and Inhalation
Reference Concentration
4 Risk assessments need to portray associated uncertainty. The following discussion identifies
5 uncertainties associated with the RfD and RfC for 1,4-dioxane. As presented earlier in this section (see
6 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
7 approach (U.S. EPA. 2002a. 1994) was used to derive the RfD and RfC for 1.4-dioxane. Using this
8 approach, the POD was divided by a set of factors to account for uncertainties associated with a number
9 of steps in the analysis^ including extrapolation from LOAEL to NOAEL. extrapolation from animals to
10 humans, a diverse population of varying susceptibilities, and to account for database deficiencies.
116
DRAFT - DO NOT CITE OR QUOTE
-------
1 Because information specific to 1.4-dioxane was unavailable to fully inform these extrapolations, default
2 factors were generally applied.
3 An adequate range of animal toxicology data are available for the hazard assessment of
4 1,4-dioxane, as described throughout the previous section (Section 4). The database of oral toxicity
5 studies includes chronic drinking water studies in rats and mice, multiple subchronic drinking water
6 studies conducted in rats and mice, and a developmental study in rats. Toxicity associated with oral
7 exposure to 1,4-dioxane is observed predominately in the liver and kidney. The database of inhalation
8 toxicity studies in animals includes two subchronic bioassays in rabbits, guinea pigs, mice, and rats, and
9 two chronic inhalation bioassays in rats. Toxicity associated with inhalation exposure to 1.4-dioxane was
10 observed predominately in the liver and nasal cavity. In addition to oral and inhalation data, there are
11 PBPK models and genotoxicity studies of 1,4-dioxane. Critical data gaps have been identified and
12 uncertainties associated with data deficiencies of 1,4-dioxane are more fully discussed below.
13 Consideration of the available dose-response data led to the selection of the two-year drinking
14 water bioassay in Sherman rats (Kocibaet al.. 1974) as the principal study and increased liver and kidney
15 degeneration as the critical effects for deriving the RfD for 1,4-dioxane. The dose-response relationship
16 for oral exposure to 1,4-dioxane and cortical tubule degeneration in Osborne-Mendel rats (NCI. 1978)
17 was also suitable for deriving a RfD, but it is associated with higher a POD and potential RfD compared
18 toKocibaetal. (1974).
19 The RfD was derived by applying UFs to a NOAEL for degenerative liver and kidney effects.
20 The incidence data for the observed effects were not reported in the principal study (Kociba et al.. 1974).
21 precluding modeling of the dose-response. However confidence in the NOAEL can be derived from
22 additional studies (JBRC. 1998: NCI. 1978: Argus etal.. 1973: Argus etal.. 1965) that observed effects
23 on the same organs at comparable dose levels and by the BMDL generated by modeling of the kidney
24 dose-response data from the chronic NCI (1978) study.
25 The RfC was derived by applying UFs to a LOAEL for atrophy and respiratory metaplasia of the
26 olfactory epithelium. The incidence data for the observed effects were not amenable to BMP modeling
27 (see Appendix F). The LOAEL for these effects was less than or equal to the LOAEL or NOAEL for
28 other effects observed in the Kasai et al. (2009) study.
29 Extrapolating from animals to humans embodies further issues and uncertainties. The effect and
30 the magnitude associated with the dose at the POD in rodents are extrapolated to human response.
31 Pharmacokinetic models are useful to examine species differences in pharmacokinetic processing;
32 however, it was determined that dosimetric adjustment using pharmacokinetic modeling to reduce
33 uncertainty following oral exposure to 1,4-dioxane was not supported. Insufficient information was
34 available to quantitatively assess toxicokinetic or toxicodynamic differences between animals and
35 humans, so a 10-fold UF was used to account for uncertainty in extrapolating from laboratory animals to
36 humans in the derivation of the RfD. A DAF was used to account for pharmacokinetic differences
37 between rodents and humans in the derivation of the RfC: however, there was no information to inform
38 pharmacodynamic differences between species, so an UF of 3 was used in derivation of the RfC to
39 account for these uncertainties.
117
DRAFT - DO NOT CITE OR QUOTE
-------
1 Heterogeneity among humans is another uncertainty associated with extrapolating doses from
2 animals to humans. Uncertainty related to human variation needs consideration. In the absence of
3 1,4-dioxane-specific data on human variation, a factor of 10 was used to account for uncertainty
4 associated with human variation in the derivation of the RfD and RfC. Human variation may be larger or
5 smaller; however, 1,4-dioxane-specific data to examine the potential magnitude of over- or
6 under-estimation are unavailable.
7 Uncertainties in the assessment of the health hazards of 1,4-dioxane are associated with
8 deficiencies in reproductive toxicity information. The oral and inhalation databases lack a multigeneration
9 reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
10 1,4-dioxane (Giavini et al.. 1985). This developmental study indicates that the developing fetus may be a
11 target of toxicity. The database of inhalation studies also lacks a developmental toxicity study.
5.4 Cancer Assessment
5.4.1 Choice of Study/Data - with Rationale and J ustification
5.4.1.1 Oral Study/Data
12 Three chronic drinking water bioassays provided incidence data for liver tumors in rats and mice,
13 and nasal cavity, peritoneal, and mammary gland tumors in rats only (Kano et al.. 2009; JBRC. 1998;
14 Yamazaki et al.. 1994; NCI. 1978; Kocibaet al., 1974). The dose-response data from each of these studies
15 are summarized in Table 5-7. With the exception of the NCI (1978) study, the incidence of nasal cavity
16 tumors was generally lower than the incidence of liver tumors in exposed rats. The Kano et al. (2009)
17 drinking water study was chosen as the principal study for derivation of an oral cancer slope factor (CSF)
18 for 1,4-dioxane. This study used three dose groups in addition to controls and characterized the
19 dose-response relationship at lower exposure levels, as compared to the high doses employed in the NCI
20 (1978) bioassay (Table 5-7). The Kociba et al. (1974) study also used three dose groups and low
21 exposures; however, the study authors only reported the incidence of hepatocellular carcinoma, which
22 may underestimate the combined incidence of rats with adenoma or carcinoma. In addition to increased
23 incidence of liver tumors, chosen as the most sensitive target organ for tumor formation, the Kano et al.
24 (2009) study also noted increased incidence of peritoneal and mammary gland tumors. Nasal cavity
25 tumors were also seen in high-dose male and female rats; however, the incidence of nasal tumors was
26 much lower than the incidence of liver tumors in both rats and mice.
27 In a personal communication, Dr. Yamazaki (2006) provided that the survival of mice was low in
28 all male groups (31/50, 33/50, 25/50 and 26/50 in control, low-, mid-, and high-dose groups, respectively)
29 and particularly low in high-dose females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-, and
30 high-dose groups, respectively). These deaths occurred primarily during the second year of the study.
118
DRAFT - DO NOT CITE OR QUOTE
-------
1 Survival at 12 months in male mice was 50/50, 48/50, 50/50, and 48/50 in control, low-, mid-, and
2 high-dose groups, respectively. Female mouse survival at 12 months was 50/50, 50/50, 48/50, and 48/50
3 in control, low-, mid-, and high-dose groups, respectively (Yamazaki. 2006). Furthermore, these deaths
4 were primarily tumor related. Liver tumors were listed as the cause of death for 31 of the 45
5 pretermination deaths in high-dose female Crj:BDFl mice (Yamazaki. 2006). Thus, the high mortality
6 rates in the female mice were still considered to be relevant for this analysis.
119
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-7 Incidence of liver, nasal cavity, peritoneal, and mammary gland tumors in rats and
mice exposed to 1,4-dioxane in drinking water for 2 years (based on survival to
12 months)
Study Species/strain/gender
Kociba et al. Sherman rats, male and
(1974) female combined3'13
rats"
Female
M^l 1 1 O7O\
Male B6C3F-I miced
Female B6C3F-I miced
Male F344/DuCrj
ratsd'e'f'g
Female F344/DuCrj
ratsd'e'f'g
Kanoetal. (2009)
Animal dose
(mg/kg-day)
0
14
121
1,307
0
240
530
0
350
640
0
720
830
0
380
860
0
11
55
274
0
18
83
429
0
49
191
677
0
66
278
964
Tumor Incidence
Liver
1/106"
0/110
1/106
10/66'
NA
NA
NA
0/31"
10/30'
11/29'
8/49"
19/50'
28/47'
0/50"
21/48'
35/37'
3/50
4/50
7/50
39/50J'K
3/50
1/50
6/50
48/50J'K
23/50
31/50
37/50'
40/50J'K
5/50
35/50J
41/50J
46/50J'K
Nasal
cavity
0/106"
0/110
0/106
3/66
0/33"
12/26
16/33'
0/34"
10/30'
8/29'
NA
NA
NA
NA
NA
NA
0/50
0/50
0/50
7/50K
0/50
0/50
0/50
8/50J'K
0/50
0/50
0/50
1/50
0/50
0/50
0/50
1/50
Peritoneal
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
2/50
2/50
5/50
28/50J'K
1/50
0/50
0/50
0/50
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
1/50
2/50
2/50
6/50K
8/50
8/50
11/50
18/50''K
NA
NA
NA
NA
NA
NA
NA
NA
"Incidence of hepatocellular carcinoma.
""Incidence of nasal squamous cell carcinoma.
Incidence of hepatocellular adenoma.
dlncidence 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 (no significant change from controls)
5.4.1.2 Inhalation Study/Data
1 Epidemiological studies of populations exposed to 1.4-dioxane are not adequate for
2 dose-response analysis and derivation of an inhalation unit risk (IUR). However, two chronic inhalation
120
DRAFT - DO NOT CITE OR QUOTE
-------
1 studies in animals are available and were evaluated for the potential to estimate an IUR (Table 5-8^. The
2 chronic inhalation study conducted by Torkelson et al. (1974) in rats did not find any treatment-related
3 tumors; however, only a single exposure concentration was used (111 ppm 1.4-dioxane vapor for
4 7 hours/day. 5 days/week for 2 years'). A chronic bioassay of 1.4-dioxane by the inhalation route reported
5 by Kasai et al. (2009) provides data adequate for dose-response modeling and was subsequently chosen as
6 the study for the derivation of an IUR for 1.4-dioxane. In this bioassay. groups of 50 male F344 rats were
7 exposed to either 0. 50. 250 or 1.250 ppm 1.4-dioxane. 6 hours/day. 5 days/week, for 2 years
8 (104-weeks). In male F344 rats. 1.4-dioxane produced a statistically significant increase in incidence
9 and/or a statistically significant dose-response trend for the following tumor types: hepatomas. nasal
10 squamous cell carcinomas, renal cell carcinomas, peritoneal mesotheliomas. mammary gland
11 fibroadenomas. Zvmbal gland adenomas, and subcutis fibromas (Kasai et al.. 2009). The incidence of
12 adenomas and carcinomas were combined in this assessment in accordance with EPA's Guidelines on
13 Carcinogen Risk Assessment which notes that etiologically similar tumor types, i.e.. benign and malignant
14 tumors of the same cell type, can be combined due to the possiblity that benign tumors could progress to
15 the malignant form (U.S. EPA. 2005a; McConnell et al., 1986). Consistent with the oral cancer
16 assessment (Appendix D). the incidence of hepatic adenomas and carcinomas (combined) and was used to
17 calculate an IUR in rodents (See Table 5-8^.
121
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-8 Incidence of liver, nasal cavity, kidney, peritoneal, and mammary gland, Zymbal
gland, and subcutis tumors in rats exposed to 1,4-dioxane vapors for 2 years.
Study
Torkelson
etal.
(1974)a
Kasai et al.
(2009)b
Species/ Animal
strain/ Exposure
gender (ppm)
Male
Wistar
rats
Female
Wistar
rats
Male
F344 rats
0
111
0
111
0
50
250
1,250
Tumor Incidence
Liverc
0/150
0/206
0/139
0/217
1/50
2/50
4/50
22/50
Nasal
cavityd
0/150
0/206
0/139
0/217
0/50
0/50
1/50
6/50m
Kidney6
0/150'
1/2061
1/1 39J
0/2 17j
0/50
0/50
0/50
4/50
Peritoneal'
NA
NA
NA
NA
2/50
4/50
14/50"
41/50"
Mammary
gland
NA
NA
11/139"
29/2 17k
1/50'
2/50'
3/50'
5/50'
Zymbal
gland9
NA
NA
NA
NA
0/50
0/50
0/50
4/50
Subcutis11
0/150
2/206
0/139
0/217
1/50
4/50
9/50"
5/50
Incidence reported based on survival to 9 months.
blncidence reported based on survival to 12 months.
Incidence of hepatocellular adenoma or carcinoma. For Kasai et al. (2009) incidence data was provided via personal communication
from Dr. Tatsuya Kasai to Dr. Reeder Sams on 12/23/2008 (2008). Statistics were not reported. Individual incidence rates for
adenomas and carcinomas are in Table 5-10.
Incidence of nasal squamous cell carcinoma.
Incidence of renal cell carcinoma.
'incidence of peritoneal mesothelioma.
Incidence of Zymbal gland adenoma.
Incidence of subcutis fibroma.
'Incidence of kidney fibroma.
'Incidence of kidney adenocarcinoma
Incidence of mammary gland adenoma.
'incidence of mammary gland fibroadenoma.
"Tumor 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
5.4.2 Dose-Response Data
5.4.2.1 Oral Data
1 Table 5-9 summarizes the incidence of hepatocellular adenoma or carcinoma in rats and mice
2 from the Kano et al. (2009) 2-year drinking water study. There were statistically significant increasing
3 trends in tumorigenic response for males and females of both species. The dose-response curve for female
4 mice is steep, with 70% incidence of liver tumors occurring in the low-dose group (66 mg/kg-day).
5 Exposure to 1,4-dioxane increased the incidence of these tumors in a dose-related manner.
6 A significant increase in the incidence of peritoneal mesothelioma was observed in high-dose
7 male rats only (28/50 rats, Table 5-7). The incidence of peritoneal mesothelioma was lower than the
8 observed incidence of hepatocellular adenoma or carcinoma in male rats (Table 5-9); therefore,
9 hepatocellular adenoma or carcinoma data were used to derive an oral CSF for 1,4-dioxane.
122
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-9 Incidence of hepatocellular adenoma or carcinoma in rats and mice exposed to
1,4-dioxane in drinking water for 2 years
Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male Crj:BDF1 mice
Female Crj:BDF1 mice
Animal dose incidence of liver tumors3
(mg/kg-day)
0
11
55
274
0
18
83
429
0
49
191
677
0
66
278
964
3/50
4/50
7/50
39/50c'c
3/50
1/50
6/50
48/50c'c
23/50
31/50
37/50"
40/50C'C
5/50
35/50c
41/50C
46/50c'c
"Incidence of either hepatocellular adenoma or carcinoma.
bp < 0.05; positive dose-related trend (Peto's test).
°Significantly different from control at p < 0.01 by Fisher's Exact test.
dSignificantly different from control at p < 0.01 by Fisher's Exact test.
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
5.4.2.2 Inhalation Data
1 Multi-tumor dose-response modeling was performed for all tumor responses from the Kasai et al.
2 (2009) bioassay. Kasai et al. (2009) reported tumor incidence data for male F344 rats exposed via
3 inhalation to 0. 50. 250. or 1.250 ppm 1.4-dioxane for 6 hours/day. Sclavs/week, for 2 years (104-weeks).
4 Statistically significant dose-response trends for the increase in tumors with increasing dose was observed
5 for the nasal cavity squamous cell carcinomas, hepatomas. renal cell carcinomas, peritoneal
6 mesotheliomas. mammary gland fibroadenomas. and Zymbal gland adenomas. Following 250 ppm
7 1.4-dioxane exposure, statistically elevated tumor incidences were found in two tissue types (peritoneal
8 mesothelioma and subcutis fibroma) compared to controls. It is important to note, for observations of
9 subcutis fibroma, the incidence was increased compared to controls at all concentrations but a decrease in
10 incidence, compared to the mid-concentration, was noted at the highest concentration (1.250 ppm).
11 However, a significantly decreased survival rate was noted in this exposure group by the study authors.
12 Interim sacrifices were not performed. Tumor incidences following 1.250 ppm inhalation exposure to
13 1.4-dioxane were statistically elevated compared to controls in three tissues (nasal cavity squamous cell
14 carcinoma, hepatomas. and peritoneal mesothelioma). Incidence data for the tumor types reported by
15 Kasai et al. (2009) are summarized in Table 5-10.
123
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-10 Incidence of tumors in F344 male rats exposed to 1,4-dioxane for 104 weeks (6
hours/day, 5 days/week)
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
Animal Exposure (ppm)
0
0/50
1/50
0/50
1/50
0/50
2/50
1/50
0/50
0/50
1/50
0/50
14/50C
0/50
9/50c
1,250
6/50a'°
21/50a'c
2/50
22/50a'c
4/50a
41/50a'c
5/50a
1/50
4/50a
"Statistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
"Tumor 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).
Statistically significant trend for increased tumor incidence by Peto's test (p S 0.05).
eProvided via personal communication from Dr. Tatsuya Kasai to Dr. Reeder Sams on 12/23/2008 (2008). Statistics were not
reported for these data by study authors, so statistical analyses were conducted by EPA.
Source: Kasai et al. (2009) and Kasai personal communication (2008)
5.4.3 Dose Adjustments and Extrapolation Method(s)
5.4.3.1 Oral
1 Human equivalent doses (HEDs) were calculated from the administered animal doses using a BW
2 scaling factor (BW°75) (U.S. EPA. 201 Ib). This was accomplished using the following equation:
HED = animal dose (mg/kg) x
animal BW (kg)
human BW (kg)
4 For all calculations, a human BW of 70 kg was used. HEDs for the principal study (Kano et al..
5 2009) are given in Table 5-11. HEDs were also calculated for supporting studies (NCI. 1978; Kociba et
6 al.. 1974) and are also shown in Table 5-11.
124
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-11 Calculated HEDs for the tumor incidence data used for dose-response modeling
ft
Study Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male Crj:BDF1 mice
Female Crj:BDF1 mice
Kociba et al. (1974) 0, .
old man ra ^
Mf^l /-I O7O\
NU (ISJ/o)
mimal BW (g)
432a
432a
432a
267a
267a
267a
47.9a
47.9a
47.9a
35.9a
35.9a
35.9a
325°
325°
285C
470°
470°
310°
310°
32°
32°
30°
30°
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
a TWA BWs were determined from BW growth curves provided for each species and gender.
bTWA BWs were determined from BW curve provided for control animals.
°BWs of high dose male and female rats were significantly lower than controls throughout the study. TWA represents the mean of
TWA for male and females (calculated separately from growth curves).
dHEDs are calculated as HED = (animal dose) x (animal BW/ human BW)025.
Sources: Kano et al. (2009): Kociba et al. (1974): and NCI (1978).
1 The U.S. EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) recommend that
2 the method used to characterize and quantify cancer risk from a chemical is determined by what is known
3 about the mode of action of the carcinogen and the shape of the cancer dose-response curve. The linear
4 approach is recommended if the mode of action of carcinogenicity is not understood (U.S. EPA. 2005a).
5 In the case of 1,4-dioxane, the mode of carcinogenic action for peritoneal, mammary, nasal, and liver
6 tumors is unknown. Therefore, a linear low-dose extrapolation approach was used to estimate human
7 carcinogenic risk associated with 1,4-dioxane exposure.
8 However, several of the external peer review panel members (Appendix A: Summary of External
9 Peer Review and Public Comments and Disposition) recommended that the mode of action data support
10 the use of a nonlinear extrapolation approach to estimate human carcinogenic risk associated with
11 exposure to 1,4-dioxane and that such an approach should be presented in the Toxicological Review. As
12 discussed in Section 4.5.1, numerous short-term in vitro and a few in vivo tests were nonpositive for
13 1,4-dioxane-induced genotoxicity. Results from two-stage mouse skin tumor bioassays demonstrated that
14 1,4-dioxane does not initiate mouse skin tumors, but it is a promoter of skin tumors initiated by DMBA
15 (King et al.. 1973). These data suggest that a potential mode of action for 1,4-dioxane-induced tumors
16 may involve proliferation of cells initiated spontaneously, or by some other agent, to become tumors
125
DRAFT - DO NOT CITE OR QUOTE
-------
1 (Mivagawa et al.. 1999; Uno et al. 1994; Goldsworthy et al.. 1991; Lundberg et al.. 1987; Bull et al.
2 1986; Stott et al.. 1981; King et al.. 1973). However, key events related to the promotion of tumor
3 formation by 1,4-dioxane are unknown. Therefore, under the U.S. EPA Guidelines for Carcinogen Risk
4 Assessment (U.S. EPA. 2005a). EPA concluded that the available information does not establish a
5 plausible mode of action for 1,4-dioxane and data are insufficient to establish significant biological
6 support for a nonlinear approach. EPA determined that there are no data available to inform the low-dose
7 region of the dose response, and thus, a nonlinear approach was not included.
8 Accordingly, the CSF for 1,4-dioxane was derived via a linear extrapolation from the POD
9 calculated by curve fitting the experimental dose-response data. The POD is the 95% lower confidence
10 limit on the dose associated with a benchmark response (BMR) near the lower end of the observed data.
11 The BMD modeling analysis used to estimate the POD is described in detail in Appendix D and is
12 summarized below in Section 5.4.4.
13 Model estimates were derived for all available bioassays and tumor endpoints (Appendix D);
14 however, the POD used to derive the CSF is based on the most sensitive species and target organ in the
15 principal study (Kano etal. 2009).
16 The oral CSF was calculated using the following equation:
BMR
17 CSF = •
BMDL
5.4.3.2 Inhalation
18 In accordance with the U.S. EPA (1994) RfC methodology, the FffiC values were calculated by
19 the application of DAFs. As discussed in Section 5.2.3. since 1.4-dioxane is miscible with water, has a
20 high partition coefficient, and induces effects throughout the body of the rat. a DAF of 1.0 was applied.
21 The lifetime continuous inhalation risk for humans is defined as the slope of the line from the POD, the
22 lower 95% bound on the exposure associated with a level of extra risk near the low end of the data range.
23 All POPs were converted to equivalent continuous exposure levels by multiplying by [(6
24 hours)/(24 hours)] x[(5 days)/(7 days)1. under the assumption of equal cumulative exposures leading to
25 equivalent outcomes.
26 Given the multiplicity of tumor sites, basing the IUR on one tumor site may underestimate the
27 carcinogenic potential of 1.4-dioxane. Also, simply pooling the counts of animals with one or more
28 tumors (i.e.. counts of tumor bearing animals) would tend to underestimate the overall risk for tumors
29 observed at independent sites and ignores potential differences in the dose-response relationships across
30 the sites (NRC. 1994; Bogen. 1990). NRC (1994) also noted that the assumption of independence across
31 tumor types is not likely to produce substantial error in the risk estimates unless tumors are known to be
32 biologically dependent.
126
DRAFT - DO NOT CITE OR QUOTE
-------
1 Kopylev et al. (2009) describe a Markov Chain Monte Caro (MCMC) computational approach to
2 calculating the dose associated with a specified composite risk under assumption of independence of
3 tumors. The Guidelines for Carcinogen Risk Assessment recommend calculation of an upper bound to
4 account for uncertainty in the estimate (U.S. EPA. 2005a). For uncertainty characterization. MCMC
5 methods have the advantage of providing information about the full distribution of risk and/or benchmark
6 dose, which can be used in generating a confidence bound. This MCMC approach which builds on the
7 re-sampling approach recommended by Bogen (1990). also provides a distribution of the combined
8 potency across sites. The Bayesian MCMC computations were conducted using WinBugs (Spiegelhalter
9 et al., 2003) and additional details of this analysis are included in Appendix G. In addition, the best fitting
10 BMDS multistage model was determined for each individual tumor type as shown in Section 5.4.4.2 and
11 Appendix H.
12 The carcinogenic MOA(s) by which 1.4-dioxane produces liver, nasal, kidney, peritoneal
13 (mesotheliomas). mammary gland. Zymbal gland, and subcutis tumors is unknown. Several hypothesized
14 MOA(s) have been proposed for liver and nasal tumors although these MOA(s) are not supported by the
15 available data (see Sections 4.7.3.3 and 4.7.3.4). Specifically, tumors occur in rodent models in the
16 absence of data to identify hypothesized key events (e.g.. cytotoxicity). Furthermore, studies evaluating
17 the kinetics of 1.4-dioxane suggest that liver carcinogenicity is related to the accumulation of the parent
18 compound following metabolic saturation: however, the in vivo metabolism of 1.4-dioxane is unknown
19 (Section 3.3). nor are data available to determine the toxic moiety (i.e.. parent compound and/or
20 metabolite(s)) (see Section 4.7.3.1.1 and 3.3.). For kidney, lung, peritoneal (mesotheliomasX mammary
21 gland. Zvmbal gland, and subcutis tumors there are no available data regarding any hypothesized
22 carcinogenic MOA(s) for 1.4-dioxane.
23
24 The EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). recommend that the
25 method used to characterize and quantify cancer risk from a chemical is determined by what is known
26 about the MOA of the carcinogen and the shape of the cancer dose-response curve. The linear
27 extrapolation approach is used as a default option if the mode of carcinogenic action is unknown. A
28 nonlinear extrapolation approach can be used for cases with sufficient data to ascertain the mode of action
29 and to conclude that it is not linear at low doses. Also, nonlinear extrapolation having significant
30 biological support may be presented in addition to a linear approach when the available data and weight
31 of evidence support a nonlinear approach. In the case of 1.4-dioxane. there is insufficient biological
32 support to identify key events and to have reasonable confidence in the sequence of events and how they
33 relate to the development of tumors following exposure to 1.4-dioxane; thus, the data are not strong
34 enough to ascertain the mode of action applying the Agency's mode of action framework (U.S. EPA.
35 2005a. Therefore. EPA concluded that a default linear extrapolation should be utilized to estimate the
36 cancer risk estimates for inhalation or oral exposure to 1.4-dioxane.
37 IUR estimates were calculated using the following equation:
38 IUR = BMR / HEC
127
DRAFT - DO NOT CITE OR QUOTE
-------
5.4.4 Oral Slope Factor and Inhalation Unit Risk
5.4.4.1 Oral Slope Factor
1 The dichotomous models available in the Benchmark Dose Software (BMDS, version 2.1.1) were
2 fit to the incidence data for "either hepatocellular carcinoma or adenoma" in rats and mice, as well as
3 mammary and peritoneal tumors in rats exposed to 1,4-dioxane in the drinking water (Kano et al.. 2009;
4 NCI. 1978; Kociba et al.. 1974) (Table 5-7). Animal doses are used for BMD modeling and RED BMD
5 and BMDL values are calculated using the animal TWAs (Table 5-12) and a human BW of 70kg. Doses
6 associated with a BMR of 10% extra risk were calculated. BMDs and BMDLs from all models are
7 reported, and the output and plots corresponding to the best-fitting model are shown (Appendix D). When
8 the best-fitting model is not a multistage model, the multistage model output and plot are also provided
9 (Appendix D). A summary of the BMDS model predictions for the Kano et al. (2009). NCI (1978). and
10 Kociba et al. (1974) studies is shown in Table 5-12.
128
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-12 BMD HED and BMDLHED values from models fit to tumor incidence data for rats and
mice exposed to 1,4-dioxane in drinking water for 2 years and corresponding oral
CSFs
Study
Kano et al.
(2009)
Kociba et al.
(1974)
NCI (1978)
Gender/strain/species
Male F344/DuCrj rats'
Female F344/DuCrj ratsc
Male Crj:BDF1 micea
Female Crj:BDF1 micea
Female Crj:BDF1 mice0'8
Female Crj:BDF1 micea'T
Female F344/DuCrj ratsg
Male F344/DuCrj ratsg
Male F344/DuCrj ratsb
Female F344/DuCrj ratsd
Male and female (combined)
Sherman ratsg
Male and female (combined)
Sherman ratsb
Male Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female B6C3F-I micec
Male B6C3Fi miceh
Tumor type BMDHE°a BMDLHEDa Oral CSF
'^ (mg/kg-day) (mg/kg-day) (mg/kg-day)
Hepalocelluldi
Nasal
squamous cell
carcinoma
Peritoneal
mesothelioma
Mammary
gland adenoma
Nasal
squamous cell
carcinomas
Hepatocellular
carcinoma
Nasal
squamous cell
carcinomas
Hepatocellular
adenoma
Hepatocellular
adenoma or
carcinoma
17
19
5
0
3.
7,
94
91
26
40
448
290
16
40
28
23
87
.43
.84
.63
.83
22e
,51T
.84
.97
.09
.01
.24
.78
.10
.07
.75
.12
.98
14
14
2
0
2.
4.
70
68
21
20
340
240
10
25
18
9
35
.33
.43
.68
.55
12e
95T
.23
.85
.39
.35
.99
.31
.66
.82
.68
.75
.67
7.
6.
3.
1.
1.
4.
4.
2.
4.
9.
3.
5.
1.
2.
0
9
,7
0
0
0
,4
5
7
,9
,9
2
,4
,9
,4
,0
8
x 10
x 10
x 10
.18
.14
.10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
-6
-3
~2
-'3
-3
-3
-3
-4
-4
-'6
-3
-3
-A
-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 > 1.
eValues associated with a BMR of 30%.
Values associated with a BMR of 50%.
9Multistage model, degree of polynomial =3.
hGamma model.
1 The multistage model did not provide an adequate fit (as determined by AIC, p-value < 0.1, and
2 %2 p > 10.11) to the data for the incidence of hepatocellular adenoma or carcinoma in female mice
3 (Appendix D). The high dose was dropped for the female mouse liver tumor dataset in an attempt to
4 achieve an adequate fit; however, an adequate fit was still not achieved. Because the female mice were
5 clearly the most sensitive group tested, other BMD models were applied to the female mouse liver tumor
6 dataset to achieve an adequate fit. The log-logistic model was the only model that provided adequate fit
7 for this data set due to the steep rise in the dose-response curve (70% incidence at the low dose) followed
8 by a plateau at near maximal tumor incidence in the mid- and high-dose regions (82 and 92% incidence,
9 respectively). The predicted BMD10 and BMDL10 for the female mouse data are presented in Table 5-12,
10 as well as BMDnED and BMDLHED values associated with BMRs of 30 and 50% .
11 The multistage model also did not provide an adequate fit to mammary tumor incidence data for
12 the female rat or male rat peritoneal tumors. The predicted BMD10 and BMDL10 for female rat mammary
129
DRAFT - DO NOT CITE OR QUOTE
-------
1 tumors and male peritoneal tumors obtained from the log-logistic and probit models, respectively, are
2 presented in Table 5-12.
3 A comparison of the model estimates derived for rats and mice from the Kano et al. (2009). NCI
4 (1978). and Kociba et al. (1974) studies (Table 5-12) indicates that female mice are more sensitive to liver
5 carcinogenicity induced by 1,4-dioxane compared to other species or tumor types. The BMDL50 HED for
6 the female mouse data was chosen as the POD and the CSF of 0.10 (mg/kg-day)"1 was calculated as
7 follows:
8 CSF= — : = 0.10 (mg/kg-day)-1
4.95 mg/kg - day (BMDL50HED for female mice)
9 Calculation of a CSF for 1,4-dioxane is based upon the dose-response data for the most sensitive
10 species and gender.
5.4.4.2 Inhalation Unit Risk
11 As stated in Section 5.4.2.2. multiple tumor types have been observed in rats following inhalation
12 exposure to 1.4-dioxane. These data have been used to develop IUR estimates for 1.4-dioxane. The
13 multistage cancer models available in the BMDS (version 2.1.1) were fit to the incidence data for each
14 tumor type observed in rats exposed to 1.4-dioxane via inhalation (Kasai et al.. 2009) to determine the
15 degree (e.g.. 1st. 2nd. or 3rd) of the multistage model that best fit the data (details in Appendix H). A
16 Bayesian MCMC analysis was performed using WinBUGS to calculate the total tumor risk. For
17 comparative purposes only, a total tumor analysis was also performed with the BMDS (version 2.2Beta)
18 MSCombo model and yielded similar results (See Appendix FD. MSCombo is a new addition to BMDS
19 that allows for multi-tumor analysis. A summary of the BMDS model predictions for the Kasai et al.
20 (2009) study is shown in Table 5-13. Experimental exposure concentrations were used for BMP
21 modeling and continuous human equivalent exposures were calculated by adjusting for duration of
22 exposure (Table 5-13) and applying an appropriate DAF (see Section 5.2.3). In accordance with the U.S.
23 EPA Guidelines for Carcinogen Risk Assessment (2005a). the BMCLjn (lower bound on the concentration
24 estimated to produce a 10% increase in tumor incidence over background) was estimated for the
25 dichotomous incidence data and the results of the model that best characterized the cancer incidences
26 were selected. BMCs and BMCLs from all models are reported, and the output and plots corresponding to
27 the best-fitting model are shown (Appendix H).
28 The IUR estimates are provided in Table 5-13. Human equivalent risks estimated from the
29 individual rat tumor sites ranged from 2 x 1Q"7 to 2 x 1Q"6 (^g/m3)"1 (rounded to one significant figure).
30 The highest IUR (2 * 1Q"6 (f^g/m3)"1) corresponded to peritoneal mesotheliomas in male rats, and the
31 lowest IUR (2 x 10"7 (fig/m3)"1) corresponded to renal cell carcinoma and Zymbal gland adenomas in male
32 rats.
130
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-13 Dose-response modeling summary results for male rat tumors associated with
inhalation exposure to 1,4-dioxane for
Tumor Type3
Nasal cavity squamous cell
carcinoma
Hepatocellular adenoma or
carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland
fibroadenoma
Zymbal gland adenoma
Bayesian Total Tumor Analysis'
Multistage
Model
Degree13
1
1
3
1
1
3
2 years
Point of Departure0
Bioassay Exposure
Concentration (ppm)
BMCio
1107
252.8
1355
82.21
1635
1355
141.8
BMCLio
629.9
182.3
1016
64.38
703.0
1016
81.91
HEC
(mg/m3)d
Ei
i.
IUR
stimate6
BMCio BMCLio »•»•""''
712.3
162.7
872
52.89
1052
872
91.21
405.3
117.3
653.7
41.42
452.4
653.7
2
8,
1
2
2
1
1
5
5
5
4
2
5
x10-7
x10-7
X10"'
xiQ-B
x10-7
x10"
x1Q-B
"Tumor incidence data from Kasai et al. (2009).
""Best-fitting multistage model degree (p>0.1, lowest AIC). See Appendix G for modeling details.
CBMC = 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.
The inhalation unit risk (|jg/m3)-1 was derived from the BMCL10, 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 Bayesian analysis using WinBUGS. Additionally, for comparative purposes only, total tumor analysis
was performed with the draft BMDS (version 2.2Beta) MSCombo model and yielded similar results (See Appendix H).
1
2 The carcinogenic MOA(s) by which 1.4-dioxane produces liver, nasal, kidney, peritoneal
3 (mesotheliomas). mammary gland. Zymbal gland, and subcutis tumors is unknown. Several hypothesized
4 MOA(s) have been proposed for liver and nasal tumors although these MOA(s) are not supported by the
5 available data (see Sections 4.7.3.3 and 4.7.3.4). Specifically, tumors occur in rodent models in the
6 absence of data to identify hypothesized key events (e.g.. cytotoxicity). Furthermore, studies evaluating
7 the kinetics of 1.4-dioxane suggest that liver carcinogenicity is related to the accumulation of the parent
8 compound following metabolic saturation: however, the in vivo metabolism of 1.4-dioxane is unknown
9 (Section 3.3). nor are data available to determine the toxic moiety (i.e.. parent compound and/or
10 metabolite^)) (see Section 4.7.3.1.1 and 3.3.). For kidney, lung, peritoneal (mesotheliomasX mammary
11 gland. Zvmbal gland, and subcutis tumors there are no available data regarding any hypothesized
12 carcinogenic MOA(s) for 1.4-dioxane.
13
14 The EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). recommend that the
15 method used to characterize and quantify cancer risk from a chemical is determined by what is known
16 about the MOA of the carcinogen and the shape of the cancer dose-response curve. The linear
17 extrapolation approach is used as a default option if the mode of carcinogenic action is unknown. A
18 nonlinear extrapolation approach can be used for cases with sufficient data to ascertain the mode of action
19 and to conclude that it is not linear at low doses. Also, nonlinear extrapolation having significant
20 biological support may be presented in addition to a linear approach when the available data and weight
21 of evidence support a nonlinear approach. In the case of 1.4-dioxane. there is insufficient biological
22 support to identify key events and to have reasonable confidence in the sequence of events and how they
131
DRAFT - DO NOT CITE OR QUOTE
-------
1 relate to the development of tumors following exposure to 1.4-dioxane; thus, the data are not strong
2 enough to ascertain the mode of action applying the Agency's mode of action framework (U.S. EPA.
3 2005a. Therefore. EPA concluded that a default linear extrapolation should be utilized to estimate the
4 cancer risk estimates for inhalation or oral exposure to 1.4-dioxane.
5 Given the multiplicity of tumor sites, basing the inhalation unit risk on one tumor site may
6 underestimate the carcinogenic potential of 1.4-dioxane. Consistent with recommendations of the NRC
7 (1994) and the EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) the total risk and
8 upper bound risk for all tumor sites in male F344 rats was estimated. This estimate of total risk describes
9 the risk of developing any combination of the tumor types considered. As shown in Table 5-13. the
10 resulting inhalation unit risk for all tumor types in male F344 rats was 5 x 1Q"6 (f^g/m3)"1 . Consideration of
11 all tumor sites approximately doubled the unit risk compared to the highest unit risk associated with any
12 individual tumor type. 2 x 10"6 (fig/m3)"1 for male peritoneal mesotheliomas.
13 The HEC BMCLui for the combined tumor estimate in male rats was chosen as the POD and the
14 lURof 5 x 1Q-6 (^g/nvV was calculated as follows:
IUR (mg/m3)-1 = °'10, 3 = 0.005 (mg/m3)-1
20.2 mg/m
15 lURCwg/m3)-1 =0.005 (mg/m3)-1 x *fg = 5 xlO'6(^g/m3)-
10 mg
IUR Cwg/m3)-1 = 5
16 Based on the analysis discussed above, the recommended upper bound estimate on human extra
17 cancer risk from continuous lifetime exposure to 1.4-dioxane is 5 x IP"6 (fig/m3)"1. The IUR reflects the
18 exposure-response relationships for the multiple tumor sites in male F344 rats.
5.4.5 Previous Cancer Assessment
19 A previous cancer assessment was posted for 1,4-dioxane on IRIS in 1988. 1,4-Dioxane was
20 classified as a Group B2 Carcinogen (probable human carcinogen; sufficient evidence from animal
21 studies and inadequate evidence or no data from human epidemiology studies (U.S. EPA. 1986aV) based
22 on the induction of nasal cavity and liver carcinomas in multiple strains of rats, liver carcinomas in mice,
23 and gall bladder carcinomas in guinea pigs. An oral CSF of 0.011 (mg/kg-day)"1 was derived from the
24 tumor incidence data for nasal squamous cell carcinoma in male rats exposed to 1,4-dioxane in drinking
25 water for 2 years (NCI. 1978). The linearized multistage extra risk procedure was used for linear low dose
26 extrapolation. An inhalation unit risk was not previously derived.
132
DRAFT - DO NOT CITE OR QUOTE
-------
5.5 Uncertainties in Cancer Risk Values
1 As in most risk assessments, extrapolation of study data to estimate potential risks to human
2 populations from exposure to 1,4-dioxane has engendered some uncertainty in the results. Several types
3 of uncertainty may be considered quantitatively, but other important uncertainties cannot be considered
4 quantitatively. Thus an overall integrated quantitative uncertainty analysis is not presented. However, the
5 sources of uncertainty and assumptions are described below and in Table 5-14.
5.5.1 Sources of Uncertainty
5.5.1.1 Choice of Low-Dose Extrapolation Approach
6 The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
7 1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible MOA(s)
8 for the observed tumors. The MOA is a key consideration in clarifying how risks should be estimated for
9 low-dose exposure. Exposure to 1,4-dioxane has been observed in animal models to induce multiple
10 tumor types, including liver adenomas and carcinomas, nasal carcinomas, mammary adenomas and
11 fibroadenomas, and mesotheliomas of the peritoneal cavity (Kano et al.. 2009; Kasai et al.. 2009; JBRC.
12 1998; NCI. 1978; Kociba et al.. 1974). MOA information that is available for the carcinogenicity of
13 1,4-dioxane has largely focused on liver adenomas and carcinomas, with little or no MOA information
14 available for the remaining tumor types. In Section 4.7.3, hypothesized MOAs were explored for
15 1,4-dioxane. Information that would provide sufficient support for any MOA is not available. In the
16 absence of a MOA(s) for the observed tumor types, a linear low-dose extrapolation approach was used to
17 estimate human carcinogenic risk associated with 1,4-dioxane exposure.
18 It is not possible to predict how additional MOA information would impact the dose-response
19 assessment for 1,4-dioxane because of the variety of tumors observed and the lack of data on how
20 1,4-dioxane or a metabolite thereof, interacts with cells starting the progression to the observed tumors.
21 In general, the Agency has preferred to use the multistage model for analyses of tumor incidence
22 and related endpoints because they have a generic biological motivation based on long-established
23 mathematical models such as the Moolgavkar-Venzon-Knudsen (MVK) model.
24 The MVK model does not necessarily characterize all modes of tumor formation, but it is a
25 starting point for most investigations and, much more often than not, has provided at least an adequate
26 description of tumor incidence data.
27 The multistage cancer model provided adequate fits for the tumor incidence data following a
28 2-year inhalation exposure to 1.4-dioxane by male rats (Kasai et al.. 2009). In the studies evaluated for the
29 oral cancer assessment (Kano et al.. 2009; NCI. 1978; Kocibaetal. 1974). the multistage model provided
30 good descriptions of the incidence of a few tumor types in male (nasal cavity) and female (hepatocellular
133
DRAFT - DO NOT CITE OR QUOTE
-------
1 and nasal cavity) rats and in male mice (hepatocellular) exposed to 1,4-dioxane (Appendix D for details).
2 The multistage model did not provide an adequate fit for the female mouse liver tumor dataset based upon
3 the following (U.S. EPA. 2000a):
4 • Goodness-of-fit/>-value was not greater than 0.10;
5 • Akaike's Information Criterion (AIC) was larger than other acceptable models;
6 • Data deviated from the fitted model, as measured by their %2 residuals (values
7 were greater than an absolute value of one).
8 BMDS software typically implements the guidance in the external peer review draft BMD
9 technical guidance document (U.S. EPA. 2000a) by imposing constraints on the values of certain
10 parameters of the models. When these constraints were imposed, the multistage model and most other
11 models did not fit the incidence data for female mouse liver adenomas or carcinomas.
12 The log-logistic model was selected because it provides an adequate fit for the female mouse data
13 (Kano et al.. 2009). A BMR of 50% was used because it is proximate to the response at the lowest dose
14 tested and the BMDL50 HED was derived by applying appropriate parameter constraints, consistent with
15 recommended use of BMDS in the BMD technical guidance document (U.S. EPA. 2000a).
16 The human equivalent oral CSFs estimated from tumor datasets with statistically significant
17 increases ranged from 4.2 x 10"4to 0.18 permg/kg-day (Table 5-12), arange of aboutthree orders of
18 magnitude, with the extremes coming from the combined male and female rat data for hepatocellular
19 carcinomas (Kocibaet al., 1974) and the female mouse combined liver adenoma and carcinomas (Kano et
20 al.. 2009).
5.5.1.2 Dose Metric
21 1,4-Dioxane is known to be metabolized in vivo. However, it is unknown whether a metabolite or
22 the parent compound, or some combination of parent compound and metabolites, is responsible for the
23 observed toxicity. If the actual carcinogenic moiety is proportional to administered exposure, then use of
24 administered exposure as the dose metric is the least biased choice. On the other hand, if this is not the
25 correct dose metric, then the impact on the CSF is unknown.
5.5.1.3 Cross-Species Scaling
26 For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied to
27 address toxicological equivalence of internal doses between each rodent species and humans, consistent
28 with the 2005 Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). It is assumed that equal
29 risks result from equivalent constant lifetime exposures.
30 Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
31 in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
134
DRAFT - DO NOT CITE OR QUOTE
-------
1 However, since similar cell types are prevalent throughout the respiratory tract of both rats and humans.
2 the tumors are considered biologically plausible and relevant to humans.
5.5.1.4 Statistical Uncertainty at the POD
3 Parameter uncertainty can be assessed through confidence intervals. Each description of
4 parameter uncertainty assumes that the underlying model and associated assumptions are valid. For the
5 log-logistic model applied to the female mouse data following oral exposure, there is a reasonably small
6 degree of uncertainty at the 10% excess incidence level (the POD for linear low-dose extrapolation). For
7 the multistage model applied for the male rat inhalation dataset. there is a reasonably small degree of
8 uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
5.5.1.5 Bioassay Selection
9 The study by Kano et al. (2009) was used for development of an oral CSF. This was a
10 well-designed study, conducted in both sexes in two species (rats and mice) with a sufficient number
11 (N=50) of animals per dose group. The number of test animals allocated among three dose levels and an
12 untreated control group was adequate, with examination of appropriate toxicological endpoints in both
13 sexes of rats and mice. Alternative bioassays (NCI. 1978; Kocibaetal.. 1974) were available and were
14 fully considered for the derivation of the oral CSF.
15 The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
16 well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
17 Three dose levels plus an untreated control group were examined following exposure to 1.4-dioxane via
18 inhalation for 2 years.
5.5.1.6 Choice of S pecies/Gender
19 The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the female
20 mouse, which was shown to be more sensitive than male mice or either sex of rats to the carcinogenicity
21 of 1,4-dioxane. While all data, both species and sexes reported from the Kano et al. (2009) study, were
22 suitable for deriving an oral CSF, the female mouse data represented the most sensitive indicator of
23 carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day or 10 mg/kg-day [HED])
24 resulted in a considerable and significant increase in combined liver adenomas and carcinomas observed.
25 Additional testing of doses within the range of control and the lowest dose (66 mg/kg-day or
26 10 mg/kg-day [HED]) could refine and reduce uncertainty for the oral CSF.
27 A personal communication from Dr. Yamazaki (2006) provided that the survival of mice was
28 particularly low in high-dose females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-, and high-dose
29 groups, respectively). These deaths occurred primarily during the second year of the study. Female mouse
135
DRAFT - DO NOT CITE OR QUOTE
-------
1 survival at 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-, mid-, and high-dose groups,
2 respectively (Yamazaki. 2006). Furthermore, these deaths were primarily tumor related. Liver tumors
3 were listed as the cause of death for 1/21, 2/21, 8/33, and 31/45 of the pretermination deaths in control,
4 low-, mid- and, high-dose female Crj:BDFl mice (Yamazaki. 2006). Therefore, because a number of the
5 deaths in female mice were attributed to liver tumors, this endpoint and species was still considered to be
6 relevant for this analysis; however, the high mortality rate does contribute uncertainty.
7 Additionally, the incidence of hepatocellular adenomas and carcinomas in historical controls was
8 evaluated with the data from Kano et al. (2009). Katagiri et al. (1998) summarized the incidence of
9 hepatocellular adenomas and carcinomas in control male and female BDF1 mice from ten 2-year
10 bioassays at the JBRC. For female mice, out of 499 control mice, the incidence rates were 4.4% for
11 hepatocellular adenomas and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported a 10%
12 incidence rate for hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in
13 control female BDF1. These incidence rates are near the historical control values and thus are appropriate
14 for consideration in this assessment.
15 Male F344 rat data were used to estimate risk following inhalation of 1,4-dioxane. Kasai et al.
16 (2008) showed that male rats were more sensitive than female rats to the effects of 1,4-dioxane following
17 inhalation; therefore, male rats were chosen to be studies in the 2-year bioassay conducted by the same
18 laboratory (Kasai et al.. 2009).
5.5.1.7 Relevance to Humans
19 The derivation of the oral CSF is derived using the tumor incidence in the liver of female mice. A
20 thorough review of the available toxicological data available for 1,4-dioxane provides no scientific
21 justification to propose that the liver adenomas and carcinomas observed in animal models due to
22 exposure to 1,4-dioxane are not relevant to humans. As such, liver adenomas and carcinomas were
23 considered relevant to humans due to exposure to 1,4-dioxane.
24 The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
25 rats. There is no information on 1,4-dioxane to indicate that the observed rodent tumors are not relevant to
26 humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
27 rodents and humans. In the absence of information to indicate otherwise and considering similar cell types
28 are prevalent throughout the respiratory tract of rats and humans, the nasal liver, renal peritoneal
29 mammary gland. Zymbal gland and subcutis tumors were considered relevant to humans.
5.5.1.8 Human Population Variability
30 The extent of inter-individual variability in 1,4-dioxane metabolism has not been characterized. A
31 separate issue is that the human variability in response to 1,4-dioxane is also unknown. Data exploring
32 whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life stages are unavailable.
136
DRAFT - DO NOT CITE OR QUOTE
-------
1 This lack of understanding about potential differences in metabolism and susceptibility across exposed
2 human populations thus represents a source of uncertainty. Also, the lack of information linking a MOA
3 for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.
Table 5-14 Summary of uncertainty in the 1,4-dioxane cancer risk estimation
Consideration/
approach
Low-dose
extrapolation
procedure
Dose metric
Cross-species
scaling
Bioassay
Species /gender
combination
Human
relevance of
mouse tumor
data
Human
population
variability in
metabolism and
response/
sensitive
subpopulations
Potential Impact
Departure from
EPA's Guidelines for
Carcinogen Risk
Assessment POD
paradigm, if justified,
could | or t unit risk
an unknown extent
Alternatives could f
or | CSF by an
unknown extent
Alternatives could J,
orf CSF [e.g.,
3.5-fold I (scaling by
BW) or } twofold
(scaling by BW067 )]
Alternatives could f
or 1 cancer potency
by an unknown
extent
Human risk could J, or
f, depending on
relative sensitivity
If rodent tumors
proved not to be
relevant to humans,
unit risk would not
apply i.e., could J,
CSF
Low-dose risk f or J,
to an unknown extent
Decision
Log-logistic model
to determine POD,
for CSF; Bavesian
multistage modeling
for IUR: linear
low-dose
extrapolation from
POD
Used administered
exposure
BW075 (default
approach)
CSF (Kano et al..
2009): IUR (Kasai
et al., 2009)
Female mouse
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
Justification
A linear low-dose extrapolation approach was used to
estimate human carcinogenic risk associated with
1,4-dioxane exposure. Where data are insufficient to
ascertain the MOA, EPA's 2005 Guidelines for
Carcinogen Risk Assessment recommend application
of a linear low-dose extrapolation approach.
Experimental evidence supports a role for metabolism
in toxicity, but it is unclear if the parent compound,
metabolite or both contribute to 1,4-dioxane toxicity.
There are no data to support alternatives. BWU 'b
scaling was used to calculate equivalent cumulative
exposures for estimating equivalent human risks.
PBPK modeling was conducted but not deemed
suitable for interspecies extrapolation.
Alternative bioassays were available and considered
for derivation of oral CSF and inhalation IUR.
There are no MOA data to guide extrapolation
approach for any choice. It was assumed that
humans are as sensitive as the most sensitive rodent
gender/species tested; true correspondence is
unknown. Calculation of the CSF for 1,4-dioxane was
based on dose-response data from the most sensitive
species and gender. The carcinogenic response
occurs across species.
1,4-dioxane is a multi-site carcinogen in rodents and
the MOA(s) is unknown; carcinogenicity observed in
the rodent studies is considered relevant to human
exposure.
No data to support range of human
variability/sensitivity, including whether children are
more sensitive.
137
DRAFT - DO NOT CITE OR QUOTE
-------
6 MAJOR CONCLUSIONS IN THE
CHARACTERIZATION OF HAZARD AND DOSE
RES PONS E
6.1 Human Hazard Potential
1 1,4-Dioxane is absorbed rapidly following oral and inhalation exposure, with much less
2 absorption occurring from the dermal route. 1,4-Dioxane is primarily metabolized to HEAA, which is
3 excreted in the urine. Liver, kidneyi and nasal toxicity are the primary noncancer health effects
4 associated with exposure to 1,4-dioxane in humans and laboratory animals. Several fatal cases of
5 hemorrhagic nephritis and centrilobular necrosis of the liver were related to occupational exposure (i.e.,
6 inhalation and dermal contact) to 1,4-dioxane (Johnstone. 1959; Barber. 1934). Neurological changes
7 were also reported in one case, including headache, elevation in blood pressure, agitation and restlessness,
8 and coma (Johnstone. 1959). Perivascular widening was observed in the brain of this worker, with small
9 foci of demyelination in several regions (e.g., cortex, basal nuclei). Severe liver and kidney degeneration
10 and necrosis were observed frequently in acute oral and inhalation studies (> 1,000 mg/kg-day oral, >
11 1,000 ppm inhalation) (JBRC. 1998: Drewetal.. 1978: David. 1964: Kestenetal.. 1939: Laug et al..
12 1939: Schrenk and Yant. 1936: deNavasquez. 1935: Fairlev et al.. 1934).
13 Liver and kidney toxicity were the primary noncancer health effects of subchronic and chronic
14 oral exposure to 1,4-dioxane in animals. Hepatocellular degeneration and necrosis were observed
15 (Kociba et al.. 1974) and preneoplastic changes were noted in the liver following chronic administration
16 of 1,4-dioxane in drinking water (Kano et al.. 2008: JBRC. 1998: Argus et al.. 1973) Liver and kidney
17 toxicity appear to be related to saturation of clearance pathways and an increase in the 1,4-dioxane
18 concentration in the blood (Kociba. etal.. 1974). Kidney damage was characterized by degeneration of
19 the cortical tubule cells, necrosis with hemorrhage, and glomerulonephritis (Argus, etal.. 1965: Argus, et
20 al.. 1973: Fairlev. etal.. 1934: Kociba. et al.. 1974: NCI. 1978). In chronic inhalation studies conducted in
21 rats, nasal and liver toxicity were the primary noncancer health effects. Degeneration of nasal tissue (i.e.
22 metaplasia, hyperplasia. atrophy, hydropic change, and vacuolic change) and preneoplastic cell
23 proliferation were observed in the nasal cavity following inhalation exposure to 1.4-dioxane for 2 years
24 (Kasai. et al.. 2009). Liver toxicity was described as necrosis of the centrilobular region and preneoplastic
25 changes were noted as well.
26 Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
27 pigs (Argus, et al.. 1965: Argus, etal.. 1973: Hoch-Ligeti & Argus. 1970: Hoch-Ligeti. et al.. 1970:
28 JBRC. 1998: Kano. et al.. 2009: Kasai. et al.. 2009: Kociba. et al.. 1974: NCI. 1978: Torkelson. et al..
29 1974). Liver tumors (hepatocellular adenomas and carcinomas) have been observed following drinking
30 water exposure in several species and strains of rats, mice, and guinea pigs and following inhalation
31 exposure in rats. Nasal (squamous cell carcinomas), peritoneal, mammary. Zvmbal gland, and
32 subcutaneous tumors were also observed in rats, but were not seen in mice. With the exception of the NCI
138
DRAFT - DO NOT CITE OR QUOTE
-------
1 (1978) study, the incidence of nasal cavity tumors was generally lower than that of tumors observed in
2 other tissues of the same study population.
3 Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is "likely
4 to be carcinogenic to humans" based on evidence of multiple tissue carcinogenicity in several 2-year
5 bioassays conducted in three strains of rats, two strains of mice, and in guinea pigs (Argus, etal.. 1965;
6 Argus, etal.. 1973; Hoch-Ligeti & Argus. 1970; Hoch-Ligeti. et al.. 1970; JBRC. 1998; Kano. et al..
7 2009; Kasai. et al.. 2009; Kociba. et al.. 1974; NCI. 1978). Studies in humans found no conclusive
8 evidence for a causal link between occupational exposure to 1,4-dioxane and increased risk for cancer;
9 however, only two studies were available and these were limited by small cohort size and a small number
10 of reported cancer cases (Buffler. etal.. 1978; Thiess. etal.. 1976).
11 The available evidence is inadequate to establish a MOA by which 1,4-dioxane induces tumors in
12 rats and mice. The genotoxicity data for 1,4-dioxane is generally characterized as negative, although
13 several studies may suggest the possibility of genotoxic effects (Galloway, et al.. 1987; Kitchin & Brown.
14 1990; Mirkova. 1994; Morita & Hayashi. 1998; Roy, et al.. 2005). A MOA hypothesis for liver tumors
15 involving sustained proliferation of spontaneously transformed liver cells has some support by evidence
16 that suggests 1,4-dioxane is a tumor promoter in mouse skin and rat liver bioassays (King, etal.. 1973;
17 Lundberg. et al.. 1987). Some dose-response and temporal evidence support the occurrence of cell
18 proliferation and hyperplasia prior to the development of liver tumors (JBRC. 1998; Kociba. et al.. 1974).
19 However, the dose-response relationship for the induction of hepatic cell proliferation has not been
20 characterized, and it is unknown if it would reflect the dose-response relationship for liver tumors in the
21 2-year rat and mouse studies. Conflicting data from rat and mouse bioassays (JBRC. 1998; Kociba. et al..
22 1974) suggest that cytotoxicity is not a required precursor event for 1,4-dioxane-induced cell
23 proliferation. Liver tumors were observed in female rats and female mice in the absence of lesions
24 indicative of cytotoxicity (JBRC. 1998; Kano. et al.. 2009; NCI. 1978). Data regarding a plausible dose
25 response and temporal progression from cytotoxicity to cell proliferation and eventual liver tumor
26 formation are not available. Hypothesized MOAs by which 1.4-dioxane induces tumors in other organ
27 systems such as the respiratory system are uncertain (See Section 4.7.3).
6.2 DOSE RESPONSE
6.2.1 Noncancer/Oral
28 The RfD of 3 x 10"2 mg/kg-day was derived based on liver and kidney toxicity in rats exposed to
29 1,4-dioxane in the drinking water for 2 years (Kociba. et al.. 1974). This study was chosen as the
30 principal study because it provides the most sensitive measure of adverse effects by 1,4-dioxane. The
31 incidence of liver and kidney lesions was not reported for each dose group. Therefore, BMD modeling
32 could not be used to derive a POD. Instead, the RfD is derived by dividing the NOAEL of 9.6 mg/kg-day
33 by a composite UF of 300 (factors of 10 for animal-to-human extrapolation and interindividual
34 variability, and an UF of 3 for database deficiencies). Information was unavailable to quantitatively
35 assess toxicokinetic or toxicodynamic differences between animals and humans and the potential
36 variability in human susceptibility; thus, the interspecies and intraspecies uncertainty factors of 10 were
139
DRAFT - DO NOT CITE OR QUOTE
-------
1 applied. In addition, a threefold database uncertainty factor was applied due to the lack of information
2 addressing the potential reproductive toxicity associated with 1,4-dioxane.
3 The overall confidence in the RfD is medium. Confidence in the principal study (Kociba. et al..
4 1974) is medium. Confidence in the database is medium due to the lack of a multigeneration reproductive
5 toxicity study. Reflecting medium confidence in the principal study and medium confidence in the
6 database, confidence in the RfD is medium.
6.2.2 Noncancer/lnhalation
7 The RfC of 3 x 1Q"2 mg/m3 was derived based on co-critical effects of olfactory epithelium
8 atrophy and respiratory metaplasia in rats exposed for 2 years to 1.4-dioxane via inhalation (Kasai. et al..
9 2009). This study was chosen as the principal study because it provides an adequate study design and the
10 most sensitive measure of adverse effects by 1.4-dioxane. The POD was derived using the LOAEL for
11 olfactory epithelium atrophy and respiratory metaplasia in male rats ( Kasai et al. 2009). A composite UF
12 of 1.000 was applied, consisting of factors of 10 for a LOAEL-to NOAEL extrapolation. 10 for
13 interindividual variability. 3 for animal-to-human extrapolation, and 3 for database deficiencies.
14 The overall confidence in the RfC is medium. Confidence in the principal study (Kasai. et al..
15 2009) is medium. Confidence in the database is medium due to the lack of supporting studies and a
16 multigeneration reproductive toxicity study. Reflecting medium confidence in the principal study and
17 medium confidence in the database, the confidence in the RfC is medium.
6.2.3 Cancer
18 Under EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is
19 "likely to be carcinogenic to humans" by all routes of exposure. This descriptor is based on evidence of
20 carcinogenicity from animal studies.
6.2.3.1 Oral
21 An oral CSF for 1,4-dioxane of 0.10 (mg/kg-day)"1 was based on liver tumors in female mice
22 from a chronic study (Kano. et al.. 2009). The available data indicate that the MOA(s) by which
23 1,4-dioxane induces peritoneal, mammary, or nasal tumors in rats and liver tumors in rats and mice is
24 unknown (see Section 4.7.3 for a more detailed discussion of 1,4-dioxane's hypothesized MOAs).
25 Therefore, based on the U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). a
26 linear low dose extrapolation was used. The POD was calculated by curve fitting the animal experimental
27 dose-response data from the range of observation and converting it to a HED (BMDL50 HED of
28 4.95 mg/kg-day).
29 The uncertainties associated with the quantitation of the oral CSF are discussed below.
6.2.3.2 Inhalation
30 The IUR for 1.4-dioxane of 5 x 10"6 (fig/m3)"1 was based on a chronic inhalation study conducted
31 by Kasai et al. (2009). Statistically significant increases in tumor incidence and positive dose-response
140
DRAFT - DO NOT CITE OR QUOTE
-------
1 trends were observed at multiple sites in the male rat including the nasal cavity (squamous cell
2 carcinoma), liver (adenoma), peritoneal (mesothelioma). and the subcutis (fibroma). Statistically
3 significant dose-response trends were also observed in the kidney (carcinoma), mammary gland
4 (fibroadenoma). and the Zymbal gland (adenoma). The available data indicate that the MOA(s) by which
5 1.4-dioxane induces tumors in rats is unknown (see Section 4.7.3 for a more detailed discussion of
6 1.4-dioxane's hypothesized MOAs). Therefore, based on the EPA's Guidelines for Carcinogen Risk
1 Assessment (U.S. EPA. 2005a). a linear low dose extrapolation was used. A Bayesian approach (see
8 Section 5.4.3.2 and Appendix G for details) was used to calculate the POD for the total tumor risk
9 following inhalation of 1.4-dioxane. The POD was calculated by curve fitting the animal experimental
10 dose-response data from the range of observation and converting it to a continuous human equivalent
11 exposure.
12 The uncertainties associated with the quantitation of the IUR are discussed below.
6.2.3.3 Choice of Low-Dose Extrapolation Approach
13 The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
14 1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible MOA(s)
15 for the observed tumors. The MOA is a key consideration in clarifying how risks should be estimated for
16 low-dose exposure. Exposure to 1,4-dioxane has been observed in animal models to induce multiple
17 tumor types, including liver adenomas and carcinomas, nasal carcinomas, mammary adenomas and
18 fibroadenomas, and mesotheliomas of the peritoneal cavity (Kano. et al., 2009). MOA information that is
19 available for the carcinogenicity of 1,4-dioxane has largely focused on liver adenomas and carcinomas,
20 with little or no MOA information available for the remaining tumor types. In Section 4.7.3,
21 hypothesized MOAs were explored for 1,4-dioxane. Data are not available to support a carcinogenic
22 MOA for 1,4-dioxane. In the absence of a MOA(s) for the observed tumor types associated with
23 exposure to 1,4-dioxane, a linear low-dose extrapolation approach was used to estimate human
24 carcinogenic risk associated with 1,4-dioxane exposure.
25 In general, the Agency has preferred to use the multistage model for analyses of tumor incidence
26 and related endpoints because they have a generic biological motivation based on long-established
27 mathematical models such as the MVK model. The MVK model does not necessarily characterize all
28 modes of tumor formation, but it is a starting point for most investigations and, much more often than not,
29 has provided at least an adequate description of tumor incidence data.
30 The multistage cancer model provided adequate fits for the tumor incidence data following a 2-
31 year inhalation exposure to 1.4-dioxane by male rats (Kasai. et al.. 2009). However, in the studies
32 evaluated for the oral cancer assessment (Kano. et al., 2009; Kociba. et al., 1974; NCI. 1978) the
33 multistage model provided good descriptions of the incidence of a few tumor types in male (nasal cavity)
34 and female (hepatocellular and nasal cavity) rats and in male mice (hepatocellular) exposed to
35 1,4-dioxane (see Appendix D for details). However, the multistage model did not provide an adequate fit
36 for female mouse liver tumor dataset based upon the following (U.S. EPA. 2000a):
• Goodness-of-fit />-value was not greater than 0.10;
141
DRAFT - DO NOT CITE OR QUOTE
-------
• AIC was larger than other acceptable models;
• Data deviated from the fitted model, as measured by their %2 residuals (values were greater than an absolute
value of one).
1 BMDS software typically implements the guidance in the BMD technical guidance document
2 (U.S. EPA. 2000a) by imposing constraints on the values of certain parameters of the models. When
3 these constraints were imposed, the multistage model and most other models did not fit the incidence data
4 for female mouse liver adenomas or carcinomas.
5 The log-logistic model was selected because it provides an adequate fit for the female mouse data
6 (Kano. et al.. 2009). A BMR of 50% was used because it is proximate to the response at the lowest dose
7 tested and the BMDL50 was derived by applying appropriate parameter constraints, consistent with
8 recommended use of BMDS in the BMD technical guidance document (U.S. EPA. 2000a).
9 The human equivalent oral CSF estimated from liver tumor datasets with statistically significant
10 increases ranged from 4.2 x 10"4to 1.0 x 10"1 per mg/kg-day, a range of about three orders of magnitude,
11 with the extremes coming from the combined male and female data for hepatocellular carcinomas
12 (Kociba. et al.. 1974) and the female mouse liver adenoma and carcinoma dataset (Kano. et al.. 2009).
6.2.3.4 Dose Metric
13 1,4-Dioxane is known to be metabolized in vivo. However, evidence does not exist to determine
14 whether the parent compound, metabolite(s), or a combination of the parent compound and metabolites is
15 responsible for the observed toxicity following exposure to 1,4-dioxane. If the actual carcinogenic moiety
16 is proportional to administered exposure, then use of administered exposure as the dose metric is the least
17 biased choice. On the other hand, if this is not the correct dose metric, then the impact on the CSF is
18 unknown.
6.2.3.5 Cross-Species Scaling
19 For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied to
20 address toxicological equivalence of internal doses between each rodent species and humans, consistent
21 with the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). It is assumed that equal risks
22 result from equivalent constant lifetime exposures.
23 Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
24 in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
6.2.3.6 Statistical Uncertainty at the POD
25 Parameter uncertainty can be assessed through confidence intervals. Each description of
26 parameter uncertainty assumes that the underlying model and associated assumptions are valid. For the
27 log-logistic model applied to the female mouse data following oral exposure, there is a reasonably small
28 degree of uncertainty at the 50% excess incidence level (the POD for linear low-dose extrapolation). For
29 the multistage model applied for the male rat inhalation dataset. there is a reasonably small degree of
30 uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
142
DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.7 Bioassay Selection
1 The study by Kano et al. (2009) was used for development of an oral CSF. This was a well-
2 designed study, conducted in both sexes in two species (rats and mice) with a sufficient number (N=50)
3 of animals per dose group. The number of test animals allocated among three dose levels and an
4 untreated control group was adequate, with examination of appropriate toxicological endpoints in both
5 sexes of rats and mice. Alternative bioassays (Kociba. et al.. 1974; NCI. 1978) were available and were
6 fully considered for the derivation of the oral CSF.
7 The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
8 well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
9 Three dose levels plus an untreated control group were examined following exposure to 1.4-dioxane via
10 inhalation for 2 years.
6.2.3.8 Choice of S pecies/Gender
11 The oral CSF for 1,4-dioxane was derived using the tumor incidence data for the female mouse,
12 which was thought to be more sensitive than male mice or either sex of rats to the carcinogenicity of
13 1,4-dioxane. While all data, from both species and sexes reported from the Kano et al. (2009) study, were
14 suitable for deriving an oral CSF, the female mouse data represented the most sensitive indicator of
15 carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day [animal dose] or
16 10 mg/kg-day [HED]) observed a considerable and significant increase in combined liver adenomas and
17 carcinomas. Additional testing of doses within the range of control and the lowest dose (66 mg/kg-day
18 [animal dose] or 10 mg/kg-day [FiED]) could refine and reduce uncertainty for the oral CSF.
19 Male F344 rat data were used to estimate risk following inhalation of 1.4-dioxane. Kasai et al.
20 (2008) showed that male rats were more sensitive than female rats to the effects of 1.4-dioxane following
21 inhalation; therefore, male rats were studied in the 2-year bioassay conducted by the same laboratory
22 (Kasai. et al.. 2009).
6.2.3.9 Relevance to Humans
23 The oral CSF was derived using the tumor incidence in the liver of female mice. A thorough
24 review of the available toxicological data available for 1,4-dioxane provides no scientific justification to
25 propose that the liver adenomas and carcinomas observed in animal models following exposure to
26 1,4-dioxane are not plausible in humans. Liver adenomas and carcinomas were considered plausible
27 outcomes in humans due to exposure to 1,4-dioxane.
28 The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
29 rats. There is no information on 1.4-dioxane to indicate that the observed rodent tumors are not relevant to
30 humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
31 rodents and humans.
143
DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.10 Human Population Variability
1 The extent of inter-individual variability in 1,4-dioxane metabolism has not been characterized.
2 A separate issue is that the human variability in response to 1,4-dioxane is also unknown. Data exploring
3 whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life stages is unavailable.
4 This lack of understanding about potential differences in metabolism and susceptibility across exposed
5 human populations thus represents a source of uncertainty. Also, the lack of information linking a MOA
6 for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.
144
DRAFT - DO NOT CITE OR QUOTE
-------
REFERENCES
Agrawal. A and Shapiro. B. (2000). Differential expression of gender-dependent hepatic isoforms of cvtochrome P-
450 by pulse signals in the circulating masculine episodic growth hormone profile of the rat. J Pharmacol
Exp Ther 292: 228-237.
Amendments to the Clean Air Act. Sec. 604. Phase-out of production and consumption of class I substances. 42.
U.S.C. §76710(1990).
Andersen. M: Clewell. H. Ill: Gargas. M: Smith. F: Reitz. R. (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. M: Arcos. J: Hoch-Ligeti. C. (1965). Studies on the carcinogenic activity of protein-denaturing agents:
Hepatocarcinogenicity of dioxane. J Natl Cancer Inst 35: 949-958.
Argus. M: Sohal. R: Bryant. G: Hoch-Ligeti. C: Arcos. J. (1973). Dose-response and ultrastructural alterations in
dioxane carcinogenesis. Influence of methylcholanthrene on acute toxicitv. Eur J Cancer 9: 237-243.
http://dx.doi.org/10.1016/0014-2964(73)90088-1.
Ashbv. J. (1994). The genotoxicitv of 1.4-dioxane. MutatRes 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). (2007). Toxicological profile for 1.4 dioxane. Draft
for public comment. Atlanta. GA. http://www.atsdr.cdc.gov/toxprofiles/tpl87.pdf.
Bannasch. P: Moore. M: 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.
Bannasch. P. (2003). Comments on R. Karbe and R.L. Kerlin (2002) Cystic degeneration/spongiosis hepatis
(Toxicol Pathol 30 (2). 216-227). Toxicol Pathol 31: 566-570.
http://dx.doi.org/10.1080/01926230390224700.
Barber. H. (1934). Haemorrhagic nephritis and necrosis of the liver from dioxan poisoning. Guv's Hosp Rep 84:
267-280.
Bogen. K. (1990). Uncertainty in environmental health risk assessment. New York: Garlend Publishing Company.
Boorman. G: Morgan. K: Uriah. L. (1990). Nose, larynx and trachea Pathology of the Fischer Rat Reference and
Atlas (pp. 315-337). San Diego. CA: Academic Press. San Diego.
Braun. W and Young. J. (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-
-------
Derosa. C: Wilbur. S: Holler. J: Richter. P: Stevens. Y. (1996). Health evaluation of 1.4-dioxane. Toxicol Ind Health
12: 1-43. http://dx.doi.org/10.1177/074823379601200101.
Drew. R: Patel J: Lin. F. (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.
Ernstgard. L: Iregren. A: Sjogren. B: Johanson. G. (2006). Acute effects of exposure to vapours of dioxane in
humans. HumExp Toxicol 25: 723-729. http://dx.doi.org/10.1177/0960327106073805.
Fairlev. A: Linton. E: Ford-Moore. A. (1934). The toxicitv to animals of 1:4 dioxan. J Hyg 34: 486-501.
http://dx.doi.org/10.1017/S0022172400043266.
Fisher. J: Mahle. D: Bankston. L: Greene. R: Gearhart J. (1997). Lactational transfer of volatile chemicals in breast
milk. AmlndHyg Assoc J 58: 425-431. 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 neurotoxicitv of solvents: Isoeffective air
concentrations of 48 compounds evaluated in rats and mice. Environ Res 66: 173-185.
http://dx.doi.org/10.1006/enrs. 1994.1053.
Galloway. S: Armstrong. M: Reuben. C: Colman. S: Brown. B: Cannon. C: Bloom. A: Nakamura. F: Ahmed. M:
Duk. S: Rimpo. J: Margolin. B: Resnick. M: Anderson. B: Zeiger. E. (1987). Chromosome aberrations and
sister chromatid exchanges in Chinese hamster ovary cells: Evaluations of 108 chemicals. Environ Mol
Mutaeen 10: 1-175. http://dx.doi.org/10.1002/em.2850100502.
Gaskell B. (1990). Nonneoplastic changes in the olfactory epithelium- experimental studies. Environ Health
Perspect 85: 275-289.
Giavinj E: Vismara. C: Broccia. M. (1985). Teratogenesis study of dioxane in rats. Toxicol Lett 26: 85-88.
http://dx.doi.org/10.1016/0378-4274(85')90189-4.
Goldberg. M: Johnson. H: Pozzani. U: Smyth. HJ. (1964). Effect of repeated inhalation of vapors of industrial
solvents on animal behavior: I. Evaluation of nine solvent vapors on pole-climb performance in rats. Am
Ind Hyg Assoc J 25: 369-375.
Goldsworthy. T: Monticello. T: Morgan. K: Bermudez. E: Wilson. D: Jackh. R: BE. B. (1991). Examination of
potential mechanisms of carcinogenicitv of 1.4-dioxane in rat nasal epithelial cells and hepatocvtes. Arch
Toxicol 65: 1-9. http://dx.doi.org/10.1007/BF01973495.
Green. T: Lee. R: Moore. R: Ashbv. J: Willis. G: Lund. V: 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.200Q. 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.
Guvatt G: Oxman. A: Kunz. R: Vist G: Falck-Ytter. Y: Schiinemann. H. (2008aX GRADE: What is "quality of
evidence" and why is it important to clinicians. BMJ 336: 995-998.
http://dx.doi.org/10.1136/bmi.39490.551019.BE.
Guvatt G: Oxman. A: Vist. G: Kunz. R: Falck-Ytter. Y: Alonso-Coello. P: Schiinemann. H. (2008bX GRADE: An
emerging consensus on rating quality of evidence and strength of recommendations. BMJ 336: 924-926.
http://dx.doi.org/10.1136/bmi.39489.470347.AD.
Hansch. C: Leo. A: Hoekman. D. (1995). Exploring QSAR: Hydrophobia electronic, and steric constants.
Washington. DC: American Chemical Society.
Harkema. J: Carey. S: Wagner. J. (2006). The nose revisited: A brief review of the comparative structure, function.
and toxicologic pathology of the nasal epithelium. Toxicol Pathol 34: 252-269.
Haseman. J: Huff. J: Boorman. G. (1984). Use of historical control data in carcinogenicitv studies in rodents.
Toxicol Pathol 12: 126-135. http://dx.doi.org/10.1177/019262338401200203.
Haseman. J and Hailev. J. (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.
Hawlev. G and Lewis. R. Sr (Eds.X (2001). Hawlev's Condensed Chemical Dictionary (14 ed.X New York. NY:
John Wiley & Sons. Inc.
Haworth. S: Lawlor. T: Mortelmans. K: Speck. W: Zeiger. E. (1983). Salmonella mutagenicitv test results for 250
chemicals. Environ Mutagen 5: 3-142. http://dx.doi.org/10.1002/em.2860050703.
Havashi. S: Watanabe. J: Kawajiri. K. (1991). Genetic polymorphisms in the 5'-flanking region change
transcriptional regulation of the human cvtochrome P450HE1 gene. JBiochem 110: 559-565.
Hellmer. L and 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/10.1016/0165-1161(92')90043-L.
146
DRAFT - DO NOT CITE OR QUOTE
-------
HF Jr. S: Seaton. J: Fischer. L. (1941). The single dose toxicitv of some glycols and derivatives. J Ind Hyg Toxicol
23:259-268.
Hill. A. (1965). The environment and disease: Association or causation? Proc R Soc Med 58: 295-300.
Hoch-Ligeti. C and Argus. M. (1970). Effect of carcinogens on the lung of guinea pigs. In P Nettlesheim. MG
Hanna Jr & JW Deatherage Jr (Eds.). Morphology of Experimental Respiratory Carcinogenesis:
Proceedings of a Biology Division. Oak Ridge National Laboratory. Conference held in Gatlinburg.
Tennessee. May 13-16. 1970 (Vol. 21. pp. 267-279). Oak Ridge. TN: United States Atomic Energy
Co mission. Division of Technical Information.
Hoch-Ligeti. C: Argus. M: Arcos. J. (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. C-Y: Huang. K-L: Cheng. T-J: Wang. J-D: Hsieh. L-L. (19971. 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 IARC Monographs on the
Evaluation of Carciongenic Risks to Humans, pp. 589-602. Lyon. France: World Health Organization.
http://monographs.iarc.fr/ENG/Monographs/vol71/mono71-25.pdf.
IARC. (International Agency for Research on Cancer). (2006). Preamble to the IARC monographs. Lyon. France.
http://monographs.iarc.fr/ENG/Preamble/.
ICRP. (International Commission on Radiological Protection). (1975). Report of the task group on reference man:
ICRP publication 23 (Vol. 23). New York. NY: International Commission of Radiological Protection.
Pergamon Press.
JBRC. (Japan Bioassav Research Center). (1998). Two-year studies of 1.4-dioxane inF344 rats and BDF1 mice
(drinking water). Kanagawa. Japan.
Johnstone. R. (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: Saito. M: Senoh. H: Ohbayashi H: Aiso. S: Yamazaki. K: Nagano. K: Fukushima. S. (2008).
Thirteen-week oral toxicitv of 1.4-dioxane in rats and mice. J Toxicol Sci 33: 141-153.
http://dx.doi.org/10.2131/its.33.141.
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.
Karbe. E and Kerlin. R. (2002). Cystic degeneration/spongiosis hepatis in rats. Toxicol Pathol 30: 216-227.
http://dx.doi.org/10.1080/019262302753559551.
Kasai. T (2008) ri.4-Dioxane toxicitv studies].
Kasai. T: Saito. M: Senoh. H: Umeda. Y: Aiso. S: Ohbavashi. H: Nishizawa. T: Nagano. K: Fukushima. S. (2008).
Thirteen-week inhalation toxicitv of 1.4-dioxane in rats. Inhal Toxicol 20: 961-971.
http://dx.doi.org/10.1080/08958370802105397.
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 carcinogenicitv and chronic toxicitv of 1.4-dioxane in male rats.
Inhal Toxicol 21: 889-897. http://dx.doi.org/10.1080/08958370802629610.
Kasper. P: Uno. Y: Mauthe. R: Asano. N: Douglas. G: Matthews. E: Moore. M: Mueller. L: Nakaiima. M: Singer. T:
Speit. G. (2007). Follow-up testing of rodent carcinogens not positive in the standard genotoxicitv testing
battery: IWGT workgroup report. 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 inBDFl mice. J Toxicol Pathol 11: 21-25.
http://dx.doi.org/10.1293/tox.ll.21.
Kesten. H: Mulinos. M: Pomerantz. L. (1939). Pathologic effects of certain glycols and related compounds. Arch
Pathol 27: 447-465.
Khudolev. V: Mizgireuv. I: Pliss. G. (1987). The study of mutagenic activity of carcinogens and other chemical
agents with Salmonella tvphimurium assays: Testing of 126 compounds. Arch Geschwulstforsch 57: 453-
462.
King. M: Shefner. A: Bates. R. (1973). Carcinogenesis bioassav of chlorinated dibenzodioxins and related
chemicals. Environ Health Perspect 5: 163-170.
147
DRAFT - DO NOT CITE OR QUOTE
-------
Kitchin. K and Brown. J. (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. P. (1935). Narcotic potency of some cyclic acetals. J Pharmacol Exp Ther 53: 440-444.
Kociba. R: McCollister. S: Park. C: Torkelson. T: Gehring. P. (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. R: Torkelson. T: Young. J: Gehring. P. (1975). 1.4-Dioxane: Correlation of the results of chronic ingestion
and inhalation studies with its dose-dependent fate in rats Proceedings of the 6th Annual Conference on
Environmental Toxicology (pp. 345-354). Wright-Patterson Air Force Base. OH: Wright-Patterson Air
Force Base. Air Force Systems Command. Aerospace Medical Division. Aerospace Medical Research
Laboratory.
Kopylev. L: John Fox. J: Chen. C. (2009). Combining risks from several tumors using Markov Chain Monte Carlo.
In RM Cooke (Ed.X Uncertainty Modeling in Dose Response (1 ed.. pp. 197-205). Hoboken. NJ: John
Wiley & Sons.
Kurl R: Poellinger. L: Lund. J: Gustafsson. J-A. (1981). Effects of dioxane on RNA synthesis in the rat liver. Arch
Toxicol 49: 29-33. http://dx.doi.org/10.1007/BF00352068.
Kwan. K: Dutka. B: Rao. S: 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)90124-11.
Laug. E: Calvery. H: Morris. H: Woodard. G. (1939). The toxicology of some glycols and derivatives. J Ind Hyg
Toxicol 21: 173-201.
Lesage. S: Jackson. R: Priddle. M: Riemann. P. (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.
Leung. H-W and Paustenbach. D. (1990). Cancer risk assessment for dioxane based upon a physiologically-based
pharmacokinetic approach. Toxicol Lett 51: 147-162.
Lewandowskj T and Rhomberg. L. (2005). A proposed methodology for selecting a trichloroethylene inhalation
unit risk value for use in risk assessment. Regul Toxicol Pharmacol 41: 39-54.
http://dx.doi.0rg/10.1016/i.vrtph.2004.09.003.
Lewis. R. Sr. (2000). Sax's Dangerous Properties of Industrial Materials (10 ed.X New York. NY: John Wiley &
Sons. Inc.
Lide. D. R.. (Ed.X (2000). Handbook of Chemistry and Physics (81 ed.X Boca Raton. FL: CRC Press LLC.
Lundberg. I: Ekdahl. M: Kronevi. T: Lidums. V: Lundberg. S. (1986). Relative hepatotoxicitv of some industrial
solvents after intraperitoneal injection or inhalation exposure in rats. Environ Res 40: 411-420.
htto://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-l.
Lyman. W: Reehl W: DH Rosenblatt (Eds.X (American Chemical Society). (1990). Handbook of Chemical
Property Estimation Methods: Environmental Behavior of Organic Compounds. Washington. DC:
American Chemical Society.
Marzulli. F: Anjo. D: Maibach. H. (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. htto://dx.doi.org/10.1016/0015-6264(81')90530-7.
McConnell E: Solleveld. H: Swenberg. J: Boorman. G. (1986). Guidelines for combining neoplasms for evaluation
of rodent carcinogenesis studies. J Natl Cancer Inst 76: 283-289.
McFee. A: Abbott M: Gulati D: Shelby. M. (1994). Results of mouse bone marrow micronucleus studies on 1.4-
dioxane. MutatRes 322: 145-148.
McGregor. D: Brown. A: Howgate. S: McBride. D: Riach. C: Caspary. W. (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. M and Bond. J. (2001). Sites and mechanisms for uptake of gases and vapors in the respiratory tract.
Toxicology 160: 165-172.
Meylan. W: Howard. P: Boethling. R: 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. M: Gorlinskava Ye. P: Solowova. T. (1990). The body distribution and biological action of xenobiotics. J
Hyg Epidemiol Microbiol Immunol 34: 329-336.
Mirkova. E. (1994). Activity of the rodent carcinogen 1.4-dioxane in the mouse bone marrow micronucleus assay.
Mutat Res 322: 142-144.
148
DRAFT - DO NOT CITE OR QUOTE
-------
Miyagawa. M: Shirotori. T: Tsuchitani. M: Yoshikawa. K. (1999). Repeat-assessment of 1.4-dioxane in a rat-
hepatocvte replicative DNA synthesis (RDS) test: Evidence for stimulus of hepatocvte proliferation. Exp
Toxicol Pathol 51: 555-558.
Morgan. K: Patterson. D: Gross. E. (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 clastogenicitv of 1.4 dioxane as examined in the mouse peripheral blood micronucleus test.
Mammalian Mutagenicitv Study Group Communications 2: 7-8.
Morita. T and Havashj 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/(SICin098-2280(1998')32:3<269::AID-EM10>3.0.CO:2-8.
Mungikar. A and Pawar. S. (1978). Induction of the hepatic microsomal mixed function oxidase system in mice by
p-dioxane. Bull Environ Contain Toxicol 20: 797-804. http://dx.doi.org/10.1007/BF01683603.
Munoz. E and Barnett B. (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 P. (2005). Effects of dioxane on cvtochrome 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.
NCI. (National Institutes of Health. National Cancer Institute). (1978). Bioassav of 1.4-dioxane for possible
carcinogenicitv. (78-1330 NCICGTR-80X Bethesda. MD: National Cancer Institute.
http://ntp.niehs.nih.gov/ntp/htdocs/LT rpts/trO80.pdf.
Nelson. N. (195IX Solvent toxicitv with particular reference to certain octyl alcohols and dioxanes. Med Bull 11:
226-238.
Nestmann. E: Otson. R: Kowbel D: Bothwell P: Harrington. T. (1984). Mutagenicitv 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.
NRC. (National Research Council). (1994). Science and judgment in risk assessment. Washington. DC: National
Academy Press.
NRC. (National Research Council). (2009). Science and decisions: Advancing risk assessment. Washington. DC:
National Academies Press.
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.
O'Neil MJ: Smith. A: Heckelmaa PE: Obenchain. JR: Gallipeau. JR: D'Arecca. MA (Eds.X (2001). The Merck
Index: An Encyclopedia of Chemicals. Drugs, and Biologicals (13th ed.X Whitehouse Station. NJ: Merck
& Co.. Inc.
Park. J: Hussam A: Couasnon. P: Fritz. D: Carr. P. (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. O: Wallington. T. (1997). Atmospheric chemistry of 1.4-dioxane.
Faraday Trans 1 93: 2855-2863. http://dx.doi.org/10.1039/a700598i.
Pozzani. U: Weil C: Carpenter. C. (1959). The toxicological basis of threshold limit values. 5: The experimental
inhalation of vapor mixtures by rats, with notes upon the relationship between single dose inhalation and
single dose oral data. Am Ind Hyg Assoc J 20: 364-369. http://dx.doi.org/10.1080/00028895909343733.
Ramsey. J and Andersen. M. (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. R: McCroskev. P: Park. C: Andersen. M: Gargas. M. (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. H andKlopman. G. (1992). 1.4-dioxane: Prediction of in vivo clastogenicitv. MutatRes 280: 245-251.
http://dx.doi.org/10.1016/0165-1218(92')90054-4.
Rothman. K and Greenland. S. (1998). Modern epidemiology (2nd ed.X Philadelphia. PA: Lippincott. Williams. &
Wilkins.
Roy. S: Thilagar. A: Eastmond. D. (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. H and Yant W. (1936). Toxicitv of dioxan. J Ind Hyg Toxicol 18: 448-460.
149
DRAFT - DO NOT CITE OR QUOTE
-------
Sheu. C: Moreland. F: Lee. J: Dunkel. V. (1988). In vitro BALB/3T3 cell transformation assay of nonoxynol-9 and
1.4-dioxane. Environ Mol Mutagen 11: 41-48. http://dx.doi.org/10.1002/em.2850110106.
Silverman. L: Schulte. H: First. M. (1946). Further studies on sensory response to certain industrial solvent vapors. J
Ind Hyg Toxicol 28: 262-266.
Sina. J: Bean. C: Dysart. G: Taylor. V: Bradley. M. (1983). Evaluation of the alkaline elution/rat hepatocvte assay as
a predictor of carcinogenic/mutagenic potential. MutatRes 113: 357-391. http://dx.doi.org/10.1016/0165-
1161(83)90228-5.
Spiegelhalter. D: Thomas. A: Best. N: Lunn. D. (2003). WinBugs version 1.4 user manual. Cambridge. UK: MRC
Biostatistics Unit.
Sticknev. J: Sager. S: Clarkson. J: Smith. L: Locev. B: Bock. M: Hartung. R: Olp. S. (2003). An updated evaluation
of the carcinogenic potential of 1.4-dioxane. Regul Toxicol Pharmacol 38: 183-195.
htto://dx.doi.org/10.1016/S0273-2300(03X)0090-4.
Stoner. G: Conran. P: Greisiger. E: Stober. J: Morgan. M: Pereira. M. (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.
Stott W: Quasi J: Watanabe. P. (1981). Differentiation of the mechanisms of oncogenicitv 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 pericvtoma: A benign neoplasm deriving from
the perisinusoidal (Ito) cells in rat liver. Am J Pathol 146: 903-913.
Surprenant K. (2002). Dioxane Ullmann's Encyclopedia of Industrial Chemistry (6th ed.. pp. 543). Weinheim.
Germany: Wilev-VCH Verlag.
Sweeney. L: Thrall K: Poet T: Corlev. R: Weber. T: Locev. B: Clarkson. J: Sager. S: Gargas. M. (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.
Thiess. A: Tress. E: Fleig. I. (1976). Arbeitsmedizinische Untersuchungsergebnisse von Dioxan-exponierten
Mitarbeitern [Industrial-medical investigation results in the case of workers exposed to dioxanel.
Arbeitsmedizin. Sozialmedizin. Umweltmedizin 11: 35-46.
Thurman. G: Simms. B: Goldstein. A: Kilian. D. (1978). The effects of organic compounds used in the manufacture
of plastics on the responsivitv of murine and human lymphocytes. Toxicol Appl Pharmacol 44: 617-641.
http://dx.doi.org/10.1016/0041-008X(78')90269-7.
Tinwell. H and Ashbv. J. (1994). Activity of 1.4-dioxane in mouse bone marrow micronucleus assays. Mutat Res
322: 148-150.
Torkelson. T: Leong. B: Kociba. R: Richter. W: Gehring. P. (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.N. Environment Programme. (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.une
p.org%2Fozone%2Fpdfs%2Fmontreal-protocol2000.pdf&ei=-c89TPXON9PRngf-i-
idDg&usg=AFOiCNH4OH15inPn5XFcYTvblPPRDZu-fO&sig2=aaSaM nuOlXlHc409kBvgw.
U.S. EPA. (U.S. Environmental Protection Agency). (1986a). Guidelines for carcinogen risk assessment.
(EPA/630/R-00/004). pp. 38. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
Forum. http://epa.gov/raf/publications/pdfs/CA%20GUIDELINES 1986.PDF.
U.S. EPA. (U.S. Environmental Protection Agency). (1986bX Guidelines for the health risk assessment of chemical
mixtures. (EPA/630/R-98/002X pp. 38. Washington. DC: U.S. Environmental Protection Agency. Risk
Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=22567.
U.S. EPA. (U.S. Environmental Protection Agency). (1986cX Guidelines for mutagenicitv risk assessment.
(EPA/630/R-98/003V pp. 23. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
Forum, http://www.epa.gov/iris/backgrd.html.
U.S. EPA. (U.S. Environmental Protection Agency). (1988). Recommendations for and documentation of biological
values for use in risk assessment. (EPA/600/6-87/008'), pp. 395. Cincinnati. OH: U.S. Environmental
Protection Agency. Environmental Criteria and Assessment Office.
http://cfpub.epa. gov/ncea/cfm/recordisplav.cfm?deid=34855.
U.S. EPA. (U.S. Environmental Protection Agency). (1991). Guidelines for developmental toxicitv risk assessment.
(EPA/600/FR-91/00IX pp. 83. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
Forum, http://www.epa.gov/iris/backgrd.html.
U.S. EPA. (U.S. Environmental Protection Agency). (1994). Methods for derivation of inhalation reference
concentrations and application of inhalation dosimetry. (EPA/600/8-90/066FX pp. 409. Research Triangle
150
DRAFT - DO NOT CITE OR QUOTE
-------
Park. NC: U.S. Environmental Protection Agency. Office of Research and Development. Office of Health
and Environmental Assessment Environmental Criteria and Assessment Office.
http://cfpub.epa. gov/ncea/cfm/recordisplav.cfm?deid=71993.
U.S. EPA. (U.S. Environmental Protection Agency). (1996). Guidelines for reproductive toxicitv risk assessment.
(EPA/630/R-96/009X pp. 143. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
Forum. http://www.epa.gov/raf/publications/pdfs/REPRO51.PDF.
U.S. EPA. (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicitv risk assessment.
(EPA/630/R-95/001FX pp. 89. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
Forum. 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/630/R-00/001). pp. 96. Washington. DC: U.S. Environmental Protection
Agency. Risk Assessment Forum, http://www.epa.gov/raf/publications^enchmark-dose-doc-draft.htm.
U.S. EPA. (U.S. Environmental Protection Agency). (200010. Science policy council handbook: Risk
characterization. (EPA 100-B-00-002). pp. 189. Washington. D.C.: U.S. Environmental Protection
Agency. Office of Research and Development Office of Science Policy.
http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf.
U.S. EPA. (U.S. Environmental Protection Agency). (20000). Supplementary guidance for conducting health risk
assessment of chemical mixtures. (EPA/630/R-00/002). pp. 209. Washington. DC: U.S. Environmental
Protection Agency. Risk Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=20533.
U.S. EPA. (U.S. Environmental Protection Agency). (2002aX A review of the reference dose and reference
concentration processes. (EPA/630/P-02/0002F). pp. 192. Washington. DC: U.S. Environmental Protection
Agency. Risk Assessment Forum. 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, from http://www.epa.gov/iur/
U.S. EPA. (U.S. Environmental Protection Agency). (2005a). Guidelines for carcinogen risk assessment.
(EPA/630/P-03/001FX pp. 166. Washington. DC: U.S. Environmental Protection Agency. Risk
Assessment Forum, http://www.epa.gov/cancerguidelines/.
U.S. EPA. (U.S. Environmental Protection Agency). (200510. Supplemental guidance for assessing susceptibility
from early-life exposure to carcinogens. (EPA/630/R-03/003F). pp. 126. Washington. DC: U.S.
Environmental Protection Agency. Risk Assessment Forum.
http://www.epa.gov/cancerguidelines/guidelines-carcinogen-supplement.htm.
U.S. EPA. (U.S. Environmental Protection Agency). (2006). Peer review handbook (3rd edition). (EPA/100/B-
06/002). pp. 190. Washington. DC: U.S. Environmental Protection Agency. Science Policy Council.
http://www.epa.gov/peerreview/pdfs/peer review handbook 2006.pdf.
U.S. EPA. (U.S. Environmental Protection Agency). (2009a). Integrated Risk Information System (IRIS) Update
Project Announcement of 2009/2010 Agenda. Fed Reg 74: 54040-54042.
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 Draftl. (EPA/635/R-09/005). pp. 1-276. Washington. DC: U.S. Environmental Protection Agency.
Integrated Risk Information System. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=199330.
U.S. EPA. (U.S. Environmental Protection Agency). (2009cX Status report: Advances in inhalation dosimetry of
gases and vapors with portal of entry effects in the upper respiratory tract. (EPA/600/R-09/072X pp. 100.
Research Triangle Park. NC: U.S. Environmental Protection Agency. National Center for Environmental
Assessment. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=212131.
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-635/R-
09-005-FX pp. 319. Washington. DC. www.epa.gov/iris/toxreviews/0326tr.pdf.
U.S. EPA. (U.S. Environmental Protection Agency). (201 la). IRIS Process, from http://epa.gov/iris/process.htm
U.S. EPA. (U.S. Environmental Protection Agency). (201 IbX Recommended use of body weight 3/4 as the default
method in derivation of the oral reference dose. (EPA/100/R11/000IX pp. 50. Washington. DC.
http://www.epa.gov/raf/publications/interspecies-extrapolation.htm.
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 hepatocvtes 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.
van Delft. J: van Agen. E: van Breda. S: Herwijnen. M: Staal Y: Klemjans. J. (2004). Discrimination of genotoxic
from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25: 1265-1276.
http://dx.doi.org/10.1093/carcin/bghl08.
151
DRAFT - DO NOT CITE OR QUOTE
-------
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.
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. D: Pampori. N: Ram. P: Agrawal A: Shapiro. B. (1991). Interpulse interval in circulating growth
hormone patterns regulates sexually dimorphic expression of hepatic cvtochrome P450. PNAS 88: 6868-
6872.
Wirth. W and Klimmer. O. (1936). [On the toxicology of organic solvents. 1.4 dioxane (diethylene dioxide)]. Archiv
fuer Gewerbepathologie und Gewerbehygiene 17: 192-206.
Wolfe. N and Jeffers. P. (2000). Hydrolysis. InRS Boethling & D Mackav (Eds.). Handbook of Property Estimation
Methods for Chemicals: Environmental and Health Sciences (pp. 311-333). Boca Raton. FL: Lewis
Publishers.
Wolford. S: Schroer. R: Gohs. F: Gallo. P: Brodeck. M: Falk. H: Ruhrea R. (1986). Reference range database for
serum chemistry and hematologv values in laboratory animals. J Toxicol Environ Health A 18: 161-188.
http://dx.doi.org/10.1080/15287398609530859.
Woo. Y-T: Arcos. J: Argus. M: Griffin. G: K. N. (1911 a). 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. Y-T: Argus. M: Arcos. J. (1977b). Tissue and subcellular distribution of 3H-dioxane in the rat and apparent
lack of microsome-catalyzed covalent binding in the target tissue. Life Sci 21: 1447-1456.
http://dx.doi.org/10.1016/0024-3205(77')90199-0.
Woo. Y-T: Argus. M: Arcos. J. (1977cX 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. Y-T: Argus. M: Arcos. J. (1978). Effect of mixed-function oxidase modifiers on metabolism and toxicitv of
the oncogen dioxane. Cancer Res 38: 1621-1625.
Yamamoto. S: Urano. K: Koizumi. H: Wakana. S: Hioki. K: Mitsumori. K: Kurokawa. Y: Havashi. Y: T. N.
(1998aX Validation of transgenic mice carrying the human prototype c-Ha-ras gene as a bioassav model for
rapid carcinogenicitv testing. Environ Health Perspect 106: 57-69.
Yamamoto. S: Urano. K: Nomura. T. (1998bX Validation of transgenic mice harboring the human prototype c-Ha-
ras gene as a bioassav model for rapid carcinogenicitv testing. Toxicol Lett 102-103: 473-478.
http://dx.doi.org/10.1016/S0378-4274(98X)0341-5.
Yamamoto. S: Ohsawa. M: Nishizawa. T: Saito. A: Kasai. T: Noguchi. 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). J Toxicol Sci 25: 347.
Yamazaki. 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.X 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.
Yamazaki. K (2006) [Correspondence between Kazunori Yamazaki and Julie Sticknevl.
Yant W: Schrenk. H: Waite. C: Patty. F. (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: Kawagoshj 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: Katamj 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. J: Mason. J: Valencia. R: Woodruff. R: Zimmering. S. (1985). Chemical mutagenesis testing in Drosophila.
IV. Results of 45 coded compounds tested for the National Toxicology Program. Environ Mutagen 7: 349-
367. http://dx.doi.org/10.1002/em.286007Q310.
Young. J: Braun. W: Gehring. P: Horvath. B: Daniel R. (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. J: Braun. W: Rampy. L: Chenoweth. M: Blau. G. (1977). Pharmacokinetics of 1.4-dioxane in humans. J
Toxicol Environ Health 3: 507-520. http://dx.doi.org/10.1080/15287397709529583.
152
DRAFT - DO NOT CITE OR QUOTE
-------
Young. J: Braun. W: Gehring. P. (1978aX Dose—dependent fate of 1.4-dioxane in rats(bX J Toxicol Environ
Health A 4: 709-726. http://dx.doi.org/10.1080/15287397809529693.
Young. J: Braun. W: Gehring. P. (1978bX The dose-dependent fate of 1.4-dioxane in rats. J Environ Pathol Toxicol
2: 263-282. http://dx.doi.org/10.1080/15287397809529693.
Zimmermann. F: Mayer. V: Scheel. I: Resnick. M. (1985). Acetone, methyl ethyl ketone. ethyl acetate, acetonitrile
and other polar aprotic solvents are strong inducers of aneuploidv in Saccharomyces cerevisiae. Mutat Res
149: 339-351. http://dx.doi.org/10.1016/0027-5107(85')90150-2.
153
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX A. SUMMARY OF EXTERNAL PEER
REVIEW AND PUBLIC COMMENTS AND
DIS POS ITION
1 Note: The comments and responses in this appendix were in regards to the oral assessment previously
2 reviewed. A summary of external peer review and public comments and disposition following review of
3 the inhalation assessment for 1.4-dioxane will be included when they become available.
4 The Toxicological Review of 1,4-Dioxane has undergone formal external peer review performed
5 by scientists in accordance with EPA guidance on peer review (U.S. EPA. 2006. 2000b). The external
6 peer reviewers were tasked with providing written answers to general questions on the overall assessment
7 and on chemical-specific questions in areas of scientific controversy or uncertainty. A summary of
8 significant comments made by the external reviewers and EPA's responses to these comments arranged
9 by charge question follow. In many cases the comments of the individual reviewers have been
10 synthesized and paraphrased for development of Appendix A. The majority of the specific observations
11 (in addition to EPA's charge questions) made by the peer reviewers were incorporated into the document
12 and are not discussed further in this Appendix. Public comments that were received are summarized and
13 addressed following the peer-reviewers' comments and disposition.
A.1 External Peer Review Panel Comments
14 The reviewers made several editorial suggestions to clarify portions of the text. These changes
15 were incorporated in the document as appropriate and are not discussed further.
16 In addition, the external peer reviewers commented on decisions and analyses in the
17 Toxicological Review ofl,4-Dioxane under multiple charge questions, and these comments were
18 organized and summarized under the most appropriate charge question.
A.1.1 General Charge Questions
19 1. Is the Toxicological Review logical, clear and concise? Has EPA accurately, clearly and objectively
20 represented and synthesized the scientific evidence for noncancer and cancer hazards?
21 Comment. All reviewers found the Toxicological Review to be logical, clear, and concise. One
22 reviewer remarked that it was an accurate, open-minded and balanced analysis of the literature.
23 Most reviewers found that the scientific evidence was presented objectively and transparently;
24 however, one reviewer suggested two things to improve the objectivity and transparency (1)
25 provide a clear description of the mode of action and how it feeds into the choice of the
26 extrapolation for the cancer endpoint and (2) provide a presentation of the outcome if internal
27 dose was used in the cancer and noncancer assessments.
A-l
DRAFT - DO NOT CITE OR QUOTE
-------
1 One reviewer commented that conclusions could not be evaluated in a few places where
2 dose information was not provided (Sections 3.2, 3.3 and 4.5.2.2). The same reviewer found the
3 MOA schematics, key event temporal sequence/dose-response table, and the POD plots to be
4 very helpful in following the logic employed in the assessment.
6 Response. The mode of action analysis and how conclusions from that analysis fed into the
7 choice of extrapolation method for the cancer assessment are discussed further under charge
8 questions C2 and C5. Because of the decision not to utilize the PBPK models, internal doses were
9 not calculated and thus were not included as alternatives to using the external dose as the POD for
10 the cancer and noncancer assessments.
11 In the sections noted by the reviewer (3.2, 3.3 and 4.5.2.2) dose information was added as
12 available. In Section 3.2, Mikheev et al. (1990) did not report actual doses, which is noted in this
13 section. All other dose information in this section was found to be present after further review by
14 the Agency. In Section 3.3, dose information for Woo et al. (1978. 1977c) was added to the
15 paragraph. In Section 4.5.2.2, study details for Nannelli et al. (2005) were provided earlier in
16 Section 3.3 and a statement referring the reader to this section was added.
17
18 2. Please identify any additional studies that should be considered in the assessment of the noncancer
19 and cancer health effects of 1,4-dioxane.
20 Comment. Five reviewers stated they were unaware of any additional studies available to add to
21 the oral toxicity evaluation of 1,4-dioxane. These reviewers also acknowledged the Kasai et al.
22 (2009; 2008) publications that may be of use to derive toxicity values following inhalation of
23 1,4-dioxane.
24 a. Kasai T; Saito H; Senoh Y; et al. (2008) Thirteen-week inhalation toxicity of 1,4-dioxane
25 in rats. Inhal Toxicol 20: 961 -971.
26 b. Kasai T; Kano Y; Umeda T; et al. (2009) Two-year inhalation study of carcinogenicity
27 and chronic toxicity of 1,4-dioxane in male rats. Inhal Toxicol in press.
28 Other references suggested by reviewers include:
29 c. California Department of Health Services (1989) Risk Specific Intake Levels for the
30 Proposition 65 Carcinogen 1, 4-dioxane. Reproductive and Cancer Hazard Assessment
31 Section. Office of Environmental Health Hazard Assessment
32 d. National Research Council (2009) Science and Decisions: Advancing Risk Assessment.
\ / O
33 Committee on Improving Risk Analysis Approaches Used by the U.S. EPA. Washington,
34 D.C., National Academy Press.
A-2
DRAFT - DO NOT CITE OR QUOTE
-------
1 e. ATSDR (2007) Toxicological Profile for 1,4-dioxane. Agency for Toxic Substances and
2 Disease Registry. Atlanta, GA.
3 f. Stickney JA; Sager SL; Clarkson JR; et al. (2003) An updated evaluation of the
4 carcinogenic potential of 1,4-dioxane. Regul Toxicol Pharmacol 38: 183-195.
5 g. Yamamoto S; Ohsawa M; Nishizawa T; et al. (2000) Long-term toxicology study of
6 1,4-dioxane in R344 rats by multiple-route exposure (drinking water and inhalation). J
7 Toxicol Sci 25: 347.
9 Response. The references a-b above will be evaluated for derivation of an RfC and IUR, which
10 will follow as an update to this oral assessment. References c and e noted above were considered
11 during development of this assessment as to the value they added to the cancer and noncancer
12 analyses. Reference g listed above is an abstract from conference proceedings from the 27th
13 Annual Meeting of the Japanese Society of Toxicology; abstracts are not generally considered in
14 the development of an IRIS assessment. Reference d reviews EPA's current risk assessment
15 procedures and provides no specific information regarding 1,4-dioxane. The Stickney et al.
16 (2003) reference was a review article and no new data were presented, thus it was not referenced
17 in this Toxicological Review but the data were considered during the development of this
18 assessment.
19 Following external peer review (as noted above) Kano et al. (2009) was added to the assessment,
20 which was an update and peer-reviewed published manuscript of the JBRC (1998) report.
21
22 3. Please discuss research that you think would be likely to increase confidence in the database for
23 future assessments of 1,4-dioxane.
24 Comment. All reviewers provided suggestions for additional research that would strengthen the
25 assessment and reduce uncertainty in several areas. The following is a brief list of questions that
26 were identified that could benefit from further research. What are the mechanisms responsible for
27 the acute and chronic nephrotoxicity? Is the acute kidney injury (AKI) multifactorial? Are there
28 both tubular and glomerular/vascular toxicities that result in cortical tubule degeneration and
29 evidence for glomerulonephritis? What are the functional correlates of the histologic changes in
30 terms of assessment of renal function? What is the exposure in utero and risk to the fetus and
31 newborn? What are the concentrations in breast milk following maternal exposure to
32 1,4-dioxane? What is the risk for use of contaminated drinking water to reconstitute infant
33 formula? What are the exposures during early human development? What is the pharmacokinetic
34 and metabolic profile of 1,4-dioxane during development? What are the susceptible populations
35 (e.g., individuals with decreased renal function or chronic renal disease, obese individuals,
36 gender, age)?
A-3
DRAFT - DO NOT CITE OR QUOTE
-------
1 Additional suggestions for future research include: evaluation of potential epigenetic mechanisms
2 of carcinogenicity, additional information on sources of exposure and biological concentrations as
3 well as human toxicokinetic data for derivation of parameter to refine PBPK model, studies to
4 determine toxic moiety, focused studies to inform mode of action, additional inhalation studies
5 and a multigeneration reproductive toxicity study.
6 One reviewer suggested additional analyses of the existing data including a combined analysis of
7 the multiple datasets and outcomes for cancer and non-cancer endpoints, evaluation of the dose
8 metrics relevant to the MOA to improve confidence in extrapolation approach and uncertainty
9 factors, and complete a Bayesian analysis of human pharmacokinetic data to estimate human
10 variability in key determinants of toxicity (e.g., metabolic rates and partition coefficients).
11
12 Response. A number of research suggestions were provided for further research that may enhance
13 future health assessments of 1,4-dioxane. Regarding the suggested additional analyses for the
14 existing data, EPA did not identify a MOA in this assessment, thus combined analysis of the
15 cancer and non-cancer endpoints as well as application of various dose metrics to a MOA is not
16 applicable. Because the human PBPK model was not implemented in this assessment for oral
17 exposure to 1,4-dioxane a Bayesian analysis was not completed. No additional changes to the
18 Toxicological Review of 1,4-Dioxane were made in response to these research recommendations.
19
20 4. Please comment on the identification and characterization of sources of uncertainty in Sections 5 and
21 6 of the assessment document. Please comment on whether the key sources of uncertainty have been
22 adequately discussed. Have the choices and assumptions made in the discussion of uncertainty been
23 transparently and objectively described? Has the impact of the uncertainty on the assessment been
24 transparently and objectively described?
25 Comment. Six reviewers stated Sections 5 and 6 adequately discussed and characterized
26 uncertainty, in a succinct, and transparent manner. One reviewer suggested adding additional
27 discussion of uncertainty relating to the critical study used in the cancer assessment and another
28 reviewer suggested adding more discussion around the uncertainty of the toxic moiety.
29 One reviewer made specific comments on uncertainty surrounding the Kociba et al. (1974) study
30 as used for derivation of the RfD, choice of the non-cancer dose metric, and use of a 10%BMR as
31 the basis for the CSF derivation. These comments and responses are summarized below under
32 their appropriate charge question.
33
34 Response. The majority of the reviewers thought the amount of uncertainty discussion was
35 appropriate. Since the external review, Kano et al. (2009) was published and this assessment was
A-4
DRAFT - DO NOT CITE OR QUOTE
-------
1 updated accordingly (previously JBRC (1998). It is assumed the uncertainty referred to by the
2 reviewer was addressed by the published Kano et al. (2009) paper.
3 Clarification regarding the uncertainty surrounding the identification of the toxic moiety was
4 added to Section 4.6.2.1 stating that the mechanism by which 1,4-dioxane induces tissue damage
5 is not known, nor is it known whether the toxic moiety is 1,4-dioxane or a metabolite of
6 1,4-dioxane. Additional text was added to Section 4.7.3 clarifying that available data also do not
7 clearly identify whether 1,4-dioxane or one of its metabolites is responsible for the observed
8 effects. The impact of the lack of evidence to clearly identify a toxic moiety related to
9 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
10 1. A chronic RfD for 1,4-dioxane has been derived from a 2-year drinking water study (Kocibaet al.,
11 1974) in rats and mice. Please comment on whether the selection of this study as the principal study
12 has been scientifically justified. Has the selection of this study been transparently and objectively
13 described in the document? Are the criteria and rationale for this selection transparently and
14 objectively described in the document? Please identify and provide the rationale for any other studies
15 that should be selected as the principal study.
16 Comment. Seven of the reviewers agreed that the use of the Kociba et al. (1974) study was the
17 best choice for the principal study.
18 One reviewer stated that Kociba et al. (1974) was not the best choice because it reported only
19 NOAEL and LOAELs without providing incidence data for the endpoints. This reviewer also
20 stated that the study should not have been selected based on sensitivity of the endpoints, but
21 rather study design and adequacy of reporting of the study results. Additionally, this reviewer
22 suggested a better principal study would be either the NCI (1978) or JBRC (1998) study.
23 Response. The reviewer is correct that Kociba et al. (1974) did not provide incidence data;
24 however, Kociba et al. (1974) identified a NOAEL (9.6 mg/kg-day) and LOAEL (94 mg/kg-day)
25 within the text of the manuscript. Kociba et al. (1974) was a well conducted chronic bioassay
26 (four dose levels, including controls, with 60 rats/sex/group) and seven of the peer reviewers
27 found this study to be appropriate as the basis for the RfD. Further support for the selection of the
28 Kociba et al. (1974) as the principal study comes from comparison of the liver and kidney
29 toxicity data reported by JBRC (1998) and NCI (1978). which was presented in Section 5.1. The
30 effects reported by JBRC (1998) and NCI (1978) were consistent with what was observed by
31 Kociba et al. (1974) and within a similar dose range. Derivation of an RfD from these datasets
32 resulted in a similar value (Section 5.1.).
33 2. Degenerative liver and kidney effects were selected as the critical effect. Please comment on whether
34 the rationale for the selection of this critical effect has been scientifically justified. Are the criteria and
35 rationale for this selection transparently and objectively described in the document? Please provide a
A-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 detailed explanation. Please comment on whether EPA's rationale regarding adversity of the critical
2 effect for the RfD has been adequately and transparently described and is scientifically supported by
3 the available data. Please identify and provide the rationale for any other endpoints that should be
4 considered in the selection of the critical effect.
5 Comment. Five of the reviewers agreed with the selection of liver and kidney effects as the
6 critical effect. One of these reviewers suggested analyzing all datasets following dose adjustment
7 (e-g-, body weight scaling or PBPK model based) to provide a better rationale for selection of a
8 critical effect.
9 One reviewer stated that 1,4-dioxane causing liver and kidney organ specific effects is logical;
10 however, with regards to nephrotoxicity, the models and limited human data have not addressed
11 the mechanisms of injury or the clinical correlates to the histologic data. Also, advances in the
12 field of biomarkers have not yet been used for the study of 1,4-dioxane.
13 One reviewer found the selection of these endpoints to be 'without merit' because of the lack of
14 incidence data to justify the NOAEL and LOAEL values identified in the study. This reviewer
15 suggested selecting the most sensitive endpoint(s) from the NCI (NCI. 1978) or JBRC (1998)
16 studies for the basis of the RfD, but did not provide a suggestion as to what effect should be
17 selected.
18 Response. The liver and kidney effects from Kociba et al. (1974) was supported as the critical
19 effect by most of the reviewers. PBPK model adjustment was not performed because the PBPK
20 model was found to be inadequate for use in the assessment. EPA acknowledges that neither the
21 mechanisms of injury nor the clinical correlates to histologic data exist for 1,4-dioxane. This type
22 of information could improve future health assessments of 1,4-dioxane.
23 As stated above, Kociba et al. (1974) identified a NOAEL (9.6 mg/kg-day) and LOAEL
24 (94 mg/kg-day) within the text of the manuscript and was a well conducted chronic bioassay (four
25 dose levels, including controls, with 60 rats/sex/group).
26 3. Kociba et al. (1974) derived a NOAEL based upon the observation of degenerative liver and kidney
27 effects and these data were utilized to derive the point of departure (POD) for the RfD. Please provide
28 comments with regard to whether the NOAEL approach is the best approach for determining the
29 POD. Has the approach been appropriately conducted and objectively and transparently described?
30 Please identify and provide rationales for any alternative approaches for the determination of the POD
31 and discuss whether such approaches are preferred to EPA's approach.
32 Comment. Seven reviewers agreed with the NOAEL approach described in the document. One of
33 these reviewers also questioned whether any attempt was made to "semi-qualitatively represent
34 the histopathological observations to facilitate a quantitative analysis".
35 One reviewer stated that data were not used to derive the POD, but rather a claim by the authors
36 of Kociba et al. (1974) of the NOAEL and LOAEL for the endpoints. This reviewer preferred the
A-6
DRAFT - DO NOT CITE OR QUOTE
-------
1 use of a BMD approach for which data include the reported incidence rather than a study reported
2 NOAEL or LOAEL.
3 Response. The suggestion to "semi-qualitatively represent the histopathological observations to
4 facilitate a quantitative analysis" was not incorporated into the document because it is unclear
5 how this would be conducted since Kociba et al. (1974) did not provide incidence data and the
6 reviewer did not illustrate their suggested approach. See responses to B1 and B2 regarding the
7 NOAEL and LOAEL approach. The Agency agrees that a Benchmark Dose approach is preferred
8 over the use of a NOAEL or LOAEL for the POD if suitable data (e.g., reflecting the most
9 sensitive sex, species, and endpoint identified) are available for modeling and, if suitable data are
10 not available, then NOAEL and LOAEL values are utilized. In this case, the data were not
11 suitable for BMD modeling and the LOAEL or NOAEL approach was used.
12 4. EPA evaluated the PBPK and empirical models available to describe kinetics following inhalation of
13 1,4-dioxane (Reitzetal.. 1990; Young etal.. 1978a: Young etal. 1978b: Young etal. 1977). EPA
14 concluded that the use of existing, revised, and recalibrated PBPK models for 1,4-dioxane were not
15 superior to default approaches for the dose-extrapolation between species. Please comment on
16 whether EPA's rationale regarding the decision to not utilize existing or revised PBPK models has
17 been adequately and transparently described and is supported by the available data. Please identify
18 and provide the rationale for any alternative approaches that should be considered or preferred to the
19 approach presented in the toxicological review.
20 Comment. Six reviewers found the decision not to utilize the available PBPK models to be
21 appropriate and supported by available data. One of these reviewers suggested presenting as part
22 of the uncertainty evaluation an adjustment of the experimental doses based on metabolic
23 saturation. Another reviewer stated Appendix B was hard to follow and that the main document
24 should include a more complete description of the model refinement effort performed by
25 Sweeney et al. (2008).
26 Two reviewers noted a complete evaluation of the models was evident; one of the reviewers
27 questioned the decision not to use the models on the basis that they were unable to fit the human
28 blood PK data for 1,4-dioxane. This reviewer suggested the rat model might fit the human blood
29 PK data, thus raising concern in the reliance on the human blood PK data to evaluate the PBPK
30 model for 1,4-dioxane. Instead, the reviewer suggested the human urinary metabolite data may be
31 sufficient to give confidence in the model. One other reviewer also questioned the accuracy of the
32 available human data. One reviewer commented that the rationale for not using the PBPK model
33 to extrapolate from high to low dose was questioned. In addition, the reviewer suggested that two
34 aspects of the model code for Reitz et al. (1990) need to be verified:
35 a. In the document, KLC is defined as a first-order rate constant and is scaled by BW°7.
36 This is inconsistent when multiplied by concentration does not result in units
37 of mg/hr. However, if the parameter is actually considered a clearance constant
38 (zero-order rate constant) then the scaling rule used, as well as the interpretations
39 provided, would be acceptable.
A-7
DRAFT - DO NOT CITE OR QUOTE
-------
1 b. It is unclear as to why AM is calculated on the basis of RAM and not RMEX. RMEX
2 seems to represent the amount metabolized per unit time.
3 Response. The U.S. EPA performed a rigorous evaluation of the PBPK models available for
4 1,4-dioxane. This effort was extensively described in Section 3.5 and in Appendix B. In short,
5 several procedures were applied to the human PBPK model to determine if an adequate fit of the
6 model to the empirical model output or experimental observations could be attained using
7 biologically plausible values for the model parameters. The re-calibrated model predictions for
8 blood 1,4-dioxane levels did not come within 10-fold of the experimental values using measured
9 tissue:air partition coefficients of (Leung and Paustenbach. 1990) or (Sweeney et al.. 2008)
10 (Figure B-8 and Figure B-9). The utilization of a slowly perfused tissue:air partition coefficient
11 10-fold lower than measured values produces exposure-phase predictions that are much closer to
12 observations, but does not replicate the elimination kinetics (Figure B-10). Re-calibration of the
13 model with upper bounds on the tissue:air partition coefficients results in predictions that are still
14 six- to sevenfold lower than empirical model prediction or observations (Figure B-12 and
15 Figure B-13). Exploration of the model space using an assumption of first-order metabolism
16 (valid for the 50 ppm inhalation exposure) showed that an adequate fit to the exposure and
17 elimination data can be achieved only when unrealistically low values are assumed for the slowly
18 perfused tissue:air partition coefficient (Figure B-16). Artificially low values for the other
19 tissue:air partition coefficients are not expected to improve the model fit, as these parameters are
20 shown in the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km.
21 In the absence of actual measurements for the human slowly perfused tissue:air partition
22 coefficient, high uncertainty exists for this model parameter value. Differences in the ability of rat
23 and human blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this is
24 expected to be evident in very different values for rat and human blood:air partition coefficients,
25 which is not the case (Table B-l). Therefore, some other, as yet unknown, modification to model
26 structure may be necessary.
27 The results of U.S. EPA model evaluation were confirmed by other investigators (Sweeney et al..
28 2008). Sweeney et al. (2008) concluded that the available PBPK model with refinements resulted
29 in an under-prediction of human blood levels for 1,4-dioxane by six- to seven fold. It is
30 anticipated that the high uncertainty in predictions of the PBPK model for 1,4-dioxane would not
31 result in a more accurate derivation of human health toxicity values.
32 Because it is unknown whether the parent or the metabolite is the toxic moiety, analyses were not
33 conducted to adjust the experimental doses on the basis of metabolic saturation.
34 The discussion of Sweeney et al. (2008) was expanded in the main document in Section 3.5.3. In
35 the absence of evidence to the contrary, the Agency cannot discount the human blood kinetic data
36 published by Young et al. (1977). Even though the PBPK model provided satisfactory fits to the
37 rodent kinetic data, it was not used to extrapolate from high dose to low dose in the animal
38 because an internal dose metric was not identified and external doses were utilized in derivation
39 of the toxicity values.
A-8
DRAFT - DO NOT CITE OR QUOTE
-------
1 KLC was implemented by the U.S. EPA during the evaluation of the model and should have been
2 described as a clearance constant (zero-order rate constant) with units of L/hr/kg070. These
3 corrections have been made in the document; however, this does not impact the model predictions
4 because it was in reference to the terminology used to describe this constant.
5 The reviewer is correct that RMEX is the rate of metabolism of 1,4-dioxane per unit time;
6 however an amount of 1,4-dioxane metabolized was not calculated in the Reitz et al. (1990)
7 model code. Thus, AM is the amount of the metabolite (i.e., HEAA) in the body rather than the
8 amount metabolized of 1,4-dioxane. RAM was published by Reitz et al. (1990) as equation 2 for
9 the change in the amount of metabolite in the body per unit time. AMEX is the amount of the
10 metabolite excreted in the urine. While the variables used are confusing, the code describes the
11 metabolism of 1,4-dioxane as published in the manuscripts. The comments in the model code
12 were updated to make this description more clear (Appendix B).
13 5. Please comment on the selection of the uncertainty factors applied to the POD for the derivation of
14 the RfD. For instance, are they scientifically justified and transparently and objectively described in
15 the document? If changes to the selected uncertainty factors are proposed, please identify and provide
16 a rationale(s). Please comment specifically on the following uncertainty factors:
17 • An interspecies uncertainty factor of 10 was used to account for uncertainties in extrapolating
18 from laboratory animals to humans because a PBPK model to support interspecies extrapolation
19 was not suitable.
20 • An intraspecies (human variability) uncertainty factor of 10 was applied in deriving the RfD
21 because the available information on the variability in human response to 1,4-dioxane is
22 considered insufficient to move away from the default uncertainty factor of 10.
23 • A database uncertainty factor of 3 was used to account for lack of adequate reproductive
24 toxicity data for 1,4-dioxane, and in particular absence of a multigeneration reproductive
25 toxicity study. Has the rationale for the selection of these uncertainty factors been transparently
26 and objectively described in the document? Please comment on whether the application of these
27 uncertainty factors has been scientifically justified.
28 Comment. One reviewer noted the uncertainty factors appear to be the standard default choices
29 and had no alternatives to suggest.
30 o Five reviewers agreed that the use of an uncertainty factor of 10 for the interspecies
31 extrapolation is fully supportable. One reviewer suggested using BW3/4 scaling rather than
32 an uncertainty factor of 10 for animal to human extrapolation. Along the same lines, one
33 reviewer suggested a steady-state quantitative analysis to determine the importance of
34 pulmonary clearance and hepatic clearance and stated that if hepatic clearance scales to
35 body surface and pulmonary clearance is negligible, then an adjusted uncertainty factor
36 based on body surface scaling would be more appropriate.
A-9
DRAFT - DO NOT CITE OR QUOTE
-------
1 o Seven reviewers stated that the uncertainty factor of 10 for interindividual variability
2 (intraspecies) is fully supportable.
3 o Six reviewers commented that the uncertainty factor of 3 for database deficiencies is fully
4 justifiable. One reviewer suggested adding text to clearly articulate the science policy for
5 the use of a factor of 3 for database deficiencies.
6 Response. The preferred approach to interspecies scaling is the use of a PBPK model; however,
7 the PBPK models available for 1,4-dioxane are not suitable for use in this health assessment as
8 outlined elsewhere. Another approach that has been commonly implemented in the cancer
9 assessments is the use of body weight scaling based on body surface area (BW3/4 scaling). It is not
10 standard practice to apply BW3/4 scaling in noncancer assessments at this time. The current
11 default approach used by the Agency when PBPK models are not available for extrapolation is
12 the application of an UFA of 10, which was implemented in this assessment.
13 The absence of a multigenerational reproductive study is why the uncertainty factor for database
14 deficiencies (UFD) was retained; however, it was reduced from 10 to 3. In the text in Section
15 5.1.3 text was included to clearly state that because of the absence of a multigenerational
16 reproductive study for 1,4-dioxane an uncertainty factor of 3 was used for database deficiencies.
17 No other changes regarding the use of the uncertainty factors were made to the document.
A.1.3 Carcinogenicity of 1,4-dioxane
18 1. Under the EPA's 2005 Guidelines for Carcinogen Risk Assessment
19 (www.epa.gov/iris/backgr-d.htm), the Agency concluded that 1,4-dioxane is likely to be carcinogenic
20 to humans. Please comment on the cancer weight of evidence characterization. Has the scientific
21 justification for the weight of evidence descriptor been sufficiently, transparently and objectively
22 described? Do the available data for both liver tumors in rats and mice and nasal, mammary, and
23 peritoneal tumors in rats support the conclusion that 1,4-dioxane is a likely human carcinogen?
24 Comment. All reviewers agreed with the Agency's conclusion that 1,4-dioxane is "likely to be
25 carcinogenic to humans". However, two reviewers also thought 1,4-dioxane could be categorized
26 as a potential human carcinogen, since low-dose environmental exposures would be unlikely to
27 result in cancer. One reviewer also suggested providing a brief recapitulation of the guidance
28 provided by the 2005 Guidelines for Carcinogen Risk Assessment regarding classification of a
29 compound as likely to be carcinogenic to humans and how a chemical falls into this category.
30 Response. The document includes a weight-of-evidence approach to categorize the carcinogenic
31 potential of 1,4-dioxane. This was included in Section 4.7.1 based upon U.S. EPA's Guidelines
32 for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-Dioxane can be described as likely to be
33 carcinogenic to humans based on evidence of liver carcinogenicity in several 2-year bioassays
34 conducted in three strains of rats, two strains of mice, and in guinea pigs. Additionally, tumors in
35 other organs and tissues have been observed in rats due to exposure to 1,4-dioxane.
A-10
DRAFT - DO NOT CITE OR QUOTE
-------
1 2. Evidence indicating the mode of action of carcinogenicity of 1,4-dioxane was considered. Several
2 hypothesized MOAs were evaluated within the Toxicological Review and EPA reached the
3 conclusion that a MOA(s) could not be supported for any tumor types observed in animal models.
4 Please comment on whether the weight of the scientific evidence supports this conclusion. Please
5 comment on whether the rationale for this conclusion has been transparently and objectively
6 described. Please comment on data available for 1,4-dioxane that may provide significant biological
7 support for a MOA beyond what has been described in the Toxicological Review. Considerations
8 should include the scientific support regarding the plausibility for the hypothesized MOA(s), and the
9 characterization of uncertainty regarding the MOA(s).
10 Comment. Three reviewers commented that the weight of evidence clearly supported the
11 conclusion that a mode of action could not be identified for any of the tumor sites. One reviewer
12 commented that there is inadequate evidence to support a specific MOA with any confidence and
13 low-dose linear extrapolation is necessary; this reviewer also pointed out that EPA should not rule
14 out a metabolite as the toxic moiety.
15 One reviewer stated this was outside of his/her area of expertise but indicated that the discussion
16 was too superficial and suggested adding statements as to what the Agency would consider
17 essential information to make a determination about a MOA.
18 Two reviewers commented that even though the MOA for 1,4-dioxane is not clear there is
19 substantial evidence that the MOA is non-genotoxic. One of these reviewers also suggested that a
20 nonlinear cancer risk assessment model should be utilized.
21 One reviewer suggested adding more text to the summary statement to fully reflect the available
22 MOA information which should be tied to the conclusion and choice of an extrapolation model.
23 Response. The Agency agrees with the reviewer not to rule out a toxic metabolite as the toxic
24 moiety. In Section 5.5.1.2 text is included relating that there is not enough information to
25 determine whether the parent compound, its metabolite(s), or a combination is responsible for the
26 observed toxicities following exposure to 1,4-dioxane.
27 It is not feasible to describe the exact data that would be necessary to conclude that a particular
28 MOA was operating to induce the tumors observed following 1,4-dioxane exposure. In general,
29 the data would fit the general criteria described in the U.S. EPA's Guidelines for Carcinogen Risk
30 Assessment (U.S. EPA. 2005a). For 1,4-dioxane, several MOA hypotheses have been proposed
31 and are explored for the observed liver tumors in Section 4.7.3. This analysis represents the extent
32 to which data could provide support for any particular MOA.
33 One reviewer suggested that the evidence indicating that 1,4-dioxane is not genotoxic supports a
34 nonlinear approach to low-dose extrapolation. In accordance with the U.S. EPA's Guidelines for
35 Carcinogen Risk Assessment (U.S. EPA, 2005a), the absence of evidence for genotoxicity does
36 not invoke the use of nonlinear low-dose extrapolation, nor does it define a MOA. A nonlinear
37 low-dose extrapolation can be utilized when a MOA supporting a nonlinear dose response is
A-ll
DRAFT - DO NOT CITE OR QUOTE
-------
1 identified. For 1,4-dioxane this is not the case; a cancer MOA for any of the tumor types observed
2 in animal models has not been elucidated. Therefore, as concluded in the Toxicological Review,
3 the application of a nonlinear low-dose extrapolation approach was not supported.
4 Additional text has been added to Section Error! Reference source not found, to relay the fact
5 that several reviewers recommended that the MOA data support the use of a nonlinear
6 extrapolation approach to estimate human carcinogenic risk associated with exposure to
7 1,4-dioxane and that such an approach should be presented in the Toxicological Review.
8 Additional text has also been added to the summary statement in Section 6.2.3 stating that the
9 weight of evidence is inadequate to establish a MOA(s) by which 1,4-dioxane induces peritoneal,
10 mammary, or nasal tumors in rats and liver tumors in rats and mice (see Section 4.7.3 for a more
11 detailed discussion of 1,4-dioxane 's hypothesized MOAs).
12 3. A two-year drinking water cancer bioassay (JBRC. 1998) was selected as the principal study for the
13 development of an oral slope factor (OSF). Please comment on the appropriateness of the selection of
14 the principal study. Has the rationale for this choice been transparently and objectively described?
15 Comment. Seven reviewers agreed with the choice of the JBRC (1998) study as the principal
16 study for the development of an OSF. However, two reviewers that agreed with the choice of
17 JBRC (1998) also commented on the description and evaluation of the study. One reviewer
18 commented the evaluation of the study should be separated from the evaluation/selection of
19 endpoints within the study. The other reviewer suggested that details on the following aspects
20 should be added to improve transparency of the study: (1) rationale for selection of doses; (2)
21 temporal information on body weight for individual treatment groups; (3) temporal information
22 on mortality rates; and (4) dosing details.
23 One reviewer thought that the complete rationale for selection of the JBRC (1998) study was not
24 provided because there was no indication of whether the study was conducted under GLP
25 conditions, and the study was not peer reviewed or published. This reviewer noted the NCI
26 (1978) study was not appropriate for use, but that the Kociba et al. (1974) study may have
27 resulted in a lower POD had they employed both sexes of mice and combined benign and
28 malignant tumors.
29 Response. Since the External Peer Review draft of the Toxicological Review of 1,4-Dioxane was
30 released (U.S. EPA. 2009b). the cancer portion of the study conducted by the JBRC laboratory
31 was published in the peer-reviewed literature as Kano et al. (2009). This manuscript was
32 reviewed by EPA. EPA determined that the data published by Kano et al. (2009) should be
33 included in the assessment of 1,4-dioxane for several reasons: (1) while the JBRC (1998) was a
34 detailed laboratory report, it was not peer-reviewed; (2) the JBRC improved the diagnosis of pre-
35 and neoplastic lesions in the liver according to the current diagnostic criteria and submitted the
36 manuscript based on this updated data; (3) the Kano et al. (2009) peer-reviewed manuscript
37 included additional information such as body weight growth curves and means and standard
38 deviations of estimated dose for both rats and mice of both sexes. Thus, the Toxicological Review
A-12
DRAFT - DO NOT CITE OR QUOTE
-------
1 was updated to reflect the inclusion of the data from Kano et al. (2009). and Appendix E was
2 added for a clear and transparent display of the data included in the multiple reports.
3 In response to the peer reviewers, dose information was updated throughout the assessment and
4 are also provided in detail in Section 4.2.1.2.6, along with temporal information on body weights
5 and mortality. Text was also added to Section 4.2.1.2.6 regarding the choice of high dose
6 selection as included in the Kano et al. (2009) manuscript. Additional discussion regarding the
7 mortality rates was also added to Section 5.4.1 in selection of the critical study for the oral cancer
8 assessment. Documentation that the study was conducted in accordance with Organization for
9 Economic Co-operation and Development (OECD) Principles of Good Laboratory Practice
10 (GLP) is provided in the manuscript (Kano et al.. 2009) and this was also added to the text in
11 Section 4.2.1.2.6.
12 4. Combined liver tumors (adenomas and carcinomas) in female CjrBDFl mice from the JBRC (1998)
13 study were chosen as the most sensitive species and gender for the derivation of the final OSF. Please
14 comment on the appropriateness of the selections of species and gender. Please comment on whether
15 the rationale for these selections is scientifically justified. Has the rationale for these choices been
16 transparently and objectively described?
17 Comment. Six reviewers agreed the female CjrBDFl mice should be used for the derivation of
18 the OSF. Five of these reviewers agreed with the rationale for the selection of the female
19 CjrBDFl mouse as the most sensitive gender and species. However, one reviewer suggested that
20 the specific rationale (i.e., that the final OSF is determined by selecting the gender/species that
21 gives the greatest OSF value) be stated clearly in a paragraph separate from the other
22 considerations of study selection.
23 One reviewer was unsure of both the scientific justification for combining benign and malignant
24 liver tumors, as well as the background incidence of the observed liver tumors in historical
25 control CjrBDFl male and female mice.
26 One reviewer commented that the scientific basis for the selection of female CjrBDFl mice was
27 unclear. This reviewer thought that the rationale for the choice of this strain/sex compared to all
28 others was not clearly articulated.
29 Response. Using the approach described in the Guidelines for Carcinogen Risk Assessment (U.S.
30 EPA. 2005a) studies were first evaluated based on their quality and suitability for inclusion in the
31 assessment. Once the studies were found to be of sufficient quality for inclusion in the
32 assessment, the dose-response analysis was performed with the goal of determining the most
33 appropriate endpoint and species for use in the derivation of an OSF. These topics are discussed
34 in detail in Section 4.7 and 5.4.
35 Benign and malignant tumors that arise from the same cell type (e.g., hepatocellular) may be
36 combined to more clearly identify the weight of evidence for a chemical. This is in accordance
37 with the U.S. EPA 2005 Guidelines for Carcinogen Risk Assessment as referenced in the
A-13
DRAFT - DO NOT CITE OR QUOTE
-------
1 lexicological Review. In the absence of a MOA (MOA analysis described in detail in Section
2 4.7.) for 1,4-dioxane carcinogenicity, it is not possible to determine which species may more
3 closely resemble humans. Text in Section 5.4.4 indicates that the calculation of an OSF for
4 1,4-dioxane is based upon the dose-response data for the most sensitive species and gender.
5 5. Has the scientific justification for deriving a quantitative cancer assessment been transparently and
6 objectively described? Regarding liver cancer, a linear low-dose extrapolation approach was utilized
7 to derive the OSF. Please provide detailed comments on whether this approach to dose-response
8 assessment is scientifically sound, appropriately conducted, and objectively and transparently
9 described in the document. Please identify and provide the rationale for any alternative approaches for
10 the determination of the OSF and discuss whether such approaches are preferred to EPA's approach.
11 Comment. Four reviewers agreed with the approach for the dose-response assessment. One
12 reviewer commented that even if a nongenotoxic MOA were identified for 1,4-dioxane it may not
13 be best evaluated by threshold modeling. One reviewer commented the use of the female mouse
14 data provided an appropriate health protective and scientifically valid approach.
15 One reviewer commented that the basic adjustments and extrapolation method for derivation of
16 the OSF were clearly and adequately described, but disagreed with the linear low-dose
17 extrapolation. This reviewer suggested that the lack of certainty regarding the MOA was not a
18 sufficient cause to default to a linear extrapolation. Another reviewer commented that the
19 rationale for a linear low-dose extrapolation to derive the OSF was not clear, but may be in
20 accordance with current Agency policy in the absence of a known MOA. This reviewer also
21 commented that 1,4-dioxane appears to be non-genotoxic and nonlinear models should be tested
22 on the available data to determine if they provide a better fit and are more appropriate.
23 One reviewer thought that the justification for a linear extrapolation was not clearly provided and
24 that a disconnect between the MOA summary and the choice of a linear extrapolation model
25 existed. In addition, this reviewer commented that the pharmacokinetic information did not
26 support the use of a linear extrapolation approach, but rather use of animal PBPK models to
27 extrapolate from high to low dose that would result in a mixture of linear and nonlinear
28 extrapolation models was warranted.
29 One reviewer suggested consideration of an integrated assessment of the cancer and noncancer
30 endpoints; however, if linear low-dose extrapolation remains the approach of choice by the
31 Agency, then the effect of choosing BMRs other than 10% was recommended to at least be
32 included in the uncertainty discussion. Using BMRs lower than 10% may allow for the
33 identification of a risk level for which the low-dose slope is 'best' estimated.
34 Response. The EPA conducted a cancer MOA analysis evaluating all of the available data for
35 1,4-dioxane. Application of the framework in the U.S. EPA Guidelines for Carcinogen Risk
36 Assessment (2005a) demonstrates that the available evidence to support any hypothesized MOA
37 for 1,4-dioxane-induced tumors does not exist. In the absence of a MOA, the U.S. EPA
38 Guidelines for Carcinogen Risk Assessment (2005a) indicate that a low dose linear extrapolation
A-14
DRAFT - DO NOT CITE OR QUOTE
-------
1 should be utilized for dose response analysis (see Section 5.4). Some of the potential uncertainty
2 associated with this conclusion was characterized in Section 5.5. Note that there is no scientific
3 basis to indicate that in the absence of evidence for genotoxicity a nonlinear low-dose
4 extrapolation should be used. As concluded in the Toxicological Review, the application of a
5 nonlinear low-dose extrapolation approach was not supported.
6 With regards to the PBPK model available for 1,4-dioxane, it is clear that there currently exist
7 deficiencies within the model and as such, the model was not utilized for interspecies
8 extrapolation. Given the deficiencies and uncertainty in the 1,4-dioxane model it also does not
9 provide support for a MOA.
10 Lastly, in the absence of a MOA for 1,4-dioxane carcinogenicity it is not possible to harmonize
11 the cancer and noncancer effects to assess the risk of health effects due to exposure. However, the
12 choice of the BMDLi0,which was more than 15-fold lower than the response at the lowest dose
13 (66 mg/kg-day), was reconsidered in response to a public comment. BMDs and BMDLs were
14 calculated using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and BMDL50). A
15 BMR of 50% was used as it resulted in a BMDL closest to the response level at the lowest dose
16 tested in the bioassay.
A.2 Public Comments
17 Comments on the Toxicological Review of 1,4-Dioxane submitted by the public are summarized
18 below in the following categories: Oral reference dose for 1,4-dioxane, carcinogenicity of
19 1,4-dioxane, PBPK modeling, and other comments.
A.2.1 Oral reference dose (RfD) for 1,4-dioxane
20 Comment: An UF for database deficiencies is not necessary because of considerable evidence
21 showing no reproductive or developmental effects from 1,4-dioxane exposure.
22 Response: Due to the lack of a multigenerational reproductive study for 1,4-dioxane an UF of 3
23 was retained for database deficiencies. Without clear evidence showing a lack of reproductive or
24 developmental effects in a multigenerational reproductive study, there is still uncertainty in this
25 area.
26
A.2.2 Carcinogenicity of 1,4-dioxane
27 Comment: Using liver tumors as the basis for the oral CSF is more appropriate than nasal tumors
28 (1988 IRIS assessment of 1,4-dioxane); however, the use of mouse liver tumor data is
A-15
DRAFT - DO NOT CITE OR QUOTE
-------
1 inappropriate because it is inconsistent with other liver models both quantitatively and in the
2 dose-response pattern. High mortality rates in the study are also a limitation. Liver tumor data
3 from rats should be used instead, which represents a better animal model for 1,4-dioxane
4 carcinogenicity assessment.
5 Response: Even though the dose-response is different for mice and rats, the female mice were
6 considered to be appropriate for the carcinogenicity assessment for several reasons. The female
7 mouse liver tumors from the Kano et al. (2009) report were found to be the most sensitive species
8 and endpoint. Section 4.2.1.2.6 was updated to include additional information on mortality rates.
9 The majority of the animals lived past 52 weeks (only 4 females died prior to 52 weeks, 2 in each
10 the mid- and high-dose groups). The cause of death in the female mice that died between 1 and 2
11 years was attributed to liver tumors.
12 Comment: The OSF was based on the most sensitive group, Crj:BDFl mice; however BDF1
13 mice have a high background rate of liver tumors. The incidence of liver tumors in historical
14 controls for this gender/species should be considered in the assessment. Sensitivity of the test
15 species/gender as well as other criteria should be considered in the selection of the appropriate
16 study, including internal and external validity as outlined in Lewandowski and Rhomberg (2005).
17 The female Crj:BDFl mice had a low survival rate that should be considered in the selection of
18 the animal model for 1,4-dioxane carcinogenicity.
19 Response. Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and
20 carcinomas in control male and female BDF1 mice from ten 2-year bioassays at the JBRC. For
21 female mice, out of 499 control mice, the incidence rates were 4.4% for hepatocellular adenomas
22 and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported a 10% incidence rate for
23 hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in control female
24 BDF1. These incidence rates are near the historical control values and thus are appropriate for
25 consideration in this assessment. Additional text regarding these historical controls was added to
26 the study description in Section 4.2.1.2.6.
27 Comment: Low-dose linear extrapolation for the oral CSF is not appropriate nor justified by the
28 data. The weight of evidence supports a threshold (nonlinear) MOA when metabolic pathway is
29 saturated at high doses. Nonlinear extrapolations should be evaluated and presented for
30 1,4-dioxane. Oral CSFs should be derived and presented using both the BW3/4 scaling as well as
31 available PBPK models to extrapolate across species.
32 Response: The absence of evidence for genotoxicity/mutagenicity does not indicate the use of
33 nonlinear low-dose extrapolation. For 1,4-dioxane, a MOA to explain the induction of tumors
34 does not exist so the nature of the low-dose region of the dose-response is unknown. The oral
35 CSF for 1,4-dioxane was derived using BW3/4 scaling for interspecies extrapolation. The PBPK
36 and empirical models available for 1,4-dioxane were evaluated and found not to be adequate for
37 use in this assessment, described in detail in Appendix B.
A-16
DRAFT - DO NOT CITE OR QUOTE
-------
1 Comment: The POD for the BDF1 female mouse is 15-fold lower than the lowest dose in the
2 bioassay, thus the POD is far below the lower limit of the data and does not follow the U.S.
3 EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a).
4 Response. The comment is correct that the animal BMDL10 was more than 15-fold lower than the
5 response at the lowest dose (66 mg/kg-day) in the bioassay. BMDs and BMDLs were calculated
6 using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and BMDL50). A BMR of
7 50% was chosen as it resulted in a BMDL closest to the response level at the lowest dose tested in
8 the bioassay.
9 Comment. The geometric mean of the oral cancer slope factors (as done with B[a]P & DDT)
10 should have been used instead of relying on the female BDF1 mouse data, since a MOA could not
11 be determined for 1,4-dioxane.
12 Response. In accordance with the BMD technical guidance document (U.S. EPA. 2000a).
13 averaging tumor incidence is not a standard or default approach. Averaging the tumor incidence
14 response diminishes the effect seen in the sensitive species/gender.
15 Comment. EPA should critically reexamine the choice of JBRC (1998) as the principal study
16 since it has not been published or peer-reviewed. A transcript of e-mail correspondence should be
17 provided.
18 Response. JBRC (1998) was published as conference proceedings as Yamazaki et al. (1994) and
19 recently in the peer-reviewed literature as Kano et al. (2009). Additional study information was
20 also gathered from the authors (Yamazaki. 2006) and is available upon request from the IRIS
21 Hotline. The peer-reviewed and published data from Kano et al. (2009) was incorporated into the
22 final version of the Toxicological Review ofl,4-Dioxane.
23 Comment. The WOE does not support a cancer descriptor of likely to be carcinogenic to humans
24 determination, but rather suggestive human carcinogen at the high dose levels used in rodent
25 studies seems more appropriate for the following reasons: 1) lack of conclusive human
26 epidemiological data; 2) 1,4-dioxane is not mutagenic; and 3) evidence at high doses it would act
27 via cell proliferation MOA.
28 Response: A cancer classification of "likely, " based on evidence of liver carcinogenicity in
29 several two-year bioassays conducted in three strains of rats, two strains of mice, and in guinea
30 pigs was chosen. Also, mesotheliomas of the peritoneum, mammary, and nasal tumors have been
31 observed in rats. The Agency agrees that human epidemiological studies are inconclusive. The
32 evidence at any dose is insufficient to determine a MOA.
33
A-17
DRAFT - DO NOT CITE OR QUOTE
-------
A.2.3 PBPK Modeling
1 Comment. EPA should have used and considered PBPK models to derive the oral toxicity values
2 (rat to human extrapolation) rather than relying on a default method. The draft did not consider
3 the Sweeney et al. (2008) model. The PBPK model should be used for both noncancer and cancer
4 dose extrapolation.
5 Response: The Agency evaluated the Sweeney et al. (2008) publication and this was included in
6 Appendix B of the document. Text was added to the main document in Section 3.5.2.4 and 3.5.3
7 regarding the evaluation of Sweeney et al. (2008). This model was determined not to be
8 appropriate for interspecies extrapolation. Additionally, see response to the external peer review
9 panel comment B4.
10 Comment: EPA should use the modified inhalation inputs used in the Reitz et al. (1990) model
11 and the updated input parameters provided in Sweeney et al. (2008) and add a compartment for
12 the kidney
13 Response: See response to previous comment regarding evaluation of Sweeney et al. (2008).
14 Modification of the model to add a kidney compartment is not within the scope of this
15 assessment.
A.2.4 Other Comments
16 Comment: EPA should consider the Kasai et al. (2009; 2008) studies for inhalation and MOA
17 relevance.
18 Response: The 13 week and 2-year inhalation studies by Kasai et al. (2009; 2008) were published
19 late in the development stage of this assessment. The IRIS Program will evaluate these recently
20 published 1,4-dioxane inhalation data for the potential to derive an RfC in a separate assessment.
21 Comment: 1,4-Dioxane is not intentionally added to cosmetics and personal care products -
22 correct sentence on page 4.
23 Response: This oversight was corrected in the document.
A-18
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX B. EVALUATION OF EXISTING PBPK
MODE LS FOR 1,4-DIOXANE
B.1 Background
1 Several pharmacokinetic models have been developed to predict the absorption, distribution,
2 metabolism, and elimination of 1,4-dioxane in rats and humans. Single compartment, empirical models
3 for rats (Young et al., 1978a; 1978b) and humans (Young et al.. 1977) were developed to predict blood
4 levels of 1,4-dioxane and urine levels of the primary metabolite, (3-hydroxyethoxy acetic acid (HEAA).
5 Physiologically based pharmacokinetic (PBPK) models that describe the kinetics of 1,4-dioxane using
6 biologically realistic flow rates, tissue volumes and affinities, metabolic processes, and elimination
7 behaviors, were also developed (Fisher et al.. 1997; Leung and Paustenbach. 1990; Reitz etal.. 1990).
8 In developing updated toxicity values for 1,4-dioxane, the available PBPK models were evaluated
9 for their ability to predict observations made in experimental studies of rat and human exposures to
10 1,4-dioxane. The model of Reitz et al. (1990) was identified for further consideration to assist in the
11 derivation of toxicity values. Issues related to the biological plausibility of parameter values in the Reitz
12 et al. (1990) human model were identified. The model was able to predict the only available human
13 inhalation data set (Young et al.. 1977) by increasing (i.e., doubling) parameter values for human alveolar
14 ventilation, cardiac output, and the blood:air partition coefficient above the measured values.
15 Furthermore, the measured value for the slowly perfused tissue:air partition coefficient (i.e., muscle) was
16 replaced with the measured liver value to improve the fit. Analysis of the Young et al. (1977) human data
17 suggested that the apparent volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold higher
18 in rats than humans, presumably due to species differences in tissue partitioning or other process not
19 represented in the model. Subsequent exercising of the model demonstrated that selecting a human slowly
20 perfused tissue:air partition coefficient much lower than the measured rat value resulted in better
21 agreement between model predictions of 1,4-dioxane in blood and experimental observations. Based upon
22 these observations, several model parameters (e.g., metabolism/elimination parameters) were
23 re-calibrated using biologically plausible values for flow rates and tissue:air partition coefficients.
24 This appendix describes activities conducted in the evaluation of the empirical models (Young et
25 al. (1978b; 1978a; 1977)). and re-calibration and exercising of the Reitz et al. (1990) PBPK model, and
26 evaluation of the Sweeney et al. (2008) model to determine the potential utility of the PBPK models for
27 1,4-dioxane for interspecies and route-to-route extrapolation.
B.2 Scope
28 The scope of this effort consisted of implementation of the Young et al. (1978b; 1978a; 1977)
29 empirical rat and human models using the acslXtreme simulation software, re-calibration of the Reitz et
30 al. (1990) human PBPK model, and evaluation of model parameters published by Sweeney et al. (2008).
B-l
DRAFT - DO NOT CITE OR QUOTE
-------
1 Using the model descriptions and equations given in Young et al. (1978b; 1978a; 1977). model code was
2 developed for the empirical models and executed, simulating the reported experimental conditions. The
3 model output was then compared with the model output reported in Young et al. (1978b; 1978a; 1977).
4 The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for cardiac and
5 alveolar flow rates and tissue:air partition coefficients. The predictions of blood and urine levels of
6 1,4-dioxane and HEAA, respectively, from the re-calibrated model were compared with the empirical
7 model predictions of the same dosimeters to determine whether the re-calibrated PBPK model could
8 perform similarly to the empirical model. As part of the PBPK model evaluation, EPA performed a
9 sensitivity analysis to identify the model parameters having the greatest influence on the primary
10 dosimeter of interest, the blood level of 1,4-dioxane. Variability data for the experimental measurements
11 of the tissue: air partition coefficients were incorporated to determine a range of model outputs bounded
12 by biologically plausible values for these parameters. Model parameters from Sweeney et al. (2008) were
13 also tested to evaluate the ability of the PBPK model to predict human data following exposure to
14 1,4-dioxane.
B.3 Implementation of the Empirical Models in aclsXtreme
15 The empirical models of Young et al. (1978b: 1978a: 1977) for 1,4-dioxane in rats and humans
16 were reproduced using acslXtreme, version 2.3 (Aegis Technologies, Huntsville, AL). Model code files
17 were developed using the equations described in the published papers. Additional files containing
18 experiment-specific information (i.e., BWs, exposure levels, and duration) were also generated.
B.3.1 Model Descriptions
19 The empirical model of Young et al. (1978b: 1978a) for 1,4-dioxane in rats is shown in
20 Figure B-l. This is a single-compartment model that describes the absorption and metabolism kinetics of
21 1,4-dioxane in blood and urine. No information is reported describing pulmonary absorption or
22 intravenous (i.v.) injection/infusion of 1,4-dioxane. The metabolism of 1,4-dioxane and subsequent
23 appearance of HEAA is described by Michaelis-Menten kinetics governed by a maximum rate (Vmax,
24 ug/mL-hour) and affinity constant (Km, ug/mL) . Both 1,4-dioxane and HEAA are eliminated via the
25 first-order elimination rate constants, ke and kme, respectively (hour"1) by which 35% of 1,4-dioxane and
26 100% of HEAA appear in the urine, while 65% of 1,4-dioxane is exhaled. Blood concentration of
27 1,4-dioxane is determined by dividing the instantaneous amount of 1,4-dioxane in blood by a Vd of 301
28 mL/kg BW.
B-2
DRAFT - DO NOT CITE OR QUOTE
-------
Inhalation (kTNH)
i.v. admin
dt
- - k, x Diox,,,
dt
„ „.
K+DlOX
-~km
tody
k x Diox
z+
Exhaled (65%)
Urine (35%)
k xHEAA
Urine
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
Figure B-l Schematic representation of empirical model for 1,4-dioxane in rats.
1 Figure B-2 illustrates the empirical model for 1,4-dioxane in humans as described in Young et al.
2 (1977). Like the rat model, the human model predicts blood 1,4-dioxane and urinary 1,4-dioxane and
3 HEAA levels using a single-compartment structure. However, the metabolism of 1,4-dioxane to HEAA in
4 humans is modeled as a first-order process governed by a rate constant, KM (hour"1). Urinary deposition of
5 1,4-dioxane and HEAA is described using the first order rate constants, ke(dlox) and kme(HEAA), respectively.
6 Pulmonary absorption is described by a fixed rate of 76.1 mg/hour (k^). Blood concentrations of
7 1,4-dioxane and HEAA are calculated as instantaneous amount (mg) divided by Vd(diOX) or
8 respectively (104 and 480 mL/kg BW, respectively).
Inhalation
Dioxane
HEAA
"d(HEAA) X
e (diox)
(HEAA)
Urine
Cumulative
Dioxane and
HEAA
Source: Reprinted with permission of Taylor & Francis, Young et al. (1977).
Figure B-2 Schematic representation of empirical model for 1,4-dioxane in humans.
B.3.2 Modifications to the Empirical Models
10 Several modifications were made to the empirical models. The need for the modifications arose in
11 some cases from incomplete reporting of the Young et al. (1978b; 1978a; 1977) studies and in other cases
12 from the desire to add capabilities to the models to assist in the derivation of toxicity values.
13 For the rat model, no information was given by Young et al. (1978b: 1978a) regarding the
14 parameterization of pulmonary absorption (or exhalation) or i.v. administration of 1,4-dioxane. Therefore,
15 additional parameters were added to simulate these processes in the simplest form. To replicate
DRAFT - DO NOT CITE OR QUOTE
B-3
-------
1 1,4-dioxane inhalation, a first-order rate constant, k^n (hour"1), was introduced. kINH was multiplied by
2 the inhalation concentration and the respiratory minute volume of 0.238 L/minute (Young et al.. 1978b:
3 1978a). The value for kWH was estimated by optimization against the blood time course data of Young et
4 al. (1978b: 1978a). Intravenous (i.v.) administration was modeled as instantaneous appearance of the full
5 dose at the start of the simulation. Rat urinary HEAA data were reported by Young et al. (1978b; 1978a)
6 in units of concentration. To simulate urinary HEAA concentration, an estimate of urine volume was
7 required. Since observed urinary volumes were not reported by Young et al. (1978b; 1978a). a standard
8 rat urine production rate of 0.00145 L/hour was used.
9 For humans, Young et al. (1977) used a fixed 1,4-dioxane inhalation uptake rate of 76.1 mg/hour,
10 which corresponded to observations during a 50 ppm exposure. In order to facilitate user-specified
11 inhalation concentrations, pulmonary absorption was modeled. The modeling was performed identically
12 to the rat model, but using a human minute volume of 7 L/minute. Urinary HEAA data were reported by
13 Young et al. (1977) as a cumulative amount (mg) of HEAA. Cumulative amount of HEAA in the urine is
14 readily calculated from the rate of transfer of HEAA from plasma to urine, so no modification was
15 necessary to simulate this dose metric for humans.
16 Neither empirical model of Young et al. (1978b: 1978a: 1977) described oral uptake of
17 1,4-dioxane. Adequate data to estimate oral absorption parameters are not available for either rats or
18 humans; therefore, neither empirical model was modified to include oral uptake.
B.3.3 Results
19 The acslXtreme implementation of the Young et al. (1978b; 1978a) rat empirical model simulates
20 the 1,4-dioxane blood levels from the i.v. experiments identically to the model output reported in the
21 published paper (Figure B-3). However, the acslXtreme version predicts urinary HEAA concentrations in
22 rats that are approximately threefold lower and reach a maximum sooner than the predicted levels
23 reported in the paper (Figure B-4). These discrepancies may be due, at least in part, to the reliance in the
24 acslXtreme implementation on a constant, standard, urine volume rather than experimental measurements,
25 which may have been different from the assumed value and may have varied over time. Unreported
26 model parameters (e.g., lag times for appearance of excreted HEAA in bladder urine) may also contribute
27 to the discrepancy.
B-4
DRAFT - DO NOT CITE OR QUOTE
-------
Observations and predictions of 1,4-dioxane in rat blood
following 3 to 1000 mg/kg IV injection
30 40 50
Time(hrs)
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
Figure B-3 Output of 1,4-dioxane blood level data from the acslXtreme
implementation (left) and published (right) empirical rat model
simulations of i.v. administration experiments.
Observations and predictions of HEAA in rat urine
following 10 or 1000 mg/kg IV injection
20 30
Time (hrs)
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
Figure B-4 Output of HEAA urine level data from acslXtreme implementation
(left) and published (right) empirical rat model simulations of i.v.
administration experiments.
The Young et al. (1978b: 1978a) report did not provide model predictions for the 50-ppm
inhalation experiment. However, the acslXtreme implementation produces blood 1,4-dioxane predictions
that are quite similar to the reported observations (Figure B-5). As with the urine data from the i.v.
B-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 experiment, the acslXtreme-predicted urinary HEAA concentrations are approximately threefold lower
2 than the observations, presumably for the same reasons discussed above for the i.v. predictions.
Observations and predictions of 1,4-dioxane in rat blood
following a 6 hour 50 ppm inhalation exposure
—
D
acsl version - Youn
(1978a, b) empirical
Young etal. (1978a
observations
etal.
model
b)
Time (hrs)
Observations and predictions of HEAA in rat urine
following a 6 hour 50 ppm inhalation exposure
25.0
20.0-
3 1
Ul
5.0-
0.0
acsl version -Young et
al. (1978a, b) empirical
model
d Young etal. (1978a, b)
observations
[
X
[
/^~
10
20 30
Time (hrs)
40
50
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
Figure B-5 acslXtreme predictions of blood 1,4-dioxane and urine HEAA levels
from the empirical rat model simulations of a 6-hour, 50-ppm inhalation
exposure.
3 Inhalation data for a single exposure level (50 ppm) are available for humans. The acslXtreme
4 predictions of the blood 1,4-dioxane observations are identical to the predictions reported in Young et al.
5 (1977) (Figure B-6). Limited blood HEAA data were reported, and the specimen analysis was highly
6 problematic (e.g., an analytical interference was sometimes present from which HEAA could not be
7 separated). For this reason, Young et al. (1977) did not compare predictions of the blood HEAA data to
8 observations in their manuscript.
Observations and predictions of 1,4-dioxane in human
blood following a 6 hour 50 ppm inhalation exposure
D
acsl version - Young et
al (1977) empirical model
observed
5 10
Time (hrs)
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
DRAFT - DO NOT CITE OR QUOTE
B-6
-------
Figure B-6 Output of 1,4-dioxane blood level data from the acslXtreme
implementation (left) and published (right) empirical human model
simulations of a 6-hour, 50-ppm inhalation exposure.
1 Data for cumulative urinary HEAA amounts are provided in Young et al. (1977). and no
2 analytical problems for these data were reported. Nevertheless, model predictions for urinary HEAA were
3 not presented in the manuscript. The acslXtreme prediction of the HEAA kinetics profile is similar to the
4 observations, although predicted values are approximately 1.5- to 2-fold lower than the observed values
5 (Figure B-7). Unlike urinary HEAA observations in the rat, human observations were reported as
6 cumulative amount produced, negating the need for urine volume data. Therefore, discrepancies between
7 model predictions and experimental observations for humans cannot be attributed to uncertainties in urine
8 volumes in the subjects. Further evaluation of the Young et al (1977) empirical model was conducted
9 against subchronic inhalation exposure data reported by Kasai et al. (2008). In the experimental study.
10 male and female F344 rats were exposed to 0. 100. 200. 400. 800. 1.600. 3.200. or 6.400ppm 1.4-dioxane
11 in a 13-week inhalation study. The simulations of the Young et al. (1977) model did not provide an
12 adequate fit (Figure B-8) for the measured plasma levels at each exposure level of 1.4-dioxane as reported
13 bv Kasai etal. (2008).
Observations and predictions of HEAA in human urine
following a 6 hour 50 ppm inhalation exposure
700.0
600.0
>> ro 500.0 -
|
II
2 <
S
O I
400.0
300.0
200.0
100.0
D
acsl version - Young et al
(1977) empirical model
observed
10 15
Time (hrs)
20
25
Source: Reprinted with permission of Taylor & Francis, Young et al. (1977).
Figure B-7 Observations and acslXtreme predictions of cumulative HEAA in
human urine following a 6-hour, 50-ppm inhalation exposure.
B-7
DRAFT - DO NOT CITE OR QUOTE
-------
3000
2500
2000
1500
•*. 1000
500
+ Male Rat Data
+ Female Rat Data
-*—Simulated Data
500
1000
1500 2000
Dose
2500
3000
3500
Figure B-8 EPA-modified Young et al. empirical model prediction (line) of plasma
1,4-dioxane levels in rats following exposure to 1,4-dioxane for 13 weeks
compared to data from Kasai et al. (2008).
B.3.4 Conclusions for Empirical Model Implementation
1 The empirical models described by Young et al. (1978b; 1978a; 1977) for rats and humans were
2 implemented using acslXtreme. The models were modified to allow for user-defined inhalation levels by
3 addition of a first-order rate constant for pulmonary uptake of 1,4-dioxane, fitted to the inhalation data.
4 No modifications were made for oral absorption as adequate data are not available for parameter
5 estimation. The acslXtreme predictions of 1,4-dioxane in the blood are identical to the published
6 predictions for simulations of 6-hour, 50-ppm inhalation exposures in rats and humans and 3 to
7 1,000 mg/kg i.v. doses in rats (Figure B-3, Figure B-5, and Figure B-6). However, the acslXtreme version
8 predicts lower urinary HEAA concentrations in rats appearing earlier than either the Young et al. (1978b:
9 1978a) model predictions or the experimental observations. The lower predicted urinary HEAA levels in
10 the acslXtreme implementation for rats is likely due to use of default values for urine volume in the
11 absence of measured volumes. The reason for differences in time-to-peak levels is unknown, but may be
12 the result of an unreported adjustment by Young et al. (1978b: 1978a) in model parameter values.
13 Additionally, the modified Young et al. (1978b; 1978a; 1977) model failed to provide adequate fit to
14 blood data reported following subchronic inhalation of 1.4-dioxane in rats (Kasai et al.. 2008). For
15 humans, Young et al. (1977) did not report model predictions of urinary HEAA levels. The urinary
16 HEAA levels predicted by acslXtreme were low relative to the observations. However, unlike the
17 situation in rats, these data are not dependent on unreported urine volumes (observations were reported as
18 cumulative HEAA amount rather than HEAA concentration), but reflect the model parameter values
19 reported by Young et al. (1977). Presently, there is no explanation for the lack of fit of the reported
20 urinary HEAA elimination rate constant to the observations.
DRAFT - DO NOT CITE OR QUOTE
B-8
-------
B.4 Initial Recalibration of the PBPK Model
1 Concern regarding adjustments made to some of the parameter values in Reitz et al. (1990)
2 prompted a re-calibration of the Reitz et al. (1990) human PBPK model using more biologically plausible
3 values for all measured parameter values. Reitz et al. (1990) doubled the measured physiological flows
4 and blood:air partition coefficient and substituted the slowly-perfused tissue:air partition coefficient with
5 the liver:air value in order to attain an adequate fit to the observations. This approach increases
6 uncertainty in these parameter values, and in the utilization of the model for cross-species dose
7 extrapolation. Therefore, the model was re-calibrated using parameter values that are more biologically
8 plausible to determine whether an adequate fit of the model to the available data can be attained.
B.4.1 Sources of Values for Flow Rates
9 The cardiac output of 30 L/hour/kg074 (Table B-l) reported by Reitz et al. (Reitz etal.. 1990) is
10 approximately double the mean resting value of 14 L/hour/kg0 74 reported in the widely accepted
11 compendium of Brown et al. (1997). Resting cardiac output was reported to be 5.2 L/minute (or 14
12 L/hour/kg074), while strenuous exercise resulted in a flow of 9.9 L/minute (or 26 L/hour/kg074) (Brown et
13 al.. 1997). Brown et al. (1997) also cite the ICRP (1975) as having a mean respiratory minute volume of
14 7.5 L/minute, which results in an alveolar ventilation rate of 5 L/minute (assuming 33% lung dead space),
15 or 13 L/minute/kg0 74. Again, this is roughly half the value of 30 L/hour/kg0 74 employed for this parameter
16 by Reitz et al. (1990). Young et al. (1977) reported that the human subjects exposed to 50 ppm for 6 hours
17 were resting inside a walk-in exposure chamber. Thus, use of cardiac output and alveolar ventilation rates
18 of 30 L/hour/kg0 74 is not consistent with the experimental conditions being simulated.
B-9
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-1 Human PBPK model parameter values for 1,4-dioxane
Parameter
Reitz et al. (1990)
Leung and
Paustenbach (1990)
Sweeney et al.
(2008)
EPAC
Physiological Flows
Cardiac output (QCCf
Alveolar ventilation (QPCf
30
30
—
—
—
—
17.0
17.7
Partition Coefficients (PCs)
Blood:air(PB)
Fatair(PFA)
Liverair (PLA)
Rapidly perfused tissue:air (PRA)
Slowly perfused tissue:air (PSA)
3,650
851
1,557
1,557
1,557
1,825 ±94
851 ± 118
1,557±114
—
997 ± 254
1,666 ±287
—
1,862 ±739"
—
1,348 ±290°
1,850
851
1,557
1,557
166
Metabolic Constants
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)d
Metabolic affinity constant (Km)e
HEAA urinary elimination rate
constant (kme)f
6.35
3.00
0.56
-
—
-
-
—
-
5.49
9.8
0.44
aL/hour/kg BWU '4
""Measurement for rat tissue
""Biologically plausible values utilized by EPA in this assessment
dmg/hour/kg BW075
emg/L
'hour'1
2 Examination of the experimental data of Young et al. (1977) yields an estimated alveolar
3 ventilation to be 7 L/minute (or 16 L/hour/kg°74) for volunteers having a mean BW of 84 kg. This rate is
4 based on the Young et al. (1977) estimate of 76.1 mg/hour for 1,4-dioxane uptake. Based on these
5 findings, the cardiac output and alveolar ventilation rates of 17.0 and 17.7 L/hour/kg°74 were biologically
6 plausible for the experimental subjects. These rate estimates are based on calculations made using
7 empirical data and are consistent with standard human values and the experimental conditions (i.e.,
8 subject exertion level) reported by Young et al. (1977). Therefore, these flow values were chosen for the
9 model re-calibration.
B.4.2 Sources of Values for Partition Coefficients
10 Two data sources are available for the tissue:air equilibrium partition coefficients for 1,4-dioxane:
11 Leung and Paustenbach (1990) and Sweeney et al. (2008). Both investigators report mean values and
12 standard deviations for human blood:air, rat liverair, and rat muscle:air (e.g., slowly perfused tissue:air),
13 while Leung and Paustenbach et al. (1990) also reported values for rat fatair (Table B-1).
B.4.3 Calibration Method
14 The PBPK model was twice re-calibrated using the physiological flow values suggested values
15 (current EPA assessment, see Table B-1) and the partition coefficients of Leung and Paustenbach (1990)
B-10
DRAFT - DO NOT CITE OR QUOTE
-------
1 and Sweeney et al. (2008) separately. For each calibration, the metabolic parameters VmaxC and Km, were
2 simultaneously fit (using the parameter estimation tool provided in the acslXtreme software) to the output
3 of 1,4-dioxane blood concentrations generated by the acslXtreme implementation of the Young et al.
4 (1977) empirical human model for a 6 hour, 50 ppm inhalation exposure. Subsequently, the HEAA
5 urinary elimination rate constant, kme, was fitted to the urine HEAA predictions from the empirical model.
6 The empirical model predictions, rather than experimental observations, were used to provide a more
7 robust data set for model fitting, as the empirical model simulation provided 240 data points (one
8 prediction every 0.1 hour) compared with hourly experimental observations, and to avoid introducing
9 error by calibrating the model to data digitally captured from Young et al. (1977).
B.4.4 Results
10 Results of the model re-calibration are provided in Table B-2. The re-calibrated values for VmaxC
11 and kme associated with the Leung and Paustenbach (1990) or Sweeney et al. (2008) tissue:air partition
12 coefficients are very similar. However, the fitted value for Km using the Sweeney et al. (2008) partition
13 coefficients is far lower (0.0001 mg/L) than that resulting from use of the Leung and Paustenbach (1990)
14 partition coefficients (2.5 mg/L). This appears to be due to the higher slowly perfused tissue:air partition
15 coefficient determined by Sweeney et al. (2008) (1,348 vs. 997), resulting in a higher apparent Vd than if
16 the Leung and Paustenbach (1990) value is used. Thus, the optimization algorithm selects a low Km,
17 artificially saturating metabolism in an effort to drive predicted blood 1,4-dioxane levels closer to the
18 empirical model output. Saturation of metabolism during a 50 ppm inhalation exposure is inconsistent
19 with the observed kinetics.
Table B-2 PBPK metabolic and elimination parameter values resulting from re-calibration of the
human model using alternative values for physiological flow rates3 and tissue:air
partition coefficients
Source of Partition Coefficients Leung and Paustenbach (1990) Sweeney et al. (2008)
Maximum rate for 1,4-dioxane metabolism (Vmaxc) 16.9 20.36
Metabolic affinity constant (Km)c 2.5 0.0001
HEAA urinary elimination rate constant (kme) 0.18 0.17
aCardiac output = 17.0 L/hour/kg BWU'4, alveolar ventilation = 17.7 L/hour/kg BWU '4
bmg/hour/kg BW075
>g/L
20 Plots of predicted and experimentally observed blood 1,4-dioxane and urinary HEAA levels are
21 shown in Figure B-9. Neither re-calibration resulted in an adequate fit to the blood 1,4-dioxane data from
22 the empirical model output or the experimental observations. Re-calibration using either the Leung and
23 Paustenbach (1990) or Sweeney et al. (2008) partition coefficients resulted in blood 1,4-dioxane
24 predictions that were at least 10-fold lower than empirical model predictions or observations.
B-ll
DRAFT - DO NOT CITE OR QUOTE
-------
Observations and predictions of 1,4-dioxane in human blood from
a 6-hour, 50 ppm exposure: VmaxC and Km fit while using PC
values from Leung and Paustenbach (1990)
100.0
0.1
10 12 14
Time (hrs)
Observations and predictions of HEAA in human urine from a
6-hour, 50 ppm exposure: kme fit while using PC values from
Leung and Paustenbach (1990)
5 10 15
Time (hrs)
Figure B-9 Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following re-calibration of the human
PBPK model with tissue:air partition coefficient values.
Source: Reprinted with permission of Elsevier, Ltd., Leung and Paustenbach (1990).
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.
Observations and predictions of 1 ,4-dioxane in human blood from
a 6-hour, 50 ppm exposure: VmaxC and Km fit while using PC
values from Sweeney et al. (2008)
Blood 1,4-Dioxane Concentration
(mg/L)
p i
-»• o o c
PPPIY" H' t H
n observed
_ T jf_ J. JP • • • • empirical predricted
•II * T
i? *'|.
l.r
f" \ M--
\ I i-
0 5 10 15
Time (hrs)
Observations and predictions of HEAA in human urine from a
6-hour, 50 ppm exposure: kme fit while using PC values from
Sweeney etal. (2008)
700.0
Figure B-10 Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) following re-calibration of the human
PBPK model with tissue:air partition coefficient values.
Source: Reprinted with permission of Oxford Journals, Sweeney et al. (2008).
4 Outputs of the blood 1,4-dioxane and urinary HEAA levels using the suggested (Table B-2)
5 parameters are shown in Figure B-l 1. These outputs rely on a very low value for the slowly perfused
6 tissue:air partition coefficient (166) that is six- to eightfold lower than the measured values reported in
7 Leung and Paustenbach (1990) and Sweeney et al. (2008). and 10-fold lower than the value used by Reitz
8 et al. (1990). While the predicted maximum blood 1,4-dioxane levels are much closer to the observations,
B-12
DRAFT - DO NOT CITE OR QUOTE
-------
1 the elimination kinetics are markedly different, producing higher predicted elimination rates compared to
2 observations during the post-exposure phase of the experiment.
Observations and predictions of 1,4-dioxane in Observations and predictions of HEAA in human
human blood from a 6-hour, 50 ppm exposure: urine from a 6-hour, 50 ppm exposure:
EPA parameter estimates used EPA parameter estimates used
Figure B-ll Predicted and observed blood 1,4-dioxane concentrations (left) and
urinary HEAA levels (right) using EPA estimated biologically plausible
parameters (Table B-l).
B.4.5 Conclusions for PBPK Model Implementation
3 Re-calibration of the human PBPK model was performed using experiment-specific values for
4 cardiac output and alveolar ventilation (Young et al.. 1977) and measured mean tissue:air 1,4-dioxane
5 partition coefficients reported by Leung and Paustenbach (1990) or Sweeney et al. (2008). The resulting
6 predictions of 1,4-dioxane in blood following a 6-hour, 50-ppm inhalation exposure were 10-fold (or
7 more) lower than either the observations or the empirical model predictions, while the predictions of
8 urinary HEAA by the PBPK and empirical models were similar to each other, but lower than observed
9 values (Figure B-9 and Figure B-10). Output from the model using biologically plausible parameter
10 values (Table B-l), Figure B-l 1 shows that application of a value for the slowly perfused tissue:air
11 partition coefficient, which is 10-fold lower than the measured value reported by Leung and Paustenbach
12 (1990). results in closer agreement of the predictions to observations during the exposure phase, but not
13 during the elimination phase. Thus, model re-calibration using experiment-specific flow rates and mean
14 measured partition coefficients does not result in an adequate fit of the PBPK model to the available data.
15 The Sweeney et al. (2008) PBPK model consisted of compartments for fat, liver, slowly perfused.
16 and other well perfused tissues. Lung and stomach compartments were used to describe the route of
17 exposure, and an overall volume of distribution compartment was used for calculation of urinary
18 excretion levels of 1.4-dioxane and its metabolite. HEAA. Metabolic constants (VmaxC and Km) for the
19 rat PBPK model were derived by optimization data from an i.v. exposure of 1.000 mg/kg data (Young et
20 al.. 1978a; 1978b) for induced metabolism. For uninduced metabolism data generated by_ i.v. exposures to
21 3. 10. 30. and 100 mg/kg were used (Young et al.. 1978a: 1978bV Data generated from the 300 mg/kg i.v.
B-13
DRAFT - DO NOT CITE OR QUOTE
-------
1 exposure was not used to estimate VmaxC and Km. The best fitting values for VmaxC to estimate the
2 blood data from the Young et al. (1978b: 1978a) study using the Sweeney et al. (2008) model resulted in
3 VmaxC values of 12.7. 10.8. 7.4 mg/kg-hr; suggesting a gradual dose dependent increase in metabolic
4 rate with dose. These estimates were for a range of doses between 3 and 1.000 mg/kg i.v. dose. Although
5 the Sweeney et al. (2008) model utilized two values for VmaxC (induced and uninduced). the PBPK
6 model does not include dose-dependent function description of the change of Vmax for i.v. doses between
7 100 and 1.000 mg/kg. PBPK model outputs were compared with other data not used in fitting model
8 parameters by visual inspection. The model predictions gave adequate match to the 1.4-dioxane
9 exhalation data after a 1.000 mg/kg i.v. dose. 1.4-Dioxane exhalation was overpredicted by a factor of
10 about 3 for the 10 mg/kg i.v. dose. Similarly, the simulations of exhaled 1.4-dioxane after oral dosing
11 were adequate at 1.000 mg/kg. and 100 mg/kg (within 50%). but poor at 10 mg/kg (model overpredicted
12 by a factor of five). The fit of the model to the human data (Young et al.. 1977) was also problematic
13 (Sweeney et al.. 2008). Using physiological parameters of Brown et al. (1997) and measured partitioning
14 parameters (Sweeney et al.. 2008; Leung and Paustenbach. 1990) with no metabolism, measured blood
15 1.4-dioxane concentrations reported by Young et al. (1977) could not be achieved unless the estimated
16 exposure concentration was increased from 53 to 100 ppm. Inclusion of any metabolism necessarily
17 decreased predicted blood concentrations. If estimated metabolism rates were used with the reported
18 exposure concentration, urinary metabolite excretion was underpredicted (Sweeney et al.. 2008). Thus.
19 the models were inadequate to use for rat to human extrapolation.
B.4.6 Sensitivity Analysis
20 A sensitivity analysis of the Reitz et al. (1990) model was performed to determine which PBPK
21 model parameters exert the greatest influence on the outcome of dosimeters of interest—in this case, the
22 concentration of 1,4-dioxane in blood. Knowledge of model sensitivity is useful for guiding the choice of
23 parameter values to minimize model uncertainty.
B.4.7 Method
24 A univariate sensitivity analysis was performed on all of the model parameters for two endpoints:
25 blood 1,4-dioxane concentrations after 1 and 4 hours of exposure. These time points were chosen to
26 assess sensitivity during periods of rapid uptake (1 hour) and as the model approached steady state
27 (4 hours) for blood 1,4-dioxane. Model parameters were perturbated 1% above and below nominal values
28 and sensitivity coefficients were calculated as follows:
Ax
B-14
DRAFT - DO NOT CITE OR QUOTE
-------
1 where x is the model parameter, f(x) is the output variable, Ax is the perturbation of the parameter from
2 the nominal value, and f (x) is the sensitivity coefficient. The sensitivity coefficients were scaled to the
3 nominal value of x and f(x) to eliminate the potential effect of units of expression. As a result, the
4 sensitivity coefficient is a measure of the proportional change in the blood 1,4-dioxane concentration
5 produced by a proportional change in the parameter value, with a maximum value of 1.
B.4.8 Results
6 The sensitivity coefficients for the seven most influential model parameters at 1 and 4 hours of
7 exposure are shown in Figure B-12. The three parameters with the highest sensitivity coefficients in
8 descending order are alveolar ventilation (QPC) (1.0), the blood:air partition coefficient (PB) (0.65), and
9 the slowly perfused tissue:air partition coefficient (PSA) (0.51). Not surprisingly, these were the
10 parameters that were doubled or given surrogate values in the Reitz et al. (1990) model in order to
11 achieve an adequate fit to the data. Because of the large influence of these parameters on the model, it is
12 important to assign values to these parameters in which high confidence is placed, in order to reduce
13 model uncertainty.
0.
QPC
-
PB
-
fe PSA
X
0
a QSC
ro
°- QCC
-
V -
Km
Sensitivity Coefficients: CV - 1hr
31 0.10 1.
I
I
|
I
I
I
DO
0.
QPC
-
PB
0 PSA
X
0
E y
ro
^ ^
PRA
QSC
Sensitivity Coefficients: CV - 4 hr
31 0.10
I
I
I
I
1.
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.5 PBPK Model Exercises Using Biologically Plausible Paramter
Boundaries
14 The PBPK model includes numerous physiological parameters whose values are typically taken
15 from experimental observations. In particular, values for the flow rates (cardiac output and alveolar
16 ventilation) and tissue:air partition coefficients (i.e., mean and standard deviations) are available from
17 multiple sources as means and variances. The PBPK model was exercised by varying the partition
DRAFT - DO NOT CITE OR QUOTE
B-15
-------
1 coefficients over the range of biological plausibility (parameter mean ± 2 standard deviations),
2 re-calibrating the metabolism and elimination parameters, and exploring the resulting range of blood
3 1,4-dioxane concentration time course predictions. Cardiac output and alveolar ventilation were not
4 varied because the experiment-specific values used did not include any measure of inter-individual
5 variation.
B.5.1 Observations Regarding the Volume of Distribution
6 Young et al. (1978b: 1978a) used experimental observations to estimate a Vd for 1,4-dioxane in
7 rats of 301 mL, or 1,204 mL/kg BW. For humans, the Vd was estimated to be 104 mL/kg BW (Young et
8 al.. 1977). It is possible that a very large volume of the slowly perfused tissues in the body of rats and
9 humans may be a significant contributor to the estimated 10-fold difference in distribution volumes for
10 the two species. This raises doubt regarding the appropriateness of using the measured rat slowly perfused
11 tissue:air partition coefficient as a surrogate values for humans in the PBPK model.
B.5.2 Defining Boundaries for Parameter Values
12 Given the possible 10-fold species differences in the apparent Vd for 1,4-dioxane in rats and
13 humans, boundary values for the partition coefficients were chosen to exercise the PBPK model across its
14 performance range to either minimize or maximize the simulated Vd. This was accomplished by defining
15 biologically plausible values for the partition coefficients as the mean ± 2 standard deviations of the
16 measured values. Thus, to minimize the simulated Vd for 1,4-dioxane, the selected blood:air partition
17 coefficient was chosen to be the mean + 2 standard deviations, while all of the other tissue:air partition
18 coefficients were chosen to be the mean - 2 standard deviations. This created conditions that would
19 sequester 1,4-dioxane in the blood, away from other tissues. To maximize the simulated 1,4-dioxane Vd,
20 the opposite selections were made: blood and other tissue:air partition coefficients were chosen as the
21 mean - 2 standard deviations and mean + 2 standard deviations, respectively. Subsequently, VmaxC, Km,
22 and kme were optimized to the empirical model output data as described in Section B.4.3. This procedure
23 was performed for both the Leung and Paustenbach (1990) and Sweeney et al. (2008) partition
24 coefficients (Table B-l). The two predicted time courses resulting from the re-calibrated model with
25 partition coefficients chosen to minimize or maximize the 1,4-dioxane Vd represent the range of model
26 performance as bounded by biologically plausible parameter values.
B.5.3 Results
27 The predicted time courses for a 6-hour, 50-ppm inhalation exposure for the re-calibrated human
28 PBPK model with mean (central tendency) and ± 2 standard deviations from the mean values for partition
29 coefficients are shown in Figure B-13 for the Leung and Paustenbach (1990) values and Figure B-14 for
30 the Sweeney et al. (2008) values. The resulting fitted values for VmaxC, Km, and kme, are given in
B-16
DRAFT - DO NOT CITE OR QUOTE
-------
1 Table B-3. By bounding the tissue:air partition coefficients with upper and lower limits on biologically
2 plausible values from Leung and Paustenbach (1990) or Sweeney et al. (2008). the model predictions are
3 still at least six- to sevenfold lower than either the empirical model output or the experimental
4 observations. The range of possible urinary HEAA predictions brackets the prediction of the empirical
5 model, but this agreement is not surprising, as the cumulative rate of excretion depends only on the rate of
6 metabolism of 1,4-dioxane, and not on the apparent Vd for 1,4-dioxane. These data show that the PBPK
7 model cannot adequately reproduce the predictions of blood 1,4-dioxane concentrations of the Young et
8 al. (1977) human empirical model or the experimental observations when constrained by biologically
9 plausible values for physiological flow rates and tissue:air partition coefficients.
1,4-Dioxane In human blood from a 6-hour, SO ppm
exposure
1000
100
„
Ltfung and PAualafibacfr ; 1 9iO| PC - UCi.
Lsuna and p*us!«ib*ai 1 1 SBQi PC - c*Mra
.....u. Lwng,jrwJP»Lalefibsch,19«JlPC-LCL
Q Voting *i il. J19TTJ obtervfiuOrt Hate
«n "°" \
' \
»° t.
'/" ~\\
/ \s
\\ " N
-• Vs
Cumulative HEAA In human urine from a 6-hour, 50 ppm
exposure
700
600
£• 0)500
S E
S I
11
I?
D 2 4
Time(hrs)
400
300
200
100
0
~ VpungelBl 11977j •mpmol mwfct
Leung nwpausKfibaerj) 1990) PC-LJCL
L«unfl«wIPaiistqtiBcml$»(PC -Cwlra
... LcL*ig srd PauileflofltM 13901 PC - L'JL
D Vwietgaal 119*71 olservabonflaifl
10 15
Time (hrs)
25
Source: Reprinted with permission of Elsevier, Ltd., Leung and Paustenbach (1990)
Figure B-13 Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients with empirical model
predictions and experimental observations for blood 1,4-dioxane
concentrations (left) and urinary HEAA levels (right) from a 6-hour,
50-ppm inhalation exposure.
B-17
DRAFT - DO NOT CITE OR QUOTE
-------
1,4-Dioxane In human blood from a 6-hour, 50 ppm
100.0 -
c
o
o 10.0-
Blood 1,4-Dioxane Cor
(mg/L)
P T-^
-^ O
exposure
,n-n- l a
*i v_
/n *
1
'/ \
! \
Young et al. (1977) empirical model
Sweeney et al. (2008) PC - UCL
Sweeney et al. (2008) PC - Central
Sweeney et al. (2008) PC - LCL
Young et al. (1977) observation
data
ta
\
] 5 10 15
Time (hrs)
Cumulative HEAA in human urine from a 6-hour, 50 ppm
exposure
700-
600 -
£•3500-
Cumulative Ur
HEAA Amount
O O O O O
D n
D
D
D
jyr — — Young et al. (1977) empirical me
j%r Sweeney et al. (2008) PC - UC
ff/
jf* Sweeney et al. (2008) PC - LCL
^/^* n Young et al. (1977) observation
del
tral
data
3 5 10 15 20 25
Time (hrs)
Source: Reprinted with permission of Oxford Journals, Sweeney et al. (2008): Used with permission of Taylor & Francis,
Young et al. (1977).
Figure B-14 Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients with empirical model
predictions and experimental observations for blood 1,4-dioxane
concentrations (left) and urinary HEAA levels (right) from a 6-hour,
50-ppm inhalation exposure.
Table B-3 PBPK metabolic and elimination parameter values resulting from recalibration of the
human model using biologically plausible values for physiological flow rates3 and
selected upper and lower boundary values for tissue:air partition coefficients
Source of partition
coefficients
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)b
Metabolic dissociation constant
(Km)c
HEAA urinary elimination rate
constant (kme)d
aCardiac output = 17.0 L/hour/kg BWU'4'
bmg/hour/kg BW075
cmg/L
dhour-1
Leung and Paustenbach (1990)
For maximal Vd
14.95
5.97
0.18
alveolar ventilation =
For minimal Vd
18.24
0.0001
0.17
17.7 L/hour/kg BWU'4
Sweeney et al. (2008)
For maximal Vd For minimal Vd
17.37 21.75
4.88 0.0001
0.26 0.19
B.5.4 Alternative Model Parameterization
1 Since the PBPK model does not predict the experimental observations of Young et al. (1977)
2 when parameterized by biologically plausible values, an exercise was performed to explore alternative
3 parameters and values capable of producing an adequate fit of the data. Since the metabolism of
4 1,4-dioxane appears to be linear in humans for a 50-ppm exposure (Young et al., 1977). the parameters
5 VmaxC and Km were replaced by a zero-order, non-saturable metabolism rate constant, kLC. This rate
6 constant was fitted to the experimental blood 1,4-dioxane data using partition coefficient values of
DRAFT - DO NOT CITE OR QUOTE
B-18
-------
1 Sweeney et al. (2008) to minimize the Vd (i.e., maximize the blood 1,4-dioxane levels). The resulting
2 model predictions are shown in Figure B-15. As before, the maximum blood 1,4-dioxane levels were
3 approximately sevenfold lower than the observed values.
16.1
c 14 1
O
1 12.1
c
u
e 10.1
o
o 3-
g o> 81
I"
*-" 4 1
•o
O
a 2.1
01
1
1 ,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure: ktc(3.0) fitted to all observations
a
>**"*«
^ »
to •
* P
-------
1,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure: kLC(0.1) fitted 1 to 6-hour observations
Young et al. (1977) empirical
model
6 8 10
Time (hrs)
12
14
Figure B-16 Predictions of blood 1,4-dioxane concentration following calibration
of a zero-order metabolism rate constant, kLC, to only the exposure
phase of the experimental data.
1 Finally, the model was re-calibrated by simultaneously fitting kLC and the slowly perfused
2 tissue:air partition coefficient to the experimental data with no bounds on possible values (except that
3 they be non-zero). The fitted slowly perfused tissue:air partition coefficient was an extremely low (and
4 biologically unlikely) value of 0.0001. The resulting model predictions, however, were closer to the
5 observations than even the empirical model predictions (Figure B-17). These exercises show that better
6 fits to the observed blood 1,4-dioxane kinetics are achieved only when parameter values are adjusted in a
7 way that corresponds to a substantial decrease in apparent Vd of 1,4-dioxane in the human, relative to the
8 rat (e.g., decreasing the slowly perfused tissue:air partition coefficient to extremely low values, relative to
9 observations). Downward adjustment of the elimination parameters (e.g., decreasing kLC) increases the
10 predicted blood concentrations of 1,4-dioxane, achieving better agreement with observations during the
11 exposure phase of the experiment; however, it results in unacceptably slow elimination kinetics, relative
12 to observations following cessation of exposure. These observations suggest that some other process not
13 captured in the present PBPK model structure is responsible for the species differences in 1,4-dioxane Vd
14 and the inability to reproduce the human experimental inhalation data with biologically plausible
15 parameter value s.
DRAFT - DO NOT CITE OR QUOTE
B-20
-------
1,4-Dioxane in human blood from a 6-hour,
50 ppm exposure
rj Young etal. (1977)
observation data
10
12
14
Time (hrs)
Figure B-17 Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a zero-order metabolism rate constant, kLC, and slowly
perfused tissue:air partition coefficient to the experimental data.
B.6 Conclusions
1 The rat and human empirical models of Young et al. (1978b: 1978a: 1977) were successfully
2 implemented in acslXtreme and perform identically to the models reported in the published papers
3 (Figures 3-3 through 3-6), with the exception of the lower predicted HEAA concentrations and early
4 appearance of the peak HEAA levels in rat urine. The early appearance of peak HEAA levels cannot
5 presently be explained, but may result from manipulations of kme or other parameters by Young et al.
6 (1978b; 1978a) that were not reported. The lower predictions of HEAA levels are likely due to reliance on
7 a standard urine volume production rate in the absence of measured (but unreported) urine volumes.
8 While the human urinary HEAA predictions were lower than observations, this is due to parameter fitting
9 of Young et al. (1977). No model output was published in Young et al. (1977) for comparison. The
10 empirical models were modified to allow for user-defined inhalation exposure levels. However, no
11 modifications were made to model oral exposures because adequate data to parameterize such
12 modifications do not exist for rats or humans. The inhalation Young et aL (1977) model failed to provide
13 adequate fits to the subchronic exposure plasma levels of 1.4-dioxane in rats using the data from the Kasai
14 et al. (2008) study.
15 Several procedures were applied to the human PBPK model to determine if an adequate fit of the
16 model to the empirical model output or experimental observations could be attained using biologically
17 plausible values for the model parameters. The re-calibrated model predictions for blood 1,4-dioxane
18 levels do not come within 10-fold of the experimental values using measured tissue: air partition
19 coefficients from Leung and Paustenbach (1990) or Sweeney et al. (2008) (Figure B-9 and Figure B-10).
20 Use of a slowly perfused tissue: air partition coefficient 10-fold lower than measured values produces
B-21
DRAFT - DO NOT CITE OR QUOTE
-------
1 exposure-phase predictions that are much closer to observations, but does not replicate the elimination
2 kinetics (Figure B-l 1). Re-calibration of the model with upper bounds on the tissue:air partition
3 coefficients results in predictions that are still six- to sevenfold lower than empirical model prediction or
4 observations (Figure B-l 3 and Figure B-l 4). Exploration of the model space using an assumption of
5 first-order metabolism (valid for the 50-ppm inhalation exposure) showed that an adequate fit to the
6 exposure and elimination data can be achieved only when unrealistically low values are assumed for the
7 slowly perfused tissue:air partition coefficient (Figure B-17). Artificially low values for the other
8 tissue:air partition coefficients are not expected to improve the model fit, because the sensitivity analysis
9 to exert less influence on blood 1,4-dioxane than VmaxC and Km. This suggests that the model structure is
10 insufficient to capture the apparent 10-fold species difference in the blood 1,4-dioxane Vd between rats
11 and humans. In the absence of actual measurements for the human slowly perfused tissue:air partition
12 coefficient, high uncertainty exists for this model parameter value. Differences in the ability of rat and
13 human blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this is expected to be
14 evident in very different values for rat and human blood:air partition coefficients, which is not the case
15 (Table B-l). Therefore, some other, as yet unknown, modification to model structure may be necessary.
16 Sweeney et al. (2008) PBPK model provided an overall improvement on previous models: however, the
17 Sweeney et al. (2008) inhalation model predictions of animal and human data were problematic.
B.7 aclsXtreme Code for the Young et al. Empricial Model for
1,4-Dioxane in Rats
18 PROGRAM: Young (1978b) rat.csl
19 !
20 ! Created by Michael Lumpkin, Syracuse Research Corporation, 08/06
21 ! This program implements the 1-compartment empirical model for 1,4-dioxane
22 ! in rats, developed by Young et al. (1978a: 1978bX Program was modified to run
23 ! in ACSL Xtreme and to include user-defined i.v. and inhalation concentrations
24 !(MLumpkin, 08/06)
9S I
Z,,J ! ————————————————————————————————————————————————————————
26
27 INITIAL
28
29 !*****Timing and Integration Commands*****
30 ALGORITHM IALG=2 IGear integration algorithm for stiff systems
31 IMERROR %%%%=0.01 IRelative error for lead in plasma
32 NSTEPS NSTP=1000 INumber of integration steps per communication interval
3 3 CINTERVAL CINT=0.1 ! Communication interval
34 CONSTANT TSTART=0.!Start of simulation (hr)
35 CONSTANT TSTOP=70. !End of simulation (hr)
36
37 !*****MODEL PARAMETERS*****
38 CONSTANT BW=0.215 Body weight (kg)
39 CONSTANT MINVOL=0.238 Irespiratory minute volume (L/min) estimated from Young et al. (1978)
40 CONSTANT IVDOSE = 0. !IV dose (mg/kg)!
41 CONSTANT CONC = 0. ! inhalation concentration (ppm)
42
43 CONSTANT MOLWT=88.105 !mol weight of 1,4-dioxane
44 CONSTANT TCHNG=6.0 Exposure pulse 1 width (hr)
45 CONSTANT TDUR=24.0 Exposure duration (hr)
B-22
DRAFT - DO NOT CITE OR QUOTE
-------
1 CONSTANT TCHNG2=120.0 Exposure pulse 2 width (hr)
2 CONSTANT TDUR2=168.0 Exposure duration 2 (hr)
3
4 CONSTANT Vmax=4.008 !(mcg/mL/hr)
5 CONSTANT Km=6.308 !(mcg/mL)
6 CONSTANT Kinh=0.43 Ipulmonary absorption constant (/hr)
7 CONSTANT Ke=0.0149 !(/hr)
8 CONSTANT Kme=0.2593 !(/hr)
9 CONSTANT Vd=0.3014 !(L)
10
11 rV=IVDOSE*BW
12 AmDIOXi=IV
13
14 END ! Of Initial Section
15
16 DYNAMIC
17 DERIVATIVE
18
19 ! * * * Dioxane inhalation concentration * * *
20 CIZONE=PULSE(0.0, TDUR, TCHNG) * PULSE(0.0, TDUR2, TCHNG2)
21 IFirst pulse is hours/day, second pulse is hours/week
22 CI=CONC*CIZONE*MOLWT/24450. !Convert to mg/L
23
24 I * * * Dioxane metabolism/1 st order elimination * * *
25 dAmDIOX=(Kinh*CI*(MINVOL*60))-((Vmax*(AmDIOX))/(Km+(AmDIOX)))-(Ke*(AmDIOX))
26 AmDIOX=INTEG(dAmDIOX,AmDIOXi)
27 ConcDIOX=AmDIOX/Vd Iplasma dioxane concentration (mcg/mL)
28 AUCDIOX=INTEG(ConcDIOX,0) Iplasma dioxane AUC
29
30 ! * * * HE AA production and 1 st order metabolism * * *
31 dAmHEAA=((Vmax*(AmDIOX))/(Km+(AmDIOX)))-(Kme*(AmHEAA))
32 AmHEAA=INTEG(dAmHEAA,0.)
33 ConcHEAA=AmHEAA/Vd Iplasma HEAA concentration
34
35 ! * * * 1 st order dioxane elimination to urine * * *
36 dAmDIOXu=(Ke*(AmDIOX))*0.35
37 AmDIOXu=INTEG(dAmDIOXu,0.)
38 ConcDIOXu=Ke*AmDIOX*0.35/1.45e-3 lurine production approx 1.45e-3 L/hr in SD rats
39
40 ! * * * 1 st order dioxane exhaled * * *
41 dAmDIOXex=(Ke*(AmDIOX))*0.65
42 AmDIOXex=INTEG(dAmDIOXex,0.)
43
44 ! * * * 1 st order HEAA elimination to urine * * *
45 dAniHEAAu=(Kme*(AmHEAA))
46 AmHEAAu=INTEG(dAmHEAAu,0.)
47 ConcHEAAu=Kme*AmHEAA/1.45e-3 lurine production approx 1.45e-3 L/hr in SD rats
48
49 END lof Derivative Section
50
51 DISCRETE
52
53 END lof Discrete Section
54
55 TERMT (T .GT. TSTOP)
56
57 END lof Dynamic Section
58
59 TERMINAL
B-23
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 END ! of Terminal Section
3
4 END ! of Program
B.8 aclsXtreme Code for the Young etal. Empricial Model for
1,4-Dioxane in Humans
5 PROGRAM: Young (1977) human.cs!
6 !
7 ! Created by Michael Lumpkin, Syracuse Research Corporation, 01/06
8 ! This program implements the 1-compartment model for 1,4-dioxane in humans,
9 ! developed by Young et al., (1977). Program was modified to run
10 ! in acslXtreme (MLumpkin, 08/06)
11 !
12
13 INITIAL
14
15 !*****Timing and Integration Commands*****
16 ALGORITHM IALG=2 ! Gear integration algorithm for stiff systems
17 IMERROR %%%%=0.01 IRelative error for lead in plasma
18 NSTEPS NSTP=1000 INumber of integration steps per communication interval
19 CINTERVALCINT=0.1 !Communication interval
20 CONSTANT TSTART=0.!Start of simulation (hr)
21 CONSTANT TSTOP= 120. !End of simulation (hr)
22
23 |*****MODEL PARAMETERS*****
24 ! CONST ANT D AT A=l lOptimization dataset
25 CONSTANT MOLWT=88.105 !mol weight for 1,4-dioxane
26 CONSTANT DOSE=0. IDose (mg/kg
27 CONSTANT CONC=0. llnhalation concentration (ppm)
28 CONSTANT BW=84.1 Body weight (kg)
29 CONSTANT MINVOL=7.0 Ipulmonary minute volume (L/min)
30 CONSTANT F=1.0 IFraction of dose absorbed
31 CONST ANT kinh= 1.06 IRate constant for inhalation (mg/hr); optimized by MHL
32 CONSTANT ke=0.0033 IRate constant for dioxane elim to urine (hr-1)
33 CONSTANT km=0.7096 IRate constant for metab of dioxane to HEAA (hr-1)
34 CONSTANT kme=0.2593 IRate constant for transfer from rapid to blood (hr-1)
35 CONSTANT VdDkg=0.104 I Volume of distribution for dioxane (L/kg BW)
36
37 CONSTANT VdMkg=0.480 I Volume of distribution for HEAA (L/kg BW)
38 CONSTANT OStart=0. I Time of first oral dose (hr)
39 CONSTANT OPeriod= 120. I Oral Dose pulse period (hr)
40 CONSTANT OWidth= 1. I Width (gavage/drink time) of oral dose (hr)
41
42 CONSTANT IStart=0. ITime of inhalation onset (hr)
43 CONSTANT IPeriod=120. llnhalation pulse period (hr)
44 CONSTANT IWidth=6. I Width (duration) of inhalation exposure (hr)
45
46 END I Of Initial Section
47
48 DYNAMIC
49
50 DERIVATIVE
51 !****VARIABLES and DEFINED VALUES*****
52 VdD=BW*VdDkg I Volume of distribution for dioxane
B-24
DRAFT - DO NOT CITE OR QUOTE
-------
1 VdM=BW*VdMkg ! Volume of distribution for HEAA
2
3 InhalePulse=PULSE(IStart,IPeriod,IWidth)
4 Inhale=CONC*InhalePulse*MOLWT/24450. !Convert to mg/L
5
6 !*****DIFFERENTIAL EQUATIONS FOR COMPARTMENTS****
7
8 ! *** Dioxane in the body (plasma) ***
9 dAMTbD=(Kinh*Inhale*(MINVOL*60))-(AMTbD*km)-(AMTbD*ke)
10 AMTbD=INTEG(dAMTbD,0.)
11 CbD=AMTbD/VdD
12 AUCbD=INTEG(CbD,0)
13
14 i * * * HEAA in the body (plasma) * * *
15 dAMTbM=AMTbD*km-AMTbM*kme
16 AMTbM=INTEG(dAMTbM,0.)
17 CbM=AMTbM/VdM
18
19 ! * * * Cumulative Dioxane in the urine * * *
20 dAMTuD=(AMTbD*ke)
21 AMTuD=INTEG(dAMTuD,0.)
22
23 ! *** Cumulative HEAA in the urine ***
24 dAMTuM=(AMTbM*kme)
25 AMTuM=INTEG(dAMTuM,0.)
26
27 END ! Of Derivative Section
28
29 DISCRETE
30
31 END !of Discrete Section
32
3 3 TERMT (T . GT. TSTOP)
34
35 END ! Of Dynamic Section
36
37 TERMINAL
38
39 END ! of Terminal Section
40
41 END ! of Program
B.9 aclsXtreme Code for the Reitz et al. PBPK Model For 1,4-Dioxane
42 (Reitz etal. 1990)
43 PROGRAM: DIOXANE.CSL (Used in Risk Estimation Procedures)
44 ! Added a venous blood compartment and 1 st order elim of metab.'
45 Mass Balance Checked OK for Inhal, IV, Oral, and Water RHR'
46 IDefined Dose Surrogates for Risk Assessment 01/04/89'
47 Modified the Inhal Route to use PULSE for exposure conditions'
48 Modifications by GLDiamond, Aug2004, marked as !**
49 !
50 Metabolism of dioxane modified by MLumpkin, Oct2006, to include 1st order
51 lor saturable kinetics. For 1st order, set VmaxC=0; for M-Menten, set K1C=0.
52 !
53 INITIAL
B-25
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 INTEGER I
3 1=1
4 ! ARRAY TDATA(20) ! CONSTANT TDATA=999, 19*1.0E-6 !**
5 CONSTANT BW = 0.40 !'Body weight (kg)'
6 CONSTANT QPC =15.! 'Alveolar ventilation rate (1/hr)'
7 CONSTANT QCC =15. ! 'Cardiac output (1/hr)'
8
9 IFlows to Tissue Compartments'
10 CONSTANT QLC = 0.25 !'Fractional blood flow to liver'
11 CONSTANT QFC = 0.05 I'Fractional blood flow to fat'
12 CONSTANT QSC = 0.18 I'Fractional blood flow to slow'
13 QRC= 1.0-(QFC + QSC + QLC)
14 CONSTANT SPDC =1.0! diffusion constant for slowly perfused tissues
15
16 ! Volumes of Tissue/Blood Compartments'
17 CONSTANT VLC = 0.04 !'Fraction liver tissue'
18 CONSTANT VFC = 0.07 I'Fraction fat tissue'
19 CONSTANT VRC = 0.05 I'Fraction Rapidly Perf tissue'
20 CONSTANT VBC = 0.05 I'Fraction as Blood'
21 VSC = 0.91 - (VLC + VFC + VRC + VBC)
22
23 IPartition Coefficients'
24 CONSTANT PLA = 1557. I'Liver/air partition coefficient'
25 CONSTANT PFA =851.1 'Fat/air partition coefficient'
26 CONSTANT PSA = 2065. I'Muscle/air (Slow Perf) partition'
27 CONSTANT PRA = 1557. I'Richly perfused tissue/air partition'
28 CONSTANT PB = 1850. I'Blood/air partition coefficient'
29
30 I Other Compound Specific Parameters'
31 CONSTANT MW = 88.1 I'Molecular weight (g/mol)'
32 CONSTANT KLC = 12.0 I temp zero-order metab constant
33 CONSTANT VMAXC = 13.8 I'Maximum Velocity of Metabol.'
34 CONSTANT KM = 29.4 I'Michaelis Menten Constant'
35 CONSTANT ORAL = 0.0 I'Oral Bolus Dose (mg/kg)'
36 CONSTANT KA = 5.0 I'Oral uptake rate (/hr)'
37 CONSTANT WATER = 0.0 I 'Cone in Water (mg/liter, ppm)'
38 CONSTANT WDOSE=0.0 I Water dose (mg/kg-day) **
39 CONSTANT IV = 0.0 1'IV dose (mg/kg)'
40 CONSTANT CONC = 0.0 I'Inhaled concentration (ppm)'
41 CONSTANT KME = 0.276 I'Urinary Elim constant for met (hr-1)'
42
43 ITiming commands'
44 CONSTANT TSTOP = 50 I 'Length of experiment (hrs)'
45 CONSTANT TCHNG = 6 I 'Length of inhalation exposure (hrs)'
46 CINTERVALCINT=0.1
47 CONSTANT WIDD=24. !**
48 CONSTANT PERD=24. !**
49 CONSTANT PERW=168. !**
50 CONSTANT WIDW=168. !**
51 CONSTANT DAT=0.017!**
52
53 ! Scaled parameters calculated in this section of Program'
54 QC=QCC*BW**0.74
55 QP=QPC*BW**0.74
56 QL=QLC*QC
57 QF=QFC*QC
58 QS=QSC*QC
59 QR=QRC*QC
B-26
DRAFT - DO NOT CITE OR QUOTE
-------
1 VL=VLC*BW
2 VF=VFC*BW
3 VS=VSC*BW
4 VR=VRC*BW
5 VB=VBC*BW
6 PL=PLA/PB
7 PR=PRA/PB
8 PS=PSA/PB
9 PF=PFA/PB
10 KL = KLC*bw**0.7 ! Zero-order metab constant
11 VMAX = VMAXC*BW**0.7
12 DOSE = ORAL*B W ! 'Initial Amount in Stomach'
13 ABO = IV*B W ! 'Initial Amount in Blood'
14 IDRINK = 0.102*BW**0.7*WATER/24 I'lnput from water (mg/hr)' ! **
15 !DRINKA = 0.102*BW**0.7*WATER/DAT I'Inputfrom water (mg/hr)' !**
16 DRTNKA=WDOSE*BW/DAT
17 CV = ABO/VB ! 'Initialize CV
18
19 END !'End of INITIAL'
20
21 DYNAMIC
22
23 ALGORITHM IALG = 2 ! 'Gear method for stiff systems'
24 TERMT(T .GE. TSTOP )
25 CR = AR/VR
26 CS = AS/VS
27 CF = AF/VF
28 BODY = AL + AR + AS + AF + AB+ TUMMY
29 BURDEN = AM + BODY
30 TMASS = BURDEN + AX + AMEX
31
3 2 ! Calculate the Interval Excretion Data here:'
33 ! DAX = AMEX-AMEX2
34 ! IF(DOSE .LE. 0.0 .AND. IV .LE. 0.0 ) GO TO SKIP1
35 ! PCTAX = 100*(AX - AX2)/(DOSE + IV*BW)
36 ! PCTMX = 100*(AMEX - AMEX2)/(DOSE + IV*BW)
37 ! SKIP!.. CONTINUE
38 ! IF(T .LT. TDATA(I) .OR. I .GE. 20 ) GO TO SKIP
39 ! AX2=AX
40 ! AMEX2=AMEX
41 ! 1=1+1
42 ! SKIP.. CONTINUE
43
44 IDISCRETE EXPOSE
45 ! CIZONE = 1.0 ! CALL LOGD(.TRUE.) Turns on inhalation exposure?
46 !END
47 IDISCRETE CLEAR
48 ! CIZONE = 0.0 ! CALL LOGD(.TRUE.)
49 !END
50
51 DERIVATIVE
52
53 !Use Zero-Crossing Form of DISCRETE Function Here'
54 ! SCHEDULE command must be in DERIVATIVE section'
55 ! DAILY = PULSE (0.0, PERI, TCHNG)
56 ! WEEKLY = PULSE (0.0, PER2, LEN2 )
57 ! SWITCHY = DAILY * WEEKLY
58 !SCHEDULE EXPOSE .XP. SWITCHY - 0.995
59 !SCHEDULE CLEAR .XN. SWITCHY - 0.005
B-27
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 DAILY=PULSE(0.0,PERD,WIDD)
3 WEEKLY=PULSE(0.0,PERW,WIDW)
4 SWITCHY = DAILY * WEEKLY
5
/T 1^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^IVTOfllfiGfi 1~TGT"G fOT" ^VOTlff^
7 CI = CONC * MW / 24451.0 * SWITCHY!**
8
9 ! CA = Concentration in arterial blood (mg/1)'
10 CA = (QC*CV+QP*CI)/(QC+(QP/PB))
11 CX=CA/PB
12
13 DRINK=DRINKA* SWITCHY !**
14
15 ! TUMMY = Amount in stomach'
16 RTUMMY = -KA*TUMMY
17 TUMMY = INTEG(RTUMMY,DOSE)
18 ! RAX = Rate of Elimination in Exhaled air'
19 RAX = QP*CX
20 AX = INTEG(RAX, 0.0)
21
22 ! AS = Amount in slowly perfused tissues (mg)'
23 RAS = SPDC*(CA-CVS) !now governed by diffusion-limited constant, SPDC, instead of QS
24 AS = INTEG(RAS,0.)
25 CVS = AS/(VS*PS)
26
27 ! AR = Amount in rapidly perfused tissues (mg)'
28 RAR = QR*(CA-CVR)
29 AR = INTEG(RAR,0.)
30 CVR = AR/(VR*PR)
31
32 ! AF = Amount in fat tissue (mg)'
33 RAF = QF*(CA-CVF)
34 AF = INTEG(RAF,0.)
35 CVF = AF/(VF*PF)
36
37 ! AL = Amount in liver tissue (mg)'
38 RAL = QL*(CA-CVL) - KL*CVL - VMAX*CVL/(KM+CVL) + KA*TUMMY + DRINK
39 AL = INTEG(RAL,0.)
40 CVL = AL/(VL*PL)
41
42 Metabolism comments updated by EDM on 2/1/10
43 ! AM = Amount metabolized (mg)'
44 RMEX = (KL*CVL)+(VMAX*CVL/(KM+CVL)) IRate of 1,4-dioxane metabolism
45 RAM = (KL*CVL)+(VMAX*CVL)/(KM+CVL) - KME* AM IRate of change of metabolite in body
46
47 AM = INTEG(RAM, 0.0) ! 'Amt Metabolite in body
48 CAM = AM/B W ! 'Cone Metabolite in body'
49 AMEX = INTEG(KME*AM, 0.0) I'Amt Metabolite Excreted via urine'
50
51 ! AB = Amount in Venous Blood'
52 RAB = QF*CVF + QL*CVL + QS*CVS + QR*CVR - QC*CV
53 AB = INTEG(RAB, ABO)
54 CV = AB/VB
55 AUCV = INTEG(CV, 0.0)
56
57 IPossible Dose Surrogates for Risk Assessment Defined Here'
58
59 CEX = 0.667*CX + 0.333*CI I'Conc inExhal Air'
B-28
DRAFT - DO NOT CITE OR QUOTE
-------
1 AVECON = PLA * (CEX+CI)/2 I'Ave Cone in Nose Tissue'
2 AUCCON = INTEG(AVECON, 0.0) I'Area under Curve (Nose)'
3
4 AUCMET = INTEG(CAM, 0.0) I'Area under Curve (Metab)'
5
6 CL = AL/VL I'Cone Liver Tissue'
7 AUCL = INTEG(CL, 0.0) I'Area under Curve (Liver)'
8 AAUCL=AUCL/TIME
9
10 I Dose Surrogates are Average Area under Time/Cone Curve per 24 hrs'
11 IF (T .GT. 0) TIME=T
12 dayS = TIME/24.0
13 NOSE = AUCCON/D AYS I 'Nasal Turbinates'
14 LIVER = AUCL/DAYS I 'Liver Tissues'
15 METAB = AUCMET/DAYS I'Stable Metabolite'
16
17 END I 'End of dynamic'
18
19 END I End of TERMINAL
20
21 END I 'End of PROGRAM
B-29
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX C. DETAILS OF BMD ANALYSIS FOR
ORAL RFD FOR 1,4-DIOXANE
C.1 Cortical Tubule Degeneration
1 All available dichotomous models in the Benchmark Dose Software (version 2.1.1) were fit to the
2 incidence data shown in Table C-l, for cortical tubule degeneration in male and female Osborne-Mendel
3 rats exposed to 1,4-dioxane in the drinking water (NCI. 1978). Doses associated with a BMR of a 10%
4 extra risk were calculated.
Table C-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)
0 240
0/3 1a 20/3 1b
(65%)
Females (mg/kg-day)
530
27/33b
(82%)
0
0/3 1a
350
0/34
640
10/32b
(31%)
Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this review.
blncidence significantly elevated compared to control by Fisher's exact test (p < 0.05) performed for this review.
Source: NCI (1978).
6 As assessed by the %2 goodness-of-fit test, several models in the software provided adequate fits
7 to the data for the incidence of cortical tubule degeneration in male and female rats (%2 p > 0.1)
8 (Table C-2). Comparing across models, a better fit is indicated by a lower AIC value (U.S. EPA. 2000a).
9 As assessed by Akaike's Information Criterion (AIC), the log-probit model provided the best fit to the
10 cortical tubule degeneration incidence data for male rats (Table C-2, Figure C-l) and could be used to
11 derive a POD of 38.5 mg/kg-day for this endpoint. The Weibull model provided the best fit to the data for
12 female rats (Table C-2, Figure C-5) and could be used to derive a POD of 452.4 mg/kg-day for this
13 endpoint. For those models that exhibit adequate fit, models with the lower AIC values are preferred.
14 Differences in AIC values of less than 1 are generally not considered important. BMDS modeling results
15 for all dichotomous models are shown in Table C-2.
C-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-2 Goodness-of-fit statistics and BMD10 and BMDL10 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
p-valuea
Scaled Residual
of Interest
BMDio
(mg/kg-day)
BMDLio
(mg/kg-day)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
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
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
41.9712
43.7495
41.7501
43.7495
48.1969
43.7495
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.
"Power restricted to > 1.
°Slope restricted to > 1.
dBetas restricted to >0.
Source: NCI (1978).
DRAFT - DO NOT CITE OR QUOTE
C-2
-------
LogProbit Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
100
200
300
400
500
dose
14:4902/01 2010
Source: NCI (1978).
Figure C-l BMD Log-probit model of cortical tubule degeneration incidence data
for male rats exposed to 1,4-dioxane in drinking water for 2 years to
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
DRAFT - DO NOT CITE OR QUOTE
-------
1 Default Initial (and Specified) Parameter Values
2 background = 0
3 intercept = -5.14038
4 slope = 1
5
6
7 Asymptotic Correlation Matrix of Parameter Estimates
8 (*** The model parameter(s) -background -slope have been estimated at a boundary
9 point, or have been specified by the user, and do not appear in the correlation
10 matrix)
11
12 intercept
13 intercept 1
14
15
16 Parameter Estimates
17
18 95.0% Wald Confidence Interval
19 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
20 background 0 NA
21 intercept -5.22131 0.172682 -5.55976 -4.88286
22 slope 1 NA
23
24 NA - Indicates that this parameter has hit a bound implied by some ineguality
25 constraint and thus has no standard error.
26
27
28
29 Analysis of Deviance Table
30
31 Model Log(likelihood) # Param's Deviance Test d.f. P-value
32 Full model -35.8087 3
33 Fitted model -36.084 1 0.550629 2 0.7593
34 Reduced model -65.8437 1 60.07 2 <.0001
35
36 AIC: 74.168
37
38
39 Goodness of Fit
40 Scaled
41 Dose Est._Prob. Expected Observed Size Residual
42
43 0.0000 0.0000 0.000 0.000 31 0.000
44 240.0000 0.6023 18.672 20.000 31 0.487
45 530.0000 0.8535 28.166 27.000 33 -0.574
46
47 ChiA2 = 0.57 d.f. = 2 P-value = 0.7532
48
49
50 Benchmark Dose Computation
51 Specified effect =0.1
52 Risk Type = Extra risk
53 Confidence level = 0.95
54 BMD = 51.4062
55 BMDL = 38.5284
C-4
DRAFT - DO NOT CITE OR QUOTE
-------
Weibull 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
C-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 Background = 0.015625
2 Slope = 1.55776e-010
3 Power = 3.33993
4
5
6 Asymptotic Correlation Matrix of Parameter Estimates
7 (*** The model parameter(s) -Background -Power have been estimated at a boundary
8 point, or have been specified by the user, and do not appear in the correlation
9 matrix)
10
11 Slope
12 Slope -1.$
13
14 Parameter Estimates
15 95.0% Wald Confidence Interval
16 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17 Background 0 NA
18 Slope 1.15454e-051 1.#QNAN 1.#QNAN 1.#QNAN
19 Power 18 NA
20
21 NA - Indicates that this parameter has hit a bound implied by some ineguality
22 constraint and thus has no standard error.
23
24 Analysis of Deviance Table
25
26 Model Log(likelihood) # Param's Deviance Test d.f. P-value
27 Full model -19.8748 3
28 Fitted model -19.875 1 0.000487728 2 0.9998
29 Reduced model -32.1871 1 24.6247 2 <.0001
30
31 AIC: 41.75
32
33
34 Goodness of Fit
35 Scaled
36 Dose Est._Prob. Expected Observed Size Residual
37
38 0.0000 0.0000 0.000 0.000 31 0.000
39 350.0000 0.0000 0.000 0.000 34 -0.016
40 640.0000 0.3125 9.999 10.000 32 0.000
41
42 ChiA2 = 0.00 d.f. = 2 P-value = 0.9999
43
44
45 Benchmark Dose Computation
46 Specified effect =0.1
47 Risk Type = Extra risk
48 Confidence level =0.95
49 BMD = 596.445
50 BMDL = 452.359
C.2 Liver hyperplasia
51 All available dichotomous models in the Benchmark Dose Software (version 2.1.1) were fit to the
52 incidence data shown in Table C-3, for liver hyperplasia in male and female F344/DuCrj rats exposed to
53 1,4-dioxane in the drinking water (Kano et al., 2009; JBRC. 1998). Benchmark doses associated with a
54 BMR of a 10% extra risk were calculated.
C-6
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-3 Incidence of liver hyperplasia in F344/DuCrj rats exposed to 1,4-dioxane in drinking
water3
Males (mg/kg-day) Females (mg/kg-day)
0 11 55 274 0 18 83 429
3/40 2/45 9/35a 12/22° 2/38b 2/37 9/38 24/24°
aDose information from Kano et al. (2009) and incidence data from sacrificed animals from JBRC (1998).
"Incidence significantly elevated compared
""Incidence significantly elevated compared 1
Sources: Kano et al. (2009): JBRC (1998).
"Incidence significantly elevated compared to control by x2 test (p < 0.05).
Incidence significantly elevated compared to control by x2 test (p < 0.01).
1 For incidence of liver hyperplasia in F344 male rats, the logistic, probit, and dichotomous-Hill
2 models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1; see Table C-4), and thus
3 should not be considered further for identification of a POD. All of the remaining models exhibited
4 adequate fit, but the AIC values for the gamma, multistage, quantal-linear, and Weibull models were
5 lower than the AIC values for the log-logistic and log-probit models. Finally, the AIC values for gamma,
6 multistage, quantal-linear, and Weibull models in Table C-4 are equivalent and, in this case, essentially
7 represent the same model. Therefore, consistent with the external review draft Benchmark Dose
8 Technical Guidance (U.S. EPA. 2000aX any of them with equal AIC values (gamma, multistage,
9 quantal-linear, or Weibull) could be used to identify a POD for this endpoint of 23.8 mg/kg-day.
10 For liver hyperplasias in F344 female rats exposed to 1,4-dioxane, the quantal-linear and
11 dichotomous-Hill models did not result in a good fit (i.e., %2/rvalue < 0.1; See Table C-4). The
12 multistage (3-degree) model had the lowest AIC value and was selected as the best-fitting model.
13 Therefore, consistent with the BMD technical guidance document (U.S. EPA. 2000a). the BMDL from
14 the multistage (3-degree) model was selected to yield a POD for this endpoint of 27.1 mg/kg-day.
C-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4 Benchmark dose modeling results based on the incidence of liver hyperplasias in
male and female F344 rats exposed to 1,4-dioxane in drinking water for 2 years
Model
AIC
Scaled „.._.
p-va,ue- Reside, (mg™^y)
BMDLio
(mg/kg-day)
Male
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage0
(2 degree)
Probit
Weibullb
Quantal-Linear
Dichotomous-Hill
114.172
117.047
115.772
115.57
114.172
116.668
114.172
114.172
117.185
0.3421
0.0706
0.1848
0.1431
0.3421
0.0859
0.3421
0.3421
NCe
0.886
1.869
0.681
1.472
0.886
1.804
0.886
0.886
-0.2398
35.90
83.56
33.39
54.91
35.90
76.69
35.90
35.90
32.01
23.81
63.29
16.96
37.05
23.81
58.57
23.81
23.81
14.84
Female
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage0
(2 degree)
Multistage0
(3 degree)
Probit
Weibullb
Quantal-Linear
Dichotomous-Hill
78.8357
77.0274
78.8357
78.8357
76.9718
76.8351
77.0308
78.8349
87.3833
2972.99
0.9783
0.9174
0.9781
0.9781
0.9563
0.9999
0.9095
0.9995
0.0245
NCe
0
-0.016
0
0
-0.107
0
0.017
0
-1.116
0
70.78
54.66
77.72
74.64
56.06
65.28
52.53
66.47
21.52
NCe
40.51
41.11
51.21
50.97
31.17
27.08
38.44
36.14
15.61
NCe
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 > 1.
°Slope restricted to 5 1.
dBetas restricted to >0.
eNC=Not calculated.
Sources: Kano et al. (2009): JBRC (1998).
DRAFT - DO NOT CITE OR QUOTE
C-8
-------
Gamma Multi-Hit Model with 0.95 Confidence Level
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
•5
I
C
o
13
ro
100
150
200
250
dose
14:35 12/042009
Figure C-3 BMD gamma model of liver hyperplasia incidence data for F344 male
rats exposed to 1,4-dioxane in drinking water for 2 years to support
results Table C-4.
Gamma Model. (Version: 2.13; Date: 05/16/2008)
Input Data File:
Z:\14Dioxane\BMDS\gam_jbrcl998_mrat_liver_hyper_Gam-BMR10-Restrict.(d)
Gnuplot Plotting File:
Z:\14Dioxane\BMDS\gam_jbrcl998_mrat_liver_hyper_Gam-BMR10-Restrict.plt
Fri Dec 04 14:35:02 2009
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 = 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
C-9
DRAFT - DO NOT CITE OR QUOTE
-------
1 Background = 0.0853659
2 Slope = 0.00479329
3 Power = 1.3
4
5
6 Asymptotic Correlation Matrix of Parameter Estimates
7 (*** The model parameter(s) -Power have been estimated at a boundary point, or have
8 been specified by the user, and do not appear in the correlation matrix )
9
10 Background Slope
11 Background 1 -0.36
12 Slope -0.36 1
13
14 Parameter Estimates
15 95.0% Wald Confidence Interval
16 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17 Background 0.0569658 0.0278487 0.00238329 0.111548
18 Slope 0.00293446 0.000814441 0.00133818 0.00453073
19 Power 1 NA
20
21 NA - Indicates that this parameter has hit a bound implied by some ineguality
22 constraint and thus has no standard error.
23
24 Analysis of Deviance Table
25
26 Model Log(likelihood) # Param's Deviance Test d.f. P-value
27 Full model -53.9471 4
28 Fitted model -55.0858 2 2.27725 2 0.3203
29 Reduced model -67.6005 1 27.3066 3 <.0001
30
31 AIC: 114.172
32
33
34 Goodness of Fit
35 Scaled
36 Dose Est._Prob. Expected Observed Size Residual
37
38 0.0000 0.0570 2.279 3.000 40 0.492
39 11.0000 0.0869 3.911 2.000 45 -1.011
40 55.0000 0.1975 6.913 9.000 35 0.886
41 274.0000 0.5780 12.715 12.000 22 -0.309
42
43 ChiA2 = 2.15 d.f. = 2 P-value = 0.3421
44
45
46 Benchmark Dose Computation
47 Specified effect =0.1
48 Risk Type = Extra risk
49 Confidence level = 0.95
50 BMD = 35.9046
51 BMDL = 23.8065
C-10
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
I
C
o
0.1
100
150
200
250
dose
14:35 12/042009
Figure C-4 BMD multistage (2 degree) model of liver hyperplasia incidence data
for F344 male rats exposed to 1,4-dioxane in drinking water for 2 years
to support results
C-ll
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4.
1 ====================================================================
2 Multistage Model. (Version: 3.0; Date: 05/16/2008)
3 Input Data File:
4 Z:\14Dioxane\BMDS\mst_jbrcl998_mrat_liver_hyper_Mst-BMR10-restrict.(d)
5 Gnuplot Plotting File:
6 Z:\14Dioxane\BMDS\mst_jbrcl998_mrat_liver_hyper_Mst-BMR10-Restrict.plt
7 Fri Dec 04 14:35:06 2009
8 ====================================================================
9 BMDS Model Run
10 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11 The form of the probability function is:
12
13 P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2) ]
14
15 The parameter betas are restricted to be positive
16
17 Dependent variable = Effect
18 Independent variable = Dose
19
20 Total number of observations = 4
21 Total number of records with missing values = 0
22 Total number of parameters in model = 3
23 Total number of specified parameters = 0
24 Degree of polynomial = 2
25
26
27 Maximum number of iterations = 250
28 Relative Function Convergence has been set to: le-008
29 Parameter Convergence has been set to: le-008
30
31
32
33 Default Initial Parameter Values
34 Background = 0.0750872
35 Beta(l) = 0.00263797
36 Beta(2) = 0
37
38
39 Asymptotic Correlation Matrix of Parameter Estimates
40 (*** The model parameter(s) -Beta(2) have been estimated at a boundary point, or have
41 been specified by the user, and do not appear in the correlation matrix)
42
43 Background Beta(l)
44 Background 1 -0.49
45 Beta(l) -0.49 1
46
47
48 Parameter Estimates
49 95.0% Wald Confidence Interval
50 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
51 Background 0.0569658 * * *
52 Beta(l) 0.00293446 * * *
53 Beta(2) 0 * * *
54
55 * - Indicates that this value is not calculated.
56
57
58
59 Analysis of Deviance Table
60
61 Model Log(likelihood) # Param's Deviance Test d.f. P-value
62 Full model -53.9471 4
63 Fitted model -55.0858 2 2.27725 2 0.3203
64 Reduced model -67.6005 1 27.3066 3 <.0001
C-12
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
AIC: 114.172
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0570 2.279 3.000 40 0.492
11.0000 0.0869 3.911 2.000 45 -1.011
55.0000 0.1975 6.913 9.000 35 0.886
274.0000 0.5780 12.715 12.000 22 -0.309
ChiA2 = 2.15 d.f. = 2 P-value = 0.3421
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 35.9046
BMDL = 23.8065
BMDU = 82.1206
Taken together, (23.8065, 82.1206) is a 90% two-sided confidence interval for the BMD
Weibull Model with 0.95 Confidence Level
o
13
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Weibull
BMDL
BMD
50
100
150
200
250
dose
14:35 12/042009
Figure C-5 BMD Weibull model of liver hyperplasia incidence data for F344 male
rats exposed to 1,4-dioxane in drinking water for 2 years to support the
results in
C-13
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4.
1 ====================================================================
2 Weibull Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
3 Input Data File:
4 Z:\14Dioxane\BMDS\wei_jbrcl998_mrat_liver_hyper_Wei-BMR10-Restrict.(d)
5 Gnuplot Plotting File:
6 Z:\14Dioxane\BMDS\wei_jbrcl998_mrat_liver_hyper_Wei-BMR10-Restrict.plt
7 Fri Dec 04 14:35:08 2009
8 ====================================================================
9 BMDS Model Run
10 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(-slope*dose/xpower)]
14
15 Dependent variable = Effect
16 Independent variable = Dose
17 Power parameter is restricted as power >=1
18
19 Total number of observations = 4
20 Total number of records with missing values = 0
21 Maximum number of iterations = 250
22 Relative Function Convergence has been set to: le-008
23 Parameter Convergence has been set to: le-008
24
25
26
27 Default Initial (and Specified) Parameter Values
28 Background = 0.0853659
29 Slope = 0.00253609
30 Power = 1
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34 (** The model parameter(s) -Power have been estimated at a boundary point, or have
35 been specified by the user, and do not appear in the correlation matrix )
36
37 Background Slope
38 Background 1 -0.36
39 Slope -0.36 1
40
41
42 Parameter Estimates
43 95.0% Wald Confidence Interval
44 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
45 Background 0.0569661 0.0278498 0.00238155 0.111551
46 Slope 0.00293445 0.000814445 0.00133816 0.00453073
47 Power 1 NA
48
49 NA - Indicates that this parameter has hit a bound implied by some ineguality
50 constraint and thus has no standard error.
51
52
53 Analysis of Deviance Table
54
55 Model Log(likelihood) # Param's Deviance Test d.f. P-value
56 Full model -53.9471 4
57 Fitted model -55.0858 2 2.27725 2 0.3203
58 Reduced model -67.6005 1 27.3066 3 <.0001
59
60 AIC: 114.172
61
62
63 Goodness of Fit
64 Scaled
C-14
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0570 2.279 3.000 40 0.492
11.0000 0.0869 3.911 2.000 45 -1.011
55.0000 0.1975 6.913 9.000 35 0.886
274.0000 0.5780 12.715 12.000 22 -0.309
ChiA2 = 2.15 d.f. = 2 P-value = 0.3421
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 35.9047
BMDL = 23.8065
Quantal Linear Model with 0.95 Confidence Level
o
13
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Quantal Linear
BMDL
BMD
50
100
150
200
250
dose
14:35 12/042009
Figure C-6 BMD quantal-linear model of liver hyperplasia incidence data for F344
male rats exposed to 1,4-dioxane in drinking water for 2 years to
support the results in
C-15
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4.
1 ===================================
2 Quantal Linear Model using Weibull Model (Version: 2.12; Date: 05/16/2008)
3 Input Data File: Z:\14Dioxane\BMDS\qln_jbrcl998_mrat_liver_hyper_Qln-BMR10.(d)
4 Gnuplot Plotting File: Z:\14Dioxane\BMDS\gln_jbrcl998_mrat_liver_hyper_Qln-BMR10.plt
5 Fri Dec 04 14:35:09 2009
6 ====================================================================
7 BMDS Model Run
8 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9 The form of the probability function is:
10
11 P[response] = background + (1-background)*[1-EXP(-slope*dose)]
12
13
14 Dependent variable = Effect
15 Independent variable = Dose
16
17 Total number of observations = 4
18 Total number of records with missing values = 0
19 Maximum number of iterations = 250
20 Relative Function Convergence has been set to: le-008
21 Parameter Convergence has been set to: le-008
22
23 Default Initial (and Specified) Parameter Values
24 Background = 0.0853659
25 Slope = 0.00253609
26 Power = 1 Specified
27 Asymptotic Correlation Matrix of Parameter Estimates
28 (*** The model parameter(s) -Power have been estimated at a boundary point, or have
29 been specified by the user, and do not appear in the correlation matrix)
30
31 Background Slope
32 Background 1 -0.36
33 Slope -0.36 1
34
35
36
37 Parameter Estimates
38 95.0% Wald Confidence Interval
39 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40 Background 0.0569665 0.02785 0.00238157 0.111551
41 Slope 0.00293447 0.000814452 0.00133818 0.00453077
42
43
44
45 Analysis of Deviance Table
46
47 Model Log(likelihood) # Param's Deviance Test d.f. P-value
48 Full model -53.9471 4
49 Fitted model -55.0858 2 2.27725 2 0.3203
50 Reduced model -67.6005 1 27.3066 3 <.0001
51
52 AIC: 114.172
53
54
55 Goodness of Fit
56 Scaled
57 Dose Est._Prob. Expected Observed Size Residual
58 ~
59 0.0000 0.0570 2.279 3.000 40 0.492
60 11.0000 0.0869 3.911 2.000 45 -1.011
61 55.0000 0.1975 6.913 9.000 35 0.886
62 274.0000 0.5780 12.716 12.000 22 -0.309
63
64 Chi^2 = 2.15 d.f. = 2 P-value = 0.3421
C-16
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 35.9044
BMDL = 23.8065
Multistage Model with 0.95 Confidence Level
O
I
c
O
0.8
0.6
0.4
0.2 -
10:3005/21 2010
Source: JBRC (1998).
Figure C-7 BMD Multistage model (third (3°)of liver hyperplasia incidence data for
F344 female rats exposed to 1,4-dioxane in drinking water for 2 years to
support the results in
C-17
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-4.
1 ====================================================================
2 Multistage Model. (Version: 3.0; Date: 05/16/2008)
3 Input Data File:
4 H:\14Dioxane\BMDS\mst_jbrcl998_frat_liver_hyper_Mst-BMR10-Restrict-3deg.(d)
5 Gnuplot Plotting File:
6 H:\14Dioxane\BMDS\mst_jbrcl998_frat_liver_hyper_Mst-BMR10-Restrict-3deg.plt
7 Fri May 21 10:30:14 2010
8 ====================================================================
9 BMDS Model Run
10 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11 The form of the probability function is:
12
13 P[response] = background +
14 (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
15
16 The parameter betas are restricted to be positive
17
18 Dependent variable = Effect
19 Independent variable = Dose
20
21 Total number of observations = 4
22 Total number of records with missing values = 0
23 Total number of parameters in model = 4
24 Total number of specified parameters = 0
25 Degree of polynomial = 3
26
27 Maximum number of iterations = 250
28 Relative Function Convergence has been set to: le-008
29 Parameter Convergence has been set to: le-008
30
31 Default Initial Parameter Values
32 Background = 0
33 Beta(l) = 0
34 Beta(2) = 0
35 Beta(3) = 1.2696e+012
36
37 Asymptotic Correlation Matrix of Parameter Estimates
38
39 (*** The model parameter(s) -Beta(l), -Beta(2) have been estimated at a boundary
40 point, or have been specified by the user, and do not appear in the correlation
41 matrix)
42
43 Background Beta(3)
44 Background 1 -0.55
45 Beta(3) -0.55 1
46
47
48 Parameter Estimates
49
50 95.0% Wald Confidence Interval
51 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
52 Background 0.0523101 * * *
53 Beta(l) 0 * * *
54 Beta(2) 0 * * *
55 Beta(3) 3.78712e-007 * * *
56
57 * - Indicates that this value is not calculated.
58
59
60 Analysis of Deviance Table
61
62 Model Log(likelihood) # Param's Deviance Test d.f. P-value
63 Full model -36.4175 4
64 Fitted model -36.4175 2 0.00016582 2 0.9999
C-18
DRAFT - DO NOT CITE OR QUOTE
-------
1 Reduced model -79.9164 1 86.9979 3 <.0001
2
3 AIC: 76.8351
4
5 Goodness of Fit
6 Scaled
7 Dose Est._Prob. Expected Observed Size Residual
9 0.0000 0.0523 1.988 2.000 38 0.009
10 18.0000 0.0544 2.013 2.000 37 -0.009
11 83.0000 0.2368 8.999 9.000 38 0.000
12 429.0000 1.0000 24.000 24.000 24 0.000
13
14 ChiA2 = 0.00 d.f. = 2 P-value = 0.9999
15
16 Benchmark Dose Computation
17 Specified effect =0.1
18 Risk Type = Extra risk
19 Confidence level = 0.95
20 BMD = 65.2814
21 BMDL = 27.0766
22 BMDU = 91.3457
23
24 Taken together, (27.0766, 91.3457) is a 90% two-sided confidence interval for the BMD
C-19
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX D. DETAILS OF BMD ANALYSIS FOR
ORAL CSF FOR 1,4-DIOXANE
1 Dichotomous models available in the Benchmark Dose Software (BMDS) (version 2.1.1) were fit
2 to the incidence data for hepatocellular carcinoma and/or adenoma for mice and rats, as well as nasal
3 cavity tumors, peritoneal mesotheliomas, and mammary gland adenomas in rats exposed to 1,4-dioxane in
4 the drinking water. Doses associated with a benchmark response (BMR) of a 10% extra risk were
5 calculated. BMD10 and BMDL10 values from the best fitting model, determined by adequate global- fit (%2
6 p > 0.1) and AIC values, are reported for each endpoint (U.S. EPA. 2000a). If the multistage cancer
7 model is not the best fitting model for a particular endpoint, the best-fitting multistage cancer model for
8 that endpoint is also presented as a point of comparison.
9 A summary of the model predictions for the Kano et al. (2009) study are shown in Table D-l. The
10 data and BMD modeling results are presented separately for each dataset as follows:
11 • Hepatic adenomas and carcinomas in female F344 rats (Table D-2 and
12 Table D-3; Figure D-l)
13 • Hepatic adenomas and carcinomas in male F344 rats (Table D-4 and Table D-5;
14 Figure D-2 and Figure D-3)
15 • Significant tumor incidence data at sites other than the liver (i.e., nasal cavity,
16 mammary gland, and peritoneal) in male and female F344 rats (Table D-6)
17 o Nasal cavity tumors in female F344 rats (Table D-7; Figure D-4)
18 o Nasal cavity tumors in male F344 rats (Table D-8; Figure D-5)
19 o Mammary gland adenomas in female F344 rats (Table D-9; Figure D-6
20 and Figure D-7)
21 o Peritoneal mesotheliomas in male F344 rats (Table D-10; Figure D-8 and
22 Figure D-9)
23 • Hepatic adenomas and carcinomas in female BDF1 mice (Table D-l 1,
24 Table D-l2, and Table D-l3; Figure D-10, Figure D-l 1, Figure D-l2, and
25 Figure D-l3)
26 • Hepatic adenomas and carcinomas in male BDF1 mice (Table D-l4 and
27 Table D-15; Figure D-14 and Figure D-15)
28 Data and BMD modeling results from the additional chronic bioassays (NCI. 1978; Kociba et al..
29 1974) were evaluated for comparison with the data from Kano et al. (2009). These results are presented as
30 follows:
31 • Summary of BMDS dose-response modeling estimates associated with liver and
32 nasal tumor incidence data resulting from chronic oral exposure to 1,4-dioxane in
33 rats and mice (Table D-l6)
D-l
DRAFT - DO NOT CITE OR QUOTE
-------
1 • Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in
2 male and female Sherman rats (combined) (Kociba et al.. 1974) treated with
3 1,4-dioxane in the drinking water for 2 years (Table D-17)
4 o BMDS dose-response modeling results for incidence of hepatocellular
5 carcinoma in male and female Sherman rats (combined) (Kociba et al..
6 1974) exposed to 1,4-dioxane in drinking water for 2 years (Table D-18;
7 Figure D-16 and Figure D-17)
8 o BMDS dose-response modeling results for incidence of nasal squamous
9 cell carcinoma in male and female Sherman rats (combined) (Kociba et
10 al.. 1974) exposed to 1,4-dioxane in the drinking water for 2 years
11 (Table D-19; Figure D-18)
12 • Incidence of nasal cavity squamous cell carcinoma and hepatocellular adenoma
13 in Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the drinking
14 water (Table D-20)
15 o BMDS dose-response modeling results for incidence of hepatocellular
16 adenoma in female Osborne-Mendel rats (NCI. 1978) exposed to
17 1,4-dioxane in the drinking water for 2 years (Table D-21; Figure D-19
18 and Figure D-20)
19 o BMDS dose-response modeling results for incidence of nasal cavity
20 squamous cell carcinoma in female Osborne-Mendel rats (NCI. 1978)
21 exposed to 1,4-dioxane in the drinking water for 2 years (Table D-22;
22 Figure D-21 and Figure D-22)
23 o BMDS dose-response modeling results for incidence of nasal cavity
24 squamous cell carcinoma in male Osborne-Mendel rats (NCI. 1978)
25 exposed to 1,4-dioxane in the drinking water for 2 years (Table D-23;
26 Figure D-23 and Figure D-24)
27 • Incidence of hepatocellular adenoma or carcinoma in male and female B6C3F]
28 mice (NCI. 1978) exposed to 1,4-dioxane in drinking water (Table D-24)
29 o BMDS dose-response modeling results for the combined incidence of
30 hepatocellular adenoma or carcinoma in female B6C3Fi mice (NCI.
31 1978) exposed to 1,4-dioxane in the drinking water for 2 years
32 (Table D-25; Figure D-25)
33 o BMDS dose-response modeling results for incidence of combined
34 hepatocellular adenoma or carcinoma in male B6C3Fi mice (NCI. 1978)
35 exposed to 1,4-dioxane in the drinking water for 2 years (Table D-26;
36 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
37 The incidence of adenomas and the incidence of carcinomas within a dose group at a site or tissue
38 in rodents are sometimes combined. This practice is based upon the hypothesis that adenomas may
D-2
DRAFT - DO NOT CITE OR QUOTE
-------
1 develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a; McConnell et al..
2 1986). The incidence at high doses of both tumors in rat and mouse liver is high in the key study (Kano et
3 al., 2009). The incidence of hepatic adenomas and carcinomas was summed without double-counting
4 them so as to calculate the combined incidence of either a hepatic carcinoma or a hepatic adenoma in
5 rodents.
6 The variable N is used to denote the total number of animals tested in the dose group. The
7 variable Y is used here to denote the number of rodents within a dose group that have characteristic X,
8 and the notation Y(X) is used to identify the number with a specific characteristic X. Modeling was
9 performed on the adenomas and carcinomas separately and the following combinations of tumor types:
10 • Y(adenomas) = number of animals with adenomas, whether or not carcinomas
11 are present;
12 • Y(carcinomas) = number of animals with carcinomas, whether or not adenomas
13 are also present;
14 • Y(either adenomas or carcinomas) = number of animals with adenomas or
15 carcinomas, not both = Y(adenomas) + Y(carcinomas) - Y(both adenomas and
16 carcinomas);
17 • Y(neither adenomas nor carcinomas) = number of animals with no adenomas and
18 no carcinomas = N - Y(either adenomas or carcinomas).
D.1.2 Model Selection Criteria
19 Multiple models were fit to each dataset. The model selection criteria used in the BMD technical
20 guidance document (U.S. EPA. 2000a) were applied as follows:
21 • p-value for goodness-of-fit > 0.10
22 • AIC smaller than other acceptable models
23 • %2 residuals as small as possible
24 • No systematic patterns of deviation of model from data
25 Additional criteria were applied to eliminate implausible dose-response functions:
26 • Monotonic dose-response functions, e.g. no negative coefficients of polynomials
27 in MS models
28 • No infinitely steep dose-response functions near 0 (control dose), achieved by
29 requiring the estimated parameters "power" in the Weibull and Gamma models
30 and "slope" in the log-logistic model to have values > 1.
31 Because no single set of criteria covers all contingencies, an extended list of preferred models are
32 presented below in Table D-l.
D-3
DRAFT - DO NOT CITE OR QUOTE
-------
D.1.3 Summary
1 The BMDS models recommended to calculate rodent BMD and BMDL values and corresponding
2 human BMDnED and BMDLHED values are summarized in Table D-l.
Table D-1 Recommended models for rodents exposed to 1,4-dioxane in drinking water (Kano et
al.. 2009)
Endpoint
Model
selection
criterion
Model Type
AIC
p-value
BMDa
mg/kg-day
BMDLa BMDHEDa BMDLHEDa
mg/kg-day mg/kg-day mg/kg-day
Female F344 Rat
Hepatic
Tumors
Mammary
Gland
Tumors
Nasal
Cavity
Tumors
Lowest AIC
Lowest AIC
Lowest AIC
Multistage
(2 degree)
LogLogistic
Multistage
(3 degree)
91.
194
42.
5898
.151
6063
0.4516
0.8874
0.9966
79.83
161.01
381.65
58
81
282
.09
.91
.61
19
40
94
.84
.01
.84
14.43
20.35
70.23
Male F344 Rat
Hepatic
Tumors
Peritoneal
Meso-thel
ioma
Nasal
Cavity
Tumors
Female BDF1
Hepatic
Tumors
Lowest AIC
Lowest AIC
Lowest AIC
Mouse
Lowest AIC
BMR 50%
Probit
Probit
Multistage
(3 degree)
LogLogistic
LogLogistic
147
138
24
176
176
.787
.869
.747
.214
.214
0.9867
0.9148
0.9989
0.1421
0.1421
62.20
93.06
328.11
5.54
49.88"
51
76
245
3
32.
.12
.32
.63
.66
93°
17
26
91
0
7.
.43
.09
.97
.83
51b
14.33
21.39
68.85
0.55
4.95°
Male BDF1 Mouse
Hepatic
Tumors
Lowest AIC
Log-Logistic
248
.839
0.3461
34.78
16
.60
5
.63
2.68
"Values for BMR 10% unless otherwise noted.
bBMR 50%.
D.2 Female F344 Rats: Hepatic Carcinomas and Adenomas
3 The incidence data for hepatic carcinomas and adenomas in female F344 rats (Kano et al.. 2009)
4 are shown in Table D-2.
D-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-2 Data for hepatic adenomas and carcinomas in female F344 rats (Kano et al.. 2009)
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
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: Used with permission from Elsevier, Ltd., Kano et al. (2009)
1 Note that the incidence of rats with adenomas, with carcinomas, and with either adenomas or
2 carcinomas are monotone non-decreasing functions of dose except for 3 female rats in the control group.
3 These data therefore appear to be appropriate for dose-response modeling using BMDS.
4 The results of the BMDS modeling forthe entire suite of models are presented in Table D-3.
D-5
DRAFT - DO NOT CITE OR QUOTE
-------
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
LogLogistic
LogProbit"
AIC
93.1067
91.7017
93.102
93.0762
p-value
0.3024
0.4459
0.3028
0.3074
BMDio
mg/kg-day
89.46
93.02
88.34
87.57
BMDLio
mg/kg-day
62.09
71.60
65.52
66.19
x2a
0.027
0.077
0.016
0.001
BMDio HED
mg/kg-day
22.23
23.12
21.95
21.76
BMDLio HED
mg/kg-day
15.43
17.79
16.28
16.45
114094
°-0001
25.58
19.92
-1.827
6.36
4.95
Multistage-Cancer
(2 degree)
79.83
58.09 -0.408
19.84
14.43
gSefanC6r 93-2682 °-2747
92.81
59.31
0.077
23.06
14.74
Probit
91.8786 0.3839
85.46
67.84 -0.116
21.24
16.86
Weibull
93.2255 0.2825
92.67
59.89
0.088
23.03
14.88
Quantal-Linear
114.094
0.0001
25.58
19.92
-1.827
6.36
4.95
Dichotomous-Hill
4458.37
0
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
""Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
Multistage Cancer Model with 0.95 Confidence Level
I
c
o
t3
(O
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
BMDL
BMD
50
100
150
200 250
dose
300
350
400
450
07:20 10/262009
Source: Used with permission of Elsevier, Ltd., 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.
1
2
3
4
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)
DRAFT - DO NOT CITE OR QUOTE
D-6
-------
1 Gnuplot Plotting File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_hepato_adcar_Msc-BMR10-2poly.plt
3 Mon Oct 26 08:20:52 2009
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7
8 The form of the probability function is:
9 P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
10
11 The parameter betas are restricted to be positive
12
13 Dependent variable = Effect
14 Independent variable = Dose
15
16 Total number of observations = 4
17 Total number of records with missing values = 0
18 Total number of parameters in model = 3
19 Total number of specified parameters = 0
20 Degree of polynomial = 2
21
22 Maximum number of iterations =250
23 Relative Function Convergence has been set to: le-008
24 Parameter Convergence has been set to: le-008
25
26 Default Initial Parameter Values
27 Background = 0.0281572
28 Beta(l) = 0
29 Beta(2) = 1.73306e-005
30
31 Asymptotic Correlation Matrix of Parameter Estimates (*** The model parameter(s)
32 -Beta(l)have been estimated at a boundary point, or have been specified by the user,
33 and do not appear in the correlation matrix )
34
35 Background Beta(2)
36 Background 1 -0.2
37 Beta(2) -0.2 1
38
39 Parameter Estimates
40 95.0% Wald Confidence Interval
41 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
42 Background 0.0362773 * * *
43 Beta(l) 0 * * *
44 Beta(2) 1.65328e-005 * * *
45
46 * - Indicates that this value is not calculated.
47
48
49 Analysis of Deviance Table
50
51 Model Log(likelihood) # Param's Deviance Test d.f. P-value
52 Full model -42.9938 4
53 Fitted model -43.7949 2 1.60218 2 0.4488
54 Reduced model -120.43 1 154.873 3 <.0001
55
56 AIC: 91.5898
57
58 Goodness of Fit
59 Scaled
60 Dose Est._Prob. Expected Observed Size Residual
61
62 0.0000 0.0363 1.814 3.000 50 0.897
63 18.0000 0.0414 2.071 1.000 50 -0.760
64 83.0000 0.1400 7.001 6.000 50 -0.408
65 429.0000 0.9540 47.701 48.000 50 0.202
66
67 ChiA2 = 1.59 d.f. = 2 P-value = 0.4516
D-7
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
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
15
16
17
18
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)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
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: Used with permission from Elservier, Ltd., Kano et al. (2009).
19
20 Note that the incidence of rats with hepatic adenomas, carcinomas, and with either adenomas or
21 carcinomas are monotone non-decreasing functions of dose. These data therefore appear to be appropriate
22 for dose-response modeling using BMDS.
23 The results of the BMDS modeling for the entire suite of models tested using the data for hepatic
24 adenomas and carcinomas for male F344 rats are presented in Table D-5.
D-8
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-5 BMDS dose-response modeling results for the combined incidence of adenomas and
carcinomas in livers of male F344 rats (Kano et al.. 2009)
Model
Gamma
Logistic
LogLogistic
LogProbit"
AIC
149.884
147.813
149.886
149.913
p-value
0.7257
0.9749
0.7235
0.6972
BMDio
mg/kg-day
62.41
68.74
62.10
61.70
BMDLio
mg/kg-day
30.79
55.39
34.61
37.49
x2a
-0.03
0.097
-0.021
-0.003
BMDio HED
mg/kg-day
17.49
19.27
17.41
17.29
BMDLio HED
mg/kg-day
8.63
15.53
9.70
10.51
Multistage-Cancer
(1 degree)
152.836 0.0978
23.82
18.34
-0.186
6.68
5.14
Multistage-Cancer
(2 degree)
149.814 0.8161
61.68
28.26
-0.063
17.29
7.92
Multistage-Cancer
(3 degree)
149.772 0.9171
63.62
27.49
-0.024
17.83
7.71
Probitc
147.787 0.9867
62.20
51.12
-0.05
17.43
14.33
Weibull
149.856 0.7576
62.63
30.11
-0.039
17.56
8.44
Quantal-Linear
152.836 0.0978
23.82
18.34
-0.186
6.68
5.14
Dichotomous-Hill
4441.71
NCT
NCT
0
0
0
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
""Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
Probit Model with 0.95 Confidence Level
I
C
o
'•
0.8
0.6
0.4
0.2
100
150
200
250
dose
07:32 10/26 2009
Source: Used with permission from Elservier, Ltd., Kano et al. (2009).
Figure D-2 Probit BMD model for the combined incidence of hepatic adenomas and
carcinomas in male F344 rats.
1
2
o
J
4
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)
DRAFT - DO NOT CITE OR QUOTE
D-9
-------
1 Gnuplot Plotting File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-BMR10.plt
3 Mon Oct 26 08:32:08 2009
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7
8 The form of the probability function is:
9 P[response] = CumNorm(Intercept+Slope*Dose),
10 where CumNorm(.) is the cumulative normal distribution function
11
12 Dependent variable = Effect
13 Independent variable = Dose
14 Slope parameter is not restricted
15
16 Total number of observations = 4
17 Total number of records with missing values = 0
18 Maximum number of iterations =250
19 Relative Function Convergence has been set to: le-008
20 Parameter Convergence has been set to: le-008
21
22
23 Default Initial (and Specified) Parameter Values
24 background = 0 Specified
25 intercept = -1.51718
26 slope = 0.00831843
27
28 Asymptotic Correlation Matrix of Parameter Estimates
29 (*** Thg model parameter(s) -background have been estimated at a boundary point, or
30 have been specified by the user, and do not appear in the correlation matrix )
31
32 intercept slope
33 intercept 1 -0.69
34 slope -0.69 1
35
36
37 Parameter Estimates
38 95.0% Wald Confidence Interval
39 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40 intercept 1.53138 0.160195 -1.84535 -1.2174
41 slope 0.00840347 0.000976752 0.00648907 0.0103179
42
43
44 Analysis of Deviance Table
45
46 Model Log(likelihood) # Param's Deviance Test d.f. P-value
47 Full model -71.8804 4
48 Fitted model -71.8937 2 0.0265818 2 0.9868
49 Reduced model -115.644 1 87.528 3 <.0001
50
51 AIC: 147.787
52
53
54 Goodness of Fit
55 Scaled
56 Dose Est._Prob. Expected Observed Size Residual
57
58 0.0000 0.0628 3.142 3.000 50 -0.083
59 11.0000 0.0751 3.754 4.000 50 0.132
60 55.0000 0.1425 7.125 7.000 50 -0.050
61 274.0000 0.7797 38.985 39.000 50 0.005
62
63 ChiA2 = 0.03 d.f. = 2 P-value = 0.9867
64
65 Benchmark Dose Computation
66
67 Specified effect =0.1
D-10
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
Risk Type = Extra risk
Confidence level = 0.95
BMD = 62.1952
BMDL = 51.1158
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
T3
-------
1
2 Maximum number of iterations =250
3 Relative Function Convergence has been set to: le-008
4 Parameter Convergence has been set to: le-008
5
6 Default Initial Parameter Values
7 Background = 0.0623822
8 Beta(l) = 0.00142752
9 Beta(2) = 0
10 Beta(3) = 5.14597e-008
11 Asymptotic Correlation Matrix of Parameter Estimates
12 (*** The model parameter(s) -Beta(2)have been estimated at a boundary point, or have
13 been specified by the user, and do not appear in the correlation matrix )
14
15 Background Beta(l) Beta(3)
16 Background 1 -0.67 0.58
17 Beta(l) -0.67 1 -0.95
18 Beta(3) 0.58 -0.95 1
19
20
21 Parameter Estimates
22
23 95.0% Wald Confidence Interval
24 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
25 Background 0.0619918 * * *
26 Beta(l) 0.001449 * * *
27 Beta(2) 0 * * *
28 Beta(3) 5.11829e-008 * * *
29
30 * - Indicates that this value is not calculated.
31
32
33
34 Analysis of Deviance Table
35
36 Model Log(likelihood) # Param's Deviance Test d.f. P-value
37 Full model -71.8804 4
38 Fitted model -71.8858 3 0.0107754 1 0.9173
39 Reduced model -115.644 1 87.528 3 <.0001
40
41 AIC: 149.772
42
43
44 Goodness of Fit
45 Scaled
46 Dose Est._Prob. Expected Observed Size Residual
47 ~
48 0.0000 0.0620 3.100 3.000 50 -0.058
49 11.0000 0.0769 3.844 4.000 50 0.083
50 55.0000 0.1412 7.059 7.000 50 -0.024
51 274.0000 0.7799 38.997 39.000 50 0.001
52
53 Chi^2 = 0.01 d.f. = 1 P-value = 0.9171
54
55
56 Benchmark Dose Computation
57
58 Specified effect =0.1
59 Risk Type = Extra risk
60 Confidence level = 0.95
61 BMD = 63.6179
62 BMDL = 27.4913
63 BMDU = 123.443
64
65 Taken together, (27.4913, 123.443) is a 90% two-sided confidence interval for the BMD
66
67 Multistage Cancer Slope Factor = 0.00363752
D-12
DRAFT - DO NOT CITE OR QUOTE
-------
D.4 F344 Rats: Tumors at Other Sites
1 The data for tumors at sites other than the liver in male and female F344 rats (Kano et al.. 2009)
2 are shown in Table D-6. Note that the incidence of rats with these endpoints are monotone non-decreasing
3 functions (except female peritoneal mesotheliomas). These data therefore appear to be appropriate for
4 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)
Tumor site and type
Nasal cavity squamous cell carcinoma
Peritoneal mesothelioma
Mammary gland adenoma
Total number per group
Female
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: Used with permission from Elsevier, Ltd., Kano et al., (2009).
5 The results of the BMDS modeling for the entire suite of models are presented in Table D-7
6 through Table D-10 for tumors in the nasal cavity, mammary gland, and peritoneal cavity.
D-13
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-7 BMDS dose-response modeling results for the incidence of nasal cavity tumors in
female F344 rats3 (Kano et al.. 2009)
Model
Gamma
Logistic
LogLogistic
LogProbitc
(7d^TCancer
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
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'B
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.
""Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
°Slope restricted > 1.
dBest-fitting model.
Multistage uancer Model with u.yt> uontidence Level
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMD
50
100
150
200 250
dose
300
350
400
450
07:28 10/26 2009
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-4 Multistage BMD model (3 degree) for nasal cavity tumors in female
F344 rats.
1 =========================================================
2 Multistage Cancer Model . (Version: 1.7; Date: 05/16/2008)
3 Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-BMR10-3poly. (d)
DRAFT - DO NOT CITE OR QUOTE
D-14
-------
1 Gnuplot Plotting File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-BMR10-3poly.plt
3 Mon Oct 26 08:28:58 2009
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7 The form of the probability function is: P[response] = background +
8 (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
9
10 The parameter betas are restricted to be positive
11
12 Dependent variable = Effect
13 Independent variable = Dose
14 Total number of observations = 4
15 Total number of records with missing values = 0
16 Total number of parameters in model = 4
17 Total number of specified parameters = 0
18 Degree of polynomial = 3
19
20 Maximum number of iterations =250
21 Relative Function Convergence has been set to: le-008
22 Parameter Convergence has been set to: le-008
23
24 Default Initial Parameter Values
25 Background = 0
26 Beta(l) = 0
27 Beta(2) = 0
28 Beta(3) = 1.91485e-009
29 Asymptotic Correlation Matrix of Parameter Estimates
30 (*** The model parameter(s) -Background -Beta(l) -Beta(2)
31 have been estimated at a boundary point, or have been specified by the user,
32 and do not appear in the correlation matrix )
33
34 Beta(3)
35 Beta (3) 1
36
37 Parameter Estimates
38
39 95.0% Wald Confidence Interval
40 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
41 Background 0 * * *
42 Beta(l) 0 * * *
43 Beta(2) 0 * * *
44 Beta(3) 1.89531e-009 * * *
45
46 * - Indicates that this value is not calculated.
47
48
49 Analysis of Deviance Table
50
51 Model Log(likelihood) # Param's Deviance Test d.f. P-value
52 Full model -20.2482 4
53 Fitted model -20.3031 1 0.109908 3 0.9906
54 Reduced model -30.3429 1 20.1894 3 0.0001551
55
56 AIC: 42.6063
57
58
59 Goodness of Fit
60 Scaled
61 Dose Est._Prob. Expected Observed Size Residual
62
63 0.0000 0.0000 0.000 0.000 50 0.000
64 18.0000 0.0000 0.001 0.000 50 -0.024
65 83.0000 0.0011 0.054 0.000 50 -0.233
66 429.0000 0.1390 6.949 7.000 50 0.021
67
D-15
DRAFT - DO NOT CITE OR QUOTE
-------
1 ChiA2 = 0.06 d.f. = 3 P-value = 0.9966
2
3
4 Benchmark Dose Computation
5
6 Specified effect = 0.1
7 Risk Type = Extra risk
8 Confidence level = 0.95
9 BMD = 381.651
10 BMDL = 282.609
11 BMDU = 500.178
12
13 Taken together, (282.609, 500.178) is a 90% two-sided confidence interval for the BMD
14
15 Multistage Cancer Slope Factor = 0.000353846
D-16
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-8 BMDS dose-response modeling results for the incidence of nasal cavity tumors in
male F344 rats3 (Kano et al.. 2009)
Model
Gamma
Logistic
LogLogistic
LogProbitc
AIC
26.6968
26.6968
26.6968
26.6968
p-value
1
1
1
1
BMDio
mg/kg-day
299.29
281.06
288.31
303.06
BMDLio
mg/kg-day
244.10
261.29
245.29
238.86
x2b
0
0
0
0
BMDio HED
mg/kg-day
83.89
78.78
80.81
84.94
BMDLio HED
mg/kg-day
68.42
73.24
68.75
66.95
Multistage-Cancer
(1 degree)
26.0279 0.8621
582.49
256.43
0.384
163.28
71.88
Multistage-Cancer
(2 degree)
24.9506
0.988
365.19
242.30
0.073
102.37
67.92
Multistage-Cancer
(3 degree)d
24.747
0.9989
328.11
245.63
0.015
91.97
68.85
Probit
26.6968
1
287.96
257.01
0
80.72
72.04
Weibull
26.6968
1
288.00
246.36
0
80.73
69.06
Quantal-Linear
26.0279 0.8621
582.49
256.43
0.384
163.28
71.88
Dichotomous-Hill
28.6968 0.9994
290.52
261.47
6.25x10"
81.44
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.
Multistage cancer ivioaei wnn u.ao (^onnaence i_evei
o
'o
CO
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMO
50
100
150 200
dose
250
300
07:34 10/26 2009
Source: Used with permission from Elsevier, Ltd., 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)
D-17
DRAFT - DO NOT CITE OR QUOTE
-------
1 Input Data File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_nasal_car_Msc-BMR10-3poly.(d)
3 Gnuplot Plotting File:
4 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_nasal_car_Msc-BMR10-3poly.plt
5 Mon Oct 26 08:34:20 2009
6 ====================================================================
7 BMDS Model Run
8 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9 The form of the probability function is: P[response] = background +
10 (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
11
12 The parameter betas are restricted to be positive
13
14 Dependent variable = Effect
15 Independent variable = Dose
16 Total number of observations = 4
17 Total number of records with missing values = 0
18 Total number of parameters in model = 4
19 Total number of specified parameters = 0
20 Degree of polynomial = 3
21
22 Maximum number of iterations =250
23 Relative Function Convergence has been set to: le-008
24 Parameter Convergence has been set to: le-008
25
26 Default Initial Parameter Values
27 Background = 0
28 Beta(l) = 0
29 Beta(2) = 0
30 Beta(3) = 3.01594e-009
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34
35 (*** The model parameter(s) -Background -Beta(l) -Beta (2)
36 have been estimated at a boundary point, or have been specified by the user,
37 and do not appear in the correlation matrix )
38
39 Beta(3)
40 Beta (3) 1
41
42
43 Parameter Estimates
44
45 95.0% Wald Confidence Interval
46 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
47 Background 0 * * *
48 Beta(l) 0 * * *
49 Beta(2) 0 * * *
50 Beta(3) 2.98283e-009 * * *
51
52 * - Indicates that this value is not calculated.
53
54
55
56 Analysis of Deviance Table
57
58 Model Log(likelihood) # Param's Deviance Test d.f. P-value
59 Full model -11.3484 4
60 Fitted model -11.3735 1 0.0502337 3 0.9971
61 Reduced model -15.5765 1 8.45625 3 0.03747
62
63 AIC: 24.747
64
65
66 Goodness of Fit
67 Scaled
D-18
DRAFT - DO NOT CITE OR QUOTE
-------
1 Dose Est._Prob. Expected Observed Size Residual
2
3 0.0000 0.0000 0.000 0.000 50 0.000
4 11.0000 0.0000 0.000 0.000 50 -0.014
5 55.0000 0.0005 0.025 0.000 50 -0.158
6 274.0000 0.0595 2.976 3.000 50 0.015
7
8 ChiA2 = 0.03 d.f. = 3 P-value = 0.9989
9
10
11 Benchmark Dose Computation
12
13 Specified effect =0.1
14 Risk Type = Extra risk
15 Confidence level = 0.95
16 BMD = 328.108
17 BMDL = 245.634
18 BMDU = 1268.48
19
20 Taken together, (245.634, 1268.48) is a 90% two-sided confidence interval for the BMD
21
22 Multistage Cancer Slope Factor = 0.00040711
D-19
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-9 BMDS dose-response modeling results for the incidence of mammary gland
adenomas in female F344 rats (Kano et al.. 2009)
Model
Gamma
Logistic
Log Logistic"
LogProbitc
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
NC°
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'b
BMDio HED
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.
Log-Logistic Model with 0.95 Confidence Level
o
OS
0.5
0.4
0.3
0.2
0.1
100 150 200 250 300 350 400 45(
11:31 02/01 2010
Source: Use with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-6 LogLogistic BMD model for mammary gland adenomas in female F344
rats.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
DRAFT - DO NOT CITE OR QUOTE
D-20
-------
1 Input Data File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.(d)
2 Gnuplot Plotting File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.plt
3 Mon Feb 01 11:31:31 2010
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7 The form of the probability function is:
8
9 P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
10
11 Dependent variable = Effect
12 Independent variable = Dose
13 Slope parameter is restricted as slope >= 1
14
15 Total number of observations = 4
16 Total number of records with missing values = 0
17 Maximum number of iterations = 250
18 Relative Function Convergence has been set to: le-008
19 Parameter Convergence has been set to: le-008
20
21 User has chosen the log transformed model
22
23 Default Initial Parameter Values
24 background = 0.12
25 intercept = -7.06982
26 slope = 1
27 Asymptotic Correlation Matrix of Parameter Estimates
28
29 (*** Thg model parameter(s) -slope have been estimated at a boundary point, or have
30 been specified by the user, and do not appear in the correlation matrix )
31
32 background intercept
33 background 1 -0.53
34 intercept -0.53 1
35
36 Parameter Estimates
37
38 95.0% Wald Confidence Interval
39 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40 background 0.130936 * * *
41 intercept -7.2787 * * *
42 slope 1 * * *
43
44 * - Indicates that this value is not calculated.
45
46
47
48 Analysis of Deviance Table
49
50 Model Log(likelihood) # Param's Deviance Test d.f. P-value
51 Full model -94.958 4
52 Fitted model -95.0757 2 0.235347 2 0.889
53 Reduced model -98.6785 1 7.4409 3 0.0591
54
55 AIC: 194.151
56
57
58 Goodness of Fit
59 Scaled
60 Dose Est._Prob. Expected Observed Size Residual
61
62 0.0000 0.1309 6.547 6.000 50 -0.229
63 18.0000 0.1416 7.080 7.000 50 -0.032
64 83.0000 0.1780 8.901 10.000 50 0.406
65 429.0000 0.3294 16.472 16.000 50 -0.142
66
67 ChiA2 = 0.24 d.f. = 2 P-value = 0.8874
D-21
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 161.012
BMDL = 81.9107
Multistage Cancer Model with 0.95 Confidence Level
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
50
100
150
200 250
dose
300
350
400
450
07:27 10/26 2009
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-7 Multistage BMD model (1 degree) for mammary gland adenomas in
female F344 rats.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_mamm_ad_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_mamm_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
D-22
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Maximum number of iterations =250
3 Relative Function Convergence has been set to: le-008
4 Parameter Convergence has been set to: le-008
5
6 Default Initial Parameter Values
7 Background = 0.136033
8 Beta(l) = 0.000570906
9 Asymptotic Correlation Matrix of Parameter Estimates
10
11 Background Beta(l)
12 Background 1 -0.58
13 Beta(l) -0.58 1
14
15
16 Parameter Estimates
17
18 95.0% Wald Confidence Interval Variable Estimate Std. Err. Lower Conf. Limit Upper
19 Conf. Limit
20 Background .133161 * * *
21 Beta(l) 0.000596394 * * *
22
23 * - Indicates that this value is not calculated.
24
25
26
27 Analysis of Deviance Table
28
29 Model Log(likelihood) # Param's Deviance Test d.f. P-value
30 Full model -94.958 4
31 Fitted model -95.111 2 0.305898 2 0.8582
32 Reduced model -98.6785 1 7.4409 3 0.0591
33
34 AIC: 194.222
35
36
37 Goodness of Fit
38 Scaled
39 Dose Est._Prob. Expected Observed Size Residual
40
41 0.0000 0.1332 6.658 6.000 50 -0.274
42 18.0000 0.1424 7.121 7.000 50 -0.049
43 83.0000 0.1750 8.751 10.000 50 0.465
44 429.0000 0.3288 16.442 16.000 50 -0.133
45
46 ChiA2 = 0.31 d.f. = 2 P-value = 0.8559
47
48
49 Benchmark Dose Computation
50
51 Specified effect =0.1
52 Risk Type = Extra risk
53 Confidence level = 0.95
54 BMD =176.663
55 BMDL = 99.1337
56 BMDU = 501.523
57
58 Taken together, (99.1337, 501.523) is a 90% two-sided confidence interval for the BMD
59
60 Multistage Cancer Slope Factor = 0.00100874
D-23
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-10 BMDS dose-response modeling results for the incidence of peritoneal mesotheliomas
in male F344 rats (Kano et al.. 2009)
Model
Gamma
Logistic
LogLogistic
LogProbit"
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
NCa
BMDio
mg/kg-day
73.52
103.52
72.56
70.29
41.04
77.73
77.73
93.06
74.77
41.04
NCa
BMDLio
mg/kg-day
35.62
84.35
36.37
52.59
30.51
35.43
35.43
76.32
35.59
30.51
NC°
x2a
0.018
0.446
0.014
0.001
-1.066
0.067
0.067
0.315
0.027
-1.066
0
BMDlOHED
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.
"Slope restricted > 1.
""Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
Probit Model with 0.95 Confidence Level
1
2
3
4
o
£5
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0 -
100
150
200
250
dose
07:41 10/262009
Source: Used with permission from Elsevier, Ltd., 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)
D-24
DRAFT - DO NOT CITE OR QUOTE
-------
1 Gnuplot Plotting File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_peri_meso_Prb-BMR10.plt
3 Mon Oct 26 08:41:29 2009
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7
8 The form of the probability function is: P[response] = CumNorm(Intercept+Slope*Dose) ,
9 where CumNorm(.) is the cumulative normal distribution function
10
11 Dependent variable = Effect
12 Independent variable = Dose
13 Slope parameter is not restricted
14
15 Total number of observations = 4
16 Total number of records with missing values = 0
17 Maximum number of iterations =250
18 Relative Function Convergence has been set to: le-008
19 Parameter Convergence has been set to: le-008
20
21 Default Initial (and Specified) Parameter Values
22 background = 0 Specified
23 intercept = -1.73485
24 slope = 0.00692801
25
26 Asymptotic Correlation Matrix of Parameter Estimates
27 (*** The model parameter(s) -background have been estimated at a boundary point, or
28 have been specified by the user, and do not appear in the correlation matrix )
29
30 intercept slope
31 intercept 1 -0.75
32 slope -0.75 1
33
34 Parameter Estimates
35 95.0% Wald Confidence Interval
36 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
37 intercept -1.73734 0.18348 -2.09695 -1.37772
38 slope 0.00691646 0.000974372 0.00500672 0.00882619
39
40 Analysis of Deviance Table
41 Model Log(likelihood) # Param's Deviance Test d.f. P-value
42 Full model -67.3451 4
43 Fitted model -67.4344 2 0.178619 2 0.9146
44 Reduced model -95.7782 1 56.8663 3 <.0001
45 AIC: 138.869
46
47 Goodness of Fit
48 Scaled
49 Dose Est._Prob. Expected Observed Size Residual
50
51 0.0000 0.0412 2.058 2.000 50 -0.041
52 11.0000 0.0483 2.417 2.000 50 -0.275
53 55.0000 0.0874 4.370 5.000 50 0.315
54 274.0000 0.5627 28.134 28.000 50 -0.038
55
56 ChiA2 = 0.18 d.f. = 2 P-value = 0.9148
57 Benchmark Dose Computation
58 Specified effect =0.1
59 Risk Type = Extra risk
60 Confidence level = 0.95
61 BMD = 93.0615
62 BMDL = 76.3242
D-25
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
o
t3
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
50
100
150
200
250
dose
07:41 10/26 2009
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-9 Multistage BMD (2 degree) model for peritoneal mesotheliomas in male
F344 rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_peri_meso_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_peri_meso_Msc-BMR10-2poly.plt
Mon Oct 26 08:41:28 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/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
Background = 0.0358706
D-26
DRAFT - DO NOT CITE OR QUOTE
-------
1 Beta(l) = 0.000816174
2 Beta(2) = 7.47062e-006
3
4
5 Asymptotic Correlation Matrix of Parameter Estimates
6
7 Background Beta(l) Beta(2)
8 Background 1 -0.67 0.59
9 Beta(l) -0.67 1 -0.98
10 Beta(2) 0.59 -0.98 1
11
12 Parameter Estimates
13 95.0% Wald Confidence Interval
14 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
15 Background 0.0366063 * * *
16 Beta(l) 0.000757836 * * *
17 Beta(2) 7.6893e-006 * * *
18
19 * - Indicates that this value is not calculated.
20
21 Analysis of Deviance Table
22
23 Model Log(likelihood) # Param's Deviance Test d.f. P-value
24 Full model -67.3451 4
25 Fitted model -67.3733 3 0.056567 1 0.812
26 Reduced model -95.7782 1 56.8663 3 <.0001
27
28 AIC: 140.747
29
30
31 Goodness of Fit
32 Scaled
33 Dose Est._Prob. Expected Observed Size Residual
34 ~
35 0.0000 0.0366 1.830 2.000 50 0.128
36 11.0000 0.0455 2.275 2.000 50 -0.186
37 55.0000 0.0972 4.859 5.000 50 0.067
38 274.0000 0.5605 28.027 28.000 50 -0.008
39
40 Chi^2 = 0.06 d.f. = 1 P-value = 0.8135
41
42
43 Benchmark Dose Computation
44
45 Specified effect =0.1
46 Risk Type = Extra risk
47 Confidence level =0.95
48 BMD = 77.7277
49 BMDL = 35.4296
50 BMDU = 118.349
51
52 Taken together, (35.4296, 118.349) is a 90% two-sided confidence interval for the BMD
53
54 Multistage Cancer Slope Factor = 0.0028225
D.5 Female BDF1 Mice: Hepatic Carcinomas and Adenomas
55 Data for female BDF1 mouse hepatic carcinomas and adenomas are shown in Table D-l 1. Note
56 that the incidence of carcinomas and the incidence of either adenomas or carcinomas are monotone
57 non-decreasing functions of dose. These data therefore appear to be appropriate for dose-response
58 modeling using BMDS. However, the incidence of adenomas clearly reaches a peak value at
D-27
DRAFT - DO NOT CITE OR QUOTE
-------
1 66 mg/kg-day and then decreases sharply with increasing dose. This cannot be modeled by a multistage
2 model using only non-negative coefficients. To some extent the incidence of "either adenomas or
3 carcinomas" retains some of the inverted-U shaped dose-response of the adenomas, which dominate
4 based on their high incidence at the lowest dose groups (66 and 278 mg/kg-day), thus is not well
5 characterized by any multistage model.
Table D-11 Data for hepatic adenomas and carcinomas in female BDF1 mice (Kano et al.. 2009)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
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: Used with permission from Elsevier, Ltd., Kano et al. (2009).
6 The results of the BMDS modeling for the entire suite of models for hepatic adenomas and
7 carcinomas in female BDF1 mice are presented in Table D-12. The multistage models did not provide
8 reasonable fits to the incidence data for hepatocellular adenoma or carcinoma in female BDF1 mice. The
9 log-logistic model provided the best-fit to the data as indicated by the AIC and/>-value as was chosen as
10 the best-fitting model to carry forward in the analysis; however, this model resulted in a BMDLio much
11 lower than the response level at the lowest dose in the study (Kano et al.. 2009). Thus, the log-logistic
12 model was run for BMRs of 30 and 50%. The output from these models are shown in Figures D-11 and
13 D-12. A summary of the BMD results for BMRs of 10, 30, and 50% are shown in Table D-13. Using a
14 higher BMR resulted in BMDLs closer to the lowest observed response data, and a BMR of 50% was
15 chosen to carry forward in the analysis.
16 The graphical output from fitting these models suggested that a simpler model obtained by
17 dropping the data point for the highest dose (964 mg/kg-day) might also be adequate. This was tested and
18 the results did not affect the choice of the model, nor significantly affect the resulting BMDs and BMDLs.
DRAFT - DO NOT CITE OR QUOTE
D-28
-------
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 Logistic"
LogProbitc
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
NCa
BMDio
mg/kg-day
26.43
58.05
5.54
26.37
26.43
26.43
26.43
69.89
26.43
26.43
NCa
BMDLio
mg/kg-day
19.50
44.44
3.66
19.57
19.50
19.50
19.50
56.22
19.50
19.50
NCa
x23
-2.654
3.201
-0.121
-1.166
-2.654
-2.654
-2.654
3.114
-2.654
-2.654
0
BMDlOHED
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.
b Best-fitting model, lowest AIC value.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
Table D-13 BMDS LogLogistic dose-response modeling results using BMRs of 10, 30, and 50%
for the combined incidence of hepatic adenomas and carcinomas in female
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
BMDLHED
mg/kg-day
0.55
2.12
4.95
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
D-29
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
Log-Logistic
Bli/IDLJBMD
200
400
600
800
1000
dose
11:2605/122010
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-10 LogLogistic 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
User has chosen the log transformed model
D-30
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Default Initial Parameter Values
3 background = 0.1
4 intercept = -4.33618
5 slope = 1
6
7 Asymptotic Correlation Matrix of Parameter Estimates
8 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
9 been specified by the user, and do not appear in the correlation matrix )
10
11 background intercept
12 background 1 -0.32
13 intercept -0.32 1
14
15 Parameter Estimates
16
17 95.0% Wald Confidence Interval
18 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
19 background 0.105265 * * *
20 intercept -3.90961 * * *
21 slope 1 * * *
22
23 * - Indicates that this value is not calculated.
24
25 Analysis of Deviance Table
26
27 Model Log(likelihood) # Param's Deviance Test d.f. P-value
28 Full model -84.3055 4
29 Fitted model -86.107 2 3.6029 2 0.1651
30 Reduced model -131.248 1 93.8853 3 <.0001
31
32 AIC: 176.214
33
34
35 Goodness of Fit
36 Scaled
37 Dose Est._Prob. Expected Observed Size Residual
38 ~
39 0.0000 0.1053 5.263 5.000 50 -0.121
40 66.0000 0.6149 30.743 35.000 50 1.237
41 278.0000 0.8639 43.194 41.000 50 -0.905
42 964.0000 0.9560 47.799 46.000 50 -1.240
43
44 Chi^2 = 3.90 d.f. = 2 P-value = 0.1421
45
46
47 Benchmark Dose Computation
48 Specified effect =0.1
49 Risk Type = Extra risk
50 Confidence level =0.95
51 BMD = 5.54218
52 BMDL = 3.65848
53
D-31
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
Log-Logistic
EMDLBI
BMD
200
400
600
800
1000
dose
11:2605/122010
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-ll LogLogistic 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
User has chosen the log transformed model
D-32
DRAFT - DO NOT CITE OR QUOTE
-------
1 Default Initial Parameter Values
2 background = 0.1
3 intercept = -4.33618
4 slope = 1
5
6 Asymptotic Correlation Matrix of Parameter Estimates
7 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
8 been specified by the user, and do not appear in the correlation matrix)
9
10 background intercept
11 background 1 -0.32
12 intercept -0.32 1
13
14 Parameter Estimates
15
16 95.0% Wald Confidence Interval
17 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
18 background 0.105265 * * *
19 intercept -3.90961 * * *
20 slope 1 * * *
21
22 * - Indicates that this value is not calculated.
23
24
25 Analysis of Deviance Table
26
27 Model Log(likelihood) # Param's Deviance Test d.f. P-value
28 Full model -84.3055 4
29 Fitted model -86.107 2 3.6029 2 0.1651
30 Reduced model -131.248 1 93.8853 3 <.0001
31
32 AIC: 176.214
33
34
35 Goodness of Fit
36 Scaled
37 Dose Est._Prob. Expected Observed Size Residual
38 ~
39 0.0000 0.1053 5.263 5.000 50 -0.121
40 66.0000 0.6149 30.743 35.000 50 1.237
41 278.0000 0.8639 43.194 41.000 50 -0.905
42 964.0000 0.9560 47.799 46.000 50 -1.240
43
44 Chi^2 = 3.90 d.f. = 2 P-value = 0.1421
45
46
47 Benchmark Dose Computation
48 Specified effect =0.3
49 Risk Type = Extra risk
50 Confidence level =0.95
51 BMD = 21.377
52 BMDL = 14.1113
D-33
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
Log-Logistic
BMDL
BMD
200
400
600
800
1000
dose
11:2605/122010
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-12 LogLogistic 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
User has chosen the log transformed model
D-34
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Default Initial Parameter Values
3 background = 0.1
4 intercept = -4.33618
5 slope = 1
6
7 Asymptotic Correlation Matrix of Parameter Estimates
8 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
9 been specified by the user, and do not appear in the correlation matrix)
10
11 background intercept
12 background 1 -0.32
13 intercept -0.32 1
14
15 Parameter Estimates
16
17 95.0% Wald Confidence Interval
18 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
19 background 0.105265 * * *
20 intercept -3.90961 * * *
21 slope 1 * * *
22
23 * - Indicates that this value is not calculated.
24
25 Analysis of Deviance Table
26
27 Model Log(likelihood) # Param's Deviance Test d.f. P-value
28 Full model -84.3055 4
29 Fitted model -86.107 2 3.6029 2 0.1651
30 Reduced model -131.248 1 93.8853 3 <.0001
31
32 AIC: 176.214
33
34 Goodness of Fit
35 Scaled
36 Dose Est._Prob. Expected Observed Size Residual
37
38 0.0000 0.1053 5.263 5.000 50 -0.121
39 66.0000 0.6149 30.743 35.000 50 1.237
40 278.0000 0.8639 43.194 41.000 50 -0.905
41 964.0000 0.9560 47.799 46.000 50 -1.240
42
43 ChiA2 = 3.90 d.f. = 2 P-value = 0.1421
44
45
46 Benchmark Dose Computation
47 Specified effect =0.5
48 Risk Type = Extra risk
49 Confidence level = 0.95
50 BMD = 49.8797
51 BMDL = 32.9263
D-35
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
liMDLB
BMD
200
400
600
800
1000
dose
11:2605/122010
Source: Used with permission from Elsevier, Ltd., 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
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
D-36
DRAFT - DO NOT CITE OR QUOTE
-------
1 Parameter Convergence has been set to: le-008
2
3 Default Initial Parameter Values
4 Background = 0.51713
5 Beta(l) = 0.00201669
6
7 Asymptotic Correlation Matrix of Parameter Estimates
8
9 Background Beta(l)
10 Background 1 -0.65
11 Beta(l) -0.65 1
12
13 Parameter Estimates
14
15 95.0% Wald Confidence Interval
16 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17 Background 0.265826 * * *
18 Beta(l) 0.00398627 * * *
19
20 * - Indicates that this value is not calculated.
21
22 Analysis of Deviance Table
23
24 Model Log(likelihood) # Param's Deviance Test d.f. P-value
25 Full model -84.3055 4
26 Fitted model -99.6653 2 30.7195 2 2.1346928e-007
27 Reduced model -131.248 1 93.8853 3 <.0001
28
29 AIC: 203.331
30
31 Goodness of Fit
32 Scaled
33 Dose Est._Prob. Expected Observed Size Residual
34 ~
35 0.0000 0.2658 13.291 5.000 50 -2.654
36 66.0000 0.4357 21.783 35.000 50 3.770
37 278.0000 0.7576 37.880 41.000 50 1.030
38 964.0000 0.9843 49.213 46.000 50 -3.651
39
40 Chi^2 = 35.65 d.f. = 2 P-value = 0.0000
41
42
43 Benchmark Dose Computation
44 Specified effect = 0.1
45
46 Risk Type = Extra risk
47 Confidence level =0.95
48 BMD = 26.4309
49 BMDL = 19.5045
50 BMDU = 37.5583
51
52 Taken together, (19.5045, 37.5583) is a 90% two-sided confidence interval for the BMD
53
54 Multistage Cancer Slope Factor = 0.00512702
D.6 Male BDF1 Mice: Hepatic Carcinomas and Adenomas
55 Data for hepatic carcinomas and adenomas in male BDF1 mice (Kano et al., 2009) are shown in
56 Table D-14. Note that the incidence of carcinomas and the incidence of either adenomas or carcinomas
57 are monotone non-decreasing functions of dose. These data therefore appear to be appropriate for
58 dose-response modeling using BMDS. However, the incidence of adenomas clearly reaches a peak value
D-37
DRAFT - DO NOT CITE OR QUOTE
-------
1 at 191 mg/kg-day and then decreases sharply with increasing dose. This cannot be modeled by a
2 multistage model using only non-negative coefficients. To some extent the incidence of "either adenomas
3 or carcinomas or both" retains some of the inverted-U shaped dose-response of the adenomas, which
4 dominate based on their high incidence at the lowest dose groups (49 and 191 mg/kg-day), thus is not
5 well characterized by any multistage model.
Table D-14 Data for hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.. 2009)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
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: Used with permission from Elsevier, Ltd., Kano et al. (2009).
6 The results of the BMDS modeling for the entire suite of models for hepatic adenomas and
7 carcinomas in male BDF1 mice are presented in Table D-15.
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 Logistic"
LogProbitc
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
NC°
BMDio
mg/kg-da
y
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-da
y
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'b
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.
"Best-fitting model.
°Slope restricted > 1.
Value unable to be calculated (NC: not calculated) by BMDS.
Values much larger than 1 are undesirable.
DRAFT - DO NOT CITE OR QUOTE
D-38
-------
"0
I
d
O
t3
cc
0.9
0.8
0.7
0.6
0.5
0.4
0.3
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
3MDL BMD
100
200
300
400
500
600
700
dose
07:30 10/262009
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-14 LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
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
intercept = -5.58909
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
D-39
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
3 been specified by the user, and do not appear in the correlation matrix )
4
5 background intercept
6 background 1 -0.69
7 intercept -0.69 1
8
9
10 Parameter Estimates
11
12 95.0% Wald Confidence Interval
13 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
14 background 0.507468 * * *
15 intercept -5.74623 * * *
16 slope 1 * * *
17
18 * - Indicates that this value is not calculated.
19
20
21 Analysis of Deviance Table
22
23 Model Log(likelihood) # Param's Deviance Test d.f. P-value
24 Full model -121.373 4
25 Fitted model -122.419 2 2.09225 2 0.3513
26 Reduced model -128.859 1 14.9718 3 0.001841
27
28 AIC: 248.839
29
30
31 Goodness of Fit
32 Scaled
33 Dose Est._Prob. Expected Observed Size Residual
34 ~
35 0.0000 0.5075 25.373 23.000 50 -0.671
36 49.0000 0.5741 28.707 31.000 50 0.656
37 191.0000 0.6941 34.706 37.000 50 0.704
38 677.0000 0.8443 42.214 40.000 50 -0.863
39
40 Chi^2 = 2.12 d.f. = 2 P-value = 0.3461
41
42
43 Benchmark Dose Computation
44 Specified effect = 0.1
45 Risk Type = Extra risk
46 Confidence level =0.95
47 BMD = 34.7787
48 BMDL = 16.5976
D-40
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with O.9b Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
o
I
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
dose
07:30 10/26 2009
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
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
Beta(l) = 0.00123152
Asymptotic Correlation Matrix of Parameter Estimates
D-41
DRAFT - DO NOT CITE OR QUOTE
-------
1 Background Beta(l)
2 Background 1 -0.58
3 Beta(l) -0.58 1
4
5
6 Parameter Estimates
7
8 95.0% Wald Confidence Interval
9 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
10 Background 0.545889 * * *
11 Beta(l) 0.00148414 * * *
12
13 * - Indicates that this value is not calculated.
14
15
16
17 Analysis of Deviance Table
18
19 Model Log(likelihood) # Param's Deviance Test d.f. P-value
20 Full model -121.373 4
21 Fitted model -123.275 2 3.80413 2 0.1493
22 Reduced model -128.859 1 14.9718 3 0.001841
23
24 AIC: 250.551
25
26
27 Goodness of Fit
28 Scaled
29 Dose Est._Prob. Expected Observed Size Residual
30 ~
31 0.0000 0.5459 27.294 23.000 50 -1.220
32 49.0000 0.5777 28.887 31.000 50 0.605
33 191.0000 0.6580 32.899 37.000 50 1.223
34 677.0000 0.8337 41.687 40.000 50 -0.641
35
36 Chi^2 = 3.76 d.f. = 2 P-value = 0.1527
37
38
39 Benchmark Dose Computation
40
41 Specified effect =0.1
42 Risk Type = Extra risk
43 Confidence level =0.95
44 BMD = 70.9911
45 BMDL = 44.0047
46 BMDU = 150.117
47
48 Taken together, (44.0047, 150.117) is a 90% two-sided confidence interval for the BMD
49
50 Multistage Cancer Slope Factor = 0.00227248
D.7 BMD Modeling Results from Additional Chronic Bioassays
51 Data and BMDS modeling results for the additional chronic bioassays (NCI. 1978; Kociba et al.,
52 1974) were evaluated for comparison with the Kano et al. (2009) study. These results are presented in the
53 following sections.
54 The BMDS dose-response modeling estimates and HEDs that resulted are presented in detail in
55 the following sections and a summary is provided in Table D-16.
D-42
DRAFT - DO NOT CITE OR QUOTE
-------
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
Lowest AIC
Probit
Multistage
(3 degree)
84.3126
26.4156
0.606
0.9999
BMDio
mg/kg-da
y
Rats
1113
1717
.94
.16
BMDLio
mg/kg-day
920.62
1306.29
BMDlOHED
mg/kg-day
290.
448.
,78
24
BMDLio HED
mg/kg-day
240.
340.
,31
,99
Female Osborne-Mendel Rats
Lowest AIC
Lowest AIC
LogLogistic
LogLogistic
84.2821
84.2235
0.7333
0.2486
111.
155.
46
32
72.41
100.08
28.
40.
,75
,07
18.
25.
68
82
Male Osborne-Mendel Rats
Lowest AIC
Female B6C3Fi
Lowest AIC,
Multistage
model
LogLogistic
Mice
Multistage
(2 degree)
92.7669
85.3511
0.7809
1
56.
160.
26
68
37.26
67.76
16.
23.
,10
,12
10.
9.
66
,75
Male B6C3Fi Mice
Lowest AIC
Gamma
177.539
0.7571
601.
69
243.92
87.
98
35.
,67
"Incidence of hepatocellular carcinoma.
blncidence of nasal squamous cell carcinoma.
Incidence of hepatocellular adenoma.
dlncidence of hepatocellular adenoma or carcinoma.
D.7.1 Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma
(Kociba etal.. 1974)
1 The incidence data for hepatocellular carcinoma and nasal squamous cell carcinoma are presented
2 in Table D-17. The predicted BMD10HED and BMDL10HED values are also presented in Table D-18 and
3 Table D-19 for hepatocellular carcinomas and nasal squamous cell carcinomas, respectively.
D-43
DRAFT - DO NOT CITE OR QUOTE
-------
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) Incidence of hepatocellular Incidence of nasal
. * i j * i j i .a squamous cell
(average of male and female dose) carcinoma . a
v a ' carcinoma
0
14
121
1,307
1/1 06b
0/110
1/106
10/66d
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: Used with permission from Elsevier, Ltd., 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
LogLogistic
LogProbit"
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
NCa
BMDio
mg/kg-day
985.13
1148.65
985.62
1036.97
940.12
1041.72
1041.72
1113.94
998.33
940.12
NCa
BMDLio
mg/kg-day
628.48
980.95
611.14
760.29
583.58
628.56
628.56
920.62
629.93
583.58
NCa
x23
-0.005
-0.004
-0.005
-0.011
0.279
-0.006
-0.006
-0.005
-0.005
0.279
0
BMDio HED
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
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
""Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
D-44
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
I
c
o
t3
ns
0.25
0.2
0.15
0.1
0.05
Probit
BMDL
BMD
200
400
600
dose
800
1000
1200
11:54 10/27 2009
Source: Used with permission from Elsevier, Ltd., Kociba et al. (1974).
Figure D-16 Probit BMD model for the incidence of hepatocellular carcinoma in
male and female Sherman rats exposed to 1,4-dioxane in drinking
water.
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.plt
Tue Oct 27 12:54:14 2009
BMDS Model Run
is the cumulative normal
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
(*** 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-45
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 intercept slope
3 intercept 1 -0.82
4 slope -0.82 1
5
6
7 Parameter Estimates
8
9 95.0% Wald Confidence Interval
10 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
11 intercept -2.55961 0.261184 -3.07152 -2.0477
12 slope 0.00117105 0.000249508 0.000682022 0.00166008
13
14
15 Analysis of Deviance Table
16
17 Model Log(likelihood) # Param's Deviance Test d.f. P-value
18 Full model -39.3891 4
19 Fitted model -40.1563 2 1.53445 2 0.4643
20 Reduced model -53.5257 1 28.2732 3 <.0001
21
22 AIC: 84.3126
23
24
25 Goodness of Fit
26 Scaled
27 Dose Est._Prob. Expected Observed Size Residual
28
29 0.0000 0.0052 0.555 1.000 106 0.598
30 14.0000 0.0055 0.604 0.000 110 -0.779
31 121.0000 0.0078 0.827 1.000 106 0.191
32 1307.0000 0.1517 10.014 10.000 66 -0.005
33
34 ChiA2 = 1.00 d.f. = 2 P-value = 0.6060
35
36
37 Benchmark Dose Computation
38
39 Specified effect =0.1
40 Risk Type = Extra risk
41 Confidence level = 0.95
42 BMD = 1,113.94
43 BMDL = 920.616
D-46
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
1
2
o
6
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
T3
•§
<
CZ
CD
CO
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMD
0
200
400
600
dose
800
1000
1200
11:54 10/272009
Source: Used with permission from Elsevier, Ltd., Kociba et al. (1974).
Figure D-17 Multistage BMD model (1 degree) for the incidence of hepatocellular
carcinoma in male and female Sherman rats exposed to 1,4-dioxane in
drinking water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_hepato_car_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_hepato_car_Msc-BMR10-lpoly.plt
Tue Oct 27 12:54:10 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]
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 = 0.000925988
D-47
DRAFT - DO NOT CITE OR QUOTE
-------
1 Beta(l) = 0.000124518
2
3
4 Asymptotic Correlation Matrix of Parameter Estimates
5 Background Beta(l)
6 Background 1 -0.44
7 Beta(l) -0.44 1
8
9
10 Parameter Estimates
11
12 95.0% Wald Confidence Interval
13 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
14 Background 0.0038683 * * *
15 Beta(l) 0.000112071 * * *
16
17 * - Indicates that this value is not calculated.
18
19
20 Analysis of Deviance Table
21
22 Model Log(likelihood) # Param's Deviance Test d.f. P-value
23 Full model -39.3891 4
24 Fitted model -40.5594 2 2.34056 2 0.3103
25 Reduced model -53.5257 1 28.2732 3 <.0001
26
27 AIC: 85.1187
28
29
30 Goodness of Fit
31 Scaled
32 Dose Est._Prob. Expected Observed Size Residual
33
34 0.0000 0.0039 0.410 1.000 106 0.923
35 14.0000 0.0054 0.597 0.000 110 -0.775
36 121.0000 0.0173 1.832 1.000 106 -0.620
37 1307.0000 0.1396 9.213 10.000 66 0.279
38
39 Chi^2 = 1.92 d.f. = 2 P-value = 0.3838
40
41
42 Benchmark Dose Computation
43
44 Specified effect = 0.1
45 Risk Type = Extra risk
46 Confidence level =0.95
47 BMD = 940.124
48 BMDL = 583.576
49 BMDU = 1,685.88
50
51 Taken together, (583.576, 1685.88) is a 90% two-sided confidence interval for the BMD
52
53 Multistage Cancer Slope Factor = 0.000171357
D-48
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-19 BMDS dose-response modeling results for the incidence of nasal squamous cell
carcinoma in male and female Sherman rats (combined) (Kocibaet al.. 1974) exposed
to 1,4-dioxane in the drinking water for 2 years
Model
Gamma
Logistic
LogLogistic
LogProbit"
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"'
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 x^
"Slope restricted > 1.
""Best-fitting model.
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
Multistage Cancer Model with 0.95 Confidence Level
f
o
OS
0.14
0.12
0.1
0.08
0.06
0.04
0.02
Multistage Cancer
Linear extrapolation
BMID
06:25 10/272009
0 200 400 600 800 1000 1200 1400 1600 1800
dose
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.
D-49
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 ====================================================================
3 Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
4 Input Data File:
5 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_nasal_car_Msc-BMR10-3poly.(d)
6 Gnuplot Plotting File:
7 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_nasal_car_Msc-BMR10-3poly.plt
8 Tue Oct 27 07:25:02 2009
9 ====================================================================
10 BMDS Model Run
11 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
12
13 The form of the probability function is:
14
15 P[response] = background +
16 (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
17
18 The parameter betas are restricted to be positive
19
20 Dependent variable = Effect
21 Independent variable = Dose
22
23 Total number of observations = 4
24 Total number of records with missing values = 0
25 Total number of parameters in model = 4
26 Total number of specified parameters = 0
27 Degree of polynomial = 3
28
29 Maximum number of iterations =250
30 Relative Function Convergence has been set to: le-008
31 Parameter Convergence has been set to: le-008
32 Default Initial Parameter Values
33 Background = 0
34 Beta(l) = 0
35 Beta(2) = 0
36 Beta(3) = 2.08414e-011
37
38
39 Asymptotic Correlation Matrix of Parameter Estimates
40
41 (*** The model parameter(s) -Background -Beta(l) -Beta(2)
42 have been estimated at a boundary point, or have been specified by the user,
43 and do not appear in the correlation matrix )
44
45 Beta(3)
46 Beta (3) 1
47
48
49 Parameter Estimates
50
51 95.0% Wald Confidence Interval
52 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
53 Background 0 * * *
54 Beta(l) 0 * * *
55 Beta(2) 0 * * *
56 Beta(3) 2.08088e-011 * * *
57
58 * - Indicates that this value is not calculated.
59
60
61
62 Analysis of Deviance Table
63
64 Model Log(likelihood) # Param's Deviance Test d.f. P-value
65 Full model -12.2039 4
66 Fitted model -12.2078 1 0.00783284 3 0.9998
67 Reduced model -17.5756 1 10.7433 3 0.0132
D-50
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 AIC: 26.4156
3
4
5 Goodness of Fit
6 Scaled
7 Dose Est._Prob. Expected Observed Size Residual
9 0.0000 0.0000 0.000 0.000 106 0.000
10 14.0000 0.0000 0.000 0.000 110 -0.003
11 121.0000 0.0000 0.004 0.000 106 -0.063
12 1307.0000 0.0454 2.996 3.000 66 0.002
13
14 ChiA2 = 0.00 d.f. = 3 P-value = 0.9999
15
16
17 Benchmark Dose Computation
18
19 Specified effect =0.1
20 Risk Type = Extra risk
21 Confidence level =0.95
22 BMD = 1,717.16
23 BMDL = 1,306.29
24 BMDU = 8,354.46
25
26 Taken together, (1306.29, 8354.46) is a 90% two-sided confidence interval for the BMD
27
28 Multistage Cancer Slope Factor = 7.65529e-005
D.7.2 Nasal Cavity Squamous Cell Carcinoma and Liver Hepatocellular
Adenoma in Osborne-Mendel Rats (NCI. 1978)
29 The incidence data for hepatocellular adenoma (female rats) and nasal squamous cell carcinoma
30 (male and female rats) are presented in Table D-20. The log-logistic model adequately fit both the male
31 and female rat nasal squamous cell carcinoma data, as well as female hepatocellular adenoma incidence
32 data. For all endpoints and genders evaluated in this section, compared to the multistage models, the
33 log-logistic model had a higher p-value, as well as both a lower AIC and lower BMDL. The results of the
34 BMDS modeling for the entire suite of models are presented in Table D-21 through Table D-23.
D-51
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-20 Incidence of nasal cavity squamous cell carcinoma and hepatocellular adenoma in
Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in the drinking water
Male rat Animal Dose (mg/kg-day)a
0
Nasal cavity squamous cell carcinoma 0/33C
240°
12/26°
530
16/33°
Female rat Animal Dose (mg/kg-day)a
0
Nasal cavity squamous cell carcinoma 0/34C
Hepatocellular adenoma 0/31C
350
10/30"
10/30°
640
8/29°
11/29°
"Tumor incidence values were adjusted for mortality (NCI. 1978).
bGroup not included in statistical analysis by NCI (1978) because the dose group was started a year earlier without
appropriate controls.
°p S 0.001; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: NCI (1978).
D-52
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-21 BMDS dose-response modeling results for the incidence of hepatocellular adenoma
in female Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in the drinking
water for 2 years
Model
Gamma
Logistic
Log Logistic"
LogProbit
AIC
84.6972
92.477
84.2821
85.957
p-value
0.5908
0.02
0.7333
0.3076
BMDio
mg/kg-day
132.36
284.09
111.46
209.47
BMDLio
mg/kg-day
94.06
220.46
72.41
160.66
x2a
0
1.727
0
1.133
BMDlOHED
mg/kg-day
34.144
73.29
28.75
54.04
BMDLio HED
mg/kg-day
24.26
56.87
18.68
41.45
Multistage-Cancer
(1 degree)
84.6972
0.5908
132.36
94.06
34.14
24.26
Multistage-Cancer
(2 degree)
84.6972
0.5908
132.36
94.06
34.14
24.26
Probit
91.7318
0.0251
267.02
207.18
1.7
68.88
53.44
Weibull
84.6972
0.5908
132.36
94.06
34.14
24.26
Quantal-Linear
84.6972
0.5908
132.36
94.06
0
34.14
24.26
"Maximum absolute x2
bBest-fitting model.
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
Log-Logistic Model with O.9b Confidence Level
T3
-------
1 ===============
2 BMDS Model Run
4 The form of the probability function is:
5 P [response] = background+ (1-background) / [1+EXP (-intercept-slope*Log (dose) ) ]
6
7 Dependent variable = Effect
8 Independent variable = Dose
9 Slope parameter is restricted as slope >= 1
10
11 Total number of observations = 3
12 Total number of records with missing values = 0
13 Maximum number of iterations =250
14 Relative Function Convergence has been set to: le-008
15 Parameter Convergence has been set to: le-008
16
17 User has chosen the log transformed model
18
19 Default Initial Parameter Values
20 background = 0
21 intercept = -6.62889
22 slope = 1
23
24 Asymptotic Correlation Matrix of Parameter Estimates
25
26 (*** The model parameter (s) -background -slope have been estimated at a boundary
27 point, or have been specified by the user, and do not appear in the correlation
28 matrix)
29
30 intercept
31 intercept 1
32
33 Parameter Estimates
34
35 95.0% Wald Confidence Interval
36 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
37 background 0 * * *
38 intercept -6.91086 * * *
39 slope 1 * * *
40
41 * - Indicates that this value is not calculated.
42
43
44 Analysis of Deviance Table
45
46 Model Log (likelihood) # Param' s Deviance Test d.f. P-value
47 Full model -40.8343 3
48 Fitted model -41.141 1 0.613564 2 0.7358
49 Reduced model -50.4308 1 19.1932 2 <.0001
50
51 AIC: 84.2821
52
53
54 Goodness of Fit
55 Scaled
56 Dose Est._Prob. Expected Observed Size Residual
57 ------------------------------------------------------------------------
58 0.0000 0.0000 0.000 0.000 31 0.000
59 350.0000 0.2587 8.536 10.000 33 0.582
60 640.0000 0.3895 12.464 11.000 32 -0.531
61
62 ChiA2 = 0.62 d.f. = 2 P-value = 0.7333
63
64
65 Benchmark Dose Computation
66
67 Specified effect =0.1
D-54
DRAFT - DO NOT CITE OR QUOTE
-------
1 Risk Type = Extra risk
2 Confidence level = 0.95
3 BMD = 111.457
4 BMDL = 72.4092
Multistage Cancer Model with 0.95 Confidence Level
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
100
200
06:32 10/27 2009
Source: NCI (1978).
300
dose
400
500
600
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
DRAFT - DO NOT CITE OR QUOTE
D-55
-------
1 Degree of polynomial = 1
2
3 Maximum number of iterations =250
4 Relative Function Convergence has been set to: le-008
5 Parameter Convergence has been set to: le-008
6
7
8 Default Initial Parameter Values
9 Background = 0.0385912
10 Beta(l) = 0.000670869
11 Asymptotic Correlation Matrix of Parameter Estimates
12
13 (*** The model parameter(s) -Background have been estimated at a boundary point, or
14 have been specified by the user, and do not appear in the correlation matrix)
15
16 Beta(l)
17 Beta(l) 1
18
19
20
21 Parameter Estimates
22
23 95.0% Wald Confidence Interval
24 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
25 Background 0 * * *
26 Beta(l) 0.00079602 * * *
27
28 * - Indicates that this value is not calculated.
29
30
31
32 Analysis of Deviance Table
33
34 Model Log(likelihood) # Param's Deviance Test d.f. P-value
35 Full model -40.8343 3
36 Fitted model -41.3486 1 1.02868 2 0.5979
37 Reduced model -50.4308 1 19.1932 2 <.0001
38
39 AIC: 84.6972
40
41
42 Goodness of Fit
43 Scaled
44 Dose Est._Prob. Expected Observed Size Residual
45
46 0.0000 0.0000 0.000 0.000 31 0.000
47 350.0000 0.2432 8.024 10.000 33 0.802
48 640.0000 0.3992 12.774 11.000 32 -0.640
49
50 ChiA2 = 1.05 d.f. = 2 P-value = 0.5908
51
52
53 Benchmark Dose Computation
54
55 Specified effect =0.1
56 Risk Type = Extra risk
57 Confidence level = 0.95
58 BMD = 132.359
59 BMDL = 94.0591
60 BMDU = 194.33
61
62 Taken together, (94.0591, 194.33 ) is a 90% two-sided confidence interval for the BMD
63
64 Multistage Cancer Slope Factor = 0.00106316
D-56
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-22 BMDS dose-response modeling results for the incidence of nasal cavity squamous
cell carcinoma in female Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in
the drinking water for 2 years
Model
Gamma
Logistic
Log Logistic"
LogProbitc
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
BMDio HED
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
"Maximum absolute x2
bBest-fitting model.
°Slope restricted > 1 .
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
Log-Logistic Model with 0.95 Confidence Level
1
2
3
4
0.5
0.4
0.3
0.2
0.1
Log-Logistic
BMDL BMD
06:30 10/27 2009
Source: NCI (1978).
100 200 300 400
dose
500
600
Figure D-21 LogLogistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA HPAG\14Dioxane\BMDS\lnl nci frat nasal car Lnl-BMRlO-Restrict.(d)
D-57
DRAFT - DO NOT CITE OR QUOTE
-------
1 Gnuplot Plotting File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_nasal_car_Lnl-BMR10-Restrict.plt
3 Tue Oct 27 07:30:09 2009
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7
8 The form of the probability function is:
9
10 P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
11
12
13 Dependent variable = Effect
14 Independent variable = Dose
15 Slope parameter is restricted as slope >= 1
16
17 Total number of observations = 3
18 Total number of records with missing values = 0
19 Maximum number of iterations =250
20 Relative Function Convergence has been set to: le-008
21 Parameter Convergence has been set to: le-008
22
23
24 User has chosen the log transformed model
25
26
27 Default Initial Parameter Values
28 background = 0
29 intercept = -6.64005
30 slope = 1
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34 (*** The model parameter(s) -background -slope have been estimated at a boundary
35 point, or have been specified by the user, and do not appear in the correlation
36 matrix)
37
38 intercept
39 intercept 1
40
41
42 Parameter Estimates
43
44 95.0% Wald Confidence Interval
45 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
46 background 0 *
47 intercept -7.24274
48 slope 1 * * *
49
50 * - Indicates that this value is not calculated.
51
52 Analysis of Deviance Table
53
54 Model Log(likelihood) # Param's Deviance Test d.f. P-value
55 Full model -39.7535 3
56 Fitted model -41.1117 1 2.71651 2 0.2571
57 Reduced model -47.9161 1 16.3252 2 0.0002851
58
59 AIC: 84.2235
60
61 Goodness of Fit
62 Scaled
63 Dose Est._Prob. Expected Observed Size Residual
64 ~
65 0.0000 0.0000 0.000 0.000 34 0.000
66 350.0000 0.2002 7.008 10.000 35 1.264
67 640.0000 0.3140 10.992 8.000 35 -1.090
D-58
DRAFT - DO NOT CITE OR QUOTE
* *
* * *
-------
1
2
3
4
5
6
7
8
9
10
11
= 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
Multistage Cancer Model with 0.95 Confidence Level
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
"8
"o
I
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
600
06:30 10/27 2009
Source: NCI (1978).
Figure D-22 Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in female Osborne-Mendel rats exposed to
1,4-dioxane in drinking water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_nasal_car_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_nasal_car_Msc-BMR10-lpoly.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
D-59
DRAFT - DO NOT CITE OR QUOTE
-------
1 Total number of records with missing values = 0
2 Total number of parameters in model = 2
3 Total number of specified parameters = 0
4 Degree of polynomial = 1
5
6 Maximum number of iterations =250
7 Relative Function Convergence has been set to: le-008
8 Parameter Convergence has been set to: le-008
9
10 Default Initial Parameter Values
11 Background = 0.0569154
12 Beta(l) = 0.00042443
13
14 Asymptotic Correlation Matrix of Parameter Estimates
15 (*** The model parameter(s) -Background have been estimated at a boundary point, or
16 have been specified by the user, and do not appear in the correlation matrix)
17
18 Beta(l)
19 Beta(l) 1
20
21 Parameter Estimates
22
23 95.0% Wald Confidence Interval
24 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
25 Background 0 * * *
26 Beta(l) 0.000597685 * * *
27
28 * - Indicates that this value is not calculated.
29
30 Analysis of Deviance Table
31
32 Model Log(likelihood) # Param's Deviance Test d.f. P-value
33 Full model -39.7535 3
34 Fitted model -41.3998 1 3.29259 2 0.1928
35 Reduced model -47.9161 1 16.3252 2 0.0002851
36
37 AIC: 84.7996
38
39 Goodness of Fit
40 Scaled
41 Dose Est._Prob. Expected Observed Size Residual
42
43 0.0000 0.0000 0.000 0.000 34 0.000
44 350.0000 0.1888 6.607 10.000 35 1.466
45 640.0000 0.3179 11.125 8.000 35 -1.134
46
47 ChiA2 = 3.44 d.f. = 2 P-value = 0.1795
48
49 Benchmark Dose Computation
50 Specified effect =0.1
51 Risk Type = Extra risk
52 Confidence level = 0.95
53 BMD = 176.281
54 BMDL = 122.274
55 BMDU = 271.474
56
57 Taken together, (122.274, 271.474) is a 90% two-sided confidence interval for the BMD
58
59 Multistage Cancer Slope Factor = 0.000817837
D-60
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-23 BMDS dose-response modeling results for the incidence of nasal cavity squamous
cell carcinoma in male Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in the
drinking water for 2 years
Model
Gamma
Logistic
Log Logistic"
LogProbitc
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
"Maximum absolute x2
bBest-fitting model.
°Slope restricted > 1 .
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
Log-Logistic Model with 0.95 Confidence Level
1
2
o
J
4
5
6
o
I
d
o
CO
0.7
0.6
0.5
0.4
0.3
0.2
0.1
100
200
300
400
500
dose
06:27 10/27 2009
Source: NCI (1978).
Figure D-23 LogLogistic BMD model for the incidence of nasal cavity squamous
cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_mrat_nasal_car_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_mrat_nasal_car_Lnl-BMR10-Restrict.plt
DRAFT - DO NOT CITE OR QUOTE
D-61
-------
1 Tue Oct 27 07:27:57 2009
2 ====================================================================
3 BMDS Model Run
4 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5
6 The form of the probability function is:
7 P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
8
9 Dependent variable = Effect
10 Independent variable = Dose
11 Slope parameter is restricted as slope >= 1
12
13 Total number of observations = 3
14 Total number of records with missing values = 0
15 Maximum number of iterations =250
16 Relative Function Convergence has been set to: le-008
17 Parameter Convergence has been set to: le-008
18
19 User has chosen the log transformed model
20
21 Default Initial Parameter Values
22 background = 0
23 intercept = -6.08408
24 slope = 1
25
26 Asymptotic Correlation Matrix of Parameter Estimates
27 (*** The model parameter(s) -background -slope have been estimated at a boundary
28 point, or have been specified by the user, and do not appear in the correlation
29 matrix)
30
31 intercept
32 intercept 1
33
34 Parameter Estimates
35
36 95.0% Wald Confidence Interval
37 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
38 background 0 * * *
39 intercept -6.2272 * * *
40 slope 1 * * *
41
42 * - Indicates that this value is not calculated.
43
44 Analysis of Deviance Table
45
46 Model Log(likelihood) # Param's Deviance Test d.f. P-value
47 Full model -45.139 3
48 Fitted model -45.3835 1 0.488858 2 0.7832
49 Reduced model -59.2953 1 28.3126 2 <.0001
50
51 AIC: 92.7669
52
53 Goodness of Fit
54 Scaled
55 Dose Est._Prob. Expected Observed Size Residual
56 ~
57 0.0000 0.0000 0.000 0.000 33 0.000
58 240.0000 0.3216 10.612 12.000 33 0.517
59 530.0000 0.5114 17.388 16.000 34 -0.476
60
61 Chi^2 = 0.49 d.f. = 2 P-value = 0.7809
D-62
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 56.2596
BMDL = 37.256
Multistage Cancer Model with 0.95 Confidence Level
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
o
't5
CO
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
;
BMDL BMD
0
06:28 10/27 2009
Source: 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 ratsexposed 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
D-63
DRAFT - DO NOT CITE OR QUOTE
-------
1 Degree of polynomial = 1
2
3 Maximum number of iterations =250
4 Relative Function Convergence has been set to: le-008
5 Parameter Convergence has been set to: le-008
6 Default Initial Parameter Values
7 Background = 0.0578996
8 Beta(l) = 0.00118058
9
10 Asymptotic Correlation Matrix of Parameter Estimates
11 (*** The model parameter(s) -Background have been estimated at a boundary point, or
12 have been specified by the user, and do not appear in the correlation matrix)
13
14 Beta(l)
15 Beta(l) 1
16
17 Parameter Estimates
18
19 95.0% Wald Confidence Interval
20 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
21 Background 0 * * *
22 Beta(l) 0.00142499 * * *
23
24 * - Indicates that this value is not calculated.
25
26 Analysis of Deviance Table
27
28 Model Log(likelihood) # Param's Deviance Test d.f. P-value
29 Full model -45.139 3
30 Fitted model -45.8002 1 1.32238 2 0.5162
31 Reduced model -59.2953 1 28.3126 2 <.0001
32
33 AIC: 93.6005
34
35 Goodness of Fit
36 Scaled
37 Dose Est._Prob. Expected Observed Size Residual
38 ~
39 0.0000 0.0000 0.000 0.000 33 -0.000
40 240.0000 0.2896 9.558 12.000 33 0.937
41 530.0000 0.5301 18.024 16.000 34 -0.695
42
43 ChiA2 = 1.36 d.f. = 2 P-value = 0.5063
44
45 Benchmark Dose Computation
46 Specified effect =0.1
47 Risk Type = Extra risk
48 Confidence level =0.95
49 BMD = 73.9379
50 BMDL = 54.7238
51 BMDU = 103.07
52
53 Taken together, (54.7238, 103.07 ) is a 90% two-sided confidence interval for the BMD
54
55 Multistage Cancer Slope Factor = 0.00182736
D.7.3 Hepatocellular Adenoma or Carcinoma in B6C3Fi Mice (NCI. 1978)
56 The incidence data for hepatocellular adenoma or carcinoma in male and female mice are
57 presented in Table D-24. The 2-degree polynomial model (betas restricted > 0) was the lowest degree
58 polynomial that provided an adequate fit to the female mouse data (Figure D-25), while the gamma model
D-64
DRAFT - DO NOT CITE OR QUOTE
-------
1 provided the best fit to the male mouse data (Figure D-26). The results of the BMDS modeling for the
2 entire suite of models are presented in Table D-25 and Table D-26 for the female and male data,
3 respectively.
D-65
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-24 Incidence of hepatocellular adenoma or carcinoma in male and female B6C3F1 mice
(NCI. 1978) exposed to 1,4-dioxane in drinking water
Male mouse Animal Dose (mg/kg-day)a
0
8/49°
720
19/50°
830
28/47°
Female mouse Animal Dose (mg/kg-day)a
0
0/50°
380
21/48C
860
35/37c
aTumor 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 B6C3FT mice (NCI. 1978) exposed to 1,4-dioxane in
the drinking water for 2 years
Model
Gamma
Logistic
Log Logistic
LogProbit"
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
BMDio
mg/kg-day
195.69
199.63
228.08
225.8
49.10
160.68
188.24
161.77
49.10
BMDLio
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
BMDlOHED
mg/kg-day
28.16
28.72
32.82
32.49
7.06
23.12
27.08
23.28
7.065
BMDLio HED
mg/kg-day
15.19
21.78
21.75
21.71
5.58
9.75
20.36
12.84
5.58
"Maximum absolute x^
"Slope restricted > 1.
""Best-fitting model.
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
D-66
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
1
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
I
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
BMDL
BMD
100
200
300
400 500
dose
600
700
800
900
06:36 10/27 2009
Source: NCI (1978).
Figure D-25 Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in female B6C3F! 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
Default Initial Parameter Values
Background = 0
Beta(l) = 2.68591e-005
D-67
DRAFT - DO NOT CITE OR QUOTE
-------
1 Beta(2) = 3.91383e-006
2
3
4 Asymptotic Correlation Matrix of Parameter Estimates
5 (*** The model parameter(s) -Background have been estimated at a boundary point, or
6 have been specified by the user, and do not appear in the correlation matrix)
7
8 Beta(l) Beta(2)
9 Beta(l) 1 -0.92
10 Beta(2) -0.92 1
11
12
13 Parameter Estimates
14
15 95.0% Wald Confidence Interval
16 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17 Background 0 * * *
18 Beta(l) 2.686e-005 * * *
19 Beta(2) 3.91382e-006 * * *
20
21 * - Indicates that this value is not calculated.
22
23
24 Analysis of Deviance Table
25
26 Model Log(likelihood) # Param's Deviance Test d.f. P-value
27 Full model -40.6756 3
28 Fitted model -40.6756 2 3.20014e-010 1 1
29 Reduced model -91.606 1 101.861 2 <.0001
30
31 AIC: 85.3511
32
33 Goodness of Fit
34 Scaled
35 Dose Est._Prob. Expected Observed Size Residual
36
37 0.0000 0.0000 0.000 0.000 50 0.000
38 380.0000 0.4375 21.000 21.000 48 0.000
39 860.0000 0.9459 35.000 35.000 37 0.000
40
41 Chi^2 = 0.00 d.f. = 1 P-value = 1.0000
42
43
44 Benchmark Dose Computation
45 Specified effect =0.1
46 Risk Type = Extra risk
47 Confidence level =0.95
48 BMD = 160.678
49 BMDL = 67.7635
50 BMDU = 186.587
51
52 Taken together, (67.7635, 186.587) is a 90% two-sided confidence interval for the BMD
53
54 Multistage Cancer Slope Factor = 0.00147572
D-68
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-26 BMDS dose-response modeling results for the combined incidence of hepatocellular
adenoma or carcinoma in male B6C3F1 mice (NCI. 1978) exposed to 1,4-dioxane in
drinking water
Model
Gammab
Logistic
LogLogistic
LogProbit"
AIC
177.539
179.9
179.443
179.443
p-value
0.7571
0.1189
NCC
NCC
BMDio
mg/kg-day
601.69
252.66
622.39
631.51
BMDLio
mg/kg-day
243.92
207.15
283.04
305.44
x2a
-0.233
0.214
0
0
BMDlOHED
mg/kg-day
87.98
36.94
91.01
92.34
BMDLio HED
mg/kg-day
35.67
30.29
41.39
44.66
Multistage-Cancer
(1 degree)
180.618
0.0762
164.29
117.37
0.079
24.02
17.16
Multistage-Cancer
(2 degree)
179.483
0.1554
354.41
126.24
0.124
51.82
18.46
Probit
179.984
0.1128
239.93
196.90
0.191
35.08
28.79
Weibull
179.443
NCC
608.81
249.71
0
89.02
36.51
Quantal-Linear
180.618
0.0762
164.29
117.37
0.079
24.02
17.16
"Maximum 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.
Gamma Multi-Hit Model with 0.95 Confidence Level
o
'•o
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Gamma Multi-Hit
BMDL
BMD
06:34 10/27 2009
Source: NCI (1978).
100 200 300 400 500 600 700
dose
800
Figure D-26 Gamma BMD model for the incidence of hepatocellular adenoma or
carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking
water.
1
2
3
4
Gamma Model. (Version: 2.13; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-BMR10-Restrict.(d)
DRAFT - DO NOT CITE OR QUOTE
D-69
-------
1 Gnuplot Plotting File:
2 L:\Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-BMR10-Restrict.plt
3 Tue Oct 27 07:34:35 2009
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7
8 The form of the probability function is:
9 P[response]= background+(1-background)*CumGamma[slope*dose,power],
10 where CumGamma(.) is the cummulative Gamma distribution function
11
12 Dependent variable = Effect
13 Independent variable = Dose
14 Power parameter is restricted as power >=1
15
16 Total number of observations = 3
17 Total number of records with missing values = 0
18 Maximum number of iterations =250
19 Relative Function Convergence has been set to: le-008
20 Parameter Convergence has been set to: le-008
21
22 Default Initial (and Specified) Parameter Values
23 Background =0.17
24 Slope = 0.000671886
25 Power =1.3
26
27 Asymptotic Correlation Matrix of Parameter Estimates
28 (*** The model parameter(s) -Power have been estimated at a boundary point, or have
29 been specified by the user, and do not appear in the correlation matrix)
30
31 Background Slope
32 Background 1 -0.52
33 Slope -0.52 1
34
35 Parameter Estimates
36 95.0% Wald Confidence Interval
37 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
38 Background 0.160326 0.0510618 0.060247 0.260405
39 Slope 0.0213093 0.000971596 0.019405 0.0232136
40 Power 18 NA
41
42 NA - Indicates that this parameter has hit a bound implied by some ineguality
43 constraint and thus has no standard error.
44
45 Analysis of Deviance Table
46
47 Model Log(likelihood) # Param's Deviance Test d.f. P-value
48 Full model -86.7213 3
49 Fitted model -86.7693 2 0.096042 1 0.7566
50 Reduced model -96.715 1 19.9875 2 <.0001
51
52 AIC: 177.539
53
54 Goodness of Fit
55 Scaled
56 Dose Est._Prob. Expected Observed Size Residual
57
58 0.0000 0.1603 7.856 8.000 49 0.056
59 720.0000 0.3961 19.806 19.000 50 -0.233
60 830.0000 0.5817 27.339 28.000 47 0.196
61
62 ChiA2 = 0.10 d.f. = 1 P-value = 0.7571
63 Benchmark Dose Computation
64 Specified effect = 0.1
65 Risk Type = Extra risk
66 Confidence level = 0.95
67 BMD = 601.692
D-70
DRAFT - DO NOT CITE OR QUOTE
-------
BMDL = 243.917
Multistage Cancer Model with 0.95 Confidence Level
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
T3
•§
C=
O
CO
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
0
06:34 10/272009
Source: NCI (1978).
100
200
300
400
dose
500
600
700
800
Figure D-27 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.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_iranouse_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
D-71
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
Background = 0.131156
Beta(l) = 0
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
= 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-72
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX E. COMPARISON OF SEVERAL DATA
REPORTS FOR THE JBRC 2-YEAR 1,4-DIOXANE
DR INKING WATE R STUDY
1 As described in detail in Section 4.2.1.2.6 of this Toxicological Review ofl, 4-Dioxane, the JBRC
2 conducted a 2-year drinking water study on the effects of 1,4-dioxane in both sexes of rats and mice. The
3 results from this study have been reported three times, once as conference proceedings (Yamazaki et al..
4 1994). once as a detailed laboratory report (JBRC. 1998). and once as a published manuscript (Kano et
5 al.. 2009). After the External Peer Review draft of the Toxicological Review of 1,4-Dioxane (U.S. EPA.
6 2009b) had been released, the Kano et al. (2009) manuscript was published; thus, minor changes to the
7 Toxicological Review of 1,4-Dioxane occurred.
8 The purpose of this appendix is to provide a clear and transparent comparison of the reporting of
9 this 2-year 1,4-dioxane drinking water study. The variations included: (1) the level of detail on dose
10 information reported; (2) categories for incidence data reported (e.g., all animals or sacrificed animals);
11 and (3) analysis of non- and neoplastic lesions. Even though the data contained in the reports varied, the
12 differences were minor and did not did not significantly affect the qualitative or quantitative cancer
13 assessment.
14 Tables contained within this appendix provide a comparison of the variations in the reported data
15 (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). Table E-l and Table E-2 show the histological
16 nonneoplastic findings provided for male and female F344 rats, respectively. Table E-3 and Table E-4
17 show the histological nonneoplastic findings provided for male and female F344 rats, respectively.
18 Table E-3 and Table E-4 show the histological neoplastic findings provided for male and female F344
19 rats, respectively. Table E-5 and Table E-6 show the histological nonneoplastic findings provided for
20 male and female F344 rats, respectively. Table E-7 and Table E-8 show the histological neoplastic
21 findings provided for male and female Crj:BDFl mice, respectively.
E-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-1 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in male 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
Lamina propria; hydropic
change
Lamina propria; sclerosis
Nasal cavity; adhesion
Nasal cavity; inflammation
Hyperplasia; liver
Spongiosis hepatis; liver
Clear cell foci; liver
Acidophilic cell foci; liver
Basophilic cell foci; liver
Mixed-cell foci; liver
Nuclear enlargement;
kidney proximal tubule
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
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
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a
JBRC 11998}"
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000 0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-da}
Not reported
Not reported
Not reported
0/50 0/50 0/50 31/50
Not reported
0/50 0/50 0/50 2/50
Not reported
0/50 0/50 0/50 5/50
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
Not reported
3/50 2/10 10/50 24/50
Not reported
12/50 20/50 25/50 40/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
r«ntmi 8- 41' 209-
Co£V 24 121 586
W [16] (81) (398)
0/50 0/50 0/50 26/50
0/40 0/45 0/35 12/226
0/50 0/50 0/50 31/50
0/40 0/45 0/35 15/226
0/50 0/50 0/50 2/50
0/40 0/45 0/35 1/22
Not reported
Not reported
0/50 0/50 5/50 38/50
0/40 0/45 4/35 20/226
12/50 11/50 20/50 43/50
10/40 11/45 17/35 22/22e
0/50 0/50 0/50 36/50
0/40 0/45 0/35 17/226
0/50 0/50 0/50 46/50
0/40 0/45 0/35 20/226
0/50 0/50 1/50 44/50
0/40 0/45 1/35 20/226
0/50 0/50 0/50 48/50
0/40 0/45 0/35 21/226
0/50 0/50 0/50 13/50
0/40 0/45 0/35 7/22e
3/50 2/50 10/50 24/50
3/40 2/45 9/35f 12/226
12/50 20/50 25/50 40/50
12/40 20/45 21/35f 21/226
3/50 3/50 9/50 8/50
3/40 3/45 9/35f 7/22e
Not reported
Not reported
7/50 11/50 6/50 16/50
7/40 11/45 6/35 8/22f
2/50 8/50 14/50 13/50
2/40 8/45 14/356 22/22e
0/50 0/50 0/50 50/50
0/40 0/45 0/35 22/22e
0 200 1,000 5,000
fir
n 11+ 55+ 274+
u 1 3 18
0/50 0/50 0/50 26/50"
Not reported
0/50 0/50 0/50 31/50"
Not reported
0/50 0/50 0/50 2/50
Not reported
Not reported
Not reported
0/50 0/50 5/50 38/50"
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
Not reported
Not reported
Not reported
3/50 3/50 9/50 8/50
Not reported
12/50 8/50 7/50 5/50
Not reported
7/50 11/50 8/50 16/50'
Not reported
2/50 8/50 14/506 13/50"
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) 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).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
JBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
E-2
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-2 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in female F344 rats
Yamazaki et al. (1994)a
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000 0
Calculated
Not reported
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
Lamina propria;
hydropic change
Lamina propria;
slerosis
Nasal cavity;
adhesion
Nasal cavity;
inflammation
Liver; hyperplasia
Liver; spongiosis
hepatis
Liver; cyst formation
Liver; clear cell foci
Liver; acidophilic cell
foci
Liver; basophilic cell
foci
Liver; mixed-cell foci
Kidney proximal
tubule; 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
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
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
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
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
Not reported
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
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
200 1,000
5,000 0 200 1,000
5,000
Dose (Intake [mg/kg-day])"'1'
12- 56-
29 149
(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
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)
13/50
7/24e
35/50
18/246
5/50
4/24f
11/50
8/24e
39/50
22/24e
42/50
24/24e
40/50
22/24e
46/50
23/24e
48/50
23/24e
46/50
24/24e
15/50
7/24e
47/50
24/24e
20/50
14/246
8/50
5/24f
Not reported
Not reported
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
n 18+ 83+
0 3 14
0/50 0/50 0/50
429+
69
13/50"
Not reported
0/50 0/50 0/50
35/50"
Not reported
0/50 0/50 0/50
5/50
Not reported
Not reported
Not reported
0/50 0/50 28/50"
39/50"
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
Not reported
Not reported
Not reported
Not reported
Not reported
1/50 1/50 5/50
4/50
Not reported
1/50 1/50 1/50
1/50
Not reported
23/50 27/50 31/50
8/50"
Not reported
1/50 1/50 3/50
11/50'
Not reported
Not reported
Not reported
DRAFT - DO NOT CITE OR QUOTE
E-3
-------
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).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
dJBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
Table E-3 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in male F344 rats
Yamazaki et al. (1994)a
0 200 1,000 5,000
JBRC(1998)P
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day
Not reported
r«ntmi 8- 41' 209-
Co,ftrol 24 121 586
1°' (16) (81) (398)
0 200 1,000 5,000
)"""
n 11+ 55+ 274+
0 1 3 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
3/50"
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
0/50
0/50
Not
0/50
Not
0/50
Not
0/50
Not
0/50
reported
0/50
reported
0/50
reported
0/50
reported
3/50"
2/50
1/50
1/50
Liver
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma
or carcinoma
Tumors at other sites
Peritoneum
mesothelioma
Subcutis fibroma
Mammary gland
fibroadenoma
Mammary gland
adenoma
Mammary gland
fibroadenoma
or adenoma
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
0/50 2/50 4/50 24/50
Not reported
0/50 0/50 0/50 14/50
Not reported
Not reported
Not reported
2/50 2/50 5/50 28/50
Not reported
5/50 3/50 5/50 12/50
Not reported
1/50 1/50 0/50 4/50
Not reported
0/50 0/50 0/50 0/50
Not reported
Not reported
Not reported
0/50 2/50 4/49 24/50"'"
Not reported
0/50 0/50 0/49 14/50"'"
Not reported
0/50 2/50 4/49 33/50"'"
Not reported
2/50 2/50 5/50 28/50"'°
Not reported
5/50 3/50 5/50 12/50"
Not reported
1/50 1/50 0/50 4/50"
Not reported
Not reported
Not reported
Not reported
Not reported
3/50 4/50 7/50 32/50"'"
Not reported
0/50 0/50 0/50 14/50"'"
Not reported
3/50 4/50 7/50 39/50"'"
Not reported
2/50 2/50 5/50 28/50"'°
Not reported
5/50 3/50 5/50 12/50"
Not reported
1/50 1/50 0/50 4/50"
Not reported
0/50 1/50 2/50 2/50
Not reported
1/50 2/50 2/50 6/50°
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 i
the external peer review draft of this document (U.S. EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
dp<0.01 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01.
DRAFT - DO NOT CITE OR QUOTE
E-4
-------
Table E-4 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in female F344 rats
Yamazaki et al. (1994)a
0 200 1,000 5,000
JBRC(1998)P
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000 | 0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day]f "
Not Reported
r«ntmi 12' 56' 307-
Co,n,V 29 149 720
W (21) (103) (514)
n 18+ 83+ 429+
u 3 14 69
Nasal cavity
Squamous cell
carcinoma
Sarcoma NOS
Rabdomyosarcoma
Esthesioneuroepithelio
ma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50
0/50
0/50
0/50
0/50 0/50
Not
reported
0/50 0/50
Not
reported
0/50 0/50
Not
reported
0/50 0/50
Not
reported
7/50
0/50
0/50
1/50
0/50 0/50
Not
Not
Not
Not
Not
0/50 0/50
Not
0/50 7/50""
reported
reported
reported
reported
reported
0/50 1/50
reported
0/bO
0/bO
0/bO
0/bO
0/50
Not
0/50
Not
0/50
Not
0/50
Not
0/50
reported
0/50
reported
0/50
reported
0/50
reported
7/50""
0/50
0/50
1/50
Liver
Hepatocellular
adenoma
Hepatocellular
carcinoma
Hepatocellular
adenoma or
carcinoma
Tumors at other sites
Peritoneum
mesothelioma
Subcutis fibroma
Mammary gland
fibroadenoma
Mammary gland
adenoma
Mammary gland
fibroadenoma
or adenoma
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
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 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
6/50 7/50 10/50 16/50
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
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
6/50 7/50 10/50 16/50d'f
Not reported
Not reported
Not reported
3/50 1/50 6/50 48/5(r
Not reported
0/50 0/50 0/50 10/5(r
Not reported
3/50 1/50 6/50 48/5(r
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
6/50 7/50 10/50 16/50"'
Not reported
8/50 8/50 11/50 18/50"'
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).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
"p< 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.
E-5
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-5 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in male Crj:BDF1 mice
Yamazaki et al.
(1994f
JBRC(1998)M
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000 1 0 500
2,000 8,000 | 0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day]p
Nasal respiratory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
atrophy
Nasal cavity; inflammation
Tracheal epithelium; atrophy
Tracheal epithelium; nuclear
enlargement
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
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
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
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
r«ntr«i 37- 144- 451-
ColF01 94 358 1086
u (66) (251) (768)
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
1/50 2/50
1/31 1/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
1/50 0/50
1/31 0/33
2/50 3/50
2/31 2/33
0/50 0/50
0/31 0/33
40/50 42/50
28/31 30/33
0/50 31/50
0/25 19/266
9/50 49/50
7/25e 26/26e
1/50 48/50
0/25 26/26e
1/50 25/50
1/25 15/266
0/50 42/50
0/25 24/26e
0/50 17/50
0/25 12/266
0/50 41/50
0/25 24/26e
0/50 43/50
0/25 26/26e
0/50 27/50
0/25 22/26e
4/50 16/50
3/25 8/26f
0/50 39/50
0/25 22/26e
38/50 34/50
24/25f 21/26f
n 49+ 191+ 677+
0 5 21 74
0/50 0/50 0/50 31/506
Not reported
0/50 0/50 9/50e 49/506
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
Not reported
Not reported
Not reported
Not reported
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).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical Review
of 1,4-Dioxane (U.S. EPA, 2010).
dJBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
E-6
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-6 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in female Crj:BDF1 mice
Yamazaki et
(1994f
al.
JBRC (1998)"
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
Nasal respiratory
epithelium; Nuclear
enlargement
Nasal olfactory
epithelium; Nuclear
enlargement
Nasal respiratory
epithelium; Atrophy
Nasal olfactory
epithelium; Atrophy
Nasal cavity;
Inflammation
Tracheal epithelium;
Atrophy
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
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
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
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control
0
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
45-
109
(77)
0/50
0/29
0/50
0/29
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)
0/50
0/17
41/50
17/176
0/50
0/17
1/50
0/17
7/50
5/1 7e
2/50
1/17
22/50
13/176
7/50
3/17
4/50
3/17
0/50
0/17
[mg/kg-day])"'1'
759-
1374
(1066)
41/50
5/5e
33/50
1/5
26/50
1/5
42/50
5/5e
42/50
5/5e
49/50
5/5e
48/50
5/5e
50/50
5/5e
45/50
5/5e
8/50
0/5
0 ^ 278 ± 40 964 ± 88
0/50 0/50 0/50 41/50"
Not reported
0/50 0/50 41/50" 33/50"
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
Not reported
Not reported
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.
'Statistical analysis was not performed for data on 'All animals' in the JBRC (1998) report.
CJBRC (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 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
ep<0.01 by chi-square test.
E-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-7 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in male Crj:BDF1 mice
Yamazaki et al. (1994)a| JBRC (1998)p
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-da
Not reported
r«ntml 37' 144' 451'
Control 94 358 1086
0 (66) (251) (768)
0 500 2,000 8,000
yir
n 49+ 191+ 677+
u 5 21 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
Either adenoma
or carcinoma
All Animals
Sacrificed
animals
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
Not reported
Not reported
7/50 16/50 22/50" 8/50
Not reported
15/50 20/50 23/50 36/50"'"
Not reported
21/50 31/50 37/50 39/50"'"
Not reported
9/50 17/50 23/50" 11/50
Not reported
15/50 20/50 23/50 36/50""
Not reported
23/50 31/50 37/50" 40/50""
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).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
dp£ 0.05 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01.
fp£0.01 by Fisher's Exact test.
E-S
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-8 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in female Crj:BDF1 mice
Yamazaki et al.
(1994f
JBRC(1998)b
Kano et al. (2009)
Drinking water concentration ppm)
0 500 2,000 8,000
0 500 2,000 8,000
0
Calculated Dose (Intake [mg/kg-day]P
Not reported
r«ntr«i 46- 192- 759'
Control 1og 454 1374
u (77) (323) (1066)
0
500 2,000 8,000
66 ± 278 ± 964 ±
10 40 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
Not
reported
reported
0/50 0/50 0/50 1/50
Not
reported
0/50
0/50
0/50
Not
0/50
Not
0/50
reported
0/50
reported
0/50
1/50
Liver
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenoma
or carcinoma
All animals
Sacrificed
animals
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
Not reported
Not reported
4/50
30/50" 20/50"
2/50"
Not reported
0/50
6/50' 30/50"
45/50"'"
Not reported
4/50
34/50" 41/50"
46/50"'"
Not reported
5/50 31/50" 20/50" 3/50
Not reported
0/50 6/50' 30/50" 45/50"'"
Not reported
5/50 35/50" 41/50" 46/50"'"
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) 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).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010).
dp<0.01 by Fisher's Exact test.
'Significantly decreased by Cochran-Armitage test for trend p < 0.05
f p < 0.05 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01
E-9
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX F. DETAILS OF BMD ANALYSIS FOR
INHALATION RFC FOR 1,4-DIOXANE
F.1 Centrilobular Necrosis of the Liver
1 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
2 incidence data shown in Table F-l. for centrilobular necrosis of the liver in male F344/DuCrj rats exposed
3 to 1.4-dioxane vapors for 2 years (Kasai et al., 2009). Doses associated with a BMR of a 10% extra risk
4 were calculated.
Table F-1 Incidence of centrilobular necrosis of the liver in F344/DuCrj rats exposed to
1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
1/50
3/50
250
6/50
(12%)
1,250
12/50a
(24%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
5 As assessed by the y2 goodness-of-fit test, several models in the software provided adequate fits
6 to the incidence data of centrilobular necrosis of the liver in male rats (y2 p > 0.1) (Table F-2). Comparing
7 across adequately Fitting models, the BMDL estimates were not within threefold difference of each
8 other. Therefore, in accordance with EPA BMD technical guidance (U.S. EPA. 2000aX the adequately
9 Fitting model that resulted in the lowest BMDL was selected as appropriate for deriving a POD which was
10 the Dichotomous-Hill model. BMDS modeling results for all dichotomous models are shown in Table F-2
11 and the model plot (Figure F-l) and output for the selected Dichotomous-Hill model are included
12 immediately after the table.
F-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-2 Goodness-of-fit statistics and BMD10 and BMDL10 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
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
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 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.
"Power restricted to > 1.
cSlope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit model based on lowest BMDL.
Source: Kasai et al. (2009).
.Q
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
16:18 01/12 2011
Dichotomous-Hill Model with 0.95 Confidence Level
Dichotomous-Hill
BMDL
BMD
0
200
400
600
dose
800
1000
1200
Figure F-l BMD Dichotomous Hill model of centrilobular necrosis incidence data
for male rats exposed to 1,4-dioxane vapors for 2 years to support the
results in Table F-2.
Dichotomous Hill Model. (Version: 1.2; Date: 12/11/2009)
F-2
DRAFT - DO NOT CITE OR QUOTE
-------
1 Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
2 files/dhl Centr necrosis liver Dhl-BMRlO-Restrict.(d)
3 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
4 files/dhl Centr necrosis liver Dhl-BMRlO-Restrict.pit
5 Wed Jan I2_ 16:34:41 2011
6 ====================================================================
7 BMDS Model Run
8 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
9 The form of the probability function is:
10
11 P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
12 where: 0 <= g < 1, 0 < v <= 1
13 v is the maximum probability of response predicted by the model,
14 and v*g is the background estimate of that probability.
15
16 Dependent variable = Effect
17 Independent variable = Dose
18 Slope parameter is restricted as slope >= _!
19
20 Total number of observations = £
21 Total number of records with missing values = 0_
22 Maximum number of iterations = 250
23 Relative Function Convergence has been set to: le-008
24 Parameter Convergence has been set to: le-008
25
26 Default Initial Parameter Values
27 v = -9999~
28 g = -9999
29 intercept = -8.08245
30 slope = I
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
35 been specified by the user, and do not appear in the correlation matrix)
36
37 v g intercept
38 v I -0.25 -0789
39 g -0.25 I 0.016
40 intercept -0.89 0.016 I
41
42
43 Parameter Estimates
44
45 95.0% Wald Confidence Interval
46 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
47 v 0.311077 0.156196 0 . 0oT9~3876 0/617216
48 g 0.0709966 0.0662298 -0.0588115 0.200805
49 intercept -6.06188 1.34538 -8.69878 -3.42498
50 slope I NA
51
52 NA 2. Indicates that this parameter has hit a. bound implied by some ineguality
53 constraint and thus has no standard error.
54
55
56 Analysis of Deviance Table
57
58 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
59 Full model -62.1506 4
60 Fitted model -62.2022 3 0.103279 I 0.7479
61 Reduced model -69.3031 I 14.305 3 0.002518
62
63 AIC: 130.404
64
65 Goodness of Fit
66 Scaled
67 Dose Est. Prob. Expected Observed Size Residual
F-3
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 0.0000 0.0221 1.104 1.000 5£ -0.100
3 50.0000 0.0522 2.612 3.000 5£ 0.247
4 250.0000 0.1285 6.423 6.000 5j3 -0.179
5 1250.0000 0.2372 11.861 12.000 5_0 0.046
6
7 Chi"2 = 0.10 d.f. = I P-value = 0.7459
8
9
10 Benchmark Dose Computation
11 Specified effect = 0.1
12 Risk Type = Extra risk
13 Confidence level = 0.95
14 BMP = 219.51
15 BMDL = 59.5598
F.2 Spongiosis Hepatis
16 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
17 incidence data shown in Table F-3i for spongiosis hepatis of the liver in male F344/DuCrj rats exposed to
18 1.4-dioxane vapors for 2 years (Kasai et al.. 2009). Doses associated with a BMR of a 10% extra risk
19 were calculated.
Table F-3 Incidence of spongiosis hepatis of the liver in F344/DuCrj rats exposed to 1,4-dioxane
via inhalation for 2 years
1 ,4-dioxane vapor concentration (ppm)
7/50
(14%)
6/50
250
13/50
(26%)
1,250
19/50a
(38%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
20 As assessed by. the £ goodness-of-fit test, several models in the software provided adequate fits
21 to the incidence data of spongiosis of the liver in male rats (y2 p > 0.1) (Table F-4). BMDL estimates for
22 all adequately Fitting models were not within threefold difference of each other (U.S. EPA. 2000a).
23 Therefore, in accordance with EPA BMP technical guidance (U.S. EPA. 2000a). the adequately fitting
24 model that resulted m the lowest BMDL was selected as appropriate for deriving a POD which was the
25 dichotomous-Hill model. However, the dichotomous-Hill model, warned that the BMDL estimate was
26 "imprecise at best" (see Figure F-2 and subsequent textual model output). Comparing across aU models
27 (excluding the dichotomous-hill model), a better fit is indicated by a lower AIC value since the BMDL
28 estimates for all appropriately Fitting models were within threefold difference of each other (U.S.
29 EPA. 2000a). As assessed by the AIC. the log-logistic model provided the best fit to the spongiosis
30 incidence data for male rats (Table F-4. Figure F-3 and subsequent textual model output) and could be
31 used to derive a POD for this endpoint.
F-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-4 Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
data for spongiosis hepatis of the liver in male F344/DuCrj rats (NCI. 1978) exposed to
1,4-dioxane vapors
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-log isticc'T
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
Dichotomous-Hillc'
e
206.472
207.141
206.229
208.147
206.472
207.06
206.472
206.472
206.364
0.4482
0.3159
0.5102
0.1825
0.4482
0.3292
0.4482
0.4482
0.4671
1.031
1.242
0.912
1.536
1.031
1.223
1.031
1.031
1.031
369.422
537.295
314.34
633.557
369.422
515.483
369.422
369.422
289.919
224.993
392.318
172.092
414.718
224.993
371.644
224.993
224.993
59.69
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.
"Power restricted to > 1.
cSlope restricted to > 1.
dBetas restricted to > 0.
eModel output warned that the BMDL estimate was "imprecise at best".
'Bold indicates best-fit model based on lowest AIC.
Source: Kasai et al. (2009).
Dichotomous-Hill Model with 0.95 Confidence Level
0.5
0.4
0.3
0.2
0.1
1200
16:5201/122011
Figure F-2 BMD Dichotomous-Hill model of spongiosis hepatis incidence data for
male rats exposed to 1,4-dioxane vapors for 2 years to support the
results in Table F-4.
2 Dichotomous Hill Model. (Version: 1.2; Date: 12/11/2009)
3 Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
4 files/dhl spong hepa liver Dhl-BMRlO-Restrict.(d)
5 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
6 files/dhl spong hepa liver Dhl-BMRlO-Restrict.pit
F-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 Wed Jan 12_ 16:52:46 2011
2 ====================================================================
3 BMDS Model Run
4 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5 The form of the probability function is:
6
7 P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
8 where: 0 <= g < 1, 0 < v <= 1
9 v is the maximum probability of response predicted by the model,
10 and v*g is the background estimate of that probability.
11
12 Dependent variable = Effect
13 Independent variable = Dose
14 Slope parameter is restricted as slope >= _!
15
16 Total number of observations = £
17 Total number of records with missing values = 0_
18 Maximum number of iterations = 250
19 Relative Function Convergence has been set to: le-008
20 Parameter Convergence has been set to: le-008
21
22 Default Initial Parameter Values
23 v = -9999~
24 g = -9999
25 intercept = -8.74962
26 slope = 1.13892
27
28 Asymptotic Correlation Matrix of Parameter Estimates
29 (*** The model parameter(s) -v -slope have been estimated at a. boundary point, or have
30 been specified by the user, and do not appear in the correlation matrix j_
31
32 g intercept
33 g I -0.53
34 intercept -0.53 I
35
36 Parameter Estimates
37
38 95.0% Wald Confidence Interval
39 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40 v I NA
41 g 0.125 0.0332679 0.0597961 0.190204
42 intercept -7.86683 0.396424 -8.6438 -7.08985
43 slope I NA
44
45 NA ^ Indicates that this parameter has hit a. bound implied by some ineguality
46 constraint and thus has no standard error.
47
48 Analysis of Deviance Table
49
50 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
51 Full model -100.45 4
52 Fitted model -101.182 2 1.46273 2 0.4813
53 Reduced model -106.633 I 12.3646 3 0.006233
54
55 AIC: 206.364
56
57 Goodness of Fit
58 Scaled
59 Dose Est. Prob. Expected Observed Size Residual
60
61 0.0000 0.1250 6.250 7.000 _50_ 0.321
62 50.0000 0.1415 7.073 6.000 5_0 -0.435
63 250.0000 0.2015 10.075 13.000 _50_ 1.031
64 1,250.0000 0.4084 20.420 19.000 _50_ -0.409
65
66 Chi"2 = 1.52 d.f. = 2 P-value = 0.4671
67
F-6
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMP = 289.919
Warning: BMDL computation is at best imprecise for these data
BMDL = 59.69
Log-Logistic Model with 0.95 Confidence Level
I
0.5
0.4
0.3
0.2
0.1
Log-Logistic
BMDL
BMD
200
400
16:5201/122011
600
dose
800
1000
1200
Figure F-3 BMD Log-Logistic model of spongiosis hepatis incidence data for male
rats exposed to 1,4-dioxane vapors for 2 years to support the results in
Table F-4.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Logistic Model. (Version: 2.13; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl spong hepa liver Lnl-BMRlO-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl spong hepa liver Lnl-BMRlO-Restrict.pit
Wed Jan I2_ 16:52:44 2011
BMDS Model Run
The form of the probability function is:
P[response] = backgrounds- (1-background)/[1+EXP(-intercept-siope*Log(dose))]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= _!
Total number of observations = £
Total number of records with missing values = 0_
Maximum number of iterations = 250
F-7
DRAFT - DO NOT CITE OR QUOTE
-------
1 Relative Function Convergence has been set to: le-008
2 Parameter Convergence has been set to: le-008
3
4 User has chosen the log transformed model
5
6 Default Initial Parameter Values
7 background = 0.14
8 intercept = -8.74962
9 slope = 1.13892
10
11 Asymptotic Correlation Matrix of Parameter Estimates
12 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
13 been specified by the user, and do not appear in the correlation matrix)
14
15 background intercept
16 background I -0.54
17 intercept -0.54 I
18
19 Parameter Estimates
20 95.0% Wald Confidence Interval
21 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
22 background 0.1376? 1 1 1~~
23 intercept -7.9477 111
24 slope 1111
25
26 1 2. Indicates that this value is not calculated.
27
28
29 Analysis of Deviance Table
30
31 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
32 Full model -100.45 £
33 Fitted model -101.115 ,2 1.3283 2_ 0.5147
34 Reduced model -106.633 I 12.3646 3_ 0.006233
35
36 AIC: 206.229
37
38
39 Goodness of Fit
40 Scaled
41 Dose Est. Prob. Expected Observed Size Residual
42
43 0.0000 0.1377 6.885 7.000 _50_ 0.047
44 50.0000 0.1527 7.633 6.000 5J) -0.642
45 250.0000 0.2077 10.385 13.000 _50_ 0.912
46 1250.0000 0.4019 20.097 19.000 5_0 -0.316
47
48 Chi"2 = 1.35 d.f. = £ P-value = 0.5102
49
50
51 Benchmark Dose Computation
52 Specified effect =0.1
53 Risk Type = Extra risk
54 Confidence level = 0.95
55 BMP = 314.34
56 BMDL = 172.092
F.3 Squamous Cell Metaplasia
57 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
58 incidence data shown in Table F-5. for squamous cell metaplasia of the respiratory epithelium in male
F-8
DRAFT - DO NOT CITE OR QUOTE
-------
1 F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
2 a 10% extra risk were calculated.
Table F-5 Incidence of squamous cell metaplasia of the respiratory epithelium in 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
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
Source: Kasai et al. (2009).
3 For incidence of squamous cell metaplasia in F344/DuCrj male rats, the logistic and probit
4 models all exhibited a statistically significant lack of fit (i.e.. y2 p-va\ue < 0.1; see Table F-6). and thus
5 should not be considered further for identification of a POD. All of the remaining models exhibited
6 adequate fit. The BMDL estimates for all appropriately Fitting models were within threefold
7 difference of each other, indicating that BMDL selection should be made based on model fit (U.S.
8 EPA. 2000a). As assessed by the AIC. the Log-probit model provided the best fit to the squamous cell
9 metaplasia data for male rats (Table F-6, Figure F-4^ and could be used to derive a POD for this
10 endpoint.
F-9
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-6 Goodness-of-fit statistics and BMD10 and BMDL10 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
Gamma"
Logistic
Log-logistic0
Log-probitc' e
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-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
1.806
0.131
0.032
0.605
1.681
-1.76
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
1 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.
cSlope restricted to > 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest AIC.
Source: Kasai et al. (2009).
LogProbit Model with 0.95 Confidence Level
"8
•
.
o
0.8
0.6
0.4
0.2
1200
13:11 01/132011
Figure F-4 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 to support the results in Table F-6.
1 Probit Model. (Version: 3.2; Date: 10/28/2009)
2 Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
3 files/lnp squ cell meta re Lnp-BMRlO-Restrict.(d)
F-10
DRAFT - DO NOT CITE OR QUOTE
-------
1 Gnuplot Plotting File: C:/Documents and Settings/pgi11esp/Desktop/BMDS
2 files/lnp sgu cell meta re Lnp-BMRlO-Restrict.pit
3 Thu Jan 13 13:11:09 2011
4 ====================================================================
5 BMDS Model Run
f) ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
7 The form of the probability function is:
8
9 P[response] = Background + (1-Background) ^_ CumNorm(Intercept+Slope*Log(Dose)),
10 where CumNorm(.) is the cumulative normal distribution function
11
12 Dependent variable = Effect
13 Independent variable = Dose
14 Slope parameter is restricted as slope >= _!
15
16 Total number of observations = £
17 Total number of records with missing values = 0_
18 Maximum number of iterations = 250
19 Relative Function Convergence has been set to: le-008
20 Parameter Convergence has been set to: le-008
21
22 User has chosen the log transformed model
23
24 Default Initial (and Specified) Parameter Values
25 background = 0_
26 intercept = -6.76507
27 slope = 1.09006
28
29 Asymptotic Correlation Matrix of Parameter Estimates
30 (*** The model parameter(s) -background have been estimated at a. boundary point, or
31 have been specified by the user, and do not appear in the correlation matrix)
32
33 intercept slope
34 intercept I -0.99
35 slope -0.99 I
36
37 Parameter Estimates
38
39 95.0% Wald Confidence Interval
40 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
41 background 0_ NA
42 intercept -8.86173 1.2226 -11.258 -6.46548
43 slope 1.40803 0.193057 1.02965 1.78642
44
45 NA 2. Indicates that this parameter has hit a. bound implied by some ineguality
46 constraint and thus has no standard error.
47
48 Analysis of Deviance Table
49
50 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
51 Full model -38.5944 4
52 Fitted model -38.615 2_ 0.041197 £ 0.9796
53 Reduced model -113.552 I 149.916 3 <.OOQ1
54
55 AIC: 81.23
56
57 Goodness of Fit
58 Scaled
59 Dose Est. Prob. Expected Observed Size Residual
60
61 0.0000 0.0000 0.000 0.000 5£ 0.000
62 50.0000 0.0004 0.020 0.000 5_0 -0.141
63 250.0000 0.1384 6.922 7.000 _50_ 0.032
64 1250.0000 0.8808 44.038 44.000 5J) -0.017
65
66 Chi"2 = 0.02 d.f. = _2 P-value = 0.9894
67
F-ll
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Benchmark Dose Computation
3 Specified effect = 0.1
4 Risk Type = Extra risk
5 Confidence level = 0.95
6 BMP = 217.79
7 BMDL = 159.619
F.4 Squamous Cell Hyperplasia
8 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
9 incidence data shown in Table F^ for squamous cell hyperplasia of the respiratory epithelium in male
10 F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
11 a 10% extra risk were calculated.
Table F-7 Incidence of squamous cell hyperplasia of the respiratory epithelium in F344/DuCrj
rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
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).
12 For incidence of squamous cell hyperplasia in F344/DuCrj male rats, the logistic, probit. and
13 quantal-linear models all exhibited a statistically significant lack of fit (i.e.. y2 p-value < 0.1; see
14 Table F-8). and thus should not be considered further for identification of a POD. All of the remaining
15 models exhibited adequate fit. The BMDL estimates for all appropriately Fitting models were within
16 threefold difference of each other, indicating that BMDL selection should be made based on model
17 fit (U.S. EPA. 2000a). As assessed by the AIC. the Log-probit model provided the best fit to the
18 squamous cell hyperplasia data for male rats (Table F-8i Figure F-5 and subsequent textual model output).
19 and could be used to derive a POD for this endpoint.
F-12
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-8 Goodness-of-fit statistics and BMD10 and BMDL10 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
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc' e
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-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
1.806
0.131
0.032
0.605
1.681
-1.76
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
1 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.
cSlope restricted to > 1.
dBetas restricted to > 0.
eBold indicates best-fit model based on lowest AIC.
Source: Kasai et al. (2009).
LogProbit Model with 0.95 Confidence Level
0.35
0.3
0.25
0.2
0.15
0.1
0.05
1200
13:2501/132011
Figure F-5 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 to support the results in Table F-8.
1 Probit Model. (Version: 3.2; Date: 10/28/2009)
2 Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
3 files/lnp squ cell hyper re Lnp-BMRlO-Restrict.(d)
4 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
5 files/lnp sgu cell hyper re Lnp-BMRlO-Restrict.pit
DRAFT - DO NOT CITE OR QUOTE
F-13
-------
1 Thu Jan 13_ 13:25:05 2011
2 ====================================================================
3 BMDS Model Run
4 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
5 The form of the probability function is:
6
7 P[response] = Background + (1-Background) ^_ CumNorm(Intercept+Slope*Log(Dose)),
8 where CumNorm(.) is the cumulative normal distribution function
9
10 Dependent variable = Effect
11 Independent variable = Dose
12 Slope parameter is restricted as slope >= _!
13
14 Total number of observations = £
15 Total number of records with missing values = 0_
16 Maximum number of iterations = 250
17 Relative Function Convergence has been set to: le-008
18 Parameter Convergence has been set to: le-008
19
20 User has chosen the log transformed model
21
22 Default Initial (and Specified) Parameter Values
23 background = £
24 intercept = -7.75604
25 slope = I
26
27 Asymptotic Correlation Matrix of Parameter Estimates
28 (*** The model parameter(s) -background -slope have been estimated at a boundary
29 point, or have been specified by the user, and do not appear in the correlation
30 matrix)
31
32 intercept
33 intercept _!
34
35 Parameter Estimates
36
37 95.0% Wald Confidence Interval
38 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
39 background C) NA
40 intercept -7.90911 0.186242 -8.27414 -7.54408
41 slope I NA
42
43 NA 2. Indicates that this parameter has hit a. bound implied by some ineguality
44 constraint and thus has no standard error.
45
46 Analysis of Deviance Table
47
48 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
49 Full model -29.9221 £
50 Fitted model -30.2589 I 0.673572 3 0.8794
51 Reduced model -42.5964 I 25.3487 3_ <.OOQ1
52
53 AIC: 62.5177
54
55 Goodness of Fit
56 Scaled
57 Dose Est. Prob. Expected Observed Size Residual
58
59 0.0000 0.0000 0.000 0.000 _50_ 0.000
60 50.0000 0.0000 0.002 0.000 5J) -0.040
61 250.0000 0.0085 0.424 1.000 50 0.
62 1250.0000 0.2182 10.911 10.000 5_0 -0.312
63
64 Chi"2 = 0.89 d.f. = 3 P-value = 0.8282
65
66
67 Benchmark Dose Computation
F-14
DRAFT - DO NOT CITE OR QUOTE
-------
1 Specified effect = 0.1
2 Risk Type = Extra risk
3 Confidence level = 0.95
4 BMP = 755.635
5 BMDL = 560.86
F.5 Respiratory Metaplasia
6 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
7 incidence data shown in Table F^ for respiratory metaplasia of the olfactory epithelium in male
8 F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
9 a 10% extra risk were calculated.
Table F-9 Incidence of respiratory metaplasia of the olfactory epithelium in F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
11/50
(22%)
34/50
250
49/50 a
(98%)
1,250
48/50a
(96%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
10 As assessed by the y2 goodness-of-fit test, no models in the software provided adequate fits to the
11 data for the incidence of respiratory metaplasia of the olfactory epithelium in male rats (y1 p > 0.1)
12 (Table F-10). However, given that first non-control dose had a response level substantially above the
13 desired BMR (i.e. 10%). the use of BMP methods included substantial model uncertainty. The model
14 uncertainty associated with this dataset is related to low-dose extrapolation and consistent with BMP
15 technical guidance document (U.S. EPA. 2000a). all available dichotomous models in the Benchmark
16 Dose Software (version 2.1.2) were fit to the incidence data shown in Table F-9 with the highest dose
17 group omitted. As assessed by the y2 goodness-of-fit test, the logistic, log-logistic, log-probit and probit
18 models all exhibited a statistically significant lack of fit (i.e.. y2 p-va\ue < 0.1:See Table F-l 1). and thus
19 should not be considered further for identification of a POD. The BMDL estimates for all appropriately
20 Fitting models were within threefold difference of each other, indicating that BMDL selection should
21 be made based on model fit (U.S. EPA. 2000a). The AIC values for gamma, multistage, quantal-linear.
22 and Weibull models in Table F-l 1 are equivalent and the lowest and, in this case, essentially represent the
23 same model. Therefore, consistent with the external review draft Benchmark Dose Technical Guidance
24 (U.S. EPA. 2000a). any of them with equal AIC values (gamma, multistage, quantal-linear. or Weibull)
25 could be used to identify a POD for this endpoint. The model plot for the gamma model (Figure F-6)
26 and output are included immediately after the table.
F-15
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-10 Goodness-of-fit statistics and BMD10 and BMDL10 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
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-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
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 x2 good ness-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.
cSlope restricted to > 1.
dBetas restricted to >0.
Source: Kasai et al. (2009).
Table F-11 Good ness-of-fit statistics and BMD10 and BMDL10 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
Gamma0' e
Logistic
Log-logistic0
Log-probitc
Multistage
(2degree)d'e
Probit
Weibull"
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
-0.106
-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
1 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.
cSlope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit models based on lowest AIC.
Source: Kasai et al. (2009).
DRAFT - DO NOT CITE OR QUOTE
F-16
-------
Gamma Multi-Hit Model with 0.95 Confidence Level
I
0.8
0.6
0.4
0.2
250
16:2401/132011
Figure F-6 BMD Gamma model of respiratory metaplasia of olfactory epithelium
incidence data for male rats exposed to 1,4-dioxane vapors for 2 years
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
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-BMRlO-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/gam resp meta no high dose Gam-BMRlO-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[slope*dose,power],
where CumGamma(.) is the cummulative Gamma distribution function
Dependent variable = Effect
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 3_
Total number of records with missing values = 0_
Maximum number of iterations = 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
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 _! -0 . 33
Slope -0.33 1
F-17
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Parameter Estimates
3
4 95.0% Wald Confidence Interval
5 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
6 Background 0.226249 Q.05'88535 0.110898 0.3416
7 Slope 0.0162883 0.00320976 0.00999729 0.0225793
8 Power I NA
9
10 NA 2. Indicates that this parameter has hit a. bound implied by some inequality
11 constraint and thus has no standard error.
12
13 Analysis of Deviance Table
14
15 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
16 Full model -62.5908 3
17 Fitted model -62.7313 ,2 0.280907 I 0.5961
18 Reduced model -99.1059 I 73.0301 2_ <.OOQ1
19
20 AIC: 129.463
21
22 Goodness of Fit
23 Scaled
24 Dose Est. Prob. Expected Observed Size Residual
25
26 0.0000 0.2262 11.312 11.000 _50_ -0.106
27 50.0000 0.6573 32.865 34.000 5J) 0.338
28 250.0000 0.9868 49.341 49.000 _50_ -0.422
29
30 Chi^2 = 0.30 d.f. = I P-value = 0.5815
31
32 Benchmark Dose Computation
33 Specified effect = 0.1
34 Risk Type = Extra risk
35 Confidence level = 0.95
36 BMP = 6.46848
37 BMDL = 4.73742
F.6 Atrophy
38 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were
39 fit to the incidence data shown in Table F-12. for atrophy of the olfactory epithelium in
40 male F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (Kasai etal.. 2009).
41 Doses associated with a BMR of a 10% extra risk were calculated.
F-18
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-12 Incidence of respiratory metaplasia of the olfactory epithelium in 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).
1 As assessed by the y2 goodness-of-fit test, the gamma, logistic, log-probit. multistage, probit.
2 Weibull. and quantal-linear models all exhibited a statistically significant lack of fit (i.e.. y2/>-value < 0.1;
3 see Table F-13). and thus should not be considered further for identification of a POD. The BMDL
4 estimates for all appropriately Fitting models were within threefold difference of each other.
5 indicating that BMDL selection should be made based on model fit (U.S. EPA. 2000a). As assessed by
6 the AIC. the Log-logistic model provided the best fit to the atrophy data for male rats (Table F-13.
7 Figure F-?X and could be used to derive a POD for this endpoint. However, given that first non-control
8 dose had a response level substantially above the desired BMR (i.e. 10%). the use of BMP methods
9 included substantial model uncertainty.
F-19
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-13 Goodness-of-fit statistics and BMD10 and BMDL10 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
Gamma"
Logistic
Log-log isticc'e
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-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
4.342
0
3.943
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
1 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.
cSlope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit model based on lowest AIC.
Source: Kasai et al. (2009).
Log-Logistic Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
0
Log-Logistic
1
BMDLBMD
200
400
09:5301/142011
600
dose
800
1000
1200
Figure F-7 BMD Log-Logistic model of atrophy of olfactory epithelium incidence
data for male rats exposed to 1,4-dioxane vapors for 2 years to support
the results in Table F-13.
1 Logistic Model. (Version: 2.13; Date: 10/28/2009)
2 Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
3 files/lnl atrophy Lnl-BMRlO-Restrict.(d)
4 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
5 files/lnl atrophy Lnl-BMRlO-Restrict.pit
DRAFT - DO NOT CITE OR QUOTE
F-20
-------
1 Fri Jan \A_ 09:53:22 2011
2 ====================================================================
3 BMDS Model Run
4 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
5 The form of the probability function is:
6 P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]
7
8 Dependent variable = Effect
9 Independent variable = Dose
10 Slope parameter is restricted as slope >= _!
11
12 Total number of observations = £
13 Total number of records with missing values = 0_
14 Maximum number of iterations = 250
15 Relative Function Convergence has been set to: le-008
16 Parameter Convergence has been set to: le-008
17
18 User has chosen the log transformed model
19
20 Default Initial Parameter Values
21 background = 0_
22 intercept = -3.48908
23 slope = I
24
25 Asymptotic Correlation Matrix of Parameter Estimates
26 (*** The model parameter(s) -background -slope have been estimated at a boundary
27 point, or have been specified by the user, and do not appear in the correlation
28 matrix)
29
30 intercept
31 intercept _!
32
33 Parameter Estimates
34 95.0% Wald Confidence Interval
35 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
36 background 0_ 1 1 1
37 intercept -2.71122 111
38 slope 1111
39
40 1 2. Indicates that this value is not calculated.
41
42 Analysis of Deviance Table
43
44 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
45 Full model -44.7657 £
46 Fitted model -45.9537 I 2.37596 3 0.4981
47 Reduced model -126.116 I 162.701 3_ <.OOQ1
48
49 AIC: 93.9074
50
51 Goodness of Fit
52 Scaled
53 Dose Est. Prob. Expected Observed Size Residual
54
55 0.0000 0.0000 0.000 0.000 _50_ 0.000
56 50.0000 0.7687 38.433 40.000 5J) 0.525
57 250.0000 0.9432 47.161 47.000 _50_ -0.099
58 1250.0000 0.9881 49.405 48.000 5_0 -1.833
59
60 Chi"2 = 3.65 d.f. = 3 P-value = 0.3023
61
62 Benchmark Dose Computation
63 Specified effect = 0.1
64 Risk Type = Extra risk
65 Confidence level = 0.95
66 BMP = 1.67195
67 BMDL = 1.01633
F-21
DRAFT - DO NOT CITE OR QUOTE
-------
F.7 Hydropic Change
1 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
2 incidence data shown in Table F-14. for hydropic change of the lamina propria in the nasal cavity of male
3 F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (Kasai et al.. 2009). Doses associated with a
4 BMR of a 10% extra risk were calculated.
Table F-14 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)
0/50
2/50
(4%)
250
36/50 a
(72%)
1,250
49/50a
ap < 0.01 by Fisher's exact test.
Source: Kasai et al., (2009).
5 For incidence of hydropic change of the lamina propria in F344/DuCrj male rats, the gamma.
6 logistic, multistage, probit. Weibull. and quantal-linear models all exhibited a statistically significant lack
7 of fit (i.e.. y2/>-value < 0.1; see Table F-16). and thus should not be considered further for identification
8 of a POD. The BMDL estimates for all appropriately Fitting models were within threefold difference
9 of each other, indicating that BMDL selection should be made based on model fit (U.S. EPA. 2000a).
10 As assessed by the AIC. the Log-logistic model provided the best fit to the hydropic change of the lamina
11 propria data for male rats (Table F-15i Figure F-8 and subsequent text output), and could be used to
12 derive a POD of for this endpoint.
F-22
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-15 Goodness-of-fit statistics and BMD10 and BMDL10 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
Gamma"
Logistic
Log-log isticc'e
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-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
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 good ness-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.
cSlope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit model based on lowest AIC.
Source: Kasai et al. (2009).
Log-Logistic Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
Log-Logistic
BMDL
BMD
200
400
10:3001/142011
600
dose
800
1000
1200
Figure F-8 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 to support the results in Table F-16.
1 Logistic Model. (Version: 2.13; Date: 10/28/2009)
2 Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
3 files/lnl hydrpic Lnl-BMRlO-Restrict.(d)
4 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
5 files/lnl hydrpic Lnl-BMRlO-Restrict.pit
DRAFT - DO NOT CITE OR QUOTE
F-23
-------
1 Fri Jan \A_ 10:30:47 2011
2 =========================
3 BMDS Model Run
4 ~~~~~~~~~~~~~~~~
5 The form of the probability function is:
6 P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]
7
8 Dependent variable = Effect
9 Independent variable = Dose
10 Slope parameter is restricted as slope >= _!
11
12 Total number of observations = £
13 Total number of records with missing values = 0_
14 Maximum number of iterations = 250
15 Relative Function Convergence has been set to: le-008
16 Parameter Convergence has been set to: le-008
17
18 User has chosen the log transformed model
19
20 Default Initial Parameter Values
21 background = 0_
22 intercep^=~-Il. 5745
23 slope = 2.19638
24
25 Asymptotic Correlation Matrix of Parameter Estimates
26 (*** The model parameter(s) -background have been estimated at a. boundary point, or
27 have been specified by the user, and do not appear in the correlation matrix)
28
29 intercept slope
30 intercept I -0.99
31 slope -0.99 I
32
33 Parameter Estimates
34 95.0% Wald Confidence Interval
35 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
36 background 0_ ^_ ^_ ^_
37 intercept -12.1316 111
38 slope 2.3501 111"
39
40 1 2. Indicates that this value is not calculated.
41
42 Analysis of Deviance Table
43
44 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
45 Full model -42.9468 £
46 Fitted model -43.2694 ,2 0. 645129 ,2 0.7243
47 Reduced model -136.935 I 187.976 3_ <.OOQ1
48
49 AIC: 90.5388
50
51 Goodness of Fit
52 Scaled
53 Dose Est. Prob. Expected Observed Size Residual
54
55 0.0000 0.0000 0.000 0.000 _50_ 0.000
56 50.0000 0.0503 2.515 2.000 5J) -0.333
57 250.0000 0.6994 34.969 36.000 _50_ 0.318
58 1250.0000 0.9903 49.515 49.000 5_0 -0.744
59
60 Chi"2 = 0.77 d.f. = 2 P-value = 0.6819
61
62 Benchmark Dose Computation
63 Specified effect = 0.1
64 Risk Type = Extra risk
65 Confidence level = 0.95
66 BMP = 68.5266
67 BMDL = 46.7808
F-24
DRAFT - DO NOT CITE OR QUOTE
-------
F.8 Sclerosis
1 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
2 incidence data shown in Table F-16. for sclerosis of the lamina propria in the nasal cavity of male
3 F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (Kasai et al., 2009). Doses associated with a
4 BMR of a 10% extra risk were calculated.
Table F-16 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)
50 250 1,250
0/50 0/50 22/50a 40/50a
(44%) (80%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
5 As assessed by the y2 goodness-of-fit test, all models with the exception of the dichotomous-hill
6 model, exhibited a statistically significant lack of fit (i.e.. y2 p-value < 0. l:See Table F-17). and thus
7 should not be considered further for identification of a POD. Since the dichotomous-hill model provided
8 the only fit to the sclerosis of the lamina propria data for male rats as assessed by the y.2 goodness-of-fit
9 test (Table F-17. Figure F-9 and subsequent text output), it could be considered to derive a POD for this
10 endpoint; however, the model output warned that the BMDL estimate was "imprecise at best".
F-25
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-17 Goodness-of-fit statistics and BMD10 and BMDL10 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
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
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 x good ness-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.
"Power restricted to > 1.
cSlope restricted to > 1.
dBetas restricted to >0.
eModel output warned that the BMDL estimate was "imprecise at best".
Source: Kasai et al. (2009).
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Dichotomous Hill Model. (Version: 1.2; Date: 12/11/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl sclerosis Dhl-BMRlO-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl sclerosis Dhl-BMRlO-Restrict.pit
Fri Jan \A_ 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 < v <= 1
v is the maximum probability of response predicted by the model,
and v*g is the background estimate of that probability.
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= _!
Total number of observations = £
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
slope = ~.?6444
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)
F-26
DRAFT - DO NOT CITE OR QUOTE
-------
1 v intercept slope
2 v 1 0.00074 -0.00078
3 intercept 0.00074 I ^1
4 slope -0.00078 ^\_ I
5
6 Parameter Estimates
7
8 95.0% Wald Confidence Interval
9 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
10 v 0.8 0.0565686 0.689128 0.'910872
11 g £ NA
12 intercept -62.1804 4133.38 -8163.46 8039.1
13 slope 11.2979 748.603 -1455.94 1478.53
14
15 NA 2. Indicates that this parameter has hit a. bound implied by some inequality
16 constraint and thus has no standard error.
17
18
19 Analysis of Deviance Table
20
21 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
22 Full model -59.3166 4
23 Fitted model -59.3166 3 1.23973e-006 I 0.9991
24 Reduced model -123.82 I 129.007 3 <.OOQ1
25
26 AIC: 124.633
27
28 Goodness of Fit
29 Scaled
30 Dose Est. Prob. Expected Observed Size Residual
31
32 0.0000 0.0000 0.000 0.000 _50_ 0.000
33 50.0000 0.0000 0.000 0.000 5_0 -0.001
34 250.0000 0.4400 22.000 22.000 _50_ 0.000
35 1250.0000 0.8000 40.000 40.000 5J) -0.000
36
37 Chi^2 = 0.00 d.f. = I P-value = 0.9994
38
39 Benchmark Dose Computation
40 Specified effect = 0.1
41 Risk Type = Extra risk
42 Confidence level = 0.95
43 BMP = 206.74
44
45 Warning: BMDL computation is at best imprecise for these data
46 BMDL = 167.46
F-27
DRAFT - DO NOT CITE OR QUOTE
-------
I
Dichotomous-Hill Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
Dichotomous-Hill
BMDL BMD
200
400
10:5301/142011
600
dose
800
1000
1200
Figure F-9 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
to support the results in Table F-18.
F-28
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX G. RFC DERIVATION: ALTERNATIVE
APPROACH IN THE APPLICATION OF THE
DOSIMETRIC ADJUSTMENT FACTOR
1 For the derivation of a RfC based upon an animal study, the selected POD must be adjusted to
2 reflect the human equivalent concentration (HEC). and uncertainty factors (UFs) must be applied to
3 account for recognized uncertainties in the use of the available data. The HEC is calculated by the
4 application of the appropriate dosimetric adjustment factor (DAF). in accordance with the U.S. EPA RfC
5 methodology (U.S. EPA. 1994). DAFs are ratios of animal and human physiological parameters, and are
6 dependent on the nature of the contaminant (particle or gas) and the target site (e.g.. respiratory tract or
7 systemic) (U.S. EPA. 1994). UFs are used as appropriate and are an order of magnitude (10) or a reduced
8 order of magnitude (3 or 1). For the derivation of the RfC. the composite UFs are applied to the HEC.
9 1.4-Dioxane is miscible with water and has a high blood:air partition coefficient. Typically.
10 highly water-soluble and directly reactive chemicals (i.e. Category 1 gases) partition greatly into the
11 upper respiratory tract, induce portal-of-entry effects, and do not accumulate significantly in the blood.
12 1.4-Dioxane induces effects throughout the respiratory tract, liver, and kidneys; and has been measured in
13 the blood after inhalation exposure (Kasai et al.. 2008). The observations of systemic (nonrespiratory)
14 effects and measured blood levels resulting from 1.4-dioxane exposure clearly indicate that this
15 compound is absorbed into the bloodstream and distributed throughout the body. Furthermore, the lack of
16 an anterior to posterior gradient for the nasal effects induced by 1.4-dioxane is not typical of chemicals
17 which are predominantly directly reactive. Thus. 1.4-dioxane might be best described as a water-soluble
18 and non-directly reactive gas. Gases such as these are readily taken up into respiratory tract tissues and
19 can also diffuse into the blood capillaries (Medinsky and Bond. 2001). The effects in the olfactory
20 epithelium may be the result of the metabolism of 1.4-dioxane to an acid metabolite; however, for the
21 reasons stated above it is unclear whether or not these effects are solely the result of portal-of-entry or
22 systemic delivery. A similar pattern of effects were observed after oral exposure to 1.4-dioxane (JBRC.
23 1998: Kanoetal.. 2009) .
24 In consideration of all the evidence, the human equivalent concentration (HEC) for 1.4-dioxane
25 was calculated in this assessment by application of the appropriate dosimetric adjustment factor (DAF)
26 for systemic acting gases (i.e. Category 3 gases) to the POD for the co-critical effects (olfactory
27 epithelium atrophy and respiratory metaplasia), and adjusted for exposure duration (POp_ADj ^
28 32.2 mg/m3). However, since 1.4-dioxane is miscible with water and may induce portal-of-entry effects.
29 an alternative calculation of the HEC for 1.4-dioxane. based upon the application of a DAF for
30 portal-of-entry acting gases (i.e.. Category 1) was derived and is provided below in Section G.I.
31 Uncertainly factors applied in this assessment included factors of 10 for LOAEL-to-NOAEL
32 extrapolation. 10 for human interindividual variability. 3 for animal-to-human extrapolation, and 3 for
33 database deficiencies (See Section 5.2.4. for details).
G-l
DRAFT - DO NOT CITE OR QUOTE
-------
G.1 Application of DAF for Category 1 Gases
1 In accordance with the guidance for deriving inhalation RfCs (U.S. EPA. 1994). a DAF based on
2 the regional gas dose ratio (RGDR) for a gas with portal-of-entry respiratory effects (i.e., extrathoracic:
3 nasal region to the larynx) was derived by using: 1) a calculated ventilation rate (VE) of 0.254 L/minute
4 ,based on the average body weight of the male F344 rats reported in the principal study (Kasai et al..
5 2009); 2) a default VE value of 13.8 L/minute for humans; and 3) default extrathoracic region surface area
6 (SA) values of 15.0 cm2 for the rat and 200 cm2 for humans. The resulting equation is as follows:
VE(rat)/SA(rat)
RGDR = EV "
VE(human)/SA (human)
_ 0.254/15
= 13.8/200
7 = 0.25
8 Applying the RGDR of 0.25 to the POD for the co-critical effects, adjusted for exposure duration:
9 (PODAnr. 32.2 mg/m3) yields a HEC (TOD^) of 8.1 mg/m3 :
10 PODggr (mg/m3) = PODAm (mg/m3) x RGDR
11 = 32.2 mg/m3 x Q.25
12 =8.1 mg/m3
G.2 Application of Uncertainty Factors
13 A composite UF of 1.000 was determined for the derivation of the RfC. As stated above, the
14 composite UF of 1.000 includes factors of 10 for LOAEL-to-NOAEL extrapolation. 10 for human
15 interindividual variability. 3 for animal-to-human extrapolation, and 3 for database deficiencies.
16 Applying the composite UF of 1.000 to the HEC (POD^) of 8.1 mg/m3 yields an RfC of 0.008
17 or 8 x 1(T3 mg/m3.
18 RfC = PODHEc/UF
19 =8.1 mg/m3/I.OOP
20 = 0.008 or 8 x 1Q~3 mg/m3
G-2
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX H. DETAILS OF BMD ANALYSIS FOR
INHALATION UNIT RISK FOR 1,4-DIOXANE
1 Multistage cancer models available in the Benchmark Dose Software (BMDS) (version 2.2beta)
2 were fit to the incidence data for hepatocellular carcinoma and/or adenoma, nasal cavity squamous cell
3 carcinoma, renal cell carcinoma, peritoneal mesothelioma. and mammary gland fibroadenoma. Zymbal
4 gland adenoma, and subcutis Fibroma in rats exposed to 1.4-dioxane vapors for 2 years (Kasai et al..
5 2009). Concentrations associated with a benchmark response (BMR) of a 10% extra risk were calculated.
6 BMCio and BMCLjo values from the best fitting model, determined by adequate global- fit (y1 p > 0.1)
7 and AIC values, are reported for each endpoint (U.S. EPA. 2000a). Given the multiplicity of tumor sites.
8 basing the IUR on one tumor site will underestimate the carcinogenic potential of 1.4-dioxane. A
9 Bayesian analysis was performed using WinBUGS (Spiegelhalter et al.. 2003). freeware developed by the
10 MRC Biostatistical Unit. Cambridge. United Kingdom (available at
11 http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml) and reported in detail in Section H.3. In
12 addition, the combined tumor analysis was also performed using the beta version of the BMDS
13 MSCombo model (BMDS Version 2.2beta) and is included in Section H.4. The results of both analyses
14 were comparable.
15 A summary of the BMDS model predictions for the Kasai et al. (2009) study are shown in
16 Table H-l.
H.1 General Issues and Approaches to BMDS and Multitumor
Modeling
H.1.1 Combining Data tumor types
17 The incidence of adenomas and the incidence of carcinomas within a dose group at a site or tissue
18 in rodents are sometimes combined. This practice is based upon the hypothesis that adenomas may
19 develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a; McConnell et al..
20 1986). In the same manner and was done for the oral cancer assessment (Appendix D). the incidence of
21 hepatic adenomas and carcinomas was summed without double-counting them so as to calculate the
22 combined incidence of either a hepatic carcinoma or a hepatic adenoma in rodents.
23 The remaining of the tumor types were assumed to occur independently.
H.I.2 Summary
24 The BMDS models recommended to calculate rodent BMCio and BMCLjo values for individual
25 tumor types and combined tumor analysis are summarized in Table H-l. The first order multistage models
H-l
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
for most tumor types were selected because they resulted in the lowest AIC values; however, for renal cell
carcinoma and Zymbal aland adenoma, the lowest AIC
model was not the first order model. In
the third order model resulted in the lowest AIC (first (1°)-, second (2°)-, and third (3°)-degree
BMDS.
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
and Zymbal gland adenoma, the second order
(WinBugs) approach and the BMDS (Version
. Thus, for renal cell carcinoma
(2°) multistage model was used in both the MCMC
2.2 beta)
MSCombo approach for direct comparison of
results. These results are shown below in Table H-l.
Table H-1 Summary of BMC10 and BMCL10 model
combined tumor analysis for male rats
2009)
Multistage
Endpoint Model
Degree
Nasal squamous cell carcinoma First (1°)
Hepatocellular First (1°)
adenoma/carcinoma 1 — _
Renal cell carcinoma Third (3°)
Peritoneal mesothelioma First (1°)
Mammary gland fibroadenoma First (1°)
Zymbal gland adenoma Third (3°)
Subcutis fibroma3 First (1 °)
WinBUGS multitumor analysis"
BMDS Version 2.2beta MSCombo
AIC
49.03
127.9
29.99
155.4
86.29
29.99
89.2
results for individual turr
exposed to 1,4-dioxane \
D-value x2 Residual
of Interest
0.9607
0.6928
0.9984
0.8509
0.7904
0.9984
0.5245
0.176
-0.763
0.017
-0.204
-0.149
0.017
0.537
ior types and
rapors (Kasai et al.,
BMCIO(ppm) B,MCL1°
(ppm)
1107.04
252.80
1355.16
82.21
1635.46
1355.16
141.762
39.2
40.4
aHigh-dose dropped. See Section H.2.6 for details.
bln MCMC approach, the simulations for the four-parameter third order(3°) multistage model did not converge for renal
and Zymbal gland adenomas. Second order (2°) multistage model was used instead.
629.95
182.26
16.15
64.38
703.03
16.15
81.9117
31.4
30.3
cell carcinomas
H.2 BMDS Model Output for Multistage Cancer Models for Inidividual
Tumor Types
9 For tumor incidence data reported in the Kasai et al. (2009) 2-year inhalation bioassay. multistage
10 cancer models of first (1°)-. second (2°)-. and third (3°)degrees were implemented BMDS (Version
11 2.2Beta). Incidence data used for BMP analysis are shown in Table H-2^ Tumor incidence for mammary
12 gland adenoma was excluded from this analysis since only 1 tumor of this type was found across all
13 doses.
H-2
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-2 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane vapor by
whole-body inhalation for 2 years
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
1,4-dioxane vapor concentration (ppm)
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
1/50
3/50
1/50
4/50
0/50
14/50a
3/50
0/50
9/50a
1,250
6/50°'c
21/50a'c
2/50
22/50a'c
4/50c
41/50a'c
5/50a
4/50c
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
cp < 0.01 by Peto's test for dose-related trend.
dp < 0.05 by Peto's test for dose-related trend.
eProvided via personal communication from Dr. Tatsuya Kasai (2008) to Dr. Reeder Sams on 12/23/2008. Statistics
were not reported for these data by study authors, so statistical analyses were conducted by EPA.
Source: Kasai et al. (2009) and Kasai personal communication (2008)
H.2.1 Nasal Squamous Cell Carcinoma
1 The incidence data for nasal squamous cell carcinoma were monotonic non-decreasing functions
2 of dose: therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
3 BMDS modeling for the multistage cancer model for first (1°)-. second (2°)-. and third (3°)-degree
4 polynomials are shown in Table H-3i The first (l°)-degree polynomial was the best fitting model based on
5 AIC. The plot (Figure H-l) and model output for the first (l°)-degree model are shown below.
H-3
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-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
\2 Residual of
Interest
0.176
-0.021
-0.021
BMC-io
(ppm)
1,107.04
1,086.94
1,086.94
BMCLio
(ppm)
629.95
642.43
642.43
Best-fitting model based on AIC.
Multistage Cancer Model with 0.95 Confidence Level
I
C
o
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
BMDL
BMD
200
400
600
dose
800
1000
1200
10:26 11/172010
Figure H-l Multistage model (First (l°)-degree) for male rat nasal squamous cell
carcinomas.
1
2
3
4
5
6
7
8
Q ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~,
10 The form of the probability function is:
11
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.plt
Wed Nov r? 10:57:55 2010
BMDS Model Run
H-4
DRAFT - DO NOT CITE OR QUOTE
-------
1 P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]
2
3 The parameter betas are restricted to be positive
4
5 Dependent variable = EFFECT
6 Independent variable = DOSE
7
8 Total number of observations = £
9 Total number of records with missing values = 0_
10 Total number of parameters in model = 2_
11 Total number of specified parameters = 0_
12 Degree of polynomial = \_
13
14 Maximum number of iterations = 250
15 Relative Function Convergence has been set to: le-008
16 Parameter Convergence has been set to: le-008
17
18
19 Default Initial Parameter Values
20 Background = 0_
21 Beta(l) = 0.000104666
22
23 Asymptotic Correlation Matrix of Parameter Estimates
24 (***The model parameter(s) -Background have been estimated at a boundary point, or
25 have been specified by the user, and do not appear in the correlation matrix _)_
26
27 Beta(l)
28 Beta(l) I
29
30 Parameter Estimates
31 95.0% Wald Confidence Interval
32 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
33 Background 0_ ^_ ^_ ^_
34 Beta(l) 9.51733e-005 ^ ^ ^
35
36 i z Indicates that this value is not calculated.
37
38 Analysis of Deviance Table
39
40 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
41 Full model -23.2482 £
42 Fitted model -23.5154 I 0.534383 3 0.9113
43 Reduced model -30.3429 I 14.1894 3_ 0.002658
44
45 AIC: 49.0308
46
47 Log-likelihood Constant 20.493267595834471
48
49
50 Goodness of Fit
51 Scaled
52 Dose Est. Prob. Expected Observed Size Residual
53
54 0.0000 0.0000 0.000 C> _50_ 0.000
55 50.0000 0.0047 0.237 0 50 -0.
56 250.0000 0.0235 1.176 I _50_ -0.164
57 1,250.0000 0.1122 5.608 _6 _50_ 0.176
58
59 Chi^2 = 0.30 d.f. = 3^ P-value = 0.9607
60
61
H-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 Benchmark Dose Computation
2
3 Specified effect = 0.1
4 Risk Type = Extra risk
5 Confidence level = 0.95
6 BMP = 1107.04
7 BMDL = 629.948
8 BMDU = 2215.11
9
10 Taken together, (629.948, 2215.11) ijs a 90% two-sided confidence interval for the BMP
H.2.2 Hepatocellular Adenoma and Carcinoma
11 The incidence data for the occurrence of either hepatocellular adenoma or carcinoma were
12 combined for this analysis as explained in H. 1.1. The incidence data were monotonic non-decreasing
13 functions of dose: therefore, these data are appropriate for dose-response modeling using BMDS. The
14 results of the BMDS modeling for the multistage cancer model for first-, second-, and third-degree
15 polynomials are shown in Table H-4. The lst-degree polynomial was the best fitting model based on AIC.
16 The plot (Figure H-2) and model output for the lst-degree model are shown below.
H-6
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-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
X^ 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
"Best-fitting model based on AIC.
I
c
o
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
200
400
600
dose
800
1000
1200
10:24 11/172010
Figure H-2 Multistage model (First-degree (1°)) for male rat hepatocellular
adenomas and carcinomas.
1 MS COMBO. (Version: 1.4; Date: 10/20/2010)
2 Input Data File: C:\Documents and
3 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
4 Gnuplot Plotting File: C:\Documents and
5 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
6 Wed Nov r? 10:57:55 2010
7
8 BMDS Model Run
Q ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10 The form of the probability function is:
11 P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]
H-7
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 The parameter betas are restricted to be positive
3
4 Dependent variable = EFFECT
5 Independent variable = DOSE
6
7 Total number of observations = £
8 Total number of records with missing values = 0_
9 Total number of parameters in model = 2_
10 Total number of specified parameters = £
11 Degree of polynomial = I_
12
13 Maximum number of iterations = 250
14 Relative Function Convergence has been set to: le-008
15 Parameter Convergence has been set to: le-008
16
17 Default Initial Parameter Values
18 Background = 0.00480969
19 Beta(l) = 0.0004548
20
21 Asymptotic Correlation Matrix of Parameter Estimates
22
23 Background Beta(1)
24 Background I -0.53
25 Beta(l) -0.53 I
26
27 Parameter Estimates
28
29 95.0% Wald Confidence Interval
30 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
31 Background 0.0176~678 * * *
32 Beta(l) 0.000416776 111
33
34 1 z Indicates that this value is not calculated.
35
36 Analysis of Deviance Table
37
38 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
39 Full model -61.5341 4
40 Fitted model -61.9302 2 0.792109 2 0.673
41 Reduced model -82.7874 I 42.5066 3 <.OOQ1
42
43 AIC: 127.86
44
45 Log-likelihood Constant 55.486699676972215
46
47 Goodness of Fit
48 Scaled
49 Dose Est. Prob. Expected Observed Size Residual
50
51 0.0000 0.0171 0.853 I _50_ 0.160
52 50.0000 0.0373 1.867 2 5J) 0.099
53 250.0000 0.1143 5.716 £ _50_ -0.763
54 1,250.0000 0.4162 20.810 22_ _50_ 0.342
55 Chi^2 = 0.73 d.f. = 2_ P-value = 0. 6928
56
57
58 Benchmark Dose Computation
59
60 Specified effect = 0.1
61 Risk Type = Extra risk
62 Confidence level = 0.95
63 BMP = 252.799
64 BMDL = 182.256
65 BMDU = 371.457
66
67 Taken together, (182.256, 371.457) ijs a 90% two-sided confidence interval for the BMP
H-
DRAFT - DO NOT CITE OR QUOTE
-------
H.2.3 Renal Cell Carcinoma and Zymbal Gland Adenoma
1 The incidence data for renal cell carcinomas and Zymbal gland adenomas were the same. These
2 data were monotonic non-decreasing functions of dose: therefore, these data are appropriate for
3 dose-response modeling using BMDS. The results of the BMDS modeling for the multistage cancer
4 model for first (1°)-. second (2°)- and third-degree (3°) polynomials are shown in Table H-5^ The
5 third-degree (3°)polynomial was the best fitting model based on AIC; however, when conducting the
6 multitumor analysis. WinBUGS was unable to converge using the third-degree (3°) model. Thus, the
7 second degree (2°) model was used in the multitumor analyses. The plots (Figure H-3 and Figure H-4)
8 and model outputs for both the second (2°)- and third-degree (3°) models are shown below.
H-9
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-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)
Polynomial Degree
(1°) First
(2°) Second
(3°) Third3
AIC
31.6629
30.2165
29.9439
p-value
0.8004
0.9817
0.9984
X^ Residual of
Interest
0.446
0.085
0.017
BMCio
(ppm)
1,974.78
1,435.28
1,355.16
BMCLio
(ppm)
957.63
999.44
1,016.15
Best-fitting model based on AIC.
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
BMD
200
400
600 800
dose
1000
1200
1400
10:1702/102011
Figure H-3 Multistage model (Second-degree (2°)) for male rat renal cell
carcinomas and Zymbal gland adenomas.
4
5
6
7
8
9
10
11
1 Multistage Cancer Model. (Version: 1.9; Date: 05/26/2010)
2 Input Data File: C:/Documents and
3
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.pit
Thu Feb K> 10:17:39 2011
BMDS Model Run
The form of the probability function is:
H-10
DRAFT - DO NOT CITE OR QUOTE
-------
1 P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2) ]
2
3 The parameter betas are restricted to be positive
4
5 Dependent variable = EFFECT
6 Independent variable = DOSE
7
8 Total number of observations = £
9 Total number of records with missing values = 0_
10 Total number of parameters in model = 3^
11 Total number of specified parameters = 0_
12 Degree of polynomial = 2_
13
14 Maximum number of iterations = 250
15 Relative Function Convergence has been set to: le-008
16 Parameter Convergence has been set to: le-008
17
18 Default Initial Parameter Values
19 Background = Q_
20 Beta(l) = 0
21 Beta(2) = 5.40386e-008
22
23 Asymptotic Correlation Matrix of Parameter Estimates
24 (*** The model parameter(s) -Background -Beta (1) have been estimated at a boundary
25 point, or have been specified by the user, and do not appear in the correlation
26 matrix)
27
28 Beta(2)
29 Beta(2) I
30
31 Parameter Estimates
32 95.0% Wald Confidence Interval
33 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
34 Background 0_ ^_ ^_ ^_
35 Beta(l) C> 1 1 1 ~
36 Beta(2) 5.11454e-008 111
37
38 1 z Indicates that this value is not calculated.
39
40 Analysis of Deviance Table
41
42 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
43 Full model -13.9385 4
44 Fitted model -14.1082 I 0.339554 3 0.9524
45 Reduced model -19.6078 I 11.3387 3 0.01003
46
47 AIC: 30.2165
48
49 Goodness of Fit
50 Scaled
51 Dose Est. Prob. Expected Observed Size Residual
52
53 0.0000 0.0000 0.000 0.000 _50_ 0.000
54 50.0000 0.0001 0.006 0.000 5_0 -0.080
55 250.0000 0.0032 0.160 0.000 _50_ -0.400
56 1250.0000 0.0768 3.840 4.000 5J) 0.085
57
58 Chi^2 = 0.17 d.f. = 3^ P-value = 0.9817
59
H-ll
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMP = 17135.28
BMDL = 999.44
BMDU = 3,666.87
Taken together, (999.44 _,_ 3, 666.87) i_s a. 90% two-sided confidence interval for the BMP
Multistage Cancer Slope Factor = 0.000100056
T3
-------
1
2 Dependent variable = EFFECT
3 Independent variable = DOSE
4
5 Total number of observations = £
6 Total number of records with missing values = 0_
7 Total number of parameters in model = _4
8 Total number of specified parameters = 0_
9 Degree of polynomial = 3_
10
11 Maximum number of iterations = 250
12 Relative Function Convergence has been set to: le-008
13 Parameter Convergence has been set to: le-008
14
15 Default Initial Parameter Values
16 Background = 0_
17 Beta(l) = C>
18 Beta(2) = £
19 Beta(3) = 4.2804e-011
20
21
22 Asymptotic Correlation Matrix of Parameter Estimates
23 (*** The model parameter(s) -Background -Beta(1) -Beta(2) have been estimated at a.
24 boundary point, or have been specified by the user, and do not appear in the
25 correlation matrix)
26
27 Beta(3)
28 Beta(3) I
29
30 Parameter Estimates
31
32 95.0% Wald Confidence Interval
33 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
34 Background 0_ ^_ ^_ ^_
35 Beta(l) C> 1 1 1 ~
36 Beta(2) £ 1 1 1
37 Beta(3) 4.23353e-011 ^ ^ ^
38
39 ^_ 2_ Indicates that this value is not calculated.
40
41 Analysis of Deviance Table
42
43 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
44 Full model -13.9385 4
45 Fitted model -13.9719 I 0.0669578 3 0.9955
46 Reduced model -19.6078 I 11.3387 3 0.01003
47
48 AIC: 29.9439
49
50 Log-likelihood Constant 12.347138085809094
51
52
53 Goodness of Fit
54 Scaled
55 Dose Est. Prob. Expected Observed Size Residual
56
57 0.0000 0.0000 0.000 0 _50_ 0.000
58 50.0000 0.0000 0.000 0_ 5_0 -0.016
59 250.0000 0.0007 0.033 £ _50_ -0.182
60 1250.0000 0.0794 3.968 4_ 5J) 0.017
61
62 Chi^2 = 0.03 d.f. = 3^ P-value = 0.9984
63
64
65 Benchmark Dose Computation
66 Specified effect = 0.1
67 Risk Type = Extra risk
H-13
DRAFT - DO NOT CITE OR QUOTE
-------
1 Confidence level = 0.95
2 BMP = 1,355.16
3 BMDL = 1,016.15
4 BMDU = 3,393.6
5
6 Taken together, (1016.15, 3393. 6 )_ is_ a 90% two-sided confidence interval for the BMP
H.2.4 Peritoneal Mesothelioma
7 The incidence data for peritoneal mesotheliomas were monotonic non-decreasing functions of
8 dose; therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
9 BMDS modeling for the multistage cancer model for 1st. 2nd. and 3rd-degree polynomials are shown in
10 Table H-6^ The lst-degree polynomial was the best fitting model based on AIC. The plot (Figure H-5] and
11 model output for the lst-degree model are shown below.
H-14
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-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)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
155.433
157.168
157.168
p-value
0.8509
0.8053
0.8053
X^ Residual of
Interest
-0.204
-0.204
0
BMCio
(ppm)
82.21
96.23
96.23
BMCLio
(ppm)
64.38
65.15
65.15
a Best-fitting model based on AIC.
Multistage Cancer Model with 0.95 Confidence Level
I
C
o
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
Figure H-5 Multistage model (First-degree (1°)) for male rat peritoneal
mesotheliomas.
1 MS COMBO. (Version: 1.4; Date: 10/20/2010)
2 Input Data File: C:\Documents and
3 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
4 Gnuplot Plotting File: C:\Documents and
5 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
6 Wed Nov r? 10:57:55 2010
7
8 BMDS Model Run
Q ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10 The form of the probability function is:
11 P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]
H-15
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 The parameter betas are restricted to be positive
3
4 Dependent variable = EFFECT
5 Independent variable = DOSE
6
7 Total number of observations = £
8 Total number of records with missing values = 0_
9 Total number of parameters in model = 2_
10 Total number of specified parameters = £
11 Degree of polynomial = I_
12 Maximum number of iterations = 250
13 Relative Function Convergence has been set to: le-008
14 Parameter Convergence has been set to: le-008
15
16 Default Initial Parameter Values
17 Background = 0.0172414
18 Beta(l) = 0.00135351
19
20 Asymptotic Correlation Matrix of Parameter Estimates
21
22 Background Beta(1)
23 Background I -0.45
24 Beta(l) -0.45 I
25
26 Parameter Estimates
27 95.0% Wald Confidence Interval
28 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
29 Background 0.033631 ^ ^ ^
30 Beta(l) 0.00128167 ^ ^ ^
31
32 i z Indicates that this value is not calculated.
33
34 Analysis of Deviance Table
35
36 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
37 Full model -75.553 £
38 Fitted model -75.7165 2_ 0.326905 2_ 0.8492
39 Reduced model -123.008 I 94.9105 3 <.OOQ1
40
41 AIC: 155.433
42
43 Log-likelihood Constant 68.666413125908832
44
45 Goodness of Fit
46 Scaled
47 Dose Est. Prob. Expected Observed Size Residual
48
49 0.0000 0.0336 1.682 2 _50_ 0.250
50 50.0000 0.0936 4.681 £ 5_0 -0.331
51 250.0000 0.2986 14.928 I4_ _50_ -0.287
52 1,250.0000 0.8053 40.265 £1 _50_ 0.263
53
54 Chi^2 = 0.32 d.f. = 2_ P-value = 0.8509
55
56 Benchmark Dose Computation
57 Specified effect = 0.1
58 Risk Type = Extra risk
59 Confidence level = 0.95
60 BMP = 82.2057
61 BMDL = 64.3808
62 BMDU = 107.497
63
64 Taken together, (64.3808, 107.497) ijs a 90% two-sided confidence interval for the BMP
H-16
DRAFT - DO NOT CITE OR QUOTE
-------
H.2.5 Mammary Gland Fibroadenoma
1 The incidence data for mammary gland fibroadenomas were monotonic non-decreasing functions
2 of dose: therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
3 BMDS modeling for the multistage cancer model for first (1°)-. second (2°). and third (3°)-degree
4 polynomials are shown in Table H-7. Since quadratic and cubic terms of the multistage models evaluated
5 resulted in the estimates on the boundary, i.e. equal to 0. the first (l°)-degree polynomial was selected
6 based on model parsimony. The plot (Figure H-6) and model output for the first d°)-degree model are
7 shown below.
H-17
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-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
X^ Residual of
Interest
-0.149
-0.149
-0.149
BMCio
(ppm)
1,635.46
1,635.46
1,635.46
BMCLio
(ppm)
703.03
703.03
703.03
aAII model fits were equivalent based on AIC. Selected 1 -degree model based on parsimony.
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
200
400
600
800
dose
1000 1200 1400 1600
10:34 11/172010
Figure H-6 Multistage model (First-degree (1°)) for male rat mammary gland
fibroadenoma.
1 MS COMBO. (Version: 1.4; Date: 10/20/2010)
2 Input Data File: C:\Documents and
3 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
4 Gnuplot Plotting File: C:\Documents and
5 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
6 Wed Nov r? 10:57:55 2010
7
8 BMDS Model Run
Q ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10 The form of the probability function is:
11 P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]
H-18
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 The parameter betas are restricted to be positive
3
4 Dependent variable = EFFECT
5 Independent variable = DOSE
6
7 Total number of observations = £
8 Total number of records with missing values = 0_
9 Total number of parameters in model = 2_
10 Total number of specified parameters = £
11 Degree of polynomial = I_
12
13 Maximum number of iterations = 250
14 Relative Function Convergence has been set to: le-008
15 Parameter Convergence has been set to: le-008
16
17 Default Initial Parameter Values
18 Background = 0.0335609
19 Beta(l) = 5.91694e-005
20
21 Asymptotic Correlation Matrix of Parameter Estimates
22
23 Background Beta(1)
24 Background I -0. 61
25 Beta(l) -0.61 I
26
27 Parameter Estimates
28
29 95.0% Wald Confidence Interval
30 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
31 Background Q.03158'36 * * *
32 Beta(l) 6.44224e-005 111
33
34 1 z Indicates that this value is not calculated.
35
36 Analysis of Deviance Table
37
38 Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
39 Full model -40.9017 4
40 Fitted model -41.145 2 0.486662 2 0.784
41 Reduced model -42.5964 I 3.3895 3 0.3354
42
43 AIC: 86.29
44
45 Log-likelihood Constant 35.472345543489602
46
47 Goodness of Fit
48 Scaled
49 Dose Est. Prob. Expected Observed Size Residual
50
51 0.0000 0.0316 1.579 I _50_ -0.468
52 50.0000 0.0347 1.735 2 5J) 0.205
53 250.0000 0.0471 2.353 3 _50_ 0.432
54 1,250.0000 0.1065 5.326 _5 _50_ -0.149
55
56 Chi"2 = 0.47 d.f. = 2 P-value = 0.7904
57
58 Benchmark Dose Computation
59 Specified effect = 0.1
60 Risk Type = Extra risk
61 Confidence level = 0.95
62 BMP = 1,635.46
63 BMDL = 703.034
64 BMDU = 1.9523e+009
65
66 Taken together, (703.034, 1.9523e+009) is^ a 90% two-sided confidence interval for the
67 BMP
H-19
DRAFT - DO NOT CITE OR QUOTE
-------
H.2.6 Subcutis Fibroma
1 The incidence data for subcutis fibroma were monotonic non-decreasing functions of dose for the
2 control (0 ppm). low (50 ppm). and mid-dose (250 ppm); however, the incidence rate at the high dose
3 (1.250 ppm) was lower than observed at the mid-dose. No BMDS model had reasonable fit to the data
4 without dropping the high dose. The results of the BMDS modeling for the multistage cancer model for
5 first (1°)-. second (2°). and third (3°)-degree polynomials with the high dose dropped are shown in
6 Table H-8. Since quadratic and cubic terms of multistage models evaluated resulted in the estimates on
7 the boundary, i.e. equal to 0. , the first (l°)-degree polynomial was selected based on model parsimony.
8 The plot (Figure H-7) and model output for the first (l°)-degree model are shown below.
H-20
DRAFT - DO NOT CITE OR QUOTE
-------
Table H-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
X^ 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 1 -degree model based on parsimony.
I
c
o
0.35
0.3
0.25
0.2
0.15
0.1
0.05
10:56 11/172010
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
50
100 150
dose
200
250
Figure H-7 Multistage model (First-degree (1°)) for male rat subcutis fibroma (high
dose dropped).
1 MS COMBO. (Version: 1.4; Date: 10/20/2010)
2 Input Data File: C:\Documents and
3 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
4 Gnuplot Plotting File: C:\Documents and
5 Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
6 Wed Nov r? 10:57:55 2010
7
8 BMDS Model Run
Q ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10 The form of the probability function is:
11 P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]
H-21
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 The parameter betas are restricted to be positive
3
4 Dependent variable = EFFECT
5 Independent variable = DOSE
6
7 Total number of observations = 3_
8 Total number of records with missing values = 0_
9 Total number of parameters in model = 2_
10 Total number of specified parameters = £
11 Degree of polynomial = I_
12
13 Maximum number of iterations = 250
14 Relative Function Convergence has been set to: le-008
15 Parameter Convergence has been set to: le-008
16
17 Default Initial Parameter Values
18 Background = 0.0327631
19 Beta(l) = 0.000673665
20
21
22 Asymptotic Correlation Matrix of Parameter Estimates
23
24 Background Beta(1)
25 Background I -0 . 68
26 Beta(l) -0.68 I
27
28 Parameter Estimates
29
30 95.0% Wald Confidence Interval
31 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
32 Background 0.0262054 111
33 Beta(l) 0.00074322 111
34
35 1 2. Indicates that this value is not calculated.
36
37 Analysis of Deviance Table
38
39 Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
40 Full model -42.4101 3
41 Fitted model -42.6047 2 0.389155 I 0.5327
42 Reduced model -46.5274 I 8.23466 2_ 0.01629
43
44 AIC: 89.2094
45
46 Log-likelihood Constant 37.900888781466982
47
48 Goodness of Fit
49 Scaled
50 Dose Est. Prob. Expected Observed Size Residual
51
52 0.0000 0.0262 1.310 I _50_ -0.275
53 50.0000 0.0617 3.086 £ 5£ 0.537
54 250.0000 0.1913 9.566 9 _50_ -0.204
55 Chi^2 = 0.41 d.f. = I P-value = 0.5245
56
57
58 Benchmark Dose Computation
59 Specified effect = 0.1
60 Risk Type = Extra risk
61 Confidence level = 0.95
62 BMP = 141.762
63 BMDL = 81.9117
64 BMDU = 364.364
65
66 Taken together, (81.9117, 364.364) i_£ a 90% two-sided confidence interval for the BMP
H-22
DRAFT - DO NOT CITE OR QUOTE
-------
H.2.7 Multitumor analysis using Bayesian Methods
1 Given the multiplicity of tumor sites, basing the IUR on one tumor site will likely underestimate
2 the carcinogenic potential of 1.4-dioxane. Simply pooling the counts of animals with one or more tumors
3 (i.e.. counts of tumor bearing animals) would tend to underestimate the overall risk when tumors are
4 independent across sites and ignores potential differences in the dose-response relationships across the
5 sites (NRC. 1994; Bogen. 1990). NRC (1994) also noted that the assumption of independence across
6 tumor types is not likely to produce substantial error in the risk estimates unless tumors are known to be
7 biologically dependent.
8 Kopylev et al. (2009) describe a Markov Chain Monte Caro (MCMC) computational approach to
9 calculating the dose associated with a specified composite risk under assumption of independence of
10 tumors. The current Guidelines for Carcinogen Risk Assessment recommend calculation of an upper
11 bound to account for uncertainty in the estimate (U.S. EPA. 2005a). For uncertainty characterization.
12 MCMC methods have the advantage of providing information about the full distribution of risk and/or
13 benchmark dose, which can be used in generating a confidence bound. This MCMC approach building on
14 the re-sampling approach recommended by Bogen (1990). and also provides a distribution of the
15 combined potency across sites.
16 For individual tumor data modeled using the multistage model:
17 P(d \q) = 1 - exp[-(qn + qLd + q.jf + ... +
18 the model for the combined tumor risk is still multistage, with a functional form that has the sum of
19 stage-specific multistage coefficients as the corresponding multistage coefficient;
20
21 The resulting equation for fixed extra risk (BMR) is polynomial in dose (when logarithms of both
22 sides are taken) and can be straightforwardly solved for a combined BMC. Computation of the confidence
23 bound on combined risk BMC can be accomplished via likelihood methods (BMDS-MSCOMBO).
24 re-sampling (bootstrap) or Bayesian methods.
25 The MCMC computations were conducted using WinBUGS (Spiegelhalter et al., 2003)(freeware
26 developed by the MRC Biostatistical Unit. Cambridge. United Kingdom, available at
27 http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml).
28 In a Bavesian analysis, the choice of the appropriate prior is important. In the examples
29 developed by Kopylev et al. (2009). a diffuse (i.e.. high variance or low tolerance) Gaussian prior
30 restricted to be nonnegative was used; such diffuse priors performed reasonably well.
31 The mean and the 5th percentile of the posterior distribution of combined BMC provide estimates
32 of the mean BMC and the lower bound on the BMC (BMCL). respectively, for the combined tumor risk.
33 The values calculated using this method were: mean BMCm 39.2ppm. and BMCLin 31.4.
H-23
DRAFT - DO NOT CITE OR QUOTE
-------
H.3 Multitumor Analysis Using BMDS MSCOMBO (BETA)
1 The combined tumor analysis was also performed with beta version of the MSCombo model in
2 BMDS (Version 2.2beta). The model resulted in similar results to the Bayesian method and model output
3 is shown below for the combined calculation.
4
5 **** start of combined BMD and BMDL Calculations.****
6 Combined Log-Likelihood -277.79874987953076
7 Combined Log-likelihood Constant 246.62591390071873
8
9
10 Benchmark Dose Computation
11 Specified effect = 0.1
12 Risk Type = Extra risk
13 Confidence level = 0.95
14 BMD = 40.4937
15 BMDL = 32.331
H-24
DRAFT - DO NOT CITE OR QUOTE
------- |