DRAFT - DO NOT CITE OR QUOTE
EPA/635/R-11/003D
United States www.epa.gov/iris
Environmental Protection
lAgency
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)
June 2013
NOTICE
This document is a Final Interagency Science Discussion/Agency Review draft. This information is
distributed solely for the purpose of pre-dissemination peer review under applicable information quality
guidelines. It has not been formally disseminated by EPA. It does not represent and should not be
construed to represent any Agency determination or policy. It is being circulated for review of its
technical accuracy and science policy implications.
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document is a preliminary draft for review purposes only. This
information is distributed solely for the purpose of pre-dissemination peer
review under applicable information quality guidelines. It has not been
formally disseminated by EPA. It does not represent and should not be
construed to represent any Agency determination or policy. Mention of trade
names or commercial products does not constitute endorsement or
recommendation for use.
DRAFT - DO NOT CITE OR QUOTE
-------
CONTENTS: TOXICOLOGICAL REVIEW OF
1,4-DIOXANE (CAS NO. 123-91-1)
LIST OF ABBREVIATIONS AND ACRONYMS,
FOREWORD
.XIII
XV
AUTHORS, CONTRIBUTORS, AND REVIEWERS.
1 INTRODUCTION
.XVI
1
CHEMICAL AND PHYSICAL INFORMATION
Figure 2-1
Table 2-1
TOXICOKINETICS
1,4-Dioxane chemical structure.
Physical properties and chemical identity of 1,4-dioxane_
3.6 Rat Nasal Exposure via Drinking Water _
HAZARD IDENTIFICATION
4.1 Studies in Humans - Epidemiology, Case Reports, Clinical Controls
4.1.1 Thiessetal.
4.1.2 Buffleretal.
4.2 Subchronic and Chronic Studies and Cancer Bioassays in Animals - Oral and
Inhalation
4.2.1 Oral Toxicity
Table 4-1
Table 4-2
Table 4-3
Table 4-4
Table 4-5
Table 4-6
Table 4-7
Table 4-8
Table 4-9
Table 4-10
Incidence of histopathological lesions in Crj:BDF1 mice exposed to
1,4-dioxane in drinking water for 13 weeks
Incidence of nasal cavity squamous cell carcinoma and liver
hepatocellular adenoma in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water
Incidence of hepatocellular adenoma or carcinoma in B6C3F-I mice
exposed to 1,4-dioxane in drinking water
Incidence of histopathological lesions in male F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years_
Incidence of histopathological lesions in female F344/DuCrj rats
exposed to 1,4-dioxane in drinking water for 2 years_
Incidence of nasal cavity, peritoneum, and mammary gland tumors in
F344/DuCrj rats exposed to 1,4-dioxane in drinking water for 2 years_
3.1 Absorption
3.2 Distribution
3.3 Metabolism
Figure 3-1 Suggested metabolic pathways of 1,4-dioxane in the rat.
Figure 3-2 Plasma 1,4-dioxane levels in rats following i.v. doses of
3-5,600 mg/kg
3.4 Elimination
3.5 Physiologically Based Pharmacokinetic Models
Figure 3-3 General PBPK model structure.
3.5.1 Available Pharmacokinetic Data
3.5.2 Published PBPK Models for 1,4-Dioxane
3.5.3 Implementation of Published PBPK Models for 1 ,4-Dioxane
6
7
8
9
10
11
12
13
13
15
19
Incidence of histopathological lesions in F344/DuCrj rats exposed to
1,4-dioxane in drinking water for 13 weeks
_22
_23
_23
_24
_25
_26
_27
_30
31
Number of incipient liver tumors and hepatomas in male
Sprague-Dawley rats exposed to 1,4-dioxane in drinking water for
13 months 34
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 36
Incidence of nonneoplastic lesions in Osborne-Mendel rats exposed
to 1,4-dioxane in drinking water_ 38
39
40
44
44
46
in
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-11 Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane in
drinking water for 2 years 46
Table 4-12 Incidence of histopathological lesions in male Crj:BDF1 mice exposed
to 1,4-dioxane in drinking water for 2 years 48
Table 4-13 Incidence of histopathological lesions in female Crj:BDF1 mice
exposed to 1,4-dioxane in drinking water for 2 years 48
Table 4-14 Incidence of tumors in Crj:BDF1 mice exposed to 1,4-dioxane in
drinking water for 2 years 49
4.2.2 Inhalation Toxicity 50
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 52
Table 4-16 Hematology and clinical chemistry of F344/DuCrj rats exposed to
1,4-dioxane vapor by whole-body inhalation for 13 weeks 53
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 54
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 58
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 58
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 59
Table 4-21 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 2 years 60
4.2.3 Initiation/Promotion Studies 60
4.3 Reproductive/Developmental Studies—Oral and Inhalation 62
4.3.1 Giavini et al. 62
4.4 Other Duration or Endpoint Specific Studies 63
4.4.1 Acute and Short-term Toxicity 63
Table 4-22 Acute and short-term toxicity studies of 1,4-dioxane 64
4.4.2 Neurotoxicity 65
4.5 Mechanistic Data and Other Studies in Support of the Mode of Action 67
4.5.1 Genotoxicity 67
Table 4-23 Genotoxicity studies of 1,4-dioxane; in vitro 71
Table 4-24 Genotoxicity studies of 1,4-dioxane; mammalian in vivo 73
4.5.2 Mechanistic Studies 74
4.6 Synthesis of Major Noncancer Effects 77
4.6.1 Oral 78
Table 4-25 Oral toxicity studies (noncancer effects) for 1,4-dioxane 79
4.6.2 Inhalation 81
Table 4-26 Inhalation toxicity studies (noncancer effects) for 1,4-dioxane 83
4.7 Evaluation of Carcinogenicity 85
4.7.1 Summary of Overall Weight of Evidence 85
4.7.2 Synthesis of Human, Animal, and Other Supporting Evidence 86
4.7.3 Mode of Action Information 88
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 90
Table 4-27 Temporal sequence and dose-response relationship for possible key
events and liver tumors in rats and mice 93
Table 4-28 Temporal sequence and dose-response relationship for possible key
events and nasal tumors in rats and mice 95
4.8 Susceptible Populations and Life Stages 100
DOSE-RESPONSE ASSESSMENTS 101
5.1 Oral Reference Dose (RfD) 101
5.1.1 Choice of Principal Studies and Critical Effect with Rationale and Justification 101
5.1.2 Methods of Analysis—Including Models (PBPK, BMD, etc.) 102
Table 5-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years 103
Table 5-2 BMD and BMDL values derived from BMD modeling of the incidence
of cortical tubule degeneration in male and female Osborne-Mendel
rats exposed to 1,4-dioxane in drinking water for 2 years 103
5.1.3 RfD Derivation - Including Application of Uncertainty Factors (UFs) 103
5.1.4 RfD Comparison Information 104
IV
DRAFT - DO NOT CITE OR QUOTE
-------
Figure 5-1 Potential points of departure (POD) based on liver toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane. 106
Figure 5-2 Potential points of departure (POD) based on kidney toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane. 106
Figure 5-3 Potential points of departure (POD) based on nasal inflammation with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane. 107
Figure 5-4 Potential points of departure (POD) based on organ-specific toxicity
endpoints with corresponding applied uncertainty factors and derived
candidate RfDs following chronic oral exposure to 1,4-dioxane. 107
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 5-3 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 5-4 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) 114
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 candidate RfCs
following chronic inhalation exposure of F344 male rats 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
Table 5-5 Incidence of liver, nasal cavity, peritoneal, and mammary gland
tumors in rats and mice exposed to 1,4-dioxane in drinking water for
2 years (based on survival to 12 months) 119
Table 5-6 Incidence of liver, nasal cavity, kidney, peritoneal, and mammary
gland, Zymbal gland, and subcutis tumors in rats exposed to
1,4-dioxane via inhalation for 2 years. 121
5.4.2 Dose-Response Data 121
Table 5-7 Incidence of hepatocellular adenomas or carcinomas in rats and mice
exposed to 1,4-dioxane in drinking water for 2 years 122
Table 5-8 Incidence of tumors in F344 male rats exposed to 1,4-dioxane via
inhalation for 104 weeks (6 hours/day, 5 days/week) 123
5.4.3 Dose Adjustments and Extrapolation Method(s) 123
Table 5-9 Calculated HEDs for the tumor incidence data used for
dose-response modeling 124
5.4.4 Oral Slope Factor and Inhalation Unit Risk 127
Table 5-10 BMD HED and BMDI_HED values from best-fit models fit to tumor
incidence data for rats and mice exposed to 1,4-dioxane in drinking
water for 2 years and corresponding oral CSFs 128
Table 5-11 Dose-response modeling summary results for male rat tumors
associated with inhalation exposure to 1,4-dioxane for 2 years 130
5.4.5 Previous Cancer Assessment 131
5.5 Uncertainties in Cancer Risk Values 131
5.5.1 Sources of Uncertainty 131
Table 5-12 Summary of uncertainty in the 1,4-dioxane cancer risk estimation 136
MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE 137
6.1 Human Hazard Potential 137
6.2 DOSE RESPONSE 138
6.2.1 Noncancer/Oral 138
6.2.2 Noncancer/lnhalation 139
6.2.3 Cancer 139
DRAFT - DO NOT CITE OR QUOTE
-------
REFERENCES 144
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC COMMENTS
AND DISPOSITION A-1
A.1 External Peer Review Panel Comments - Oral Assessment A-1
A.1.1 General Charge Questions A-1
A. 1.2 Oral reference dose (RfD) for 1,4-dioxane A-5
A.1.3 Carcinogenicity of 1,4-dioxane and derivation of an oral slope factor A-10
A.2 Public Comments - Oral Assessment 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-17
A.2.4 Other Comments A-18
A.3 External Peer Review Panel Comments - Inhalation Update A-18
A.3.1 General Charge Questions A-19
A.3.2 Inhalation reference concentration (RfC) for 1,4-dioxane A-21
A.3.3 Carcinogenicity of 1,4-dioxane and derivation of an inhalation unit risk A-25
A.4 Public Comments - Inhalation Update A-30
A.4.1 Inhalation reference concentration (RfC) for 1,4-dioxane A-30
A.4.2 Carcinogenicity of 1,4-dioxane A-31
A.4.3 PBPK modeling A-34
A.4.4 Other comments A-36
APPENDIX B. EVALUATION OF EXISTING PHARMACOKINETIC MODELS FOR
1,4-DIOXANE 1
B.1 Background
B.2 Implementation of the Empirical Models in acsIX
B.2.1 Model Descriptions
Figure B-1 Schematic representation of empirical model for 1,4-dioxane in rats.
Figure B-2 Schematic representation of empirical model for 1 ,4-dioxane in
humans.
B.2. 2 Modifications to the Empirical Models
B.2. 3 Results
B-1
B-2
B-2
B-2
B-3
B-3
B-4
Figure B-3 Output of 1,4-dioxane blood level data from the acsIX implementation
(left) and published (right) empirical rat model simulations of i.v.
administration experiments. B-4
Figure B-4 Output of HEAA urine level data from acslXtreme implementation of
the empirical rat model (left) and published (right)data following i.v.
administration experiments. B-5
Figure B-5 acsIX empirical rat model predictions of blood 1,4-dioxane
concentration and total amount of HEAA levels in the urine for a
6-hour, 50-ppm 1,4-dioxane inhalation exposure. B-6
Figure B-6 acsIX predictions of blood 1,4-dioxane levels using the Young et al.
(1978a,b) model compared with data from Kasai et al. (2008). B-7
Figure B-7 Output of 1,4-dioxane and HEAA blood concentrations from the acsIX
implementation (left) and published (right) data of a 6-hour, 50-ppm
inhalation exposure. B-8
Figure B-8 Observations and acsIX predictions of the cumulative amount of
HEAA in human urine following a 6-hour, 50-ppm inhalation exposure. B-9
B.2.4 Conclusions for Empirical Model Implementation B-9
B.3 Initial Evaluation of the PBPK Models B-9
B.3.1 Initial Recalibration of the Reitz et al. PBPK Model B-10
B.3.2 Flow Rates B-10
Table B-1 Human PBPK model parameter values published in literature and
values used by EPA in this assessment for 1,4-dioxane B-11
B.3.3 Partition Coefficients B-12
B.3.4 Calibration Method B-12
B.3.5 Results B-13
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 13
Figure B-9 Human predicted and observed blood 1,4-dioxane concentrations
(left) and urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-
VI
DRAFT - DO NOT CITE OR QUOTE
-------
dioxane exposure and re-calibration of the PBPK model with
tissue:air partition coefficient values B-14
Figure B-10 Human predicted and observed blood 1,4-dioxane concentrations
(left) and urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-
dioxane exposure and re-calibration of the PBPK model with tissue:air
partition coefficient values B-14
Figure B-11 Human predicted and observed blood 1,4-dioxane concentrations
(left) and urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-
dioxane exposure B-15
using EPA estimated biologically plausible parameters (Table B-1). B-15
B.3.6 Conclusions for PBPK Model Implementation B-15
B.3.7 Sensitivity Analysis B-16
B.3.8 Method B-16
B.3.9 Results B-17
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-17
B.4 PBPK Model Exercises Using Biologically Plausible Parameter Boundaries B-17
B.4.1 Observations Regarding the Volume of Distribution B-18
B.4.2 Defining Boundaries for Parameter Values B-18
B.4.3 Results B-18
Figure B-13 Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients from Leung & Paustenbach
(1990) with empirical model predictions and experimental
observations for blood 1,4-dioxane concentrations (left) and amount
of HEAA in urine (right) from a 6-hour, 50-ppm inhalation exposure. B-19
Figure B-14 Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients from Sweeney et al (2008)
with empirical model predictions and experimental observations for
blood 1,4-dioxane concentrations (left) and amount of HEAA in urine
(right) from a 6-hour, 50-ppm inhalation exposure. B-20
Table B-3 PBPK metabolic and elimination parameter values resulting from
recalibration of the human model using biologically plausible values
for physiological flow rates3 and selected upper and lower boundary
values fortissue:air partition coefficients B-20
B.4.4 Alternative Model Parameterization B-20
Figure B-15 Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, ki_c (1.2 hour"1), to the
experimental data. B-21
Figure B-16 Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, ki_c(0.1 hour"1), to only the
exposure phase of the experimental data. B-22
Figure B-17 Predictions of blood 1,4-dioxane concentration following simultaneous
calibration of a first-order metabolism rate constant ( kLc) and slowly
perfused tissue:air partition coefficient (PSA) to the experimental
data. B-23
B.5 Conclusions B-23
B.6 acsIX Model Code B-24
APPENDIX C. DETAILS OF BMD ANALYSIS FOR ORAL RFD FOR 1,4-DIOXANE C-1
C.1 Cortical Tubule Degeneration C-1
Table C-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water for 2 years C-1
Table C-2 Goodness-of-fit statistics and 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 C-2
Figure C-1 BMD Log-probit model of cortical tubule degeneration incidence data
for male rats exposed to 1,4-dioxane in drinking water for 2 years to C-3
Figure C-2 BMD Weibull model of cortical tubule degeneration incidence data for
female rats exposed to 1,4-dioxane in drinking water for 2 years to
support the results in Table C-2. C-5
APPENDIX D. DETAILS OF BMD ANALYSIS FOR ORAL CSF FOR 1,4-DIOXANE D-1
vn
DRAFT - DO NOT CITE OR QUOTE
-------
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 LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 10%.
LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 30%.
LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 50%.
Multistage BMD model (1 degree) for the combined incidence of
hepatic adenomas and carcinomas in female BDF1 mice.
Figure D-11
Figure D-12
Figure D-13
D.6 Male BDF1 Mice: Hepatic Carcinomas and Adenomas
Table D-14 Data for hepatic adenomas and carcinomas in male BDF1 mice
(Kano et al., 2009)
D-2
D-2
"D-3
-D-4
D-4
-D-4
D-5
-D-5
D-6
"D-7
BMDS dose-response modeling results for the combined incidence of
adenomas and carcinomas in livers of male F344 rats (Kano et al.,
2009) D-8
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-13
.D-14
.D-16
.D-16
.D-18
.D-19
.D-21
D-23
^D-23
D-25
.D-26
D-27
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-28
.D-28
.D-29
.D-31
.D-33
D-35
^D-36
D-37
Vlll
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-15 BMDS dose-response modeling results for the combined incidence of
hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.,
2009)
Figure D-14 LogLogistic BMD model forthe combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice.
Figure D-15 Multistage BMD model (1 degree) forthe combined incidence of
hepatic adenomas and carcinomas in male BDF1 mice.
D.7 BMD Modeling Results from Additional Chronic Bioassays
Table D-16 Summary of BMDS dose-response modeling estimates associated
with liver and nasal tumor incidence data resulting from chronic oral
exposure to 1,4-dioxane in rats and mice
D.7.1
Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma (Kociba et al.,
1974)
Table D-17
Table D-18
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 forthe 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
Table D-19
BMDS dose-response modeling results forthe 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
Table D-21
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
BMDS dose-response modeling results forthe 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 forthe 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 forthe 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 forthe incidence of nasal cavity squamous
cell carcinoma in male Osborne-Mendel rats
exposed to 1,4-dioxane in drinking water.
Figure D-24 Multistage BMD model (1 degree) for the incidence of nasal cavity
squamous cell carcinoma in male Osborne-Mendel ratsexposed to
1,4-dioxane in drinking water.
D.7.3 Hepatocellular Adenoma or Carcinoma in B6C3Fi Mice (NCI, 1978)
_D-37
.D-38
D-40
"D-41
D-42
D-42
D-43
D-43
Figure D-16 Probit BMD model forthe incidence of hepatocellular carcinoma in
male and female Sherman rats exposed to 1,4-dioxane in drinking
water. D-44
Figure D-17 Multistage BMD model (1 degree) forthe incidence of hepatocellular
carcinoma in male and female Sherman rats exposed to 1,4-dioxane
in drinking water. D-46
D-48
D-48
D-50
D-51
D-51
Figure D-19 LogLogistic BMD model forthe incidence of hepatocellular adenoma
in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water. D-52
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-54
BMDS dose-response modeling results forthe 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-56
D-56
D-58
.D-60
D-60
"D-60
D-62
D-63
IX
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-24
Table D-25
Incidence of hepatocellular adenoma or carcinoma in male and
female B6C3F-: mice (NCI, 1978) exposed to 1,4-dioxane in drinking
water D-64
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-64
Figure D-25 Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in female B6C3F-: mice exposed to
1,4-dioxane in drinking water.
Table D-26
Figure D-26
Figure D-27
BMDS dose-response modeling results for the combined incidence of
hepatocellular adenoma or carcinoma in male B6C3F-: mice (NCI,
1978) exposed to 1,4-dioxane in drinking water_
Gamma BMD model for the incidence of hepatocellular adenoma or
carcinoma in male B6C3F-: mice exposed to 1,4-dioxane in drinking
water.
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.
APPENDIX E. COMPARISON OF SEVERAL DATA REPORTS FOR THE JBRC 2-YEAR
1,4-DIOXANE DRINKING WATER STUDY
Table E-1
Table E-2
Table E-3
Table E-4
Table E-5
Table E-6
Table E-7
Table E-8
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinking water study in male F344 rats
Nonneoplastic lesions: Comparison of histological findings reported
for the 2-year JBRC drinking water study in female F344 rats
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in male F344 rats
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in female F344 rats
Neoplastic lesions: Comparison of histological findings reported for
the 2-year JBRC drinking water study in female Crj:BDF1 mice
APPENDIX F.
DETAILS OF BMD ANALYSIS FOR INHALATION RFC FOR 1,4-DIOXANE
F.1 Centrilobular Necrosis of the Liver
D-65
D-67
D-67
D-69
E-1
E-2
E-3
E-4
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
E-9
.F-1
F-1
Table F-1 Incidence of centrilobular necrosis of the liver in male F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years F-1
Table F-2 Goodness-of-fit statistics and BMDio 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) F-2
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. F-2
F.2 Squamous Cell Metaplasia_ F-4
Table F-5 Incidence of squamous cell metaplasia of the respiratory epithelium in
male F344/DuCrj rats exposed to 1,4-dioxane via inhalation for
2 years F-4
Table F-6 Goodness-of-fit statistics and BMDio and BMDL-io values from models
fit to incidence data for squamous cell metaplasia of the respiratory
epithelium in male F344/DuCrj rats exposed to 1,4-dioxane vapors
(Kasai et al., 2009) F-5
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. F-5
F.3 Squamous Cell Hyperplasia F-7
Table F-7 Incidence of squamous cell hyperplasia of the respiratory epithelium
in male F344/DuCrj rats exposed to 1,4-dioxane via inhalation for
2 years F-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-8
Goodness-of-fit statistics and BMD-io and BMDL-io 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)
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.
F.4 Respiratory Metaplasia_
Table F-9
Table F-10
Table F-11
Figure F-6
F.5 Atrophy_
Table F-12
TableF-13
Figure F-7
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
Incidence of atrophy of the olfactory epithelium in male F344/DuCrj
rats exposed to 1,4-dioxane via inhalation for 2 years
Goodness-of-fit statistics and 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.
Figure F-8
F.7 Sclerosis
F-8
_F-8
F-10
Incidence of respiratory metaplasia of the olfactory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years F-10
Goodness-of-fit statistics and BMDio and BMDL-io values from models
fit to incidence data for respiratory metaplasia of olfactory epithelium
in male F344/DuCrj rats (Kasai et al., 2009) exposed to 1,4-dioxane
vapors F-11
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-11
F-12
.F-13
F-14
F-14
F-15
Hydropic Change F-16
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 F-17
Table F-15 Goodness-of-fit statistics and BMDio 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 etal.,2009)_ F-18
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-18
.F-20
F-20
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-21
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-23
APPENDIX G. DETAILS OF BMD ANALYSIS FOR INHALATION UNIT RISK FOR
1,4-DIOXANE
G.1 General Issues and Approaches to BMDS and Multitumor Modeling
G.1.1 Combining Data tumor types
G.1.2 Summary
Table G-1 Summary of BMC-io and BMCLio model results for individual tumor
G.2
types and combined tumor analysis for male rats exposed to
1,4-dioxane vapors (Kasai et al., 2009)
BMDS Model Output for Multistage Cancer Models for Individual Tumor Types
Table G-2 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
vapor by whole-body inhalation for 2 years
G.2.1 Nasal Squamous Cell Carcinoma
G-1
_G-2
_G-2
_G-3
G-3
XI
DRAFT - DO NOT CITE OR QUOTE
-------
Table G-3 BMDS Multistage cancer dose-response modeling results for the
incidence of nasal squamous cell carcinomas in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai etal., 2009) G-4
Figure G-1 Multistage model (First (1°)-degree) for male rat nasal squamous cell
carcinomas. G-4
G.2.2 Hepatocellular Adenoma and Carcinoma G-6
Table G-4 BMDS Multistage cancer dose-response modeling results for the
incidence of either hepatocellular adenoma or carcinoma in male rats
exposed to 1,4-dioxane vapors for 2-years (Kasai et al., 2009) G-7
Figure G-2 Multistage model (First-degree (1°)) for male rat hepatocellular
adenomas and carcinomas. G-7
G.2.3 Renal Cell Carcinoma and Zymbal Gland Adenoma G-9
Table G-5 BMDS Multistage cancer dose-response modeling results for the
incidence of renal cell carcinomas and Zymbal gland adenomas in
male rats exposed to 1,4-dioxane vapors for 2-years (Kasai et al.,
2009) G-10
Figure G-3 Multistage model (Second-degree (2°)) for male rat renal cell
carcinomas and Zymbal gland adenomas. G-10
Figure G-4 Multistage model (Third-degree (3°)) for male rat renal cell
carcinomas. G-12
G.2.4 Peritoneal Mesothelioma G-14
Table G-6 BMDS Multistage cancer dose-response modeling results for the
incidence of peritoneal mesothelioma in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai etal., 2009) G-14
Figure G-5 Multistage model (First-degree (1°)) for male rat peritoneal
mesotheliomas. G-15
G.2.5 Mammary Gland Fibroadenoma G-16
Table G-7 BMDS Multistage cancer dose-response modeling results for the
incidence of mammary gland fibroadenoma in male rats exposed to
1,4-dioxane vapors for 2-years (Kasai etal., 2009) G-17
Figure G-6 Multistage model (First-degree (1°)) for male rat mammary gland
fibroadenoma. G-17
G.2.6 Subcutis Fibroma G-19
Table G-8 BMDS Multistage cancer dose-response modeling results for the
incidence of subcutis fibromas in male rats exposed to 1,4-dioxane
vapors for 2-years (Kasai et al., 2009) G-19
Figure G-7 Multistage model (First-degree (1°)) for male rat subcutis fibroma
(high dose dropped). G-20
G.3 Multitumor Analysis Using BMDS MS_Combo G-21
G.4 Multitumor analysis using Bayesian Methods G-22
xn
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
BMCL-io 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
CFD computational fluid dynamic
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 glutathione S-transferase, placental form
HEAA (3-hydroxyethoxy acetic acid
HED(s) human equivalent dose(s)
HPLC high-performance liquid chromatography
HSDB Hazardous Substances Data Bank
Hz Hertz
IARC International Agency for Research on Cancer
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 mean corpuscular volume
MOA mode of action
Xlll
DRAFT - DO NOT CITE OR QUOTE
-------
MS mass spectrometry, multi-stage
MTD maximum tolerated dose
MVK Moolgavkar-Venzon-Knudsen (model)
NCE normochromatic erythrocyte
NCI National Cancer Institute
ND no data, not detected
NE not estimated
NOAEL no-observed-adverse-effect-level
NRC National Research Council
NTP National Toxicology Program
OCT ornithine carbamyl transferase
ODC ornithine decarboxylase
OECD Organization for Economic Co-operation and Development
PB blood:air partition coefficient
PBPK physiologically based pharmacokinetic
PC partition coefficient
PCS polychlorinated biphenyl
PCE polychromatic erythrocyte
PFA fatair 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
xiv
DRAFT - DO NOT CITE OR QUOTE
-------
FOREWORD
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
14 through the inhalation route of exposure became available during the fmalization of the 1,4-dioxane oral
15 assessment that was posted on the IRIS database in 2010 (U.S. EPA. 2010a). In this version of the
16 toxicological review, these studies have been incorporated into the previously posted assessment (U.S.
17 EPA. 2010a). Although the focus of the most recent peer review was on the inhalation toxicity following
18 exposure to 1,4-dioxane, a few comments were received on the oral assessment and were addressed to
19 ensure scientific consistency between both routes of exposure. These comments did not impact the final
20 conclusions of the oral assessment.
DRAFT - DO NOT CITE OR QUOTE xv
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
Assessment Team
Patricia Gillespie, Ph.D. (Chemical Manager, Inhalation)
Eva D. McLanahan, Ph.D. (Chemical Manager, Oral/Inhalation)
Reeder Sams, Ph.D. (Chemical Manager, Oral)
John Stanek, Ph.D.
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
Scientific Support Team
Lyle Burgoon, Ph.D.
J. Allen Davis, MSPH
JeffS. Gift, Ph.D.
Nagu Keshava, Ph.D.
Allan Marcus, Ph.D.
Connie Meacham, Ph.D.
Andrew Rooney, Ph.D.
Paul Schlosser, Ph.D.
John Vandenberg, Ph.D.
Jason Lambert, Ph.D.
Karen Hogan
Leonid Kopylev, Ph.D.
Susan Rieth
Anthony DeAngelo, Ph.D.
Hisham El-Masri, Ph.D.
William Lefew, Ph.D.
Douglas Wolf, Ph.D.
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
U.S. EPA/ORD/NCEA
Cincinnati, OH
U.S. EPA/ORD/NCEA
Washington, DC
U.S. EPA/ORD/NHEERL
Research Triangle Park, NC
Production Team
Ellen Lorang, M.S.
Connie Meacham, M.S.
Deborah Wales
J. Sawyer Lucy
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
U.S. EPA/ORD/NCEA
Student Services Contractor
Research Triangle Park, NC
Contractor Support
Fernando Llados
Michael Lumpkin, Ph.D.
Mark Odin, Ph.D.
Julie Stickney, Ph.D.
Environmental Science Center
Syracuse Research Corporation
Syracuse, NY
Executive Direction
Kenneth Olden, Ph.D., Sc.D., L.H.D.
Lynn Flowers, Ph.D., DABT
Vincent Cogliano, Ph.D.
Samantha Jones, Ph.D.
Lyle Burgoon, Ph.D.
Reeder Sams, Ph.D.
John Vandenberg, Ph.D
Debra Walsh, M.S.
U.S. EPA/ORD/NCEA
Washington, DC
U.S. EPA/ORD/NCEA
Research Triangle Park, NC
DRAFT - DO NOT CITE OR QUOTE
xvi
-------
Oral Assessment Reviewers
1 The oral assessment was provided for review to scientists in EPA's Program and Region Offices.
2 Comments were submitted by:
Office of Water, Washington, DC
Region 2, New York City, NY
Region 3, Philadelphia, PA
Region 6, Dallas, TX
Region 8, Denver, CO
Office of Policy, Economics, and Innovation, Washington, DC
Office of Pesticide Programs, Washington, DC
Office of Air Quality and Planning Standards, Research Triangle Park, NC
3 The oral assessment was provided for review to other federal agencies and Executive Offices of the
4 President. Comments were submitted by:
Department of Defense
National Aeronautics and Space Administration
Office of Management and Budget
5 The oral assessment was released for public comment in May 2009. A summary and EPA's disposition
6 of the comments from the public is included in Appendix A and is also available on the IRIS Web site.
7 Comments were received from the following entities:
The Alliance for Environmental Responsibility
and Openness (AERO)
Betty Locey, Ph.D., DABT ARCADIS
Ted Simon, Ph.D., DABT Novi, MI
Lu Yu, Ph.D.
P. Stephen Finn
Gregory J. Garvey Colder Associates, Inc.
Theresa Repaso-Subang, DABT Mt. Laurel, NJ
Lorenz R. Rhomberg, Ph.D. Gradient Corporation
Cambridge, MA
John E. Bailey, Ph.D. Personal Care Products Council
8 The oral assessment was peer reviewed by independent expert scientists external to EPA and a peer-
9 review meeting was held on August 17, 2009. The external peer-review comments are available on the
10 IRIS Web site. A summary and EPA's disposition of the comments received from the independent
11 external peer reviewers and from the public is included in Appendix A and is also available on the IRIS
12 Web site.
George V. Alexeeff, Ph.D., DABT California Environmental Protection Agency
Sacramento, CA
Bruce C. Allen, M.S. Bruce Allen Consulting
Chapel Hill, NC
James V. Bruckner, Ph.D. University of Georgia
Athens, GA
Harvey J. Clewell III. Ph.D., DABT The Hamner Institutes for Health Sciences
DRAFT - DO NOT CITE OR QUOTE xvii
-------
Research Triangle Park, NC
Lena Ernstgard, Ph.D. Karolinska Institute!
Location
Frederick J. Kaskel, M.D., Ph.D. Children's Hospital at Montefiore
Albert Einstein College of Medicine of Yeshiva University
Location
Universite de Montreal
Kannan Krishnan, Ph.D., DABT Montreal, Canada
University of Georgia (retired)
Raghubir P. Sharma, DVM, Ph.D. Athens, GA
Inhalation Assessment Reviewers
2 The assessment with the inhalation update was provided for review to scientists in EPA's Program and
3 Region Offices. Comments were submitted by:
Office of Policy, Washington, DC
Office of Water, Washington, DC
Office of Solid Waste and Emergency Response, Washington, DC
Region 2, New York City, NY
4 The assessment with the inhalation update was provided for review to other federal agencies and
5 Executive Offices of the President. Comments were submitted by:
Agency for Toxic Substances Disease Registry, Centers for Disease Control and
Prevention, Department of Health & Human Services
Department of Defense
National Aeronautics and Space Administration
National Toxicology Program, National Institutes for Environmental Health Sciences,
National Institutes of Health, Department of Health & Human Services
Office of Management and Budget
6 The assessment with the inhalation update was released for public comment in September 2011 and
7 comments were due on November 15, 2011. The public comments are available on the IRIS Web site. A
8 summary and EPA's disposition of the comments from the public is included in Appendix A and is also
9 available on the IRIS Web site. Comments were received from the following entities:
Michael Dourson Toxicology Excellence for Risk Assessment
Patricia Nance Cincinnati, OH
John Reichard
Mahta Mahdavi National Association of Manufacturers
Washington, DC
Lisa Goldberg Aerospace Industries Association
Arlington, VA
10
DRAFT - DO NOT CITE OR QUOTE xviii
-------
1 The assessment with the inhalation update was peer reviewed by independent expert scientists external to
2 EPA and a peer-review meeting was held on March 19, 2012. The external peer-review comments are
3 available on the IRIS Web site. A summary and EPA's disposition of the comments received from the
4 independent external peer reviewers and from the public is included in Appendix A and is also available
5 on the IRIS Web site.
James V. Bruckner, Ph.D. University of Georgia
Athens, GA
Harvey J. Clewell III. Ph.D., DABT The Hamner Institutes for Health Sciences
Research Triangle Park, NC
David C. Dorman, DVM, Ph.D. DABVT, NCSU-College of Veterinary Medicine
DABT Raleigh, NC
Ron Melnick Consulting, LLC
Ronald L. Melnick, Ph.D. Chapel Hill, NC
Fred J. Miller & Associates, LLC
Frederick J. Miller, Ph.D., Fellow ATS Cary, NC
University of Georgia (retired)
Raghubir P. Sharma, DVM, Ph.D. Athens, GA
DRAFT - DO NOT CITE OR QUOTE xix
-------
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. 1986c).
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. 1994c). Methods for Derivation of Inhalation
34 Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA. 1994a). 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.
37 1998). Science Policy Council Handbook: Risk Characterization (U.S. EPA. 2012a. 2000b). Benchmark
1
DRAFT - DO NOT CITE OR QUOTE
-------
1 Dose Technical Guidance Document (U.S. EPA. 2012a. 2000a), Supplementary Guidance for Conducting
1 Health Risk Assessment of Chemical Mixtures (U.S. EPA. 2000c). A Review of the Reference Dose and
3 Reference Concentration Processes (U.S. EPA. 2002b). Guidelines for Carcinogen Risk Assessment (U.S.
4 EPA, 2005a). Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to
5 Carcinogens (U.S. EPA. 2005cX Science Policy Council Handbook: Peer Review (U.S. EPA. 2006bX and
6 A Framework for Assessing Health Risks of Environmental Exposures to Children (U.S. EPA. 2006c).
7 In 2010, an updated health assessment for oral exposures to 1,4-dioxane was released (U.S. EPA.
8 201 Ob). During the development of the 2010 health assessment, new studies (Kasai et al.. 2009; Kasai et
9 al.. 2008) regarding the toxicity of 1,4-dioxane through the inhalation route of exposure became available
10 that were not included in the 1,4-dioxane assessment that was posted on the IRIS database in 2010. These
11 new inhalation studies have now been incorporated into the previously posted assessment and inhalation
12 toxicity values are presented in this toxicological review.
13 The literature search strategy employed for 1,4-dioxane was initially based on the chemical name,
14 Chemical Abstracts Service Registry Number (CASRN), and multiple common synonyms. A subsequent
15 search was completed which focused on the toxicology and toxicokinetics of 1,4-dioxane, particularly as
16 they pertain to target tissues, effects at low doses, mode of action (non-cancer and cancer), and sensitive
17 populations. Following peer review of the assessment, a more targeted search was carried out based on
18 comments received from expert peer reviewers. Additionally, any pertinent scientific information
19 submitted by the public to the IRIS Submission Desk and by external peer reviewers during the
20 Independent Expert Peer Review meeting was also considered in the development of this document.
21 Selection of studies for inclusion in the Toxicological Review was based on consideration of the
22 extent to which the study was informative and relevant to the assessment, and general study
23 considerations as outlined in EPA guidance documents (A Review of the Reference Dose and Reference
24 Concentration Processes (U.S. EPA. 2002c) and Methods for Derivation of Inhalation Reference
25 Concentrations and Application of Inhaled Dosimetry (U.S. EPA. 1994b)).
26 Primary, peer-reviewed-literature was reviewed through September 2009 for the oral assessment
27 and through April 2013 for the inhalation assessment and was included where the literature was
28 determined to be critical to the assessment. The relevant literature included publications on 1,4-dioxane
29 which were identified through Toxicology Literature Online (TOXLINE), PubMed, the Toxic Substance
30 Control Act Test Submission Database (TSCATS), the Registry of Toxic Effects of Chemical Substances
31 (RTECS), the Chemical Carcinogenesis Research Information System (CCRIS), the Developmental and
32 Reproductive Toxicology/Environmental Teratology Information Center (DART/ETIC), the
33 Environmental Mutagens Information Center (EMIC) and Environmental Mutagen Information Center
34 Backfile (EMICBACK) databases, the Hazardous Substances Data Bank (HSDB), the Genetic
35 Toxicology Data Bank (GENE-TOX), Chemical abstracts, and Current Contents. Other peer-reviewed
36 information, including health assessments developed by other organizations, review articles, and
37 independent analyses of the health effects data were retrieved and may be included in the assessment
38 where appropriate. Studies that had not been peer-reviewed, were not included in the assessment.
DRAFT - DO NOT CITE OR QUOTE
-------
1 The references considered and cited in this document, including bibliographic information and
2 abstracts, can be found on the Health and Environmental Research Online (HERO) website1
3 (http://hero.epa.gov). For other general information about this assessment or other questions relating to
4 IRIS, the reader is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
5 hotline.iris@epa.gov.
1HERO is a database of scientific studies and other references used to develop EPA's risk assessments aimed at
understanding the health and environmental effects of pollutants and chemicals. It is developed and managed in
EPA's Office of Research and Development (ORD) by the National Center for Environmental Assessment (NCEA).
The database includes more than 400,000 scientific articles from the peer-reviewed literature. New studies are
added continuously to HERO.
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 g/mL at 20°C (CRC Handbook (Lide. 2000))
Vapor density:
3.03 (air =1) (Lewis. 2000)
Water solubility:
Miscible with water (Hawlev and Lewis. 2001)
Other solubilities:
Miscible with ethanol, ether, acetone (CRC Handbook (Lide. 2000))
Log K0.
-0.27 (Hansch et al.. 1995)
Henry's Law constant:
4.80 x 10"b atm-irf/molecule at 25°C (Parketal.. 1987)
OH reaction rate constant:
1.09 x 10'11 cma/molecule sec at 25°C (Atkinson. 1989)
17 (estimated using log Kow) (ACS Handbook (Lvman et al.. 1990))
Bioconcentration factor:
0.4 (estimated using log Kow) (Mevlan et al.. 1999)
Conversion factors (in air):
1 ppm = 3.6 mg/m ; 1 mg/m = 0.278 ppm
(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. 2002a).
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. 2002a).
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; 2001; IARC. 1999). 1,4-Dioxane has
13 been used as a solvent in the formulation of inks, coatings, and adhesives and in the extraction of animal
14 and vegetable oil (Surprenant. 2002). Reaction products of 1,4-dioxane are used in the manufacture of
15 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; Lyman et
25 al.. 1990). This substance may volatilize from dry soil surfaces based on its vapor pressure (HSDB.
26 2007). The estimated bioconcentration factor value indicates that 1,4-dioxane will not bioconcentrate in
27 aquatic or marine organisms (Meylan et al.. 1999; Franke et al.. 1994). 1,4-Dioxane is not expected to
28 undergo hydrolysis or to biodegrade readily in the environment (ATSDR. 2007; HSDB. 2007). Therefore,
29 volatilization is expected to be the dominant removal process for moist soil and surface water. Based on a
30 Henry's Law constant of 4.8x 10"6 atm-m3/mole, the half-life for volatilization of 1,4-dioxane from a
31 model river is 5 days and that from a model lake is 56 days (HSDB. 2007; Lyman et al.. 1990; Park et al..
32 1987). 1,4-Dioxane may be more persistent in groundwater where volatilization is hindered.
33 Recent environmental monitoring data for 1,4-dioxane are lacking. Existing data indicate that
34 1,4-dioxane may leach from hazardous waste sites into drinking water sources located nearby (Yasuhara
35 et al.. 2003; Yasuhara et al.. 1997; Lesage et al.. 1990). 1,4-Dioxane has been detected in contaminated
36 surface and groundwater samples collected near hazardous waste sites and industrial facilities (Derosa et
37 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.. 1976a). 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. (1977a) 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 et al.. 1978b: 1978a).
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]-1,4-dioxane given as a single dose or as 17 consecutive daily doses (Young et al.. 1978b: 1978a).
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 (1982a) reported an in vitro study in which 1,4-dioxane penetrated excised human skin 10
3 times more under occluded conditions (3.2% of applied dose) than unoccluded conditions (0.3% of
4 applied dose). [14C]-l,4-Dioxane was dissolved in lotion, applied to the excised skin in occluded and
5 unoccluded diffusion cells, and absorption of the dose was recorded 205 minutes after application.
6 Bronaugh (1982a) also reported observing rapid evaporation, which further decreased the small amount
7 available for skin 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 determined by gas chromatography reached peak percentages at 6 hours after dosing in
35 liver (18.5%), spleen (22.6%), and colon (19.5%). At 16 hours after dosing, peak covalent binding
36 percentages were observed in whole blood (3.1%), kidney (9.5%), lung (11.2%), and skeletal muscle
7
DRAFT - DO NOT CITE OR QUOTE
-------
1 (11.2%). Within hepatocytes, radiolabel distribution at 6 hours after dosing was greatest in the cytosolic
2 fraction (43.8%) followed by the microsomal (27.9%), mitochondrial (16.6%), and nuclear (11.7%)
3 fractions. While little covalent binding of radiolabel was measured in the hepatic cytosol (4.6%), greater
4 binding was observed at 16 hours after dosing in the nuclear (64.8%), mitochondrial (45.7%), and
5 microsomal (33.4%) fractions. Pretreatment with inducers of mixed-function oxidase activity did not
6 significantly change the extent of covalent 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: 1977a)) 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 (Wooetal.. 1978. 1977bV 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 et al.. 1977c).
DRAFT - DO NOT CITE OR QUOTE
-------
O. OH ^OH O
(c) '-' " °
[V] [VI] x~x
° ' (a) HOH2C CH2OH HOH2C COOH
' " S>J V
[II]
(bf """•""- ••- -H20
O
[IV]
Source: Adapted with permission of Elsevier Ltd., Woo et al. (1977a: 1977b).
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 five 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. (1976a) 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.. 1977a). 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 1978b: 1978a). 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.. 1978b: 1978a). 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) estimated from
11 experimentally determined Km values that metabolic saturation occurred near plasma levels of 100
12 ug/mL. Kociba et al. (1975b) also estimated metabolic saturation near plasma levels of 100 ug/mL in rats
13 following a single i.v. dose. The lack of the metabolic saturation of 1,4-dioxane found in the Kasai et al.
14 (2008) study is likely attributed to enhanced metabolism by the induction of P450 enzymes, including
15 CYP2E1, by 13 weeks of repeated inhalation exposure to 1,4-dioxane at concentrations up to 3,200 ppm
16 (Kasai et al.. 2008).
10.000
0 s> 10 15 20 25 30 3S 40 45 50 55 60 65 7C
Source: Reprinted with permission of Taylor and Francis, Young et al. (1_978b).
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)]
10
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.. 2005a). 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. Takano et al. (2010) recently tested liver microsome contents from
7 male Sprague-Dawley rats treated with 500 mg 1,4-dioxane/kg BW intraperitoneally (i.p.) for 3 days for
8 CYP450 activities. CYP2B and CYP2E activities were significantly increased (p <0.05) compared to
9 control activity levels, while CYP2C activity was significantly decreased to approximately 50% of control
10 values. This is in contrast to Nannelli et al. (2005b) where CYP2C values increased.
11 The increase in CYP2C or specifically, CYP2C11 activity reported by Nanelli et al. (2005b) was
12 unexpected, as that isoform has been observed to be under hormonal control and was typically suppressed
13 in the presence of 2B1/2 and 2E1 induction. In the male rat, hepatic 2C11 induction is associated with
14 masculine pulsatile plasma profiles of growth hormone (compared to the constant plasma levels in the
15 female), resulting in masculinization of hepatocyte function (Waxman et al.. 1991). The authors
16 postulated that 1,4-dioxane may alter plasma growth hormone levels, resulting in the observed 2C11
17 induction. However, growth hormone induction of 2C11 is primarily dependent on the duration between
18 growth hormone pulses and secondarily on growth hormone plasma levels (Agrawal and Shapiro. 2000;
19 Waxman etal.. 1991). Thus, the induction of 2C11 by 1,4-dioxane may be mediated by changes in the
20 time interval between growth hormone pulses rather than changes in growth hormone levels. This may be
21 accomplished by 1,4-dioxane temporarily influencing the presence of growth hormone cell surface
22 binding sites (Agrawal and Shapiro. 2000). However, no studies are available to confirm the influence of
23 1,4-dioxane on either growth hormone levels or changes in growth hormone pulse interval.
24 In nasal and renal mucosal cell microsomes, CYP2E1 activity, but not CYP2B1/2 activity, was
25 increased. Pulmonary mucosal CYP450 activity levels were not significantly altered. Observed increases
26 in 2E1 mRNA in rats exposed by gavage and i.p. injection suggest that 2E1 induction in kidney and nasal
27 mucosa is controlled by a transcriptional activation of 2E1 genes. The lack of increased mRNA in
28 hepatocytes suggests that induction is regulated via a post-transcriptional mechanism. Differences in 2E1
29 induction mechanisms in liver, kidney, and nasal mucosa suggest that induction is controlled in a
30 tissue-specific manner.
3.4 Elimination
31 In workers exposed to a TWA of 1.6 ppm for 7.5 hours, 99% of 1,4-dioxane eliminated in urine
32 was in the form of HEAA (Young et al.. 1976a). The elimination half-life was 59 minutes in adult male
33 volunteers exposed to 50 ppm 1,4-dioxane for 6 hours, with 90% of urinary 1,4-dioxane and 47% of
34 urinary HEAA excreted within 6 hours of onset of exposure (Young et al.. 1977a). There are no data for
35 1,4-dioxane elimination in humans from oral exposures.
11
DRAFT - DO NOT CITE OR QUOTE
-------
1 Elimination of 1,4-dioxane in rats (Young et al.. 1978b: 1978a). was primarily via urine. As
2 comparably assessed in humans, the elimination half-life in rats exposed to 50 ppm 1,4-dioxane for
3 6 hours was calculated to be 1.01 hours. In Sprague Dawley rats given single daily doses of 10, 100, or
4 1,000 mg [14C]-l,4-dioxane/kg or multiple doses of 10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary
5 radiolabel ranged from 99% down to 76% of total radiolabel. Fecal elimination was less than 2% for all
6 doses. The effect of saturable metabolism on expired 1,4-dioxane was apparent, as expired 1,4-dioxane in
7 singly dosed rats increased with dose from 0.4 to 25% while expired 14CO2 changed little (between 2 and
8 3%) across doses. The same relationship was seen in Sprague Dawley rats dosed i.v. with 10 or 1,000 mg
9 [14C]-l,4-dioxane/kg. Higher levels of 14CO2 relative to 1,4-dioxane were measured in expired air of the
10 10 mg/kg group, while higher levels of expired 1,4-dioxane relative to 14CO2 were measured in the
11 1,000 mg/kg group.
3.5 Physiologically Based Pharmacokinetic Models
12 Physiologically based pharmacokinetic models (PBPK) models have been developed for
13 1.4-dioxane in rats (Sweeney et al.. 2008a: Leung and Paustenbach. 1990a: Reitz et al.. 1990a). mice
14 (Reitz etal.. 1990). humans (Sweeney et al.. 2008a: Leung and Paustenbach. 1990a: Reitz etal. 1990a).
15 and lactating women (Fisher etal.. 1997). Each of the models simulates the body as a series of
16 compartments representing tissues or tissue groups that receive blood from the central vascular
17 compartment (Figure 3-3). Modeling was conducted under the premise that transfers of 1,4-dioxane
18 between blood and tissues occur sufficiently fast to be effectively blood flow-limited, which is consistent
19 with the available data (Ramsey and Andersen. 1984). Blood time course and metabolite production data
20 in rats and humans suggest that absorption and metabolism are accomplished through common
21 mechanisms in both species (Young et al. (1978b: 1978a: 1977a)). allowing identical model structures to
22 be used for both species (and by extension, for mice as well). In all three models, physiologically
23 relevant, species-specific parameter values for tissue volume, blood flow, and metabolism and elimination
24 are used. The models and supporting data are reviewed below, from the perspective of assessing their
25 utility for predicting internal dosimetry and for cross-species extrapolation of exposure-response
26 relationships for critical neoplastic and nonneoplastic endpoints (also see Appendix B).
12
DRAFT - DO NOT CITE OR QUOTE
-------
IV
infusion inhalation
t
J
O
_O
CD
CD
Z3
O
£Z
QJ
Lung blood
I
f Rapidly
[__ perfused
f Slowly
^ perfused
« Fat
\. j
f
)
-^
L
r
L
r
i.
o
_o
m
QJ
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: 1977aX Mikheev et al. (1990). and Woo et al. (1977a: 1977cV Young et al. (1978b: 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.. 1978b: 1978a). Although it may be
13
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.. 1978b: 1978a). and metabolic induction was observed following exposure to high oral daily
6 doses (1,000 mg/kg-day) of 1,4-dioxane. Elimination of 1,4-dioxane from plasma appeared to be linear
7 following i.v. doses of 3-30 mg/kg, but was nonlinear following doses of 100-1,000 mg/kg. Accordingly,
8 10 mg/kg i.v. doses resulted in higher concentrations of 14CO2 (from metabolized 1,4-dioxane) in expired
9 air relative to unchanged 1,4-dioxane, while 1,000 mg/kg i.v. doses resulted in higher concentrations of
10 expired 1,4-dioxane relative to 14CO2. Thus, at higher i.v. doses, a higher proportion of unmetabolized
11 1,4-dioxane is available for exhalation. Taken together, the i.v. plasma and expired air data from Young et
12 al. (1978b: 1978a) corroborate previous studies describing the saturable nature of 1,4-dioxane metabolism
13 in rats (1977a: Woo et al.. 1977c) and are useful for optimizing metabolic parameters (Vmax and Km) in a
14 PBPK model.
15 Similarly, increasing single or multiple oral doses of 10-1,000 mg/kg resulted in increasing
16 percentage of 1,4-dioxane in exhaled air and decreasing percentage of radiolabel (either as 1,4-dioxane or
17 a metabolite) in the urine, with significant differences in both metrics being observed between doses of 10
18 and 100 mg/kg (Young et al.. 1978b: 1978a). These data identify the region (10-100 mg/kg) in which oral
19 exposures will result in nonlinear metabolism of 1,4-dioxane and could be used to test whether metabolic
20 parameter value estimates derived from i.v. dosing data are adequate for modeling oral exposures.
21 Post-exposure plasma data from a single 6-hour, 50 ppm inhalation exposure in rats were reported
22 (Young et al.. 1978b: 1978a). The observed linear elimination of 1,4-dioxane after inhalation exposure
23 suggests that, via this route, metabolism follows a first-order process at this exposure level.
24 The only human data adequate for use in PBPK model development (Young et al.. 1977a) come
25 from adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours. Plasma 1,4-dioxane and HEAA
26 concentrations were measured both during and after the exposure period, and urine concentrations were
27 measured following exposure. Plasma levels of 1,4-dioxane approached steady-state at 6 hours. HEAA
28 data were insufficient to describe the appearance or elimination of HEAA in plasma. Data on elimination
29 of 1,4-dioxane and HEAA in the urine up to 24 hours from the beginning of exposure were reported. At
30 6 hours from onset of exposure, approximately 90% and 47% of the cumulative (0-24 hours) urinary
31 1,4-dioxane and HEAA, respectively, were measured in the urine. The ratio of HEAA to 1,4-dioxane in
32 urine 24 hours after onset of exposure was 192:1 (similarto the ratio of 118:1 observed by Young et al.
33 (1976a) in workers exposed to 1.6 ppm for 7.5 hours), indicating extensive metabolism of 1,4-dioxane.
34 As with Sprague Dawley rats, the elimination of 1,4-dioxane from plasma was linear across all
35 observations (6 hours following end of exposure), suggesting that human metabolism of 1,4-dioxane is
36 linear for a 50 ppm inhalation exposure to steady-state. Thus, estimation of human Vmax and Km from
37 these data will introduce uncertainty into internal dosimetry performed in the nonlinear region of
38 metabolism.
14
DRAFT - DO NOT CITE OR QUOTE
-------
1 Further data were reported for the tissue distribution of 1,4-dioxane in rats. Mikheev et al. (1990)
2 administered i.p. doses of [14C]-l,4-dioxane to white rats (strain not reported) and reported time-to-peak
3 blood, liver, kidney, and testes concentrations. They also reported ratios of tissue to blood concentrations
4 at various time points after dosing. Woo et al. (1977a: 1977c) administered i.p. doses of [14C]-l,4-dioxane
5 to Sprague Dawley rats and measured radioactivity levels in urine. However, since i.p. dosing is not
6 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
7 Leung and Paustenbach (1990a) developed a PBPK model for 1,4-dioxane and its primary
8 metabolite, HEAA, in rats and humans. The model, based on the structure of a PBPK model for styrene
9 (Ramsey and Andersen. 1984). consists of a central blood compartment and four tissue compartments:
10 liver, fat, slowly perfused tissues (mainly muscle and skin), and richly perfused tissues (brain, kidney, and
11 viscera other than the liver). Tissue volumes were calculated as percentages of total BW, and blood flow
12 rates to each compartment were calculated as percentages of cardiac output. Equivalent cardiac output
13 and alveolar ventilation rates were allometrically scaled to a power (0.74) of BW for each species. The
14 concentration of 1,4-dioxane in alveolar blood was assumed to be in equilibrium with alveolar air at a
15 ratio equal to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane
16 between blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
17 between blood and tissue, governed by tissue:blood equilibrium partition coefficients. The latter were
18 derived from the quotient of blood:air and tissue:air partition coefficients, which were measured in vitro
19 (Leung and Paustenbach. 1990a) for blood, liver, fat, and skeletal muscle (slowly perfused tissue).
20 Blood:air partition coefficients were measured for both humans and rats. Rat tissue:air partition
21 coefficients were used as surrogate values for humans, with the exception of slowly perfused tissue :blood,
22 which was estimated by optimization to the plasma time-course data. Portals of entry included i.v.
23 infusion (over a period of 36 seconds) into the venous blood, inhalation by diffusion from the alveolar air
24 into the lung blood at the rate of alveolar ventilation, and oral administration via zero-order absorption
25 from the gastrointestinal tract to the liver. Elimination of 1,4-dioxane was accomplished through
26 pulmonary exhalation and saturable hepatic metabolism. Urinary excretion of HEAA was assumed to be
27 instantaneous with the generation of HEAA from the hepatic metabolism of 1,4-dioxane.
28 The parameter values for hepatic metabolism of 1,4-dioxane, Vmax and Km, were optimized and
29 validated against plasma and/or urine time course data for 1,4-dioxane and HEAA in rats following i.v.
30 and inhalation exposures and humans following inhalation exposure (Young et al. (1978b: 1978a:
31 1977a)): the exact data (i.e., i.v., inhalation, or both) used for the optimization and calibration were not
32 reported. Although the liver and fat were represented by tissue-specific compartments, no tissue-specific
33 concentration data were available for model development, raising uncertainty as the model's ability to
34 adequately predict exposure to these tissues. The human inhalation exposure of 50 ppm for 6 hours
15
DRAFT - DO NOT CITE OR QUOTE
-------
1 (Young et al.. 1977a) was reported to be in the linear range for metabolism; thus, uncertainty exists in the
2 ability of the allometrically-scaled value for the human metabolic Vmax to accurately describe 1,4-dioxane
3 metabolism from exposures resulting in metabolic saturation. Nevertheless, these values resulted in the
4 model producing good fits to the data. For rats, the values for Vmax had to be adjusted upwards by a factor
5 of 1.8 to reasonably simulate exposures greater than 300 mg/kg. The model authors attributed this to
6 metabolic enzyme induction by high doses of 1,4-dioxane.
3.5.2.2 Reitzetal.
7 Reitz et al. (1990a) developed a model for 1,4-dioxane and HEAA in the mouse, rat, and human.
8 This model, also based on the styrene model of Ramsey and Andersen (1984). included a central blood
9 compartment and compartments for liver, fat, and rapidly and slowly perfused tissues. Tissue volumes
10 and blood flow rates were defined as percentages of total BW and cardiac output, respectively.
11 Physiological parameter values were similar to those used by Andersen et al. (1987). except that flow
12 rates for cardiac output and alveolar ventilation were doubled in order to produce a better fit of the model
13 to human blood level data (Young et al.. 1977a). Portals of entry included i.v. injection into the venous
14 blood, inhalation, oral bolus dosing, and oral dosing via drinking water. Oral absorption of 1,4-dioxane
15 was simulated, in all three species, as a first-order transfer to liver (halftime approximately 8 minutes).
16 Alveolar blood levels of 1,4-dioxane were assumed to be in equilibrium with alveolar air at a
17 ratio equal to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane
18 between blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
19 between blood and tissue, governed by tissue:blood equilibrium partition coefficients. These coefficients
20 were derived by dividing experimentally measured (Leung and Paustenbach. 1990a) in vitro blood:air and
21 tissue:air partition coefficients for blood, liver, fat. Blood:air partition coefficients were measured for both
22 humans and rats. The mouse blood:air partition coefficient was different from rat or human values; the
23 source of the partition coefficient for blood in mice was not reported. Rat tissue:air partition coefficients
24 were used as surrogate values for humans. Rat tissue partition coefficient values were the same values as
25 used in the Leung and Paustenbach (1990a) model (with the exception of slowly perfused tissues) and
26 were used in the models for all three species. The liver value was used for the rapidly perfused tissues, as
27 well as slowly perfused tissues. Although slowly perfused tissue:air partition coefficients for rats were
28 measured, the authors suggested that 1,4-dioxane in the muscle and air may not have reached equilibrium
29 in the highly gelatinous tissue homogenate (Reitz et al.. 1990a). Substitution of the liver value provided
30 much closer agreement to the plasma data than when the muscle value was used. Further, doubling of the
31 measured human blood:air partition coefficient improved the fit of the model to the human blood level
32 data compared to the fit resulting from the measured value (Reitz et al.. 1990a). The Reitz et al. (1990a)
33 model simulated three routes of 1,4-dioxane elimination: pulmonary exhalation, hepatic metabolism to
34 HEAA, and urinary excretion of HEAA. The elimination of HEAA was modeled as a first-order transfer
35 of 1,4-dioxane metabolite to urine.
16
DRAFT - DO NOT CITE OR QUOTE
-------
1 Values for the metabolic rate constants, Vmax and Km, were optimized to achieve agreement with
2 various observations. Reitz et al. (1990a) optimized values for human Vmax and Km against the
3 experimental human 1,4-dioxane inhalation data (Young et al.. 1977a). As noted previously, because the
4 human exposures were below the level needed to exhibit nonlinear kinetics, uncertainty exists in the
5 ability of the optimized value of Vmax to simulate human 1,4-dioxane metabolism above the concentration
6 that would result in saturation of metabolism. Rat metabolic rate constants were obtained by optimization
7 to simulated data from a two compartment empirical pharmacokinetic model, which was fitted to i.v.
8 exposure data (Young etal. 1978b: 1978a).
9 The Leung and Paustenbach model (1990a) and the Reitz et al. (1990a) model included
10 compartments for the liver and fat, although no tissue-specific concentration data were available to
11 validate dosimetry for these organs. The derivations of human and rat HEAA elimination rate constants
12 were not reported. Since no pharmacokinetics data for 1,4-dioxane in mice were available, mouse
13 metabolic rate constants were allometrically scaled from rat and human values.
3.5.2.3 Fisher et al.
14 A PBPK model was developed by Fisher et al. (1997) to simulate a variety of volatile organic
15 compounds (VOCs, including 1,4-dioxane) in lactating humans. This model was similar in structure to
16 those of Leung and Paustenbach (1990a) and Reitz et al. (1990a) with the addition of elimination of
17 1,4-dioxane to breast milk. Experimental measurements were made for blood:air and milk:air partition
18 coefficients. Other partition coefficient values were taken from Reitz et al. (1990a). The model was not
19 optimized, nor was performance tested against experimental exposure data. Thus, the ability of the model
20 to simulate 1,4-dioxane exposure data is unknown.
3.5.2.4 Sweeney et al.
21 The Sweeney et al. (2008a) model consisted of fat, liver, slowly perfused, and other well perfused
22 tissue compartments. Lung and stomach compartments were used to describe the route of exposure, and
23 an overall volume of distribution compartment was used for calculation of urinary excretion levels of
24 1,4-dioxane and HEAA. Blood, saline, and tissue to air partition coefficient values for 1,4-dioxane were
25 experimentally determined for rats and mice. Average values of the rat and mouse partition coefficients
26 were used for humans. Metabolic constants (VmaxC and Km) for the rat were derived by optimization of
27 data from an i.v. exposure of 1,000 mg/kg (Young etal. 1978b) for inducible metabolism. For uninduced
28 VmaxC estimation, data generated by i.v. exposures to 3, 10, 30, and 100 mg/kg were used (Young et al..
29 1978b). Sweeney et al. (2008a) determined best fit values for VmaxC by fitting to blood data in Young et
30 al. (1978b). The best fit VmaxC values were 7.5, 10.8, and 12.7 mg/hr-kg0 75 for i.v. doses of 3 to 100,
31 300, and 1,000 mg/kg, suggesting a gradual dose dependent increase in metabolic rate over i.v. doses
32 ranging from 3 to 1,000 mg/kg. Although the Sweeney et al. (2008a) model utilized two values for
33 VmaxC (induced and uninduced), the PBPK model does not include a dose-dependent function
17
DRAFT - DO NOT CITE OR QUOTE
-------
1 description of the change of Vmax for i.v. doses between metabolic induced and uninduced exposures.
2 Mouse VmaxC and absorption constants were derived by optimizing fits to the blood 1,4-dioxane
3 concentrations in mice administered nominal doses of 200 and 2,000 mg/kg 1,4-dioxane via gavage in a
4 water vehicle (Young et al. 1978b). The in vitro Vmax values for rats and mice determined by Sweeney
5 et al. (2008a) were scaled to estimate in vivo rates. The scaled and optimized rat VmaxC values were
6 similar. The discrepancy between the scaled and optimized mouse values was larger, which was attributed
7 to possible induction in mice at the lowest dose tested (200 mg/kg). The ratio of optimized/scaled values
8 for the rat was used to adjust the scaled human VmaxC and Km values to projected in vivo values.
9 The Sweeney et al. (2008a) model outputs were compared, by visual inspection, with data not
10 used in fitting model parameters. The model predictions gave adequate match to the 1,4-dioxane
11 exhalation data in rats after a 1,000 mg/kg i.v. dose. 1,4-Dioxane exhalation was overpredicted by a factor
12 of about 3 after a 10 mg/kg i.v. dose. Similarly, the simulations of exhaled 1,4-dioxane after oral dosing
13 were adequate at 1,000 mg/kg and 100 mg/kg (within 50%), but poor at 10 mg/kg (model over predicted
14 by a factor of 5). The model did not adequately fit the human data (Young et al.. 1977a). Using
15 physiological parameters of Brown et al. (1997) and measured partitioning parameters (Sweeney et al..
16 2008a: Leung and Paustenbach. 1990a) with no metabolism, measured blood 1,4-dioxane concentrations
17 reported by Young et al. (1977a) could not be achieved unless the estimated exposure concentration was
18 increased by 2-fold. As expected, inclusion of any metabolism resulted in a decrease in predicted blood
19 concentrations. If estimated metabolism rates were used with the reported exposure concentration, urinary
20 metabolite excretion was also underpredicted (Sweeney et al.. 2008a).
3.5.2.5 Takano et al.
21 More recently, Takano et al. (2010) reported the development of a simplified rat and human
22 pharmacokinetic model. The purpose of this model was to provide a platform for a forward dosimetry
23 calculation using in vivo animal data and in vitro human and animal microsome data to predict the 1,4-
24 dioxane concentrations in humans. The model had three nonphysiological compartments: absorption
25 compartment, metabolizing compartment, and a central compartment. Human metabolic parameters were
26 determined from in vitro data using liver microsomes, coefficients (octanol-water partition coefficient,
27 plasma unbound fraction) derived in silico, and physiological parameters (e..g, hepatic volume and blood
28 flow rate) obtained from the literature. Clearance was described as a first order rate of metabolism from
29 both the metabolizing compartment (e.g. hepatic metabolism) and the central compartment (e.g., renal
30 clearance). This is in contrast to the saturable metabolism used in previous models (Sweeney et al..
31 2008b: Reitz et al.. 1990bV
32 The rat model outputs of Takano et al. (2010) were compared with 1,4-dioxane blood data at the
33 end of exposure in rats treated for 14 days with an oral dose of 500 mg/kg. The model adequately
34 predicted these rat data and showed a minimal amount of 1,4-dioxane remained in the blood 24hrs after
35 the last exposure. The authors performed an in vitro to in vivo extrapolation to estimate human hepatic
36 intrinsic clearance for the human pharmacokinetic model. The ratio of rat in vivo/in vitro measurements
18
DRAFT - DO NOT CITE OR QUOTE
-------
1 (0.0244/0.313) was multiplied by the human in vitro determination (22.9 L/h) to yield 1.76 L/h used in
2 the human pharmacokinetic model. The model was then used to simulate hypothetical human exposures;
3 however, no data were compared with model outputs. Thus, the ability of this model to adequately
4 simulate the available human data is unknown.
3.5.3 Implementation of Published PBPK Models for 1,4-Dioxane
5 As previously described, several pharmacokinetic models have been developed to predict the
6 absorption, distribution, metabolism, and elimination of 1,4-dioxane in rats and humans. Single
7 compartment, empirical models for rats (Young et al.. 1978b: 1978a) and humans (Young et al.. 1977a)
8 were developed to predict blood levels of 1,4-dioxane and urine levels of the primary metabolite, HEAA.
9 PBPK models that describe the kinetics of 1,4-dioxane using biologically realistic flow rates, tissue
10 volumes, enzyme affinities, metabolic processes, and elimination behaviors were also developed
11 (Sweeney et al.. 2008a: Fisher et al.. 1997; Leung and Paustenbach. 1990a: Reitz et al.. 1990a). Most
12 recently, Takano et al. (2010) published a pharmacokinetic model utilizing hepatic volume, blood flow,
13 and an in vitro to in vivo extrapolation method for human intrinsic hepatic clearance.
14 In developing updated toxicity values for 1,4-dioxane the available PBPK models were evaluated
15 for their ability to predict observations made in experimental studies of rat and human exposures to
16 1,4-dioxane (Appendix B). The Reitz et al. (199Qa) and Leung and Paustenbach (199Qa) PBPK models
17 were both developed from a PBPK model of styrene (Ramsey and Andersen. 1984). with the exception of
18 minor differences in the use of partition coefficients and biological parameters. The model code for Leung
19 and Paustenbach (1990a) was unavailable in contrast to Reitz et al. (1990a). The model of Reitz et al.
20 (1990a) was identified for further consideration to assist in the derivation of toxicity values, and the
21 Sweeney et al. (2008a) and Takano et al. (2010) models were also evaluated.
22 The biological plausibility of parameter values in the Reitz et al. (1990a) human model were
23 examined. The model published by Reitz et al. (1990a) was able to predict the only available human
24 inhalation data (50 ppm 1,4-dioxane for 6 hours; Young et al., (1977a)) by increasing (i.e., approximately
25 doubling) the parameter values for human alveolar ventilation (30 L/hour/kg°74), cardiac output (30
26 L/hour/kg°74), and the blood:air partition coefficient (3,650) above the measured values of
27 13 L/minute/kg074 (Brown etal. 1997). 14 L/hour/kg074 (Brown etal. 1997). and 1,825 (Leung and
28 Paustenbach. 1990a). respectively. Furthermore, Reitz et al. (1990a) replaced the measured value for the
29 slowly perfused tissue:air partition coefficient (i.e., muscle—value not reported in manuscript) with the
30 measured liver value (1,557) to improve the fit. Analysis of the Young et al. (1977a) human data
31 suggested that the apparent volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold higher
32 in rats than humans, presumably due to species differences in tissue partitioning or other process not
33 represented in the model. Based upon these observations, several model parameters (e.g.,
34 metabolism/elimination parameters) were re-calibrated using biologically plausible values for flow rates
35 and tissue:air partition coefficients.
19
DRAFT - DO NOT CITE OR QUOTE
-------
1 Appendix B describes all activities that were conducted in the evaluation of the empirical models
2 and the re-calibration and evaluation of the Reitz et al. (1990a) PBPK model to determine the adequacy
3 and preference for the potential use of the models.
4 The evaluation consisted of implementation of the Young et al. (1978b: 1978a: 1977a) empirical
5 rat and human models using the acslXtreme simulation software, re-calibration of the Reitz et al. (1990a)
6 human PBPK model, and evaluation of the model parameters published by Sweeney et al. (2008a). Using
7 the model descriptions and equations given in Young et al. (1978b: 1978a: 1977a). model code was
8 developed for the empirical models and executed, simulating the reported experimental conditions. The
9 model output was then compared with the model output reported in Young et al. (1978b: 1978a: 1977a).
10 The PBPK model of Reitz et al. (1990a) was re-calibrated using measured values for cardiac and
11 alveolar flow rates and tissue:air partition coefficients. The predictions of blood and urine levels of
12 1,4-dioxane and HEAA, respectively, from the re-calibrated model were compared with the empirical
13 model predictions of the same dosimeters to determine whether the re-calibrated PBPK model could
14 perform similarly to the empirical model. As part of the PBPK model evaluation, EPA performed a
15 sensitivity analysis to identify the model parameters having the greatest influence on the primary
16 dosimeter of interest, the blood level of 1,4-dioxane. Variability data for the experimental measurements
17 of the tissue:air partition coefficients were incorporated to determine a range of model outputs bounded
18 by biologically plausible values for these parameters. Model parameters from Sweeney et al. (2008a)
19 were also tested to evaluate the ability of the PBPK model to predict human data following exposure to
20 1,4-dioxane.
21 The rat and human empirical models of Young et al. (1978c. d_; 1977b) were successfully
22 implemented in acslX and perform identically to the models reported in the published papers (Figures
23 B-3, B-4, B-5, B-7, and B-8), with the exception of the lower predicted HEAA concentrations and early
24 appearance of the peak HEAA levels in rat urine. The early appearance of peak HEAA levels cannot
25 presently be explained, but may result from manipulations of kme or other parameters by Young et al.
26 (1978c. d) that were not reported. The lower predictions of HEAA levels are likely due to reliance on a
27 standard urine volume production rate in the absence of measured (but unreported) urine volumes. While
28 the human urinary HEAA predictions were closer to the observed data of Young et al. (1977b). no model
29 output was published in Young et al. (1977b) for comparison. The empirical models were modified to
30 allow for user-defined inhalation exposure levels; however, they were not modified to describe oral
31 exposures due to a lack of adequate human or animal data for parameterization. Additionally, the
32 inhalation Young et al. (1977b) model did not provide adequate fits to the subchronic exposure plasma
33 levels of 1,4-dioxane in rats using the data from the Kasai et al. (2008) study, which is likely due to the
34 absence of a model description for metabolic induction.
35 Several procedures were applied to the human PBPK model to determine if an adequate fit of the
36 model to the empirical model output or experimental observations could be attained using biologically
37 plausible values for the model parameters. The re-calibrated model predictions for blood 1,4-dioxane did
38 not adequately fit the experimental values using measured tissue:air partition coefficients from Leung and
39 Paustenbach (1990b) or Sweeney et al. (2008b) (Figure B-6 and Figure B-7). Use of a slowly perfused
20
DRAFT - DO NOT CITE OR QUOTE
-------
1 tissue:air partition coefficient 4- to 7-fold lower than measured values produces exposure-phase
2 predictions that are much closer to observations, but does not replicate the elimination kinetics
3 (Figure B-8). Re-calibration of the model with upper bounds on the tissue:air partition coefficients results
4 in predictions that are still 2- to 4-fold lower than empirical model prediction or observations
5 (Figure B-10 and Figure B-l 1). Exploration of the model space using an assumption of first-order
6 metabolism (valid for the 50-ppm inhalation exposure) showed that an adequate fit to the exposure and
7 elimination data can be achieved only when unrealistically low values are assumed for the slowly
8 perfused tissue:air partition coefficient (Figure B-14). Artificially low values for the other tissue:air
9 partition coefficients are not expected to improve the model fit, because blood 1,4-dioxane is less
10 sensitive to these parameters than it is to VmaxC and Km. This suggests that the model structure is
11 insufficient to capture the apparent species difference in the blood 1,4-dioxane Vd between rats and
12 humans. Differences in the ability of rat and human blood to bind 1,4-dioxane may contribute to the
13 difference in Vd. However, this is expected to be evident in very different values for rat and human
14 blood:air partition coefficients, which is not the case (). Additionally, the models do not account for
15 induction in metabolism, which may be present in animals repeatedly exposed to 1,4-dioxane. Therefore,
16 some other modification(s) to the Reitz et al. (1990b) model structure may be necessary.
17 Similarly, Sweeney et al. (2008a) also evaluated the available PBPK models (Leung and
18 Paustenbach. 1990a: Reitz et al.. 1990a) for 1,4-dioxane. To address uncertainties and deficiencies in
19 these models, the investigators conducted studies to fill data gaps and reduce uncertainties pertaining to
20 the pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. The following studies were
21 performed:
22 • Partition coefficients, including measurements for mouse blood and tissues (liver, kidney, fat, and
23 muscle) and confirmatory measurements for human blood and rat blood and muscle.
24 • Blood time course measurements in mice conducted for gavage administration of nominal single
25 doses (20, 200, or 2,000 mg/kg) of 1,4-dioxane administered in water.
26 • Metabolic rate constants for rat, mouse, and human liver based on incubations of 1,4-dioxane
27 with rat, mouse, and human hepatocytes and measurement of HEAA.
28 The studies conducted by Sweeney et al. (2008a) resulted in partition coefficients that were
29 consistent with previously measured values and those used in the Leung and Paustenbach (1990a) model.
30 Of noteworthy significance, the laboratory results of Sweeney et al. (2008a) did not confirm the human
31 blood:air partition coefficient Reitz et al. (1990a) reported. Furthermore, Sweeney et al. (2008a) estimated
32 metabolic rate constants (VmaxC and Km) within the range used in the previous models (Leung and
33 Paustenbach. 1990a: Reitz et al.. 1990a). Overall, the Sweeney et al. (2008a) model utilized more rodent
34 in vivo and in vitro data in model parameterization and refinement; however, the model was still unable to
35 adequately predict the human blood data from Young et al. (1977a). The Takano (2010) model was only
36 tested by the authors using a single dose and route of exposure in rats, so the ability of the model to
37 predict over a range of exposures or exposure routes is unknown. Additionally, the human model
38 (Takano et al.. 2010) was not compared to the available published data (1978c. d; 1977b: Young et al..
39 1976b).
21
DRAFT - DO NOT CITE OR QUOTE
-------
3.6 Rat Nasal Exposure via Drinking Water
2 Sweeney et al. (2008a) conducted a rat nasal exposure study to explore the potential for direct
3 contact of nasal tissues with 1,4-dioxane-containing drinking water under bioassay conditions. Two
4 groups of male Sprague Dawley rats (5/group) received drinking water in 45-mL drinking water bottles
5 containing a fluorescent dye mixture (Cell Tracker Red/FluoSpheres). The drinking water for one of these
6 two groups also contained 0.5% 1,4-dioxane, a concentration within the range used in chronic toxicity
7 studies. A third group of five rats received tap water alone (controls). Water was provided to the rats
8 overnight. The next morning, the water bottles were weighed to estimate the amounts of water consumed.
9 Rats were sacrificed and heads were split along the midline for evaluation by fluorescence microscopy.
10 One additional rat was dosed twice by gavage with 2 mL of drinking water containing fluorescent dye
11 (the second dose was 30 minutes after the first dose; total of 4 mL administered) and sacrificed 5 hours
12 later to evaluate the potential for systemic delivery of fluorescent dye to the nasal tissues.
13 The presence of the fluorescent dye mixture had no measurable impact on water consumption;
14 however, 0.5% 1,4-dioxane reduced water consumption by an average of 62% of controls following a
15 single, overnight exposure. Fluorescent dye was detected in the oral cavity and nasal airways of each
16 animal exposed to the Cell Tracker Red/FluoSpheres mixture in their drinking water, including numerous
17 areas of the anterior third of the nose along the nasal vestibule, maxillary turbinates, and dorsal
18 nasoturbinates. Fluorescent dye was occasionally detected in the ethmoid turbinate region and
19 nasopharynx. 1,4-Dioxane had no effect on the detection of the dye. Little or no fluorescence at the
20 wavelength associated with the dye mixture was detected in control animals or in the single animal that
21 received the dye mixture by oral gavage. The investigators concluded that the findings indicate rat nasal
22 tissues are exposed by direct contact with drinking water under bioassay conditions.
22
DRAFT - DO NOT CITE OR QUOTE
-------
4 HAZARD IDENTIFICATION
4.1 Studies in Humans - Epidemiology, Case Reports, Clinical
Controls
1 Case reports of acute occupational poisoning with 1,4-dioxane indicated that exposure to 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 et al.. 1934a). 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.. 1934a: Yant et al.. 1930) did not provide information on the human subjects research ethics
23
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. (1977a) 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.. 1976a).
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.. 1978a: Thiess et al.. 1976a). These studies did not provide
26 evidence 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. (1976a) 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.
24
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. (1978a) conducted a mortality study on workers exposed to 1,4-dioxane at a
27 chemical manufacturing facility in Texas. 1,4-Dioxane exposure was known to occur in a manufacturing
28 area and in a processing unit located 5 miles from the manufacturing plant. Employees who worked
29 between April 1, 1954, and June 30, 1975, were separated into two cohorts based on at least 1 month of
30 exposure in either the manufacturing plant (100 workers) or the processing area (65 workers). Company
31 records and follow-up techniques were used to compile information on name, date of birth, gender,
32 ethnicity, job assignment and duration, and employment status at the time of the study. Date and cause of
33 death were obtained from copies of death certificates and autopsy reports (if available). Exposure levels
34 for each job category were estimated using the 1974 Threshold Limit Value for 1,4-dioxane (i.e., 50 ppm)
35 and 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
25
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 1934a) and two chronic inhalation studies (Kasai et al.. 2009; Torkelson et al.. 1974a). The effects
31 following oral and inhalation exposures are described in detail below.
26
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.1 Oral Toxicity
4.2.1.1 Subchronic 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.. 1934a). 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 Stoner et al. 1,4-Dioxane was evaluated by Stoner et al. (1986) for its
15 ability to induce lung adenoma formation in A/J mice. Six- to 8-week-old male and female A/J mice
16 (16/sex/group) were given 1,4-dioxane by gavage or i.p. injection, 3 times/week for 8 weeks. Total
17 cumulative dose levels were given as 24,000 mg/kg (oral), and 4,800, 12,000, or 24,000 mg/kg (i.p.).
18 Average daily dose estimates were calculated to be 430 mg/kg-day (oral), and 86, 210, or 430 mg/kg-day
19 (i.p.) by assuming an exposure duration of 56 days. The authors indicated that i.p. doses represent the
20 maximum tolerated dose (MTD), 0.5 times the MTD, and 0.2 times the MTD. Mice were killed 24 weeks
21 after initiation of the bioassay, and lungs, liver, kidney, spleen, intestines, stomach, thymus, salivary, and
22 endocrine glands were examined for gross lesions. Histopathology examination was performed if gross
23 lesions were 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 Stott et al. 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. [3H]-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
27
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 Kano et al. In the Kano et al. (2008) study, groups of 6-week-old
7 F344/DuCrj rats (10/sex/group) and Crj:BDFl mice (10/sex/group) were administered 1,4-dioxane (>99%
8 pure) in the drinking water for 13 weeks. The animals were observed daily for clinical signs of toxicity.
9 Food 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
28
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. Nuclear enlargement has not been described elsewhere in published literature; there is a lack of
19 information concerning the nature, severity, and significance of this observation. Thus, the toxicological
20 significance of nuclear enlargement is unknown. Based on these histopathological findings the study
21 authors identified the LOAEL as 126 mg/kg-day and the NOAEL as 52 mg/kg-day.
29
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
9/1 0D
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°
"Data are presented for sacrificed animals.
°p<0.01 by x 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
30
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. As noted above, the toxicological
12 significance of nuclear enlargement is unknown. The incidence data for histopathological lesions in mice
13 are presented in Table 4-2. Based on the changes in the bronchial epithelium in female mice, the authors
14 identified the dose level of 387 mg/kg-day as the LOAEL for mice; the NOAEL was 170 mg/kg-day
15 (Kano et al. 2008).
Table 4-2 Incidence of histopathological lesions in Crj:BDF1 mice exposed to 1,4-dioxane in
drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
86
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
170
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
1/10
0/10
231
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
Female dose
387
0/10
1/10
0/10
0/10
0/10
2/10
0/10
10/10C
0/10
1/10
0/10
585
2/10
0/10
9/1 Oc
0/10
0/10
7/1 Oc
0/10
9/1 Oc
0/10
10/10C
5/1 Ob
882
5/1 Ob
0/10
10/10C
0/10
0/10
9/1 Oc
0/10
9/1 Oc
0/10
10/10C
10/10C
1,570
0/9
5/9b
9/9c
6/9c
9/9c
9/9c
6/9c
9/9c
8/9c
9/9c
9/9c
(mg/kg-day)a
898
3/10
1/10
6/1 Ob
1/1 Oc
0/10
9/1 Oc
0/10
10/10C
0/10
10/10C
7/1 Oc
1,620
3/10
5/1 Ob
10/10C
6/1 Ob
2/10
10/10C
10/10C
10/10C
7/1 Oc
10/10C
10/10C
2,669
7/1 Oc
9/1 Oc
10/10C
6/1 Ob
8/1 Oc
10/10C
10/10C
10/10C
10/10C
9/1 Ob
9/1 Oc
31
DRAFT - DO NOT CITE OR QUOTE
-------
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 et al. Studies (Yamamoto et al.. 1998a: Yamamoto et al..
2 1998b) in rasH2 transgenic mice carrying the human prototype c-Ha-ras gene have been investigated as a
3 bioassay model for rapid carcinogenicity testing. As part of validation studies of this model, 1,4-dioxane
4 was one of many chemicals that were evaluated. RasH2 transgenic mice were Fl offspring of transgenic
5 male C57BLr6J and normal female BALB/cByJ mice. CB6Fi 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 Argus et al. Twenty-six adult male Wistar rats (Argus et al.. 1965a)
2 weighing between 150 and 200 g were exposed to 1,4-dioxane (purity not reported) in the drinking water
3 at a 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
14 organs, particularly the liver and spleen, with large, round, isolated neoplastic cells. In the liver, the
32
DRAFT - DO NOT CITE OR QUOTE
-------
1 distribution of cells in the sinusoids was suggestive of myeloid leukemia. The dose of 640 mg/kg-day
2 tested in this study was a free-standing LOAEL, identified by EPA, for glomerulonephritis in the kidney
3 and histological changes in the liver (hepatocytes with enlarged hyperchromic nuclei, large cells with
4 reduced cytoplasmic basophilia).
5 4.2.1.2.2 Argus et al.; Hoch-Ligeti et al. Five groups (28-32/dose group) of male
6 Sprague Dawley rats (2-3 months of age) weighing 110-230 g at the beginning of the experiment were
7 administered 1,4-dioxane (purity not reported) in the drinking water for up to 13 months at concentrations
8 of 0, 0.75, 1.0, 1.4, or 1.8% (Argus etal. 1973a: Hoch-Ligeti et al.. 1970a). The drinking water intake
9 was determined for each group over a 3-day measurement period conducted at the beginning of the study
10 and twice during the study (weeks were not specified). The rats were killed with ether at 16 months or
11 earlier if nasal tumors were clearly observable. Complete necropsies were apparently performed on all
12 animals, but only data from the nasal cavity and liver were presented and discussed. The nasal cavity was
13 studied histologically only from rats in which gross tumors in these locations were present; therefore,
14 early tumors may have been missed and pre-neoplastic changes were not studied. No statistical analysis of
15 the results was conducted. Assuming a BW of 0.523 kg for an adult male Sprague Dawley rat (U.S. EPA.
16 1988) and a drinking water intake of 30 mL/day as reported by the study authors, dose estimates were 0,
17 430, 574, 803, and 1,032 mg/kg-day. The progression of liver tumorigenesis was evaluated by an
18 additional group of 10 male rats administered 1% 1,4-dioxane in the drinking water (574 mg/kg-day), 5 of
19 which were sacrificed after 8 months of treatment and 5 were sacrificed after 13 months of treatment.
20 Liver tissue from these rats and control rats was processed for electron microscopy examination.
5 Nasal cavity tumors were observed upon gross examination in six rats (1/30 in the 0.75% group,
6 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
7 the tumors visible either at the tip of the nose, bulging out of the nasal cavity, or on the back of the nose
8 covered by intact or later ulcerated skin. As the tumors obstructed the nasal passages, the rats had
9 difficulty breathing and lost weight rapidly. No neurological signs or compression of the brain were
10 observed. In all cases, the tumors were squamous cell carcinomas with marked keratinization and
11 formation of keratin pearls. Bony structure was extensively destroyed in some animals with tumors, but
12 there was no invasion into the brain. In addition to the squamous carcinoma, two adenocarcinomatous
13 areas were present. One control rat had a small, firm, well-circumscribed tumor on the back of the nose,
14 which proved to be subcutaneous fibroma. The latency period for tumor onset was 329-487 days.
15 Evaluation of the latent periods and doses received did not suggest an inverse relationship between these
16 two parameters.
17 Argus et al. (1973a) studied the progression of liver tumorigenesis by electron microscopy of
18 liver tissues obtained following interim sacrifice at 8 and 13 months of exposure (5 rats/group,
19 574 mg/kg-day). The authors reported qualitatively that the first change observed in the liver was an
20 increase in the size of the nucleus of the hepatocytes, mostly in the periportal area. Precancerous changes
21 were characterized by disorganization of the rough endoplasmic reticulum, an increase in smooth
22 endoplasmic reticulum, and a decrease in glycogen and increase in lipid droplets in hepatocytes. These
23 changes increased in severity in the hepatocellular carcinomas in rats exposed to 1,4-dioxane for
24 13 months.
33
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 study authors provided quantitation for the numbers of incipient liver
9 tumors and hepatomas in rats from this study (treated for 13 months and observed at 13-16 months) as
10 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. (1973a).
11 Treatment with all dose levels of 1,4-dioxane induced marked kidney alterations, but quantitative
12 incidence data were not provided. Qualitatively, the changes indicated glomerulonephritis and
13 pyelonephritis, with characteristic epithelial proliferation of Bowman's capsule, periglomerular fibrosis,
14 and distension of tubules. No kidney tumors were found. No tumors were found in the lungs. One rat at
15 the 1.4% treatment level showed early peripheral adenomatous change of the alveolar epithelium and
16 another rat in the same group showed papillary hyperplasia of the bronchial epithelium. The lowest dose
17 tested (430 mg/kg-day) was considered a LOAEL by EPA for hepatic and renal effects in this study.
18 4.2.1.2.3 Hoch-Ligeti and Argus. Hoch-Ligeti and Argus (1970a) provided a brief
19 account of the results of exposure of guinea pigs to 1,4-dioxane. A group of 22 male guinea pigs (neither
20 strain nor age provided) was administered 1,4-dioxane (purity not provided) in the drinking water for at
21 least 23 months and possibly up to 28 months. The authors stated that the concentration of 1,4-dioxane
22 was regulated so that normal growth of the guinea pigs was maintained, and varied 0.5-2% (no further
23 information provided). The investigators further stated that the amount of 1,4-dioxane received by the
24 guinea pigs over a 23-month period was 588-635 g. Using a reference BW of 0.89 kg for male guinea
25 pigs in a chronic study (U.S. EPA. 1988) and assuming an exposure period of 700 days (23 months), the
26 guinea pigs received doses between 944 and 1,019 mg 1,4-dioxane/kg-day. A group often untreated
27 guinea pigs served as controls. All animals were sacrificed within 28 months, but the scope of the
28 postmortem examination was not provided.
34
DRAFT - DO NOT CITE OR QUOTE
-------
1 Nine treated guinea pigs showed peri- or intrabronchial epithelial hyperplasia and nodular
2 mononuclear infiltration in the lungs. Also, two guinea pigs had carcinoma of the gallbladder, three had
3 early hepatomas, and one had an adenoma of the kidney. Among the controls, four guinea pigs had
4 peripheral mononuclear cell accumulation in the lungs, and only one had hyperplasia of the bronchial
5 epithelium. One control had formation of bone in the bronchus. No further information was presented in
6 the brief narrative of this study. Given the limited reporting of the results, a NOAEL or LOAEL value
7 was not provided for this study.
8 4.2.1.2.4 Kociba et al. Groups of 6-8-week-old Sherman rats (60/sex/dose level)
9 were administered 1,4-dioxane (purity not reported) in the drinking water at levels of 0 (controls), 0.01,
10 0.1, or 1.0% for up to 716 days (Kociba et al.. 1974a). The drinking water was prepared twice weekly
11 during the first year of the study and weekly during the second year of the study. Water samples were
12 collected periodically and analyzed for 1,4-dioxane content by routine gas liquid chromatography. Food
13 and water were available ad libitum. Rats were observed daily for clinical signs of toxicity, and BWs
14 were measured twice weekly during the first month, weekly during months 2-7, and biweekly thereafter.
15 Water consumption was recorded at three different time periods during the study: days 1-113, 114-198,
16 and 446-460. Blood samples were collected from a minimum of five male and five female control and
17 high-dose rats during the 4th, 6th, 12th, and 18th months of the study and at termination. Each sample
18 was analyzed for packed cell volume, total erythrocyte count, hemoglobin, and total and differential WBC
19 counts. Additional endpoints evaluated included organ weights (brain, liver, kidney, testes, spleen, and
20 heart) and gross and microscopic examination of major tissues and organs (brain, bone and bone marrow,
21 ovaries, pituitary, uterus, mesenteric lymph nodes, heart, liver, pancreas, spleen, stomach, prostate, colon,
22 trachea, duodenum, kidneys, esophagus, jejunum, testes, lungs, spinal cord, adrenals, thyroid,
23 parathyroid, nasal turbinates, and urinary bladder). The number of rats with tumors, hepatic tumors,
24 hepatocellular carcinomas, and nasal carcinomas were analyzed for statistical significance with Fisher's
25 Exact test (one-tailed), comparing each treatment group against the respective control group. Survival
26 rates were compared using %2 Contingency Tables and Fisher's Exact test. Student's test was used to
27 compare hematological parameters, body and organ weights, and water consumption of each treatment
28 group with the respective control group.
8 Male and female rats in the high-dose group (1% in drinking water) consumed slightly less water
9 than controls. BW gain was depressed in the high-dose groups relative to the other groups almost from
10 the beginning of the study (food consumption data were not provided). Based on water consumption and
11 BW data for specific exposure groups, Kociba et al. (1974a) calculated mean daily doses of 9.6, 94, and
12 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
13 the 0.01, 0.1, and 1.0% concentration levels, respectively. Treatment with 1,4-dioxane significantly
14 increased mortality among high-dose males and females beginning at about 2-4 months of treatment.
15 These rats showed degenerative changes in both the liver and kidneys. From the 5th month on, mortality
16 rates of control and treated groups were not different. There were no treatment-related alterations in
17 hematological parameters. At termination, the only alteration in organ weights noted by the authors was a
18 significant increase in absolute and relative liver weights in male and female high-dose rats (data not
19 shown). Histopathological lesions were restricted to the liver and kidney from the mid- and high-dose
35
DRAFT - DO NOT CITE OR QUOTE
-------
1 groups and consisted of variable degrees of renal tubular epithelial and hepatocellular degeneration and
2 necrosis (no quantitative incidence data were provided). Rats from these groups also showed evidence of
3 hepatic regeneration, as indicated by hepatocellular hyperplastic nodule formation and evidence of renal
4 tubular epithelial regenerative activity (observed after 2 years of exposure). These changes were not seen
5 in controls or in low-dose rats. The authors determined a LOAEL of 94 mg/kg-day based on the liver and
6 kidney effects in male rats. The corresponding NOAEL value was 9.6 mg/kg-day.
7 Histopathological examination of all the rats in the study revealed a total of 132 tumors in
8 114 rats. Treatment with 1% 1,4-dioxane in the drinking water resulted in a significant increase in the
9 incidence of hepatic tumors (hepatocellular carcinomas in six males and four females). In addition, nasal
10 carcinomas (squamous cell carcinoma of the nasal turbinates) occurred in one high-dose male and two
11 high-dose females. Since 128 out of 132 tumors occurred in rats from the 12th to the 24th month, Kociba
12 et al. (1974a) assumed that the effective number of rats was the number surviving at 12 months, which
13 was also when the first hepatic tumor was noticed. The incidences of liver and nasal tumors from Kociba
14 et al. (1974a) are presented in Table 4-4. Tumors in other organs were not elevated when compared to
15 control incidence and did not appear to be related to 1,4-dioxane administration.
Table 4-4 Incidence of liver and nasal tumors in male and female Sherman rats (combined)
treated with 1,4-dioxane in the drinking water for 2 years
Dose in mg/kg-day
(average of male and
female dose)
0
14
121
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. (1974a).
16 The high-dose level was the only dose that increased the formation of liver tumors over control
17 (males 1,015 mg/kg-day; females 1,599 mg/kg-day) and also caused significant liver and kidney toxicity
18 in these animals. The mid-dose group (males 94 mg/kg-day; females 148 mg/kg-day) experienced hepatic
19 and renal degeneration and necrosis, as well as regenerative proliferation in hepatocytes and renal tubule
20 epithelial cells. No increase in tumor formation was seen in the mid-dose group. No toxicity or tumor
21 formation was observed in either sex in the low-dose (males 9.6 mg/kg-day; females 19 mg/kg-day) group
22 of rats.
36
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.
37
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.
blncidence significantly elevated compared to control by Fisher's Exact test (p < 0.05) performed for this review.
Source: NCI (19781
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)).
38
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%)e
1/33 (3%)
Females (mg/kg-day)a
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/34 (0%)a
0/31 (0%)T
350
10/35(29%)c
10/33(30%)e
640
8/35 (23%)c
11/32(34%)e
"Tumor 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 (19781
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) yieldedp-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 observed in dosed and control mice but did not comment on the significance
26 of the effect. EPA concluded the LOAEL for 1,4-dioxane in mice was 380 mg/kg-day based on the
39
DRAFT - DO NOT CITE OR QUOTE
-------
1 increased incidence of pneumonia and rhinitis in female mice; a NOAEL was not established in this
2 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%)D
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%)D
0/50 (0%)D
380
12/48(25%)c
21/48 (44%)c
860
29/37 (78%)c
35/37 (95%)c
"Tumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
0 p £ 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (19781
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 B6C3Fi 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.
40
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.2.1.2.6 Kano etal.; 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.. 1994a). 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 (1994a) 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 of 1,4-Dioxane (U.S. EPA. 2009a). 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. (1994a) 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.. 1994a) were re-examined and
19 recategorized as appropriate according to current pathological diagnostic criteria (see references in Kano
20 et al. (2009)).
21 Groups of F344/DuCrj rats (50/sex/dose level) were exposed to 1,4-dioxane (>99% pure) in the
22 drinking water at levels of 0, 200, 1,000, or 5,000 ppm for 2 years. Groups of Crj:BDFl mice
23 (50/sex/dose level) were similarly exposed in the drinking water to 0, 500, 2,000, or 8,000 ppm of
24 1,4-dioxane. The high doses were selected based on results from the Kano et al. (2008) 13-week drinking
25 water study so as not to exceed the maximum tolerated dose (MTD) in that study. Both rats and mice
26 were 6 weeks old at the beginning of the study. Food and water were available ad libitum. The animals
27 were observed daily for clinical signs of toxicity; and BWs were measured once per week for 14 weeks
28 and once every 2 weeks until the end of the study. Food consumption was measured once a week for
29 14 weeks and once every 4 weeks for the remainder of the study. The investigators used data from water
30 consumption and BW to calculate an estimate of the daily intake of 1,4-dioxane (mg/kg-day) by male and
31 female rats and mice. Kano et al. (2009) reported a calculated mean ± standard deviation for the daily
41
DRAFT - DO NOT CITE OR QUOTE
-------
1 doses of 1,4-dioxane for the duration of the study. Male rats received doses of approximately 0, 11±1,
2 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
3 received doses of 0, 49±5, 191±21, or 677±74 mg/kg-day and female mice received 0, 66±10, 278±40, or
4 964±88 mg/kg-day. For the remainder of this document, including the dose-response analysis, the mean
5 calculated intake values are used to identify dose groups. The Kano et al. (2009) study was conducted in
6 accordance with the Organization for Economic Co-operation and Development (OECD) Principles for
7 Good Laboratory Practice (GLP).
8 No information was provided as to when urine samples were collected. Blood samples were
9 collected only at the end of the 2-year study (Yamazaki. 2006). Hematology analysis included RBCs,
10 hemoglobin, hematocrit, MCV, platelets, WBCs and differential WBCs. Serum biochemistry included
11 total protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only), phospholipid, ALT, AST,
12 LDH, LAP, ALP, y-glutamyl transpeptidase (GGT), CPK, urea nitrogen, creatinine (rat only), sodium,
13 potassium, chloride, calcium, and inorganic phosphorous. Urinalysis parameters were pH, protein,
14 glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ weights (brain, lung,
15 liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and gross necropsy and
16 histopathologic examination of tissues and organs were performed on all animals (skin, nasal cavity,
17 trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart, tongue, salivary glands, esophagus,
18 stomach, small and large intestine, liver, pancreas, kidney, urinary bladder, pituitary, thyroid, adrenal,
19 testes, epididymis, seminal vesicle, prostate, ovary, uterus, vagina, mammary gland, brain, spinal cord,
20 sciatic nerve, eye, Harderian gland, muscle, bone, and parathyroid). Dunnett's test and %2 test were used to
21 assess the statistical significance of changes in continuous and discrete variables, respectively.
22 For rats, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
23 study. Both male and female rats in the high dose groups (274 and 429 mg/kg-day, respectively) exhibited
24 slower growth rates and terminal body weights that were significantly different (p < 0.05) compared to
25 controls. A statistically significant reduction in terminal BWs was observed in high-dose male rats (5%, p
26 < 0.01) and in high-dose female rats (18%, p < 0.01) (Kano et al.. 2009). Food consumption was not
27 significantly affected by treatment in male or female rats; however, water consumption in female rats
28 administered 18 mg/kg-day was significantly greater (p < 0.05) .
29 All control and exposed rats lived at least 12 months following study initiation (Yamazaki. 2006);
30 however, survival at the end of the 2-year study in the high dose group of male and female rats (274 and
31 429 mg/kg-day, respectively) was approximately 50%, which was significantly different compared to
32 controls. The investigators attributed these early deaths to the increased incidence in nasal tumors and
33 peritoneal mesotheliomas in male rats and nasal and hepatic tumors in female rats. (Yamazaki. 2006).
34 Several hematological changes were noted in the JBRC report (1998): Decreases in RBC (male
35 rats only), hemoglobin, hematocrit, and MCV; and increases in platelets in high-dose groups were
36 observed (JBRC. 1998). These changes (except for MCV) also occurred in mid-dose males. With the
37 exception of a 23% decrease in hemoglobin in high-dose male rats and a 27% increase in platelets in
38 high-dose female rats, hematological changes were within 15% of control values. Significant changes in
39 serum chemistry parameters occurred only in high-dose rats (males: increased phospholipids, AST, ALT,
42
DRAFT - DO NOT CITE OR QUOTE
-------
1 LDH, ALP, GOT, CPK, potassium, and inorganic phosphorus and decreased total protein, albumin, and
2 glucose; females: increased total bilirubin, cholesterol, phospholipids, AST, ALT, LDH, GGT, ALP,
3 CPK, and potassium, and decreased blood glucose) (JBRC. 1998). Increases in serum enzyme activities
4 ranged from <2- to 17-fold above control values, with the largest increases seen for ALT, AST, and GGT.
5 Urine pH was significantly decreased at 274 mg/kg-day in male rats (not tested at other dose levels) and
6 at 83 and 429 mg/kg-day in female rats (JBRC. 1998). Also, blood in the urine was seen in female rats at
7 83 and 429 mg/kg-day (JBRC. 1998). In male rats, relative liver weights were increased at 55 and
8 274 mg/kg-day (Kano et al. 2009). In female rats, relative liver weight was increased at 429 mg/kg-day
9 (Kano et al.. 2009).
10 Microscopic examination of the tissues showed nonneoplastic alterations in the nasal cavity, liver,
11 and kidneys mainly in high-dose rats and, in a few cases, in mid-dose rats (Table 4-8 and Table 4-9).
12 Alterations in high-dose (274 mg/kg-day) male rats consisted of nuclear enlargement and metaplasia of
13 the olfactory and respiratory epithelia, atrophy of the olfactory epithelium, hydropic changes and sclerosis
14 of the lamina propria, adhesion, and inflammation. In female rats, nuclear enlargement of the olfactory
15 epithelium occurred at doses > 83 mg/kg-day, and nuclear enlargement and metaplasia of the respiratory
16 epithelium, squamous cell hyperplasia, respiratory metaplasia of the olfactory epithelium, hydropic
17 changes and sclerosis of the lamina propria, adhesion, inflammation, and proliferation of the nasal gland
18 occurred at 429 mg/kg-day. Alterations were seen in the liver at > 55 mg/kg-day in male rats (spongiosis
19 hepatis, and clear and mixed cell foci) and at 429 mg/kg-day in female rats (spongiosis hepatis, cyst
20 formation, and mixed cell foci). Nuclear enlargement of the renal proximal tubule occurred in males at
21 274 mg/kg-day and in females at > 83 mg/kg-day (JBRC. 1998). As noted previously, nuclear
22 enlargement has not been described elsewhere in published literature; there is a lack of information
23 concerning the nature, severity, and significance of this observation. Thus, the toxicological significance
24 of nuclear enlargement is unknown.
43
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8 Incidence of histopathological lesions in male F344/DuCrj rats exposed to 1,4-dioxane
in drinking water for 2 years
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
Spongiosis hepatis; liver
Clear cell foci; liver0'9
Acidophilic cell foci; liver0'0
Basophilic cell foci; liver0'0
Mixed-cell foci; liver0'0
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
12/50
3/50
12/50
7/50
2/50
0/50
11
0/50
0/50
0/50
0/50
11/50
0/50
0/50
0/50
0/50
0/50
20/50
3/50
8/50
11/50
8/50
0/50
55
0/50
0/50
0/50
5/50T
20/50
0/50
0/50
1/50
0/50
0/50
25/50T
9/50
7/50
8/50
14/508
0/50
274
26/508
31/508
2/50
38/508
43/50
36/50
46/50
44/50
48/50
13/50
40/50
8/50
5/50
16/50T
13/508
50/50
Data presented for all animals, including animals that became moribund or died before the end of the study.
bDose levels from Kano et al. (2009).
°Data from Kano et al. (20091
Data 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.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1_994a) and JBRC (1998) were re-
examined according to updated criteria for liver lesions and were afterwards classified as either hepatocellular adenoma
or altered hepatocellular foci in Kano et al. (2009).
Sources: Kano et al. (2009) and JBRC (1998).
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 epithelium0
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
Spongiosis hepatis; liver0
Cyst formation; liver0
Acidophilic cell foci; liver0'0
Basophilic cell foci; liver0'0
Clear cell foci; liver0'0
Mixed-cell foci; liver0'0
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
0/50
0/50
1/50
23/50
1/50
1/50
0/50
18
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
1/50
27/50
1/50
1/50
0/50
83
0/50
0/50
0/50
28/508
2/50
1/50
0/50
0/50
0/50
1/50
0/50
1/50
1/50
1/50
31/50
5/50
3/50
6/50
429
13/508
35/508
5/50
39/50
42/50
40/50
46/50
48/50
46/50
15/50
11/50
20/50
8/50
1/50
8/508
4/50
11/50T
39/50
DRAFT - DO NOT CITE OR QUOTE
44
-------
aData presented for all animals, including animals that became moribund or died before the end of the study.
"Dose levels from Kano et al. (2QQ9).
°Data from Kano et al. (2QQ9).
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.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1_994a) and JBRC (1998) were re-
examined according to updated criteria for liver lesions and were afterwards classified as either hepatocellular adenoma
or altered hepatocellular foci in Kano et al. (2009).
Sources: Kano et al. (2009) and JBRC (1998).
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, as
7 noted previously, it is unclear whether these alterations represent adverse toxicological effects.
8 Hematological effects noted in male rats given 55 and 274 mg/kg-day (decreased RBCs, hemoglobin,
9 hematocrit, increased platelets) were within 20% of control values. In female rats decreases in
10 hematological effects were observed in the high dose group (429 mg/kg-day). A reference range database
11 for hematological effects in laboratory animals (Wolford et al.. 1986) indicates that a 20% change in these
12 parameters may fall within a normal range (10th-90th percentile values) and may not represent a
13 treatment-related effect of concern. Liver lesions were also seen at a dose of 55 mg/kg-day in male rats;
14 these changes are likely to be associated with liver tumorigenesis. Clear and mixed-cell foci are
15 commonly considered preneoplastic changes and would not be considered evidence of noncancer toxicity.
16 The nature of spongiosis hepatis as a preneoplastic change is less well understood (Bannasch. 2003;
17 Karbe and Kerlin. 2002a: Stroebel et al.. 1995). Spongiosis hepatis is a cyst-like lesion that arises from
18 the perisinusoidal (Ito) cells (PSC) of the liver. It is commonly seen in aging rats, but has been shown to
19 increase in incidence following exposure to hepatocarcinogens. Spongiosis hepatis can be seen in
20 combination with preneoplastic foci in the liver or with hepatocellular adenoma or carcinoma and has
21 been considered a preneoplastic lesion (Bannasch. 2003; Stroebel et al.. 1995). This change can also be
22 associated with hepatocellular hypertrophy and liver toxicity and has been regarded as a secondary effect
23 of some liver carcinogens (Karbe and Kerlin. 2002a). In the case of the JBRC (1998) study, spongiosis
24 hepatis was associated with other preneoplastic changes in the liver (clear and mixed-cell foci). No other
25 lesions indicative of liver toxicity were seen in this study; therefore, spongiosis hepatis was not
26 considered indicative of noncancer effects. Serum chemistry changes (increases in total protein, albumin,
27 and glucose; decreases in AST, ALT, LDH, and ALP, potassium, and inorganic phosphorous) were
28 observed in both male and female rats (JBRC. 1998) in the high dose groups, 274 and 429 mg/kg-day,
29 respectively.
30 Significantly increased incidences of liver tumors (adenomas and carcinomas) and tumors of the
31 nasal cavity occurred in high-dose male and female rats (Table 4-10 and Table 4-11) treated with
32 1,4-dioxane for 2 years (Kano et al.. 2009). The first liver tumor was seen at 85 weeks in high-dose male
33 rats and 73 weeks in high-dose female rats (vs. 101-104 weeks in lower dose groups and controls)
45
DRAFT - DO NOT CITE OR QUOTE
-------
1 (Yamazaki. 2006). In addition, a significant increase (p < 0.01, Fisher's Exact test) in mesotheliomas of
2 the peritoneum was seen in high-dose males (28/50 versus 2/50 in controls). Mesotheliomas were the
3 single largest cause of death among high-dose male rats, accounting for 12 of 28 pretermination deaths
4 (Yamazaki. 2006). Also, in males, there were increasing trends in mammary gland fibroadenoma and
5 fibroma of the subcutis, both statistically significant (p < 0.01) by the Peto test of dose-response trend.
6 Females showed a significant increasing trend in mammary gland adenomas (p < 0.01 by Peto's test). The
7 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'°
0/50
0/50
1/50
Peritoneum
Mesothelioma
2/50
2/50
5/50
28/50a'D
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
"Statistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
""Significantly different from control by Fisher's exact test (p < 0.01).
°Significantly different from control by Fisher's exact test (p < 0.05).
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
Table 4-11 Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane in drinking water
for 2 years
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'D
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'D
10/50a'D
48/50a'D
aSignificantly different from control by Fisher's exact test (p < 0.01).
bStatistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
8 For mice, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
9 study. Similar to rats, the growth rates of male and female mice were slower than controls and terminal
10 body weights were lower for the mid (p < 0.01 for males administered 191 mg/kg-day and p < 0.05 for
11 females administered 278 mg/kg-day) and high doses (p < 0.05 for males and females administered 677
12 and 964 mg/kg-day, respectively). There were no differences in survival rates between control and treated
13 male mice; however, survival rates were significantly decreased compared to controls for female mice in
14 the mid (278 mg/kg-day, approximately 40% survival) and high (964 mg/kg-day, approximately 20%
46
DRAFT - DO NOT CITE OR QUOTE
-------
1 survival) dose groups. The study authors attributed these early female mouse deaths to the significant
2 incidence of hepatic tumors, and Kano et al. (2009) reported tumor incidence for all animals in the study
3 (N=50), including animals that became moribund or died before the end of the study. Additional data on
4 survival rates of mice were provided in a personal communication from Dr. Yamazaki (2006). who
5 reported that the survival of mice was low in all male groups (31/50, 33/50, 25/50 and 26/50 in control,
6 low-, mid-, and high-dose groups, respectively) and particularly low in high-dose females (29/50, 29/50,
7 17/50, and 5/50 in control, low-, mid-, and high-dose groups, respectively). These deaths occurred
8 primarily during the second year of the study. Survival at 12 months in male mice was 50/50, 48/50,
9 50/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively. Female mouse survival at
10 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively
11 (Yamazaki. 2006). Furthermore, these deaths were primarily tumor related. Liver tumors were listed as
12 the cause of death for 31 of the 45 pretermination deaths in high-dose female Crj:BDFl mice (Yamazaki.
13 2006). For mice, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
14 study. Similar to rats, the growth rates of male and female mice were slower than controls and terminal
15 body weights were lower for the mid (p < 0.01 for males administered 191 mg/kg-day and p < 0.05 for
16 females administered 278 mg/kg-day) and high doses (p < 0.05 for males and females administered 677
17 and 964 mg/kg-day, respectively).
18 Food consumption was not significantly affected, but water consumption was reduced 26% in
19 high-dose male mice and 28% in high-dose female mice. Final BWs were reduced 43% in high-dose male
20 mice and 15 and 45% in mid- and high-dose female mice, respectively. Male mice showed increases in
21 RBC counts, hemoglobin, and hematocrit, whereas in female mice, there was a decrease in platelets in
22 mid- and high-dose rats. With the exception of a 60% decrease in platelets in high-dose female mice,
23 hematological changes were within 15% of control values. Serum AST, ALT, LDH, and ALP activities
24 were significantly increased in mid- and high-dose male mice, whereas LAP and CPK were increased
25 only in high-dose male mice. AST, ALT, LDH, and ALP activities were increased in mid- and high-dose
26 female mice, but CPK activity was increased only in high-dose female mice. Increases in serum enzyme
27 activities ranged from less than two- to sevenfold above control values. Glucose and triglycerides were
28 decreased in high-dose males and in mid- and high-dose females. High-dose female mice also showed
29 decreases in serum phospholipid and albumin concentrations (not reported in males). Blood calcium was
30 lower in high-dose females and was not reported in males. Urinary pH was decreased in high-dose males,
31 whereas urinary protein, glucose, and occult blood were increased in mid- and high-dose female mice.
32 Relative and absolute lung weights were increased in high-dose males and in mid- and high-dose females
33 (JBRC. 1998). Microscopic examination of the tissues for nonneoplastic lesions showed significant
34 alterations in the epithelium of the respiratory tract, mainly in high-dose animals, although some changes
35 occurred in mid-dose mice
36 Table 4-12 and Table 4-13). Commonly seen alterations included nuclear enlargement
37 (toxicological significance unknown), atrophy, and inflammation of the epithelium. Other changes
38 observed included nuclear enlargement of the proximal tubule of the kidney and angiectasis in the liver in
39 high-dose male mice.
47
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
31/50e
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
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. (2QQ9).
°Data from Kano et al. (2QQ9).
dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
ep < 0.01 by x2 test.
Sources: Kano et al. (2009) and JBRC (1998).
Table 4-13 Incidence of histopathological 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
"Data presented for all animals, including animals that became moribund or died before the end of the study.
"Dose levels from Kano et al. (2QQ9).
°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 Liver angiectasis, an abnormal dilatation and/or lengthening of a blood or lymphatic vessel, was seen in
6 male mice given 1,4-dioxane at a dose of 677 mg/kg-day.
48
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Treatment with 1,4-dioxane resulted in an increase in the formation of liver tumors (adenomas
and carcinomas) in male and female mice. The incidence of hepatocellular adenoma was statistically
increased in male mice in the mid-dose group only. The incidence of male mice with hepatocellular
carcinoma or either tumor type (adenoma or carcinoma) was increased in the low, mid, and high-dose
groups. The appearance of the first liver tumor occurred in male mice at 64, 74, 63, and 59 weeks in the
control, low- mid-, and high-dose groups, respectively (Yamazaki. 2006). In female mice, increased
incidence was observed for hepatocellular carcinoma in all treatment groups, while an increase in
hepatocellular adenoma incidence was only seen in the 66 and 278 mg/kg-day dose groups (Table 4-14).
The appearance of the first liver tumor in female mice occurred at 95, 79, 71, and 56 weeks in the control,
low-, mid-, and high-dose groups, respectively (Yamazaki. 2006). The tumor incidence data presented for
male and female mice in Table 4-14 are based on reanalyzed sample data presented in Kano et al. (2009)
that included lesions in animals that became moribund or died prior to the completion of the 2-year study.
Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and carcinomas in
control male and female BDF1 mice from ten 2-year bioassays at the JBRC. For female mice, out of 499
control mice, the incidence rates were 4.4% for hepatocellular adenomas and 2.0% for hepatocellular
carcinomas. Kano et al. (2009) reported a 10% incidence rate for hepatocellular adenomas and a 0%
incidence rate for hepatocellular carcinomas in control female BDF1. The background incidence rates for
male BDF1 mice were 15% and 22.8% for hepatocellular adenomas and carcinomas, respectively, out of
500 control mice in ten 2-year bioassays (Katagiri et al.. 1998). Background rates for B6C3Fi mice
evaluated by the National Toxicology Program are similar (10.3% and 21.3% for hepatocellular
adenomas and carcinomas in male mice, respectively; 4.0% and 4.1% for hepatocellular adenomas and
carcinomas in female mice, respectively) to the BDF1 mice background rates observed by JBRC
(Haseman et al.. 1984). Thus, the BDF1 mouse is not particularly sensitive compared to the commonly
used B6C3Fi strain and indicates that the results obtained by JBRC are reasonable.
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'D
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'°
46/50a'b
25
26
Significantly different from control by Fisher's exact test (p < 0.01).
""Statistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
""Significantly different from control by Fisher's exact test (p < 0.05).
Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
A weight of evidence evaluation of the carcinogenicity studies presented in Section 4.2.1.2is
located in Section 4.7 and Table 4-19.
49
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.2 Inhalation Toxicity
4.2.2.1 Subchronic Inhalation Toxicity
1 4.2.2.1.1 Fairley et al. 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) (Fairley et al.. 1934a).
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.
1 4.2.2.1.2 Kasai et al. Male and female 6-week-old F344/DuCrj rats (10/sex/group)
2 were exposed to nominal concentrations of 0 (clean air), 100, 200, 400, 800, 1,600, 3,200, or 6,400 ppm
3 (0, 360, 720, 1,400, 2,900, 5,800, 1,2000, and 23,000 mg/m3, respectively) of vaporized 1,4-dioxane
4 (>99% pure) for 6 hours/day, 5 days/week, for 13 weeks in whole body inhalation chambers (Kasai et al..
5 2008). Each inhalation chamber housed 20 individual cages for 10 males and 10 females. During
6 exposure, the concentration of 1,4-dioxane vapor was determined every 15 minutes by gas
7 chromatography. In addition, during exposure, animals received food and water ad libitum and the
8 following data were collected: 1) clinical signs and mortality (daily); 2) BW and food intake (weekly); 3)
9 urinary parameters using Ames reagent strips (measured during week 13 of the exposure); and 4)
10 1,4-dioxane content in plasma from three rats of both sexes (measured on the third day of exposure during
11 weeks 12 and 13 at 1 hour after termination). At the end of the 13-week exposure period or at the time of
12 an animal's death during exposure, all organs were collected, weighed, and evaluated for macroscopic
13 lesions. Histopathological evaluations of organs and tissues were conducted in accordance with the
14 OECD test guidelines, including all tissues of the respiratory tract. Liver sections from male and female
15 rats exposed to 800, 1,600 and 3,200 ppm of 1,4-dioxane were also analyzed for foci (in the absence of
16 tumor formation) by immunohistochemical expression of glutathione S-transferase placental form
50
DRAFT - DO NOT CITE OR QUOTE
-------
1 (GST-P). Hematological and clinical chemistry parameters were measured using blood collected from the
2 abdominal aorta of rats following an overnight fasting at the end of the 13-week exposure period. The
3 measured hematological and clinical chemistry parameters included: red blood cell count, hemoglobin,
4 hematocrit, MCV, AST, ALT, glucose, and triglyceride. Statistically significant differences (p-value of
5 0.05) between 1,4-dioxane and clean air exposed groups were determined by study authors using
6 Dunnett's test or %2 test.
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
7 Table 4-16. Briefly, terminal BWs were significantly decreased in both sexes at 200 ppm; and
8 additionally in females at 800 and 1,600 ppm. Statistically significant increases in several organ weights
9 were observed, including lung (> 1,600 ppm, males; >200 ppm, females); liver (>800 ppm, both sexes),
10 and kidneys (3,200 ppm, males; >800 ppm, females). Statistically significant changes in hematological
11 parameters and clinical chemistry were observed in both sexes at 3,200 ppm including increased levels of
12 hemoglobin ALT, RBC, AST ,and MCV. In females only, at 3,200 ppm, increased levels of hematocrit
13 was noted; and in males at this exposure concentration decreased levels of glucose and triglyceride were
14 observed, in addition to slightly decreased urinary protein. However, the urinary protein data were not
15 shown in this study. At 200 ppm, an increased AST level in females was noted. Blood plasma levels of
16 1,4-dioxane were also evaluated and in both sexes, a linear increase in 1,4-dioxane levels was detected at
17 exposure concentrations of 400 ppm and above. The highest blood levels of 1,4-dioxane were detected in
18 females.
19 Exposure and/or sex-related histopathology findings also reported by the study authors included
20 nuclear enlargement of the nasal respiratory, nasal olfactory, tracheal, and bronchial epithelium; vacuolic
21 change in the olfactory and bronchial epithelium; atrophy of the nasal epithelium; hydropic change in the
22 proximal tubules of the kidney; and single-cell necrosis and centrilobular swelling in the liver. Table 4-17
23 presents a summary of these histopathological lesions, including incidence and severity data. Further
24 microscopic evaluation of liver tissue revealed GST-P positive liver foci in both sexes at 3,200 ppm (3/10
25 males, 2/10 females) and in females at 1,600 ppm (4/10).
26 The study authors determined nuclear enlargement in the respiratory epithelium as the most
27 sensitive lesion and a LOAEL value of 100 ppm was identified by the study authors based on the
28 incidence data of this lesion in both male and female rats. Nuclear enlargement has only been reported by
29 this research center; there is a lack of information concerning the nature, severity, and significance of this
30 observation. Thus, as with the oral assessment, the toxicological significance of nuclear enlargement is
31 unknown.
51
DRAFT - DO NOT CITE OR QUOTE
-------
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 (%)
0 (clean air)
323 ± 14
0.310 ± 0.011
2.610 ±0.069
0.589± 0.016
0 (clean air)
187 ±5
0.402 ±0.013
2.353 ±0.081
0.647± 0.014
1
100
323 ± 14
0.312 ±
0.007
2.697 ±
0.092
0.596 ±
0.021
1
100
195 ±8
0.402 ±
0.015
2.338 ±
0.092
0.631 ±
0.019
Males3
,4-dioxane vapor concentration (ppm)
200
304 ± 11C
0.325 ±
0.008C
2.613 ±
0.084
0.612 ±
0.013
400
311 ± 19
0.320 ±
0.009
2.666 ±
0.080
0.601 ±
0.020
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
,4-dioxane vapor concentration (ppm)
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 ±
0.1 39b
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.
°p < 0.05 by Dunnett's test.
Source: Kasai et al. (2008)
DRAFT - DO NOT CITE OR QUOTE
52
-------
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/ul)
Hemoglobin (g/dl)
Hematocrit (%)
MCV (fl)
AST (IU/I)
ALT (IU/I)
Glucose (mg/dl)
Triglyceride (mg/dl)
Females
Red blood cell (10B/ul)
Hemoglobin (g/dl)°
Hematocrit (%)a
MCV(fl)a
AST (IU/I) a
ALT (IU/I) a
Glucose (mg/dl) a
Triglyceride (mg/dl)
0 (clean air)
9.55 ±0.17
16.0 ±0.2
46.2 ± 1.2
48.4 ±0.7
73 ±8
27 ±3
197 ± 17
125± 17
0 (clean air)
8.77 ±0.23
16.2 ±0.3
46.0 ± 1.5
52.5 ±0.7
64 ±6
23 ±3
143 ± 18
45±5
1
100
9.53 ±0.24
16.1 ±0.4
46.3 ± 1.3
48.6 ±0.7
75 ± 14
27 ±4
206 ± 13
148 ±37
1
100
8.69 ±0.21
16.0 ±0.3
45.5 ± 1.2
52.3 ±0.7
65 ±3
21 ±2
144 ± 18
48 ±6
Males3
,4-dioxane vapor concentration (ppm)
200
9.54 ±0.18
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
,4-dioxane vapor concentration (ppm)
200
8.73 ±0.25
16.3 ±0.4
45.8± 1.7
52.4 ±0.7
74 ± 14C
26 ± 10
137 ±9
42 ±4
400
8.88 ±0.21
16.2 ±0.4
46.5 ± 1.5
52.4 ±0.8
69 ±5
25 ±3
140 ± 15
47 ±8
800
8.68 ±0.69
16.2 ±0.6
45.4 ±3.6
52.3 ±0.6
68 ±6
24 ±4
141 ± 15
42 ±6
1,600
8.86 ±0.16
16.3 ±0.2
46.2 ±0.7
52.1 ±0.5
70 ±5
25 ±3
139± 11
39 ±7
3,200
9.15±0.12D
16.6±0.2C
47.5±0.6C
52.0 ±0.7
76 ±5°
30 ±3°
139± 18
42 ±7
aData are presented for 10 sacrificed animals.
bp < 0.01 by Dunnett's test.
°p < 0.05 by Dunnett's test.
dData were reported for 9/10 female rats.
Source: Kasai et al. (2008)
DRAFT - DO NOT CITE OR QUOTE
53
-------
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+)
5/1 Oa
(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
6/10°
(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
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
Females3
1,4-dioxane vapor concentration (ppm)
0 (clean air)
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
100
5/1 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+)
5/1 Oa
(5, 1+)
0/10
0/10
0/10
1,600
10/10C
(10,2+)
10/10C
(7, 1+;
3, 2+)
7/1 Oc
(7, 1+)
0/10
9/1 Oc
(9, 1+)
3/10
(3, 1+)
5/1 Oa
(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+)
6/10°
(6, 1+)
"Data are presented for sacrificed animals.
bValues listed are the number of animals with the indicated lesion. Values in parentheses, are the number of lesion bearing animals 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. (2008)
54
DRAFT - DO NOT CITE OR QUOTE
-------
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.. 1974a). 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.
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.
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 tumors were seen in
17 any rat and the presence or absence of nasal cavity tumors was not evaluated.
55
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)
2 weighing 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.. 1994a). 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, %2test, 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.
56
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 by the study authors to be unaltered. 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. As noted earlier, the
16 toxicological significance of nuclear enlargement is unknown.
17 Tumor development was observed in the nasal cavity (squamous cell carcinoma), liver
18 (hepatocellular adenoma and carcinoma), peritoneum (peritoneal mesothelioma), kidney (renal cell
19 carcinoma), mammary gland (fibroadenoma and adenoma), Zymbal gland (adenoma), and subcutaneous
20 tissue (subcutis fibroma). Tumor incidences with a dose-dependent, statistically significant positive trend
21 (Peto's test) included nasal squamous cell carcinoma, hepatocellular adenoma, peritoneal mesothelioma,
22 mammary gland fibroadenoma, and Zymbal gland adenoma. Renal cell carcinoma was also identified as
23 statistically significant with a positive dose-dependent trend; however, no tumor incidences were reported
24 at 50 and 250 ppm. At 1,250 ppm, significant increases in nasal squamous cell carcinoma, hepatocellular
25 adenoma, and peritoneal mesothelioma were observed. At 250 ppm, significant increases in peritoneum
26 mesothelioma and subcutis fibroma were observed. Table 4-21 presents a summary of tumor incidences
27 found in this study. Further characterizations of neoplasms revealed nasal squamous cell carcinoma
28 occurred at the dorsal area of the nose (levels 1-3) marked by keratinization and the progression of growth
29 into surrounding tissue. Peritoneal mesotheliomas were characterized by complex branching structures
30 originating from the mesothelium of the scrotal sac. Invasive growth into surrounding tissues was
31 occasionally observed for peritoneal mesotheliomas.
57
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
250
29
376 ± 38
0.45 ±0.18
3.58 ±0.52
0.81 ±0.13
1250
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. (2008)
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
250
28
7.9± 1.0
13.4± 1.9
40.7 ±5.1
51.8±2.3
17.1 ± 1.2
95± 116
49 ±30
145 ± 171
10±8
7.1 ±0.6
1250
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. (2008)
DRAFT - DO NOT CITE OR QUOTE
58
-------
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)
0 (clean air)
50
250
1,250
Nuclear enlargement; nasal respiratory epithelium
0/50
50/50a 48/50a
38/50a
Squamous cell metaplasia; nasal respiratory epithelium
0/50
0/50
7/50^
44/50a
Squamous cell hyperplasia; nasal respiratory epithelium
0/50
0/50
1/50
10/50a
Inflammation; nasal respiratory epithelium
13/50
9/50
7/50
39/50a
Nuclear enlargement; nasal olfactory epithelium
0/50
48/50a 48/50a
45/50a
Respiratory metaplasia; nasal olfactory epithelium
11/50
34/50a 49/50a 48/50a
Atrophy; nasal olfactory epithelium
0/50
40/50a 47/50a 48/50a
Inflammation; nasal olfactory epithelium
0/50
2/50
32/50a
34/50a
Hydropic change; lamina propria
0/50
2/50
36/50a
49/50a
Sclerosis; lamina propria
0/50
0/50
22/50a
40/50a
Proliferation; nasal gland
0/50
1/50
0/50
6/50°
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 x2 test.
bp < 0.05 by x2 test.
Source: Kasai et al. (2QQ9).
DRAFT - DO NOT CITE OR QUOTE
59
-------
Table 4-21 Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane vapor by
whole-body inhalation for 2 years
1 ,4-dioxane vapor concentration (ppm)
Effect
Nasal squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Mammary gland adenoma
Zymbal gland adenoma
Subcutis fibroma
0 (clean air)
0/50
1/50
0/50
0/50
2/50
1/50
0/50
0/50
1/50
50
0/50
2/50
0/50
0/50
4/50
2/50
0/50
0/50
4/50
250
1/50
3/50
1/50
0/50
14/50a
3/50
0/50
0/50
9/50a
1,250
6/50°'c
21/50a'c
2/50
4/50c
41/50a'c
5/50°
1/50
4/50c
5/50
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
°p < 0.01 by Peto's test for dose-related trend.
dp < 0.05 by Peto's test for dose-related trend.
Source: Kasai et al. (2QQ9).
4.2.3 Initiation/Promotion Studies
1 Bronaugh et al. (1982b) reported more 1,4-dioxane absorption from occluded than unoccluded
2 surfaces. Due to the volatility of 1,4-dioxane, the unoccluded skin paint studies are unreliable; however,
3 all of the available skin paint initiation/promotion studies are summarized below.
4.2.3.1 Bulletal.
4 Bull et al. (1986) tested 1,4-dioxane as a cancer initiator in mice using oral, subcutaneous, and
5 topical routes of exposure. A group of 40 female SENCAR mice (6-8 weeks old) was administered a
6 single dose of 1,000 mg/kg 1,4-dioxane (purity >99%) by gavage, subcutaneous injection, or topical
7 administration (vehicle was not specified). A group of rats was used as a vehicle control (number of
8 animals not specified). Food and water were provided ad libitum. Two weeks after administration of
9 1,4-dioxane, 12-O-tetradecanoylphorbol-13-acetate (TPA) (1.0 (ig in 0.2 mL of acetone) was applied to
10 the shaved back of mice 3 times/week for a period of 20 weeks. The yield of papillomas at 24 weeks was
11 selected as a potential predictor of carcinoma yields at 52 weeks following the start of the promotion
12 schedule. Acetone was used instead of TPA in an additional group of 20 mice in order to determine
13 whether a single dose of 1,4-dioxane could induce tumors in the absence of TPA promotion.
14 1,4-Dioxane did not increase the formation of papillomas compared to mice initiated with vehicle
15 and promoted with TPA, indicating lack of initiating activity under the conditions of the study. Negative
16 results were obtained for all three exposure routes. A single dose of 1,4-dioxane did not induce tumors in
17 the absence of TPA promotion.
60
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.3.2 Kingetal.
1 1,4-Dioxane was evaluated for complete carcinogenicity and tumor promotion activity in mouse
2 skin (King et al. 1973a). In the complete carcinogenicity study, 0.2 mL of a solution of 1,4-dioxane
3 (purity not specified) in acetone was applied to the shaved skin of the back of Swiss Webster mice
4 (30/sex) 3 times/week for 78 weeks. Acetone was applied to the backs of control mice (30/sex) for the
5 same time period. In the promotion study, each animal was treated with 50 ug of dimethylbenzanthracene
6 1 week prior to the topical application of the 1,4-dioxane solution described above (0.2 mL, 3 times/week,
7 78 weeks) (30 mice/sex). Acetone vehicle was used in negative control mice (30/sex). Croton oil was
8 used as a positive control in the promotion study (30/sex). Weekly counts of papillomas and suspect
9 carcinomas were made by gross examination. 1,4-Dioxane was also administered in the drinking water
10 (0.5 and 1%) to groups of Osborne-Mendel rats (35/sex/group) and B6C3Fi mice for 42 weeks (control
11 findings were only reported for 34 weeks).
12 1,4-Dioxane was negative in the complete skin carcinogenicity test using dermal exposure. One
13 treated female mouse had malignant lymphoma; however, no papillomas were observed in male or female
14 mice by 60 weeks. Neoplastic lesions of the skin, lungs, and kidney were observed in mice given the
15 promotional treatment with 1,4-dioxane. In addition, the percentage of mice with skin tumors increased
16 sharply after approximately 10 weeks of promotion treatment. Significant mortality was observed when
17 1,4-dioxane was administered as a promoter (only 4 male and 5 female mice survived for 60 weeks), but
18 not as a complete carcinogen (22 male and 25 female mice survived until 60 weeks). The survival of
19 acetone-treated control mice in the promotion study was not affected (29 male and 26 female mice
20 survived until 60 weeks); however, the mice treated with croton oil as a positive control experienced
21 significant mortality (0 male and 1 female mouse survived for 60 weeks). The incidence of mice with
22 papillomas was similar for croton oil and 1,4-dioxane; however, the tumor multiplicity (i.e., number of
23 tumors/mouse) was higher for the croton oil treatment.
24 Oral administration of 1,4-dioxane in drinking water caused appreciable mortality in rats, but not
25 mice, and increased weight gain in surviving rats and male mice. Histopathological lesions (i.e.,
26 unspecified liver and kidney effects) were also reported in exposed male and female rats; however, no
27 histopathological changes were indicated for mice.
28 1,4-Dioxane was demonstrated to be a tumor promoter, but not a complete carcinogen in mouse
29 skin, in this study. Topical administration for 78 weeks following initiation with dimethylbenzanthracene
30 caused an increase in the incidence and multiplicity of skin tumors in mice. Tumors were also observed at
31 remote sites (i.e., kidney and lung), and survival was affected. Topical application of 1,4-dioxane for
32 60 weeks in the absence of the initiating treatment produced no effects on skin tumor formation or
33 mortality in mice.
61
DRAFT - DO NOT CITE OR QUOTE
-------
4.2.3.3 Lundberg et al.
1 Lundberg et al. (1987) evaluated the tumor promoting activity of 1,4-dioxane in rat liver. Male
2 Sprague Dawley rats (8/dose group, 19 for control group) weighing 200 g underwent a partial
3 hepatectomy followed 24 hours later by an i.p. injection of 30 mg/kg diethylnitrosamine (DEN) (initiation
4 treatment). 1,4-Dioxane (99.5% pure with 25 ppm butylated hydroxytoluene as a stabilizer) was then
5 administered daily by gavage (in saline vehicle) at doses of 0, 100, or 1,000 mg/kg-day, 5 days/week for
6 7 weeks. Control rats were administered saline daily by gavage, following DEN initiation. 1,4-Dioxane
7 was also administered to groups of rats that were not given the DEN initiating treatment (saline used
8 instead of DEN). Ten days after the last dose, animals were sacrificed and liver sections were stained for
9 GOT. The number and total volume of GGT-positive foci were determined.
10 1,4-Dioxane did not increase the number or volume of GGT-foci in rats that were not given the
11 DEN initiation treatment. The high dose of 1,4-dioxane (1,000 mg/kg-day) given as a promoting
12 treatment (i.e., following DEN injection) produced an increase in the number of GGT-positive foci and
13 the total foci volume. Histopathological changes were noted in the livers of high-dose rats. Enlarged,
14 foamy hepatocytes were observed in the midzonal region of the liver, with the foamy appearance due to
15 the presence of numerous fat-containing cytoplasmic vacuoles. These results suggest that cytotoxic doses
16 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.
17 Pregnant female Sprague Dawley rats (18-20 per dose group) were given 1,4-dioxane (99% pure,
18 0.7% acetal) by gavage in water at doses of 0, 0.25, 0.5, or 1 mL/kg-day, corresponding to dose estimates
19 of 0, 250, 500, or 1,000 mg/kg-day (density of 1,4-dioxane is approximately 1.03 g/mL) (Giavini et al..
20 1985a). The chemical was administered at a constant volume of 3 mL/kg on days 6-15 of gestation. Food
21 consumption was determined daily and BWs were measured every 3 days. The dams were sacrificed with
22 chloroform on gestation day 21 and the numbers of corpora lutea, implantations, resorptions, and live
23 fetuses were recorded. Fetuses were weighed and examined for external malformations prior to the
24 evaluation of visceral and skeletal malformations (Wilson's free-hand section method and staining with
25 Alizarin red) and a determination of the degree of ossification.
26 Maternal weight gain was reduced by 10% in the high-dose group (1,000 mg/kg-day). Food
27 consumption for this group was 5% lower during the dosing period, but exceeded control levels for the
28 remainder of the study. No change from control was observed in the number of implantations, live
29 fetuses, or resorptions; however, fetal birth weight was 5% lower in the highest dose group (p < 0.01).
30 1,4-Dioxane exposure did not increase the frequency of major malformations or minor anomalies and
31 variants. Ossification of the sternebrae was reduced in the 1,000 mg/kg-day dose group (p < 0.05). The
32 study authors suggested that the observed delay in sternebrae ossification combined with the decrease in
62
DRAFT - DO NOT CITE OR QUOTE
-------
1 fetal birth weight indicated a developmental delay related to 1,4-dioxane treatment. NOAEL and LOAEL
2 values of 500 and 1,000 mg/kg-day were identified from this study by EPA and based on delayed
3 ossification of the sternebrae and reduced fetal BWs.
4.4 Other Duration or Endpoint Specific Studies
4.4.1 Acute and Short-term Toxicity
4 The acute (< 24 hours) and short-term toxicity studies (<30 days) of 1,4-dioxane in laboratory
5 animals are summarized in Table 4-22. Several exposure routes were employed in these studies, including
6 dermal application, drinking water exposure, gavage, vapor inhalation, and i.v. or i.p. injection.
4.4.1.1 Oral Toxicity
7 Mortality was observed in many acute high-dose studies, and LD50 values for 1,4-dioxane were
8 calculated for rats, mice, and guinea pigs (Pozzani et al.. 1959; HF Jr et al.. 1941; Laug et al.. 1939).
9 Clinical signs of CNS depression were observed, including staggered gait, narcosis, paralysis, coma, and
10 death (Nelson. 1951; Laug et al.. 1939; Schrenk and Yant. 1936; deNavasquez. 1935). Severe liver and
11 kidney degeneration and necrosis were often seen in acute studies (JBRC. 1998; David. 1964; Kesten et
12 al.. 1939: Laugetal. 1939: Schrenk and Yant. 1936: deNavasquez. 1935). JBRC (1998) additionally
13 reported histopathological lesions in the nasal cavity and the brain of rats following 2 weeks of exposure
14 to 1,4-dioxane in the drinking water.
4.4.1.2 Inhalation Toxicity
15 Acute and short-term toxicity studies (all routes) are summarized in Table 4-18. Mortality
16 occurred in many high-concentration studies (Pozzani et al.. 1959: Nelson. 1951: Wirth and Klimmer.
17 1936). Inhalation of 1,4-dioxane caused eye and nasal irritation, altered respiration, and pulmonary edema
18 and congestion (Yant et al.. 1930). Clinical signs of CNS depression were observed, including staggered
19 gait, narcosis, paralysis, coma, and death (Nelson. 1951: Wirth and Klimmer. 1936). Liver and kidney
20 degeneration and necrosis were also seen in acute and short-term inhalation studies (Drew et al.. 1978:
21 Fairlev et al.. 1934aV
63
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 LD5o
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
3-10 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)
LD50= 6,400 mg/kg
(14,200 ppm)
LDso (mg/kg):
rat = 7, 120
LDso (mg/kg):
mouse = 5,900
rat = 5,400
guinea pig - 4,030
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
(1990a)
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
rS-fi-Paco
1,000 ppm
Drew et al.
(1978)
exposure
DRAFT - DO NOT CITE OR QUOTE
64
-------
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
LC5o = 51.3 mg/L
8,400 ppm
0.5% by volume
10,000 ppm
Nelson
(1951)
Pozzani et
al. (1959)
Klimmer
(1936)
Yant et al.
(1930)
Fairley et
al.(1934a)
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
aLowest 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
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: Kanada et al.. 1994: Goldberg et al.. 1964: Knoefel. 1935).
65
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 et al.
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.
66
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
67
DRAFT - DO NOT CITE OR QUOTE
-------
1 Table 4-24 for in vitro and in vivo tests, respectively. 1,4-Dioxane has been tested for genotoxic
2 potential using in vitro assay systems with prokaryotic organisms, non-mammalian eukaryotic organisms,
3 and mammalian cells, and in vivo assay systems using several strains of rats and mice. In the large
4 majority of in vitro systems, 1,4-dioxane was not genotoxic. Where a positive genotoxic response was
5 observed, it was generally observed in the presence of toxicity. Similarly, 1,4-dioxane was not genotoxic
6 in half of the available in vivo studies. 1,4-Dioxane did not bind covalently to DNA in a single study with
7 calf thymus DNA. Several investigators have reported that 1,4-dioxane caused increased DNA synthesis
8 indicative of cell proliferation. Overall, the available literature indicates that 1,4-dioxane is nongenotoxic
9 or weakly genotoxic. However, it is important to note that three of the negative studies reported using
10 closed systems to control for evaporation of the test substance (McGregor et al.. 1991b: Zimmermann et
11 al.. 1985b: Nestmann et al.. 1984R
12 Negative findings were reported for mutagenicity in in vitro assays with the prokaryotic
13 organisms Salmonella typhimurium, Escherichia coli, and Photobacterium phosphoreum (Mutatox assay)
14 (Morita and Hayashi. 1998: Hellmer and Bolcsfoldi. 1992: Kwanetal. 1990: Khudoley et al.. 1987:
15 Nestmann et al.. 1984a: Haworth et al.. 1983: Stottetal. 1981) (Table 4-23). In in vitro assays with
16 nonmammalian eukaryotic organisms, negative results were obtained for the induction of aneuploidy in
17 yeast (Saccharomyces cerevisiae) and in the sex-linked recessive lethal test in Drosophila melanogaster
18 (Yoon et al.. 1985: Zimmermann et al.. 1985a). In the presence of toxicity, positive results were reported
19 for meiotic nondisjunction in Drosophila (Munoz and Barnett. 2002).
20 The ability of 1,4-dioxane to induce genotoxic effects in mammalian cells in vitro has been
21 examined in model test systems with and without exogenous metabolic activation and in hepatocytes that
22 retain their xenobiotic-metabolizing capabilities. 1,4-Dioxane was reported as negative in the mouse
23 lymphoma cell forward mutation assay (Morita and Hayashi. 1998: McGregor et al.. 1991a). 1,4-Dioxane
24 did not produce chromosomal aberrations or micronucleus formation in Chinese hamster ovary (CHO)
25 cells (Morita and Hayashi. 1998: Galloway et al.. 1987a). Results were negative in one assay for sister
26 chromatid exchange (SCE) in CHO (Morita and Hayashi. 1998) and were weakly positive in the absence
27 of metabolic activation in another (Galloway et al.. 1987a). In rat hepatocytes, 1,4-dioxane exposure in
28 vitro caused single-strand breaks in DNA at concentrations also toxic to the hepatocytes (Sina et al..
29 1983) and produced a positive genotoxic response in a cell transformation assay with BALB/3T3 cells
30 also in the presence of toxicity (Sheu et al.. 1988).
31 1,4-Dioxane was not genotoxic in the majority of available in vivo mammalian assays (Table 4-
32 24). Studies of micronucleus formation following in vivo exposure to 1,4-dioxane produced mostly
33 negative results, including studies of bone marrow micronucleus formation in B6C3Fi, BALB/c, CBA,
34 and C57BL6 mice (McFee et al.. 1994: Mirkova. 1994a: Tinwell and Ashby. 1994) and micronucleus
35 formation in peripheral blood of CD1 mice (Morita and Hayashi. 1998: Morita. 1994). Mirkova (1994a)
36 reported a dose-related increase in the incidence of bone marrow micronuclei in male and female C57BL6
37 mice 24 or 48 hours after administration of 1,4-dioxane. At a sampling time of 24 hours, a dose of
38 450 mg/kg produced no change relative to control, while doses of 900, 1,800, and 3,600 mg/kg increased
39 the incidence of bone marrow micronuclei by approximately two-, three-, and fourfold, respectively. A
40 dose of 5,000 mg/kg also increased the incidence of micronuclei by approximately fourfold at 48 hours.
68
DRAFT - DO NOT CITE OR QUOTE
-------
1 This compares with the negative results for BALB/c male mice tested in the same study at a dose of
2 5,000 mg/kg and sampling time of 24 hours. Tinwell and Ashby (1994) could not explain the difference
3 in response in the mouse bone marrow micronucleus assay with C57BL6 mice obtained in their
4 laboratory (i.e., non-significant 1.6-fold increase over control) with the dose-related positive findings
5 reported by Mirkova (Mirkova. 1994a) using the same mouse strain, 1,4-dioxane dose (3,600 mg/kg) and
6 sampling time (24 hours). Morita and Hayashi (1998) demonstrated an increase in micronucleus
7 formation in hepatocytes following 1,4-dioxane dosing and partial hepatectomy to induce cellular mitosis.
8 DNA single-strand breaks were demonstrated in hepatocytes following gavage exposure to female rats
9 (Kitchin and Brown. 1990a).
10 Roy et al. (2005a) examined micronucleus formation in male CD1 mice exposed to 1,4-dioxane
11 to confirm the mixed findings from earlier mouse micronucleus studies and to identify the origin of the
12 induced micronuclei. Mice were administered 1,4-dioxane by gavage at doses of 0, 1,500, 2,500, and
13 3,500 mg/kg-day for 5 days. The mice were also implanted with 5-bromo-2-deoxyuridine
14 (BrdU)-releasing osmotic pumps to measure cell proliferation in the liver and to increase the sensitivity of
15 the hepatocyte assay. The frequency of micronuclei in the bone marrow erythrocytes and in the
16 proliferating BrdU-labeled hepatocytes was determined 24 hours after the final dose. Significant
17 dose-related increases in micronuclei were seen in the bone-marrow at all the tested doses (>
18 1,500 mg/kg-day). In the high-dose (3,500-mg/kg) mice, the frequency of bone marrow erythrocyte
19 micronuclei was about 10-fold greater than the control frequency. Significant dose-related increases in
20 micronuclei were also observed at the two highest doses (> 2,500 mg/kg-day) in the liver.
21 Antikinetochore (CREST) staining or pancentromeric fluorescence in situ hybridization (FISH) was used
22 to determine the origin of the induced micronuclei. The investigators determined that 80-90% of the
23 micronuclei in both tissues originated from chromosomal breakage; small increase in micronuclei
24 originating from chromosome loss was seen in hepatocytes. Dose-related statistically significant
25 decreases in the ratio of bone marrow polychromatic erythrocytes (PCE):normochromatic erythrocytes
26 (NCE), an indirect measure of bone marrow toxicity, were observed. Decreases in hepatocyte
27 proliferation were also observed. Based on these results, the authors concluded that at high doses
28 1,4-dioxane exerts genotoxic effects in both the mouse bone marrow and liver; the induced micronuclei
29 are formed primarily from chromosomal breakage; and 1,4-dioxane can interfere with cell proliferation in
30 both the liver and bone marrow. The authors noted that reasons for the discrepant micronucleus assay
31 results among various investigators was unclear, but could be related to the inherent variability present
32 when detecting moderate to weak responses using small numbers of animals, as well as differences in
33 strain, dosing regimen, or scoring criteria.
34 1,4-Dioxane did not affect in vitro or in vivo DNA repair in hepatocytes or in vivo DNA repair in
35 the nasal cavity (Goldsworthy et al.. 1991; Stott et al.. 1981). but increased hepatocyte DNA synthesis
36 indicative of cell proliferation in several in vivo studies (Miyagawa et al.. 1999; Uno et al.. 1994;
37 Golds worthy et al.. 1991; Stott etal. 1981). 1,4-Dioxane caused a transient inhibition of RNA
38 polymerase A and B in the rat liver (Kurl etal.. 1981). indicating a negative impact on the synthesis of
39 ribosomal and messenger RNA (DNA transcription). Intravenous administration of 1,4-dioxane at doses
69
DRAFT - DO NOT CITE OR QUOTE
-------
1 of 10 or 100 mg/rat produced inhibition of both polymerase enzymes, with a quicker and more complete
2 recovery of activity for RNA polymerase A, the polymerase for ribosomal RNA synthesis.
3 1,4-Dioxane did not covalently bind to DNA under in vitro study conditions (Woo et al.. 1977c).
4 DNA alkylation was also not detected in the liver 4 hours following a single gavage exposure
5 (1,000 mg/kg) in male Sprague Dawley rats (Stott et al.. 1981).
6 Rosenkranz and Klopman (1992) analyzed 1,4-dioxane using the computer automated structure
7 evaluator (CASE) structure activity method to predict its potential genotoxicity and carcinogenicity. The
8 CASE analysis is based on information contained in the structures of approximately 3,000 chemicals
9 tested for endpoints related to mutagenic/genotoxic and carcinogenic potential. CASE selects descriptors
10 (activating [biophore] or inactivating [biophobe] structural fragments) from a learning set of active and
11 inactive molecules. Using the CASE methodology, Rosenkranz and Klopman (1992) predicted that
12 1,4-dioxane would be inactive for mutagenicity in several in vitro systems, including Salmonella,
; O J J ? O "
13 induction of chromosomal aberrations in CHO cells, and unscheduled DNA synthesis in rat hepatocytes.
14 1,4-Dioxane was predicted to induce SCE in cultured CHO cells, micronuclei formation in rat bone
15 marrow, and carcinogenicity in rodents.
16 Gene expression profiling in cultured human hepatoma HepG2 cells was performed using DNA
17 microarrays to discriminate between genotoxic and other carcinogens (van Delft et al.. 2004). Van Delft
18 et al. (2004) examined this method using a training set of 16 treatments (nine genotoxins and seven
19 nongenotoxins) and a validation set (three and three), with discrimination models based on Pearson
20 correlation analyses for the 20 most discriminating genes. As reported by the authors (van Delft et al..
21 2004). the gene expression profile for 1,4-dioxane indicated a classification of this chemical as a
22 "nongenotoxic" carcinogen, and thus, 1,4-dioxane was included in the training set as a "nongenotoxic"
23 carcinogen. The accuracy for carcinogen classification using this method ranged from 33 to 100%,
24 depending on which chemical data sets and gene expression signals were included in the analysis.
70
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. (1984a)
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 NDd
injection
4.75%
2% in sucrose
media
35,000 ppm in
feed, 7 days or
50,000 ppm
(5% in water)
by injection
Zimmerman
etal.
(1985a)
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 _
(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. (1991a)
Morita and
Hayashi
(1998)
DRAFT - DO NOT CITE OR QUOTE
71
-------
Test system
Endpoint
Test conditions
Results3
Without With
activation activation
Dose0
Source
BALB/3T3 cells
Cell
transformation
48-Hour exposure
followed by 4 weeks
incubation; 13 day
exposure followed by
2.5 weeks incubation
+Tf
ND°
0.5 mg/mL
Sheu et al.
(1988)
CHO cells
BrdU was added 2 hours
after 1,4-dioxane addition;
SCE chemical treatment was
2 hours with S9 and
25 hours without S9
CHO cells
Chromosomal
aberration
Cells were harvested 8-
12 hours or 18-26 hours
after treatment (time of
first mitosis)
•m son .m/mi Gallowayet
10,520 ug/mL
CHO cells
SCE
3 hour pulse treatment;
followed by continuous
treatment of BrdU for
23 or 26 hours
Morita and
5,000 ug/mL 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
Calf thymus
DMA
5 hour pulse treatment or
... . 44 hour continuous
Micronucleus . . . ..
, .. treatment; cells were
formation harvested 42 hours
following exposure
„ . . Incubation with
,. ^OVatenMA microsomes from
binding to DMA 3.metny|cno|anthrene
treated rats
5,000 ug/mL
0.04 pmol/mg
DMA (bound)
Morita and
Hayashi
(1998)
Woo et al.
(1977c)
a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity, ND = no data. Endogenous metabolic activation
is not applicable for in vivo studies.
bLowest effective dose for positive results/highest dose tested for negative results; ND = no data.
°A dose-related decrease in viability was observed with 0, 2.4, 8.1, 51.7, and 82.8% mortality at concentrations of 1, 1.5,
2, 3, and 3.5%, respectively. In mature oocytes, meiotic nondisjunction was decreased at 2, 3, and 3.5%; however, a
dose-response trend was not evident.
dExogenous metabolic activation not used for most tests of fungi and many mammalian cell types in vitro, or in vivo
studies in mammals, due to endogenous metabolic ability in many of these systems.
eCell viability was 98, 57, 54, 31, and 34% of control at concentrations 0, 0.03, 0.3, 10, and 30 mM. DMA damage was
observed at 0.3, 3, 10, and 30 mM; however, no dose-response trend was observed for the extent of DMA damage
(severity score related to the elution rate).
'For the 13-day exposure, relative survival was 92, 85, 92, and 61 % of control for concentrations of 0.25, 0.5, 1, and 2
mg/mL, respectively. A significant increase in transformation frequency was observed at the highest dose level (2
mg/mL). Similar results were observed for the 48-hour exposure, with increased transformation frequency seen at
concentrations of 2, 3, and 4 mg/mL. Concentrations >2 mg/mL also caused a significant decrease in cell survival
(relative survival ranged between 6 and 52% of control).
9The highest concentration tested (10,520 ug/L) produced a 27% increase in the number of SCE/cell in the absence of
S9mix. No effect was seen at lower doses (1050 and 3500 ug/L) in the absence of S9 mix or at any concentration
level (1050, 3500, 10,500 ug/L) tested in the presence of S9.
DRAFT - DO NOT CITE OR QUOTE
72
-------
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
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
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 + (C57BL6)d
polychromatic erythrocytes /DAI R/ \
24 or 48 hours after dosing ^ '
Two i.p. injections (1/day);
micronucleated
reticulocytes measured 24,
48, and 72 hours after the
2nd dose
Gavage, partial
hepatectomy 24 hours after
dosing, hepatocytes +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 +g
dosing
Drinking water; thymidine
incorporation with
hydroxyurea to repress
normal DMA synthesis
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
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
(1990a)
Stott et al.
(1981)
McFee et al.
(1994)
Mirkova
(1994a)
Morita (1994)
Morita and
Hayashi
(1998)
Morita and
Hayashi
(1998)
Tinwell and
Ashby (1994)
Roy et al.
(2005a)
Roy et
al.(2005a)
Stott et al.
(1981)
Goldsworthy
et al. (1991)
Goldsworthy
et al. (1991)
DRAFT - DO NOT CITE OR QUOTE
73
-------
Test system
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Endpoint
Replicative DNA
synthesis (i.e., cell
proliferation) in
hepatocytes
Replicative DNA
synthesis (i.e., cell
proliferation) in nasal
epithelial cells
RNA synthesis;
inhibition of RNA
polymerase A and B
DNA synthesis in
hepatocytes
DNA synthesis in
hepatocytes
DNA synthesis in
hepatocytes
Test Conditions
Gavage and drinking water
exposure; thymidine
incorporation
Drinking water exposure;
thymidine incorporation
i.v. injection; activity
measured in isolated
hepatocytes
Gavage; thymidine and
BrdU incorporation
Thymidine incorporation
Drinking water; thymidine
incorporation
Results3 Dose0
+h 1,000 mg/kg for
M o ,,,QQL- 24 or48 hours;
(1—2 -week „ .-„„ „ .
exoosure) 1.500 mg/kg-day
exposure) for1or2 weeks
1,500 mg/kg-day
for 2 weeks
+' 10 mg/rat
+J 1,000 mg/kg
±k 2,000 mg/kg
+i 1,000 mg/kg-day
for 1 1 weeks
Source
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, ND = no data. Endogenous metabolic
activation is not applicable for in vivo studies.
bLowest effective dose for positive results/highest dose tested for negative results; ND = no data.
°Rats were given doses of 0, 168, 840, 2,550, or 4,200 mg/kg at 4 and 21 hours prior to sacrifice. A 43 and 50%
increase in the fraction of DMA eluted was observed for doses of 2,550 and 4,200 mg/kg, respectively. Alkaline
elution of DMA was not significantly different from control in the two lowest dose groups (168 and 840 mg/kg).
dA dose-related increase in the incidence of bone marrow micronuclei was observed in male and female C57BL6 mice
24 or 48 hours after administration of 1,4-dioxane. A dose of 450 mg/kg produced no change relative to control,
while doses of 900, 1,800, 3,600, and 5,000 mg/kg increased the incidence of bone marrow micronuclei by
approximately two-,three-, four- and fourfold, respectively.
eA dose-related increase in the incidence of hepatocyte micronuclei was observed in partially hepatectomized mice
6 days after administration of 1,4-dioxane. A dose of 1,000 mg/kg produced no change relative to control, while
doses of 2,000 and 3,000 mg/kg increased the incidence of hepatocyte micronuclei by 2.4- and 3.4-fold,
respectively.
'Significant increases in the frequency of micronucleated erythrocytes were observed at each test dose of 1,4-dioxane (1,500,
2,500 and 3,500 mg/kg-day, 5 days/week).
9A dose-related increase in the frequency of micronuclei was observed in proliferating cells with micronuclei at 2,500 and
3,500 mg/kg-day, 5 days/week. No increase in the frequency of micronuclei was seen in the non-proliferating cells.
hNo increase in the hepatocyte labeling index was observed 24 or 48 hours following a single gavage exposure of
1,000 mg/kg. Continuous administration of 1 % 1,4-dioxane in the drinking water for up to 2 weeks produced a
twofold increase in the hepatocyte labeling index.
'A similar pattern of RNA polymerase inhibition was observed at doses of 10 and 100 mg/rat. Inhibition was more
pronounced at the higher dose.
'Hepatocyte viability was 86, 89, 87, 88, 78, and 86% 24 hours following exposure to 0, 1,000, 1,500, 2,000, or
4,000 mg/kg. The incidence (%) of replicative DNA synthesis was increased by 2.5-fold (1,000 mg/kg) or 4.5-fold
(1,500 and 2,000 mg/kg). No increase in replicative DNA synthesis was observed at the highest dose (4,000 mg/kg).
kReplicative DNA synthesis was measured 24, 39, and 48 hours following a single dose of 0, 1,000, or 2,000 mg/kg.
Hepatocyte viability ranged from 71 to 82%. The only increase in replicative DNA synthesis was observed 24 hours
after administration of 2,000 mg/kg (threefold increase). Cell viability for this group was 79%.
'Replicative DNA synthesis was increased 1.5-fold in rats given 1,000 mg/kg of 1,4-dioxane for 11 weeks. No change
from control was observed in rats exposed to 10 mg/kg for 11 weeks or rats acutely exposed to 10, 100, or
1,000 mg/kg.
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
74
DRAFT - DO NOT CITE OR QUOTE
-------
1 100 mg/m3 of 1,4-dioxane vapor for 4 hours/day, 5 days/week, for 1 month. Rats were sacrificed during
2 the morning or evening following exposure and the ovaries and brain cortex were removed and frozen.
3 Tissue preparations were analyzed for catalase activity, glutathione peroxidase activity, and protein
4 peroxidation. Inhalation of 100 mg/m3 of 1,4-dioxane resulted in a significant increase (p < 0.05) in
5 glutathione peroxidase activity, and activation of free radical processes were apparent in both the rat
6 ovary and brain cortex. No change in catalase activity or protein peroxidation was observed at either
7 concentration. A circadian rhythm for glutathione peroxidase activity was absent in control rats, but
8 occurred in rat brain and ovary following 1,4-dioxane exposure.
4.5.2.2 Induction of Metabolism
9 The metabolism of 1,4-dioxane is discussed in detail in Section 3.3. 1,4-Dioxane has been shown
10 to induce its own metabolism (Young et al. 1978b: 1978a). Nannelli et al. (2005a) (study details provided
11 in Section 3.3) characterized the CYP450 isozymes that were induced by 1,4-dioxane in the liver, kidney,
12 and nasal mucosa of the rat. In the liver, the activities of several CYP450 isozymes were increased (i.e.,
13 CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was inducible in the kidney and nasal mucosa.
14 CYP2E1 mRNA was increased approximately two- to threefold in the kidney and nasal mucosa, but
15 mRNA levels were not increased in the liver, suggesting that regulation of CYP2E1 is organ-specific.
16 Induction of hepatic CYPB1/2 and CYP2E1 levels by phenobarbital or fasting did not increase the liver
17 toxicity of 1,4-dioxane, as measured by hepatic glutathione content or serum ALT activity. This result
18 suggested that highly reactive and toxic intermediates did not play a large role in the liver toxicity of
19 1,4-dioxane, even under conditions where metabolism was enhanced. This finding is similar to an earlier
20 conclusion by Kociba et al. (1975a) who evaluated toxicity from a chronic drinking water study alongside
21 data providing a pharmacokinetic profile for 1,4-dioxane. Kociba et al. (1975a) concluded that liver
22 toxicity and eventual tumor formation occurred only at doses where clearance pathways were saturated
23 and elimination of 1,4-dioxane from the blood was reduced. Nannelli et al. (2005a) further suggested that
~ \ / oo
24 a sustained induction of CYP2E1 may lead to generation of reactive oxygen species contributing to target
25 organ toxicity and regenerative cell proliferation; however, no data were provided to support this
26 hypothesis.
4.5.2.3 Mechanisms of Tumor Induction
27 Several studies have been performed to evaluate potential mechanisms for the carcinogenicity of
28 1,4-dioxane (Goldsworthy et al.. 1991: Kitchin and Brown. 1990a: Stottetal. 1981). Stott et al. (1981)
29 evaluated 1,4-dioxane in several test systems, including salmonella mutagenicity in vitro, rat hepatocyte
30 DNA repair activity in vitro, DNA synthesis determination in male Sprague Dawley rats following acute
31 gavage dosing or an 11-week drinking water exposure (described in Section 4.2.1), and hepatocyte DNA
32 alkylation and DNA repair following a single gavage dose. This study used doses of 0, 10, 100, or
33 1,000 mg/kg-day, with the highest dose considered to be a tumorigenic dose level. Liver histopathology
75
DRAFT - DO NOT CITE OR QUOTE
-------
1 and liver to BW ratios were also evaluated in rats from acute gavage or repeated dose drinking water
2 experiments.
3 The histopathology evaluation indicated that liver cytotoxicity (i.e., centrilobular hepatocyte
4 swelling) was present in rats from the 1,000 mg/kg-day dose group that received 1,4-dioxane in the
5 drinking water for 11 weeks (Stott et al.. 1981). An increase in the liver to BW ratio accompanied by an
6 increase in hepatic DNA synthesis was also seen in this group of animals. No effect on histopathology,
7 liver weight, or DNA synthesis was observed in acutely exposed rats or rats that were exposed to a lower
8 dose of 10 mg/kg-day for 11 weeks. 1,4-Dioxane produced negative findings in the remaining
9 genotoxicity assays conducted as part of this study (i.e., Salmonella mutagenicity, in vitro and in vivo rat
10 hepatocyte DNA repair, and DNA alkylation in rat liver). The study authors suggested that the observed
11 lack of genotoxicity at tumorigenic and cytotoxic dose levels indicates an epigenetic mechanism for
12 1,4-dioxane hepatocellular carcinoma in rats.
13 Goldsworthy et al. (1991) evaluated potential mechanisms for the nasal and liver carcinogenicity
14 of 1,4-dioxane in the rat. DNA repair activity was evaluated as a measure of DNA reactivity and DNA
15 synthesis was measured as an indicator of cell proliferation or promotional activity. In vitro DNA repair
16 was evaluated in primary hepatocyte cultures from control and 1,4-dioxane-treated rats (1 or 2% in the
17 drinking water for 1 week). DNA repair and DNA synthesis were also measured in vivo following a
18 single gavage dose of 1,000 mg/kg, a drinking water exposure of 1% (1,500 mg/kg-day) for 1 week, or a
19 drinking water exposure of 2% (3,000 mg/kg-day) for 2 weeks. Liver to BW ratios and palmitoyl CoA
20 oxidase activity were measured in the rat liver to determine whether peroxisome proliferation played a
21 role in the liver carcinogenesis of 1,4-dioxane. In vivo DNA repair was evaluated in rat nasal epithelial
22 cells derived from either the nasoturbinate or the maxilloturbinate of 1,4-dioxane-treated rats. These rats
23 received 1% 1,4-dioxane (1,500 mg/kg-day) in the drinking water for 8 days, followed by a single gavage
24 dose of 10, 100, or 1,000 mg/kg 12 hours prior to sacrifice. Archived tissues from the NCI (1978)
25 bioassay were reexamined to determine the primary sites for tumor formation in the nasal cavity
26 following chronic exposure in rats. Histopathology and cell proliferation were determined for specific
27 sites in the nasal cavity that were related to tumor formation. This evaluation was performed in rats that
28 were exposed to drinking water containing 1% 1,4-dioxane (1,500 mg/kg-day) for 2 weeks.
29 1,4-Dioxane and its metabolite l,4-dioxane-2-one did not affect in vitro DNA repair in primary
30 hepatocyte cultures (Goldsworthy et al.. 1991). In vivo DNA repair was also unaffected by acute gavage
31 exposure or ingestion of 1,4-dioxane in the drinking water for a 1- or 2-week period. Hepatocyte cell
32 proliferation was not affected by acute gavage exposure, but was increased approximately twofold
33 following a 1-2-week drinking water exposure. A 5-day drinking water exposure to 1% 1,4-dioxane
34 (1,500 mg/kg-day) did not increase the activity of palmitoyl coenzyme A or the liver to BW ratio,
35 suggesting that peroxisome proliferation did not play a role in the hepatocarcinogenesis of 1,4-dioxane.
36 Nannelli et al. (2005a) also reported a lack of hepatic palmitoyl CoA induction following 10 days of
37 exposure to 1.5% 1,4-dioxane in the drinking water (2,100 mg/kg-day).
38 Treatment of rats with 1% (1,500 mg/kg-day) 1,4-dioxane for 8 days did not alter DNA repair in
39 nasal epithelial cells (Goldsworthy et al.. 1991). The addition of a single gavage dose of up to
76
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,000 mg/kg 12 hours prior to sacrifice also did not induce DNA repair. Reexamination of tissue sections
2 from the NCI (1978) bioassay suggested that the majority of nasal tumors were located in the dorsal nasal
3 septum or the nasoturbinate of the anterior portion of the dorsal meatus (Golds worthy et al.. 1991). No
4 histopathological lesions were observed in nasal section of rats exposed to drinking water containing 1%
5 1,4-dioxane (1,500 mg/kg-day) for 2 weeks and no increase was observed in cell proliferation at the sites
6 of highest tumor formation in the nasal cavity.
7 Female Sprague Dawley rats (three to nine per group) were given 0, 168, 840, 2,550, or
8 4,200 mg/kg 1,4-dioxane (99% purity) by corn oil gavage in two doses at 21 and 4 hours prior to sacrifice
9 (Kitchin and Brown. 1990a). DNA damage (single-strand breaks measured by alkaline elution), ODC
10 activity, reduced glutathione content, and CYP450 content were measured in the liver. Serum ALT
11 activity and liver histopathology were also evaluated. No changes were observed in hepatic reduced
12 glutathione content or ALT activity. Light microscopy revealed minimal to mild vacuolar degeneration in
13 the cytoplasm of hepatocytes from three of five rats from the 2,550 mg/kg dose group. No
14 histopathological lesions were seen in any other dose group, including rats given a higher dose of
15 4,200 mg/kg. 1,4-Dioxane caused 43 and 50% increases in DNA single-strand breaks at dose levels of
16 2,550 and 4,200 mg/kg, respectively. CYP450 content was also increased at the two highest dose levels
17 (25 and 66% respectively). ODC activity was increased approximately two-, five-, and eightfold above
18 control values at doses of 840, 2,550, and 4,200 mg/kg, respectively. The results of this study
19 demonstrated that hepatic DNA damage can occur in the absence of significant cytotoxicity. Parameters
20 associated with tumor promotion (i.e., ODC activity, CYP450 content) were also elevated, suggesting that
21 promotion may play a role in the carcinogenesis of 1,4-dioxane.
4.6 Synthesis of Major Noncancer Effects
22 Liver, kidney, and nasal toxicity were the primary noncancer health effects associated with
23 exposure to 1,4-dioxane. In humans, several fatal cases of hemorrhagic nephritis and centrilobular
24 necrosis of the liver were related to occupational exposure (i.e., inhalation and dermal contact) to
25 1,4-dioxane (Johnstone. 1959; Barber. 1934). Neurological changes were also reported in one case;
26 including, headache, elevation in blood pressure, agitation and restlessness, and coma (Johnstone. 1959).
27 Perivascular widening was observed in the brain of this worker, with small foci of demyelination in
28 several regions (e.g., cortex, basal nuclei). In laboratory animals, following oral and inhalation exposure
29 to 1,4-dioxane, liver and kidney degeneration and necrosis were observed(JBRC. 1998; Drew et al.. 1978;
30 David. 1964: Kesten et al.. 1939: Laugetal.. 1939: Schrenk and Yant. 1936: deNavasquez. 1935: Fairlev
31 et al.. 1934a). in addition to changes in the nasal epithelium (Kano et al.. 2009: Kasai et al.. 2009: Kano et
32 al.. 2008: Kasai et al.. 2008: JBRC. 1998). The results of subchronic and chronic studies are discussed
33 below.
77
DRAFT - DO NOT CITE OR QUOTE
-------
4.6.1 Oral
1 Table 4-25 presents a summary of the noncancer results for the subchronic and chronic oral
2 studies of 1,4-dioxane toxicity in experimental animals. Liver and kidney toxicity were the primary
3 noncancer health effects of oral exposure to 1,4-dioxane in animals. Kidney damage at high doses was
4 characterized by degeneration of the cortical tubule cells, necrosis with hemorrhage, and
5 glomerulonephritis (NCI. 1978; Kocibaetal. 1974a: Argus etal. 1965a: Fairley et al. 1934a). Renal cell
6 degeneration generally began with cloudy swelling of cells in the cortex (Fairley et al.. 1934a). Nuclear
7 enlargement of proximal tubule cells was observed at doses below those producing renal necrosis (Kano
8 et al.. 2008; JBRC. 1998). but is of uncertain toxicological significance. The lowest dose reported to
9 produce kidney damage was 94 mg/kg-day, which produced renal degeneration and necrosis of tubule
10 epithelial cells in male rats in the Kociba et al. (1974a) study. Cortical tubule degeneration was seen at
11 higher doses in the NCI (1978) bioassay (240 mg/kg-day, male rats), and glomerulonephritis was reported
12 for rats given doses of > 430 mg/kg-day (Argus et al.. 1973a: Argus et al.. 1965a).
78
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-25 Oral toxicity studies (noncancer effects) for 1,4-dioxane
Species Dose/duration (mXday)
LOAEL
(mg/kg-day) LTTect
Reference
Subchronic studies
Rats
Rat and Mouse 0 or 1,900 mg/kg-day;
(6/species); Mice NA
unknown strain 0 or 3,300 mg/kg-day
for 67 days
Male
SpragueDawley 0, 10, or 1,000 mg/kg-day 10
(4-6/group) ^ 11 weeks
Rats
Males 0, 52, 126, 274,
F344/DuCrj Rat 657, or 1,554 mg/kg-day; g2
(10/sex/group) Females 0, 83, 185, 427,
756, or 1,614 mg/kg-day
for 13 weeks
Mice
Males 0, 86, 231, 585,
Cri-RDFI MOUSP 882, or 1,570 mg/kg-day;
Crj.BDFI Mouse Fema|es 0 170 387 170
(10/sex/group) ftc.ft . fi_ '
2,669 mg/kg-day
for 13 weeks
1 QDD t Renal cortical degeneration
3 300 ra S and necrosis' hemorrhage;
hepatocellular degeneration
Minimal centrilobular
1,000 hepatocyte swelling;
increased DNA synthesis
Nuclear enlargement of
.„„ nasal respiratory
epithelium; hepatocyte
swelling
387 Nuclear enlargement of
bronchial epithelium
Fairley et al.
(1934a)
Stott et al.
(1981)
Kano et al.
(2008)
Kano et al.
(2008)
Chronic studies
Male
Wistar n c,m * *
0 or 640 mg/kg-day NA
M I RatS
0, 430, 574, 803, or
Rat (30/group) for 1 3 ^^ ^
Rats
oh p Males 0, 9.6, 94, or
(60/sex/dose .-' . n *r\ */m 9-6
nmnni Females 0, 19, 14o, or
9 p; 1,599 mg/kg-day
for 2 years
Rats
Osborne-Mendel Males 0, 240, or
Rat 530 mg/kg-day;
(35/sex/dose Females 0, 350, or
level) 640 mg/kg-day
for 110 weeks
Mice
DR^QC M Males °. 72°. or
B6C3Fi Mouse oon ,. ,
,.-,,, ,, oou mq/Kq-aay; ...
(50/sex/dose .- , S On^ NA
i ., Females 0, 380, or
' 860 mg/kg-day
for 90 weeks
Rats
Males 0, 1 1 , 55, or
Hepatocytes with enlarged
640 hyperchromic nuclei;
glomerulonephritis
430 Hepatocytomegaly;
glomerulonephritis
Degeneration and necrosis
94 of renal tubular cells and
hepatocytes
Pneumonia, gastric ulcers,
240 and cortical tubular
degeneration in the kidney
380 Pneumonia and rhinitis
Atrophy of nasal olfactory
274 epithelium; nasal adhesion and
inflammation
Argus et al.
(1965a)
Argus et al.
(1973a)
Kociba et al.
NCI (1978)
NCI (1978)
JBRC (1998):
Kano et al.
(2009)
429 mg/kg-day for 2 years
DRAFT - DO NOT CITE OR QUOTE
79
-------
Species
Dose/duration
NOAEL LOAEL
(mg/kg-day) (mg/kg-day)
Effect
Reference
F344/DuCrj Rat
(50/sex/dose
level)
Rats
Males 0, 11, 55, or
274 mg/kg-day;
Females 0, 18, 83, or
429 mg/kg-day
for 2 years
11
55
Mixed cell liver foci
JBRC (1998):
Kano et al.
(2009)
F344/DuCrj Rat
(50/sex/dose
level)
Rats
Males 0, 11, 55, or
274 mg/kg-day;
Females 0, 18, 83, or
429 mg/kg-day
for 2 years
55
Increases in serum liver JBRC (1998):
274 enzymes (GOT, GPT, LDH, Kano et al.
and ALP) (2009)
Crj:BDF1 Mouse
(50/sex/dose
level)
Mice
Males 0, 49, 191 or
677 mg/kg-day;
Females 0, 66, 278, or
964 mg/kg-day
for 2 years
66
278
Nasal inflammation
JBRC (1998):
Kano et al.
(2009)
Crj:BDF1 Mouse
(50/sex/dose
level)
Mice
Males 0, 49, 191 or
677 mg/kg-day;
Females 0, 66, 278, or
964 mg/kg-day
for 2 years
49
Increases in serum liver JBRC (1998):
191 enzymes (GOT, GPT, LDH, Kano et al.
and ALP) (2009)
Developmental studies
Sprague Dawley
Rat
(18-20/group)
Rats
Pregnant dams 0, 250,
500, or 1,000 mg/kg-day
on gestation days 6-15
Delayed ossification of the
500 1,000 sternebrae and reduced
fetal BWs
Giavini et al.
(1985a)
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; Kociba et al.. 1974a: Argus etal. 1973a: Argus et al.. 1965a: Fairley et al.. 1934a). Hepatocellular
4 degeneration and necrosis were seen at high doses in a subchronic study (1,900 mg/kg-day in rats)
5 (Fairley et al.. 1934a) and at lower doses in a chronic study (94 mg/kg-day, male rats) (Kociba et al..
6 1974a). Argus et al. (1973a) described a progression of preneoplastic effects in the liver of rats exposed to
7 a dose of 575 mg/kg-day. Early changes (8 months exposure) were described as an increased nuclear size
8 of 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 and clear and mixed-cell foci were also observed in the liver of rats (doses
11 >55 mg/kg-day in male rats) (Kano et al.. 2009; JBRC. 1998). Clear and mixed-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. 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. 2002a: 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 2002a). 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
80
DRAFT - DO NOT CITE OR QUOTE
-------
1 hepatis was associated with other preneoplastic changes in the liver (clear and mixed-cell foci). No other
2 lesions indicative of liver toxicity were seen in this study; therefore, spongiosis hepatis was not
3 considered indicative of noncancer effects. The activity of serum enzymes (i.e., AST, ALT, LDH, and
4 ALP) was increased in rats and mice exposed to 1,4-dioxane, although only in groups with high incidence
5 of liver tumors. Blood samples were collected only at the end of the 2-year study, so altered serum
6 chemistry may be associated with the tumorigenic changes in the liver.
7 Hematological changes were reported in the JBRC (1998) study only. Mean doses are reported
8 based on information provided in Kano et al. (2009). Observed increases in RBCs, hematocrit,
9 hemoglobin in high-dose male mice (677 mg/kg-day) may be related to lower drinking water
10 consumption (74% of control drinking water intake). Hematological effects noted in male rats given
11 55 mg/kg-day (decreased RBCs, hemoglobin, hematocrit, increased platelets) were within 20% of control
12 values. A reference range database for hematological effects in laboratory animals (Wolford et al.. 1986)
13 indicates that a 20% change in these parameters may fall within a normal range (10th-90th percentile
14 values) and may not represent a treatment-related effect of concern.
15 Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
16 dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice). The
17 JBRC (1998) study also demonstrates atrophy of the nasal epithelium and adhesion in rats and mice.
18 Nasal inflammation may be a response to direct contact of the nasal mucosa with drinking water
19 containing 1,4-dioxane (Sweeney et al.. 2008a: Goldsworthy et al.. 1991) or could result from systemic
20 exposure. Regardless, inflammation may indicate toxicity due to 1,4-dioxane exposure. A significant
21 increase in the incidence of pneumonia was reported in mice from the NCI (1978) study. The significance
22 of this effect is unclear, as it was not observed in other studies that evaluated lung histopathology (Kano
23 et al.. 2008; JBRC. 1998; Kocibaetal. 1974a). No studies were available regarding the potential for
24 1,4-dioxane to cause immunological effects. Metaplasia and hyperplasia of the nasal epithelium were also
25 observed in high-dose male and female rats (JBRC. 1998): however, these effects are likely to be
26 associated with the formation of nasal cavity tumors in these dose groups. Nuclear enlargement of the
27 nasal olfactory epithelium was observed at a dose of 83 mg/kg-day in female rats (Kano et al.. 2009):
28 however, it is unclear whether this alteration represents an adverse toxicological effect. Nuclear
29 enlargement of the tracheal and bronchial epithelium and an accumulation of foamy cells in the lung were
30 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
31 Two subchronic (Kasai et al.. 2008: Fairley et al.. 1934a) and two chronic inhalation studies
32 (Kasai et al.. 2009: Torkelson et al.. 1974a) were identified. Nasal, liver, and kidney toxicity were the
33 primary noncancer health effects of inhalation exposure to 1,4-dioxane in rodents. Table 4-26 presents a
34 summary of the noncancer results for the subchronic and chronic inhalation studies of 1,4-dioxane
35 toxicity in laboratory animals.
81
DRAFT - DO NOT CITE OR QUOTE
-------
1 Of the inhalation studies, nasal tissue was only evaluated in rat studies conducted by Kasai et al.
2 (2009; 2008). Adverse effects in nasal tissue were observed frequently in these studies, and statistically
3 significant changes were noted at vapor after concentrations as low as 50 ppm. Nasal effects included
4 deformity of the nose and histopathological changes characterized by enlarged epithelial nuclei
5 (respiratory epithelium, olfactory epithelium, trachea, and bronchus), atrophy (olfactory epithelium),
6 vacuolic change (olfactory epithelium and bronchial epithelium), squamous cell metaplasia and
7 hyperplasia (respiratory epithelium), respiratory metaplasia (olfactory epithelium), inflammation
8 (respiratory and olfactory epithelium), hydropic change (lamina propria), and sclerosis (lamina propria).
9 In both studies, a concentration-dependent, statistically significant incidence of enlarged nuclei of the
10 respiratory epithelium were reported by the study authors; however, the toxicological significance of
11 nuclear enlargement is uncertain.
12 At high doses, liver damage was characterized by hepatocellular degeneration which varied from
13 swelling (Kasai et al.. 2008: Fairlev et al.. 1934a) to necrosis (Kasai et al.. 2009: Kasai et al.. 2008:
14 Fairlev et al.. 1934a). spongiosis hepatis (Kasai et al.. 2009). nuclear enlargement of centrilobular cells
15 (Kasai et al.. 2009) and basophilic and acidophilic cell foci (Kasai et al.. 2009). GST-P positive cell foci
16 are commonly considered preneoplastic changes and would not be considered evidence of noncancer
17 toxicity when observed in conjunction with liver tumor formation (Bannasch et al.. 1982a). Since
18 exposure to 1,4-dioxane resulted in tumor formation in the liver, these lesions are not considered as
19 potential noncancer toxicity.
20 At concentrations ranging from 200 to 3,200 ppm, altered liver enzymes (i.e., AST, ALT, ALP,
21 and y-GTP), increased liver weights, and induction of GST-P were also observed (Kasai et al.. 2009:
22 Kasai et al.. 2008). Changes in the activity of serum enzymes were mostly observed in exposed rat groups
23 at high 1,4-dioxane concentrations (Kasai et al.. 2009: Kasai et al.. 2008). Induction of GST-P positive
24 hepatocytes was observed in female rats at 1,600 ppm and male and female rats at 3,200 ppm following
25 13 weeks of exposure. GST-P is considered a good enzymatic marker for early detection of chemical
26 hepatocarcinogenesis (Sato. 1989). Although, GST-P positive liver foci were not observed in the 2-year
27 bioassay, the focally and proliferating GST-P positive hepatocytes noted in the 13- week study suggest
28 eventual progression to hepatocellular tumors after 2 years of exposure and therefore would not be
29 considered a potential noncancer effect.
30 The lowest vapor concentration reported to produce liver lesions after 2 years of exposure was
31 1,250 ppm. The lesions were characterized by necrosis of centrilobular cells, spongiosis hepatis, and
32 nuclear enlargement in the Kasai et al. (2009) study. However, as previously stated, the toxicological
33 significance of nuclear enlargement is uncertain.
34 Kidney effects were reported less frequently in these inhalation studies and were generally
35 observed at higher exposure concentrations than nasal and liver effects. Kidney damage was described as
36 patchy degeneration of cortical tubules with vascular congestion and hemorrhage (Fairlev et al.. 1934a).
37 hydropic change of proximal tubules (Kasai et al.. 2009: Kasai et al.. 2008). and as nuclear enlargement in
38 proximal tubule cells (Kasai et al.. 2009). Changes in serum chemistry and urinalysis indices were also
39 noted as evidence of renal damage. In a 13-week inhalation study of male and female rats (Kasai et al..
82
DRAFT - DO NOT CITE OR QUOTE
-------
1 2008) kidney toxicity was only observed in female rats exposed to 3,200 ppm of 1,4-dioxane (i.e.
2 hydropic change in the renal proximal tubules), which suggests a possible greater susceptibility of female
3 rats to renal damage following inhalation of 1,4-dioxane.
4 Other noted noncancer effects in laboratory animals included acute vascular congestion of the
5 lungs (Fairley et al.. 1934a): changes in relative lung weights (Kasai et al.. 2008): and decrease in body
6 weight gain (Kasai et al.. 2009; Kasai et al.. 2008). Following a 13-week exposure, higher 1,4-dioxane
7 plasma levels were found in female rats than male rats (Kasai et al.. 2008). 1,4-Dioxane was measured in
8 plasma along with systemic effects following subchronic inhalation exposure to 1,4-dioxane in rats (Kasai
9 et al.. 2008).
Table 4-26 Inhalation toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration N(°^}L
*(ppm)L Effect
Reference
Subchronic studies
Rat, mouse, rabbit,
and guinea pig
(3-6/species/group);
unknown strains
F344/DuCrj rat
(10/sex/group)
0, 1,000,2,000, 5,000,
or 10,000 ppm for 7
days. Days 1-5, two 1.5 ...
hour exposures; day 6,
one 1.5 hour exposure;
and day 7, no exposure
0, 100, 200, 400, 800,
1,600, 3,200, or
6,400 ppm 6 hours/day 5
days/wk, for 13 wk
Renal cortical
degeneration and
1,000 hemorrhage;
hepatocellular
degeneration and necrosis
Respiratory epithelium:
1 00 nuclear enlargement of
epithelial cells
Fairley et al.
(1934a)
Kasai et al.
(2008)
Chronic studies
Wistar rat (288/sex)
F344/DuCrj male
rat
(50/group)
111 ppm for 7hours/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,
., . survival, organ weights,
hematology, clinical
chemistry, or
histopathology
Respiratory epithelium:
50 nuclear enlargement of
epithelial cells, atrophy,
and metaplasia
Torkelson et
al. (1974a)
Kasai et al.
(2009)
4.6.2.1 Mode of Action Information
10 The metabolism of 1,4-dioxane in humans was extensive at low doses (<50 ppm). The linear
11 elimination of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism was a
12 nonsaturated, first-order process at this exposure level (Young et al.. 1977a: 1976a). Like humans, rats
13 extensively metabolized a single 50 ppm inhalation exposure to 1,4-dioxane; however, plasma data from
14 rats given single i.v. doses of 3, 10, 30, 100, or 1,000 mg [14C]-l,4-dioxane/kg demonstrated a
15 dose-related shift from linear, first-order to nonlinear, saturable metabolism of 1,4-dioxane (Young et al..
16 1978b: 1978a). Using the Young et al. (1978b: 1978a) rat kinetic model, the metabolism of 1,4-dioxane in
83
DRAFT - DO NOT CITE OR QUOTE
-------
1 rats that were exposed to 400, 800, 1,600, and 3,200 ppm via inhalation for 13 weeks could not be
2 accurately predicted due to a lack of knowledge on needed model parameters and biological processes
3 (See Section 3.5.3 and Appendix B). It appears, following prolonged inhalation exposure to 1,4-dioxane
4 at concentrations up to 3,200 ppm, that metabolism is induced (Appendix B).
5 1,4-Dioxane oxidation appeared to be CYP450-mediated, as CYP450 induction with
6 phenobarbital or Aroclor 1254 and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine or
7 cobaltous chloride was effective in significantly increasing and decreasing, respectively, the appearance
8 of HEAA in the urine of rats (Wooetal.. 1978. 1977b). 1,4-Dioxane itself induced CYP450-mediated
9 metabolism of several barbiturates in Hindustan mice given i.p. injections of 25 and 50 mg/kg of
10 1,4-dioxane (Mungikar and Pawar. 1978). The differences between single and multiple doses in urinary
11 and expired radiolabel support the notion that 1,4-dioxane may induce its own metabolism. High doses of
12 1,4-dioxane were shown to induce several isoforms of CYP450 in various tissues following acute oral
13 administration by gavage or drinking water (Nannelli et al. 2005a). In the liver, the activity of several
14 CYP450 isozymes was increased (i.e., CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was
15 inducible in the kidney and nasal mucosa. CYP2E1 mRNA was increased approximately two- to threefold
16 in the kidney and nasal mucosa, but mRNA levels were not increased in the liver, suggesting that
17 regulation of CYP2E1 was organ-specific.
18 Nannelli et al. (2005a) investigated the role of CYP450 isozymes in the liver toxicity of
19 1,4-dioxane. Hepatic CYP2B1/2 and CYP2E1 levels were induced by phenobarbital or fasting and liver
20 toxicity was measured as hepatic glutathione content or serum ALT activity. No increase in glutathione
21 content or ALT activity was observed, suggesting that highly reactive and oxidative intermediates did not
22 play a large role in the liver toxicity of 1,4-dioxane, even under conditions where metabolism was
23 enhanced. Pretreatment with inducers of mixed-function oxidases also did not significantly change the
24 extent of covalent binding in subcellular fractions (Woo et al.. 1977c). Covalent binding was measured in
25 liver, kidney, spleen, lung, colon, and skeletal muscle 1-12 hours after i.p. dosing with 1,4-dioxane.
26 Covalent binding was highest in liver, spleen, and colon. Within hepatocytes, 1,4-dioxane distribution
27 was greatest in the cytosolic fraction, followed by the microsomal, mitochondrial, and nuclear fractions.
28 The absence of an increase in toxicity following an increase in metabolism suggests that the
29 parent compound may be responsible for 1,4-dioxane toxicity. This hypothesis is supported by a
30 comparison of the pharmacokinetic profile of 1,4-dioxane with the toxicology data from a chronic
31 drinking water study (Kociba et al.. 1975a). This analysis indicated that liver toxicity did not occur unless
32 clearance pathways were saturated and elimination of 1,4-dioxane from the blood was reduced. A
33 dose-dependent increase of 1,4-dioxane concentration in the blood was seen, which correlated to the
34 observed dose-dependent increase in incidences of nasal, liver, and kidney toxicities (Kasai et al.. 2008).
35 Alternative metabolic pathways (i.e., not CYP450 mediated) may be present at high doses of 1,4-dioxane;
36 however, the available studies have not characterized these pathways or identified any possible reactive
37 intermediates. Thus, the mechanism by which 1,4-dioxane induces tissue damage is not known, nor is it
38 known whether the toxic moiety is 1,4-dioxane or a transient or terminal metabolite.
84
DRAFT - DO NOT CITE OR QUOTE
-------
4.7 Evaluation of Carcinogenicity
4.7.1 Summary 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. 1994a: NCI. 1978: Kocibaetal. 1974a: Argus etal. 1973a: Hoch-
5 Ligeti and Argus. 1970a: Hoch-Ligeti et al.. 1970a: Argus et al.. 1965a). Tissue sites where tumors have
6 been observed in these laboratory animals due to exposure to 1,4-dioxane include, peritoneum (Kano et
7 al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a). mammary gland (Kano et al.. 2009:
8 Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a). liver (Kano et al.. 2009: Kasai et al.. 2009).
9 kidney (Kasai et al.. 2009). Zymbal gland (Kasai et al.. 2009). subcutaneous (Kasai et al.. 2009). nasal
10 tissue (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a: NCI. 1978: Kociba et
11 al.. 1974a: Argus etal.. 1973a: Hoch-Ligeti et al.. 1970a). and lung (Hoch-Ligeti and Argus. 1970a).
12 Studies in humans are inconclusive regarding evidence for a causal link between occupational exposure to
13 1,4-dioxane and increased risk for cancer; however, only two studies were available and these were
14 limited by small cohort size and a small number of reported cancer cases (Buffler et al.. 1978a: Thiess et
15 al.. 1976a).
16 A MOA hypothesis involving sustained proliferation of spontaneously transformed liver cells has
17 some support from data indicating that 1,4-dioxane acts as a tumor promoter in mouse skin and rat liver
18 bioassays (Lundberg et al.. 1987: King et al.. 1973a). Dose-response and temporal data support the
19 occurrence of cell proliferation prior to the development of liver tumors (JBRC. 1998: Kociba et al..
20 1974a) in the rat model. However, the dose-response relationship for induction of hepatic cell
21 proliferation has not been characterized, and it is unknown if it would reflect the dose-response
22 relationship for liver tumors in the 2-year rat and mouse studies. Conflicting data from rat and mouse
23 bioassays (JBRC. 1998: Kocibaetal.. 1974a) suggest that cytotoxicity may not be a required precursor
24 event for 1,4-dioxane-induced cell proliferation. Data regarding a plausible dose response and temporal
25 progression (see Table 4-21) from cytotoxicity and cell proliferation to eventual liver tumor formation are
26 not available. Also, Kociba et al. (1974b) reported renal degeneration, necrosis, and regenerative
27 proliferation in exposed rats, but no increase in the incidence of kidney tumors, which does not support a
28 cytotoxicity/cell proliferation MOA.
29 For nasal tumors, there is a hypothesized MOA that includes metabolic induction, cytotoxicity,
30 and regenerative cell proliferation (Kasai et al.. 2009). The induction of CYP450 has some support from
31 data illustrating that following acute oral administration of 1,4-dioxane by gavage or drinking water,
32 CYP2E1 was inducible in nasal mucosa (Nannelli et al.. 2005a). CYP2E1 mRNA was increased
33 approximately two- to threefold in nasal mucosa (and in the kidney, see section 3.3) in the Nannelli et al.
34 (2005a) study. While cell proliferation was observed following 1,4-dioxane exposure in both a 2-year
35 inhalation study in male rats (1,250 ppm) (Kasai et al.. 2009) and a 2-year drinking water study in male
85
DRAFT - DO NOT CITE OR QUOTE
-------
1 (274 mg/kg-day) and female rats (429 mg/kg-day), no evidence of cytotoxicity in the nasal cavity was
2 observed (Kasai et al.. 2009): therefore, cytotoxicity, as a key event, is not supported. Nasal lesions,
3 including inflammation, hyperplasia, and metaplasia, were frequently seen in inhalation studies conducted
4 by the NTP with no evidence of nasal carcinogenicity (Haseman and Hailey. 1997b: Ward et al.. 1993).
5 Following a 13-week inhalation study in rats, a concentration-dependent increase of 1,4-dioxane in the
6 blood was observed (Kasai et al.. 2008). Studies have shown that water-soluble, gaseous irritants cause
7 nasal injuries such as squamous cell carcinomas (Morgan et al.. 1986). Similarly, 1,4-dioxane, which has
8 been reported as a miscible compound (Hawley and Lewis. 2001). also caused nasal injuries that were
9 concentration-dependent, including nasal tumors (Kasai et al.. 2009). Additionally, it has been suggested
10 that in vivo genotoxicity may contribute to the carcinogenic MOA for 1,4-dioxane (Kasai et al.. 2009)
11 (see Section 4.7.3.6 for further discussion). Collectively, these data are insufficient to support the
12 hypothesized MOAs.
13 There are no data available regarding any hypothesized MOA by which 1,4-dioxane produces
14 kidney, lung, peritoneal (mesotheliomas), mammary gland, Zymbal gland, and subcutis tumors.
15 U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) indicate that for
16 tumors occurring at a site other than the initial point of contact, the weight of evidence for carcinogenic
17 potential may apply to all routes of exposure that have not been adequately tested at sufficient doses. An
18 exception occurs when there is convincing information (e.g., toxicokinetic data) that absorption does not
19 occur by other routes. Information available on the carcinogenic effects of 1,4-dioxane via the oral route
20 demonstrates that tumors occur in tissues remote from the site of absorption. In addition, information on
21 the carcinogenic effects of 1,4-dioxane via the inhalation route in animals also demonstrates that tumors
22 occur at tissue sites distant from the portal of entry. Information on the carcinogenic effects of
23 1,4-dioxane via the inhalation and dermal routes in humans and via the dermal route in animals is absent.
24 If sufficient external dose is applied, it is assumed that an internal dose will be achieved regardless of the
25 route of exposure. Therefore, based on the observance of systemic tumors following oral and inhalation
26 exposure, 1,4-dioxane is "likely to be carcinogenic to humans" by all routes of exposure.
4.7.2 Synthesis of Human, Animal, and Other Supporting Evidence
27 Human studies of occupational exposure to 1,4-dioxane were inconclusive; in each case, the
28 cohort size was limited and number of reported cases was small (Buffler et al.. 1978a: Thiess et al..
29 1976a).
30 Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
31 pigs (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a: NCI. 1978: Kociba et al..
32 1974a: Torkelson et al.. 1974a: Argus et al.. 1973a: Hoch-Ligeti and Argus. 1970a: Hoch-Ligeti et al..
33 1970a: Argus et al.. 1965a). Liver tumors have been observed following drinking water exposure in male
34 Wistar rats (Argus et al.. 1965a). male guinea pigs (Hoch-Ligeti and Argus. 1970a). male Sprague
35 Dawley rats (Argus et al.. 1973a: Hoch-Ligeti et al.. 1970a). male and female Sherman rats (Kociba et al..
36 1974a). female Osborne-Mendel rats (NCI. 1978). male and female F344/DuCrj rats (Kano et al.. 2009:
86
DRAFT - DO NOT CITE OR QUOTE
-------
1 JBRC. 1998: Yamazaki et al. 1994a). male and female B6C3FJ mice (NCI. 1978). and male and female
2 Crj:BDF 1 mice (Kano et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a): and following inhalation
3 exposure in male F344 rats (Kasai et al.. 2009). In the earliest cancer bioassays, the liver tumors were
4 described as hepatomas (Argus et al.. 1973a: Hoch-Ligeti and Argus. 1970a: Hoch-Ligeti et al.. 1970a:
5 Argus et al.. 1965a): however, later studies made a distinction between hepatocellular carcinoma and
6 hepatocellular adenoma (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a: NCI.
7 1978: Kociba et al.. 1974a). Both tumor types have been seen in rats and mice exposed to 1,4-dioxane via
8 drinking water and inhalation.
9 Kociba et al. (1974a) noted evidence of liver toxicity at or below the dose levels that produced
10 liver tumors but did not report incidence data for these effects. Hepatocellular degeneration and necrosis
11 were observed in the mid- and high-dose groups of male and female Sherman rats exposed to 1,4-dioxane,
12 while tumors were only observed at the highest dose. Hepatic regeneration was indicated in the mid- and
13 high-dose groups by the formation of hepatocellular hyperplastic nodules. Kasai et al. (2009) noted
14 evidence of liver toxicity and tumor incidences (i.e. hepatocellular adenoma) in male F344/DuCrj rats
15 following inhalation exposures to 1,250 ppm. Increased liver toxicities included hepatocellular necrosis,
16 spongiosis hepatis, and acidophilic and basophilic cell foci.
17 Nasal cavity tumors were also observed in Sprague Dawley rats (Argus et al.. 1973a: Hoch-Ligeti
18 etal. 1970a). Osborne-Mendel rats (NCI. 1978). Sherman rats (Kociba et al.. 1974a). and F344/DuCrj
19 rats (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994a). Most tumors were
20 characterized as squamous cell carcinomas. Nasal tumors were not elevated in B6C3Fi or Crj:BDFl mice.
21 Kano et al. (2009) and Kasai et al. (2009) were the only studies that evaluated nonneoplastic changes in
22 nasal cavity tissue following prolonged exposure to 1,4-dioxane via oral and inhalation routes,
23 respectively.
24 Histopathological lesions in female F344/DuCrj rats following oral exposure to 1,4-dioxane were
25 suggestive of toxicity and regeneration in nasal tissue (i.e., atrophy, adhesion, inflammation, nuclear
26 enlargement, and hyperplasia and metaplasia of respiratory and olfactory epithelium). Some of these
27 effects occurred at a lower dose (83 mg/kg-day) than that shown to produce nasal cavity tumors
28 (429 mg/kg-day) in female rats. Re-examination of tissue sections from the NCI (1978) bioassay
29 suggested that the majority of nasal tumors were located in the dorsal nasal septum or the nasoturbinate of
30 the anterior portion of the dorsal meatus.
31 Histopathological lesions in male F344/DuCrj rats following exposure to 1,4-dioxane via
32 inhalation were also suggestive of toxicity and regeneration in nasal tissue (i.e. atrophy, inflammation,
33 nuclear enlargement, hyperplasia and metaplasia of the respiratory and olfactory epithelium, and
34 inflammation). Some of these effects occurred at lower concentrations (50 ppm and 250 ppm) than those
35 shown to produce nasal cavity tumors (1,250 ppm) in male rats. Nasal squamous cell carcinomas were
36 observed in the dorsal area of levels 1-3 of the nasal cavity and were characterized as well-differentiated
37 and keratinized. In two cases, invasive growth into adjacent tissue was noted, marked by carcinoma
38 growth out of the nose and through a destroyed nasal bone.
87
DRAFT - DO NOT CITE OR QUOTE
-------
1 In addition to the liver and nasal tumors observed in several studies, a statistically significant
2 increase in mesotheliomas of the peritoneum was seen in male rats from the Kano et al. (2009) study
3 (JBRC. 1998; Yamazaki et al.. 1994a) and the Kasai et al. (2009) study. Female rats dosed with
4 429 mg/kg-day in drinking water for 2 years also showed a statistically significant increase in mammary
5 gland adenomas (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994a). In male rats, exposed via
6 inhalation, a statistically significant positive trend of mammary gland adenomas was observed by Kasai et
7 al. (2009). A statistically significant increase and/or trend of subcutis fibroma, Zymbal gland adenoma,
8 and renal cell carcinoma incidences was also observed in male rats exposed for 2 years via inhalation
9 (Kasai et al.. 2009). A significant increase in the incidence of these tumors was not observed in other
10 chronic oral or inhalation bioassays of 1,4-dioxane (NCI. 1978; Kocibaetal.. 1974a: Torkelson et al..
11 1974a).
4.7.3 Mode of Action Information
12 The hypothesized MOAs for 1,4-dioxane carcinogenicity are discussed below within the context
13 of the modified Hill criteria of causality as recommended in the most recent Agency guidelines (U.S.
14 EPA. 2005a). MOA analyses were not conducted for kidney, peritoneal, mammary gland, Zymbal gland,
15 or subcutis tumors due to the absence of any chemical specific information for these tumor types.
DRAFT - DO NOT CITE OR QUOTE
-------
4.7.3.1 Identification of Key Events for Carcinogenicity
1 4.7.3.1.1 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 (Kociba et al.. 1974a). however, this study did not report incidence data for
6 these effects. Other studies suggest that cell proliferation can occur in the absence of liver cytotoxicity.
7 Liver tumors were observed in female rats and female mice in the absence of lesions indicative of
8 cytotoxicity (Kano et al.. 2008; JBRC. 1998; NCI. 1978). Figure 4-1 presents a schematic representation
9 of possible key events in the MOA for 1,4-dioxane liver carcinogenicity. These include: (1) oxidation by
10 CYP2E1 and CYP2B1/2 (i.e., detoxification pathway for 1,4-dioxane), (2) saturation of
11 metabolism/clearance leading to accumulation of the parent 1,4-dioxane, (3) liver damage followed by
12 regenerative cell proliferation, or (4) cell proliferation in the absence of cytotoxicity (i.e., mitogenesis),
13 (5) hyperplasia, and (6) tumor formation. It is suggested that liver toxicity is related to the accumulation
14 of the parent compound following metabolic saturation at high doses (Kociba et al.. 1975a): however,
15 since no in vivo or in vitro assays have identified the toxic moiety resulting from 1,4-dioxane exposure,
16 liver toxicity due to metabolites cannot be ruled out. Therefore, this hypothesis is not supported. Nannelli
17 et al. (2005a) demonstrated that an increase in the oxidative metabolism of 1,4-dioxane via CYP450
18 induction using phenobarbital or fasting does not result in an increase in liver toxicity. This result
19 suggested that the highly reactive intermediates did not play a large role in the liver toxicity of
20 1,4-dioxane, even under conditions where metabolism was enhanced. Alternative metabolic pathways
21 (e.g., not CYP450 mediated) may be present at high doses of 1,4-dioxane; although the available studies
22 have not characterized these pathways nor identified any possible reactive intermediates. Tumor
23 promotion studies in mouse skin and rat liver suggest that 1,4-dioxane may enhance the growth of
24 previously initiated cells (Lundberg et al.. 1987; King et al.. 1973a). This is consistent with the increase in
25 rat hepatocyte cell proliferation observed in several studies (Miyagawa et al.. 1999; Uno et al.. 1994;
26 Golds worthy et al.. 1991; Stott et al.. 1981). No studies have been conducted that specifically examine
27 mouse liver, thus precluding any determination on whether 1,4-dioxane acts as a tumor promoter in the
28 mouse liver. These mechanistic studies provide evidence of cell proliferation but do not indicate whether
29 mitogenesis or cytotoxicity is responsible for increased cell turnover.
1 The doses in the hepatotoxicity studies where cytotoxicity and cell proliferation were observed
2 are not equivalent to the doses used in the cancer bioassays. Although Kociba et al. (1974) noted evidence
3 of liver toxicity at or below the dose levels that produced liver tumors, they did not report incidence data
4 for these effects. Thus, a dose-response relationship is unable to be established using the available studies
5 linking cytotoxicity and cell proliferation observations with tumorigenesis. Additionally, conflicting data
6 from rat and mouse bioassays suggest that cytotoxicity may not be a required precursor event for
7 1,4-dioxane-induced cell proliferation.
89
DRAFT - DO NOT CITE OR QUOTE
-------
Toxicokiiietics"
Oral-absorption- of
1.4-dioxane'
Metabolism-by
CYP2El-and
CYP2B1 2'
HE A A -elimin ati on •
in-the-urine1
Metabolic1
saturation-and
accumulation- of
1.4-dioxane-in-the
blood
I Hypothesized- MOA-for-Lirer-Tumorsr
H ep at o cellular •
cvtotoxicitV
Regenerative- cell-
proliferation1
Hyperplasia1
Tumor- formation'
Cell-proliferation- in
absence-of
cvtotoxicitv1
Hvperplasia1
Tumor- promotion1
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
90
-------
1 4.7.3.1.2 Nasal cavity. A possible key event in the MO A hypothesis for nasal
2 tumors is sustained proliferation of spontaneously transformed nasal epithelial cells, resulting in the
3 eventual 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. Nasal lesions, including inflammation,
8 hyperplasia, and metaplasia, were frequently seen in inhalation studies conducted by the NTP with no
9 evidence of nasal carcinogenicity (Haseman and Hailey. 1997b: Wardet al.. 1993). Kasai et al. (2009;
10 2008) suggest that nasal toxicity is related to the accumulation of the parent compound following
11 metabolic induction at high doses up to 3,200 ppm; however, since no in vivo or in vitro assays have
12 examined the toxic moiety resulting from 1,4-dioxane exposure, nasal toxicity due to metabolites cannot
13 be ruled out. Nannelli et al. (2005a) demonstrated that CYP2E1 was inducible in nasal mucosa following
14 acute oral administration of 1,4-dioxane by gavage and drinking water, which could potentially lead to an
15 increase in the oxidative metabolism of 1,4-dioxane and nasal toxicity. However, Nannelli et al. (2005a)
16 neither characterized this pathway nor identified 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,
2 with 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.. 1974a) 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. (1974a). 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. (1974a). liver tumors were
9 observed in female rats and female mice in the absence of reported lesions indicative of cytotoxicity
10 (Kano et al.. 2008; JBRC. 1998; NCI. 1978). This suggests that cytotoxicity may not be a requisite step in
11 the MOA for liver cancer. Mechanistic and tumor promotion studies suggest that enhanced cell
12 proliferation without cytotoxicity may be a key event; however, data showing a plausible dose response
13 and temporal progression from cell proliferation to eventual liver tumor formation are not available (see
14 Sections 4.7.3.3 and 4.7.3.4). Mechanistic studies that demonstrated cell proliferation after short-term
15 exposure did not evaluate liver cytotoxicity (Miyagawa et al.. 1999; Uno et al.. 1994; Golds worthy et al..
16 1991). Studies have not investigated possible precursor events that may lead to cell proliferation in the
17 absence of cytotoxicity (i.e., genetic regulation of mitogenesis).
91
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.2.2 Nasal cavity. Nasal cavity tumors have been demonstrated in several rat
2 strains (Kano et al. 2009: Kasai et al. 2009: JBRC. 1998: Yamazaki et al. 1994a: NCI. 1978: Kociba et
3 al.. 1974a). but were not elevated in two strains of mice (Kano et al.. 2009: JBRC. 1998: Yamazaki et al..
4 1994a: NCI. 1978). Irritation of the nasal cavity of rats was indicated in studies by the observation of
5 inflammation (2009: Kasai et al.. 2008) and also rhinitis (JBRC. 1998). The Kasai et al. (2009: 2008)
6 studies also showed atrophy of the nasal epithelium in rats, and the JRBC (1998) study also observed
7 atrophy of the nasal epithelium as well as adhesion in rats. Regeneration of the nasal epithelium is
8 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.. 1994a). 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. (2005a) 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 cytotoxicity, 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 (i.e., the table columns in
2 sequential order from 1,4-dioxane metabolism, to liver damage, cell proliferation, hyperplasia, and the
3 formation of adenomas and/or carcinomas) and dose-response relationship for possible key events in the
4 liver carcinogenesis of 1,4-dioxane. Dose-response information provides some support for enhanced cell
5 proliferation as a key event in the liver tumorigenesis of 1,4-dioxane; however, the role of cytotoxicity as
6 a required precursor event is not supported by data from more than one study. Kociba et al. (1974a)
7 demonstrated that liver toxicity and hepatocellular regeneration occurred at a lower dose level than tumor
8 formation. Hepatocellular degeneration and necrosis were observed in the mid- and high-dose groups of
9 Sherman rats exposed to 1,4-dioxane, although it is not possible to discern whether this effect was
10 observed in both genders due to the lack of incidence data (Kociba et al.. 1974a). Hepatic tumors were
11 only observed at the highest dose (Kociba et al.. 1974a). Hepatic regeneration was indicated in the mid-
12 and high-dose group by the formation of hepatocellular hyperplastic nodules. Liver hyperplasia was also
13 reported in rats from the JBRC (1998) study, at or below the dose level that resulted in tumor formation
14 (Kano et al.. 2009): however, hepatocellular degeneration and necrosis were not reported. The liver
15 hyperplasia reported in JBRC (1998) was later reclassified to hepatocellular adenoma or altered
16 hepatocellular foci (Kano et al.. 2009). These results suggest that hepatic cell proliferation may occur in
17 the absence of significant cytotoxicity. Liver angiectasis (i.e., dilation of blood or lymphatic vessels) was
18 observed in male mice at the same dose that produced liver tumors; however, the relationship between
19 this vascular abnormality and tumor formation is unclear.
92
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) 1 4 H~ ° 'Sm
Cell
Liver damage pro|iferatjon
Hyperplasia
Adenomas
and/or
carcinomas
Kociba et al., (1974a) — 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
NCI, (1978) — male Osborne-Mendel rats
0 mg/kg-day — a
240 mg/kg-day +b
530 mg/kg-day +b
a
a
+c
+c
a
a
a
a
a
a
a
a
a
a
a
a
+c
+c
a
a
a
a
a
a
+c
a
a
a
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
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
a
a
a
F344/DuCrj rats
a
a
a
+c'a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
+c
+c
a
+c
+c
a
a
a
+c,e
Kano et al.. (2009): JBRC. (1998)— female F344/DuCrj rats
0 mg/kg-day — a
1 8 mg/kg-day +b
83 mg/kg-day +b
429 mg/kg-day +b
Kano et al.. (2009): JBRC. (1998)— male
0 mg/kg-day — a
49 mg/kg-day +b
191 mg/kg-day +b
677 mg/kg-day +b
a
a
a
a
Crj:BDF1 mice
a
a
a
+c'a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
+c,e
a
+c,e
+c,e
+c,e
Kano et al.. (2009): JBRC. (1998)— female Crj:BDF1 mice
0 mg/kg-day — a
66 mg/kg-day +b
278 mg/kg-day +b
964 mg/kg-day +b
Kasai et al. (2008)— F344 rats (male and
0 ppm — a
100 ppm — a
200 ppm — a
400 ppm — a
800 ppm — a
1,600 ppm — a
3,200 ppm — a
a
a
a
+c'a
female combined)
a
a
a
a
a
a
+'
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
+c,e
+c,e
+c,e
a
a
a
a
a
a
a
DRAFT - DO NOT CITE OR QUOTE
93
-------
6,400 ppm
Kasai et al., (2009)—male F344 rats
0 ppm
50 ppm
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.
0 Statistically significant increase noted.
d Single cell necrosis was observed in a 13 week bioassay for male rats (274 mg/kg-day), male mice (585 mg/kg-day), and female
mice (898 mg/kg-day) exposed to 1,4-dioxane in drinking water (Xano et al.. 20081
e+ Kano et al. (2009) reported incidence rates for hepatocellular adenomas and carcinomas.
f+ Kasai et al. (2008) reported significant incidence rates for single cell necrosis in female rats only (3,200 ppm) following a 2 year
bioassay.
9AII rats died during the first week of the 13-week bioassay (Kasai et al.. 20081
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 (i.e., the table columns in sequential order from
2 1,4-dioxane metabolism, to liver damage, cell proliferation, hyperplasia, and the formation of adenomas
3 and/or carcinomas) and dose-response relationship for possible key events in the nasal tissue
4 carcinogenesis of 1,4-dioxane. Toxicity and regeneration in nasal epithelium (i.e., atrophy, adhesion,
5 inflammation, and hyperplasia and metaplasia of respiratory and olfactory epithelium) was evident in one
6 study at the same dose levels that produced nasal cavity tumors (Kano et al.. 2009; JBRC. 1998). In
7 another study, dose-response information provided some support for nasal toxicity and regeneration in
8 nasal epithelium occurring before tumor development (Kasai et al.. 2009). However, the role of
9 cytotoxicity as a required precursor event is not supported by data from any of the reviewed studies. The
10 accumulation of parent 1,4-dioxane as a key event has some support since concentration-dependent
11 increases were noted for 1,4-dioxane in plasma concurrent with toxicities observed that are possible
12 precursor events (i.e., regeneration in nasal epithelium) (Kasai et al.. 2008). In a subsequent study by
13 Kasai et al. (2009) some of these same possible precursor events were observed at 50, 250, and 1,250 ppm
14 with evidence of nasal tumors at the highest concentration (1,250 ppm).
94
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 — >)
or Exposure Metabolism Nasal Cell
(ppm) 1,4-dioxane cytotoxicity proliferation
Hyperplasia
Adenomas
and/or
carcinomas
Kociba et al.. (1974a) — 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
a
a
a
a
a
a
a
a
a
a
a
a
a
a
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
a
a
NCI. (1978)— male B6C3Fi mice
0 mg/kg-day — a
720 mg/kg-day +b
830 mg/kg-day +b
NCI. (1978)— female B6C3Fi mice
0 mg/kg-day — a
380 mg/kg-day +b
860 mg/kg-day +b
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
Kano et al., (2009); JBRC, (1998)— male F344/DuCrj rats
0 mg/kg-day — a
1 1 mg/kg-day +b
55 mg/kg-day +b
274 mg/kg-day +b
Kano et al.. (2009): JBRC. (1998)— female
0 mg/kg-day — a
1 8 mg/kg-day +b
83 mg/kg-day +b
429 mg/kg-day +b
a
a
a
a
F344/DuCrj rats
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
+c,d
a
a
a
+c,d
a
a
a
+c,d
a
a
a
+c,d
Kano et al.. (2009): JBRC. (1998)— male Crj:BDF1 mice
0 mg/kg-day — a
49 mg/kg-day +b
191 mg/kg-day +b
677 mg/kg-day +b
Kano et al.. (2009): JBRC. (1998)— female
0 mg/kg-day — a
66 mg/kg-day +b
278 mg/kg-day +b
964 mg/kg-day +b
a
a
a
a
Crj:BDF1 mice
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
a
Kasai et al. (2008) — F344 rats (male and female combined)
0 ppm — a
100 ppm +b
200 ppm +b
400 ppm +c
800 ppm +c
1,600 ppm +c
3,200 ppm +c
6,400 ppm +a'b''
a
a
a
a
a
a
a
a,t
a
a
a
a
a
a
a
a,t
a
a
a
a
a
a
a
a,t
a
a
a
a
a
a
a
a,t
Kasai et al. (2009)— male F344 rats
0 ppm — a
50 ppm +b
250 ppm +b
a
a
a
a
a
a
a
a
a
a
a
a
DRAFT - DO NOT CITE OR QUOTE
95
-------
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.
°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 (Xasai et al.. 2008).
4.7.3.4 Temporal Relationship
1 4.7.3.4.1 Liver. Available information regarding temporal relationships between the
2 key event (sustained proliferation of spontaneously transformed liver cells) and the eventual formation of
3 liver tumors is limited. A comparison of 13-week and 2-year studies conducted in F344/DuCrj rats and
4 Crj:BDFl mice at the same laboratory revealed that tumorigenic doses of 1,4-dioxane produced liver
5 toxicity by 13 weeks of exposure (Kano et al.. 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 1974a). 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. (1973a) 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.
96
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.7.3.4.2 Nasal cavity. No information was available regarding the temporal
2 relationship between toxicity in the nasal epithelium and the formation of nasal cavity tumors. Sustained
3 nasal damage may lead to regenerative cell proliferation and tumor formation following chronic exposure.
4 As discussed above (Section 4.2.2.2.1), no evidence of cytotoxicity has been observed following exposure
5 to 1,4-dioxane, despite histopathological evidence of regenerative cell proliferation and nasal tumors at
6 the highest exposure concentration (Kano et al.. 2009; Kasai et al.. 2009) (See Table 4-28). Other
7 incidences of nasal damage may have occurred before tumor formation; however, temporal information
8 regarding 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
2 transformed liver cells is a key event within a MOA is possible based on supporting evidence indicating
3 that 1,4-dioxane is a tumor promoter of mouse skin and rat liver tumors (Lundberg et al.. 1987; Bull et al..
4 1986; King et al.. 1973a). Further support for this hypothesis is provided by studies demonstrating that
5 1,4-dioxane increased hepatocyte DNA synthesis, indicative of cell proliferation (Miyagawa et al.. 1999;
6 Uno et al.. 1994; Golds worthy et al.. 1991; Stott et al.. 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
2 toxicity may be related to the formation of nasal cavity tumors; however, this MOA is also not
3 established. Nasal carcinogens are generally characterized as potent genotoxins (Ashby. 1994): however,
4 other MOAs have been proposed for nasal carcinogens that induce effects through other mechanisms
5 (Kasper et al.. 2007: 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. 1997a). 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.. 1974a).
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 1 month up to 41
7 years, reported negative findings regarding increased tumor risk (Buffler et al.. 1978a: Thiess et al..
8 1976a). It is important to note, neither nasal tumors in the human studies nor genotoxicity in human or
9 animal studies were evaluated following inhalation exposure to 1,4-dioxane
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. In general, rats may be more susceptible to nasal lesions than humans due to differences in the
97
DRAFT - DO NOT CITE OR QUOTE
-------
1 anatomy and geometry of the upper respiratory tract (e.g., larger fraction of inspired air ventilates rat
2 nasal cavity compared to the human) and resulting differences in absorption (e.g., rat nasal cavity is more
3 efficient at scrubbing gases than human) or in local respiratory system effects; however, there is not as
4 much known about other respiratory tract lesions (e.g., trachea or lower respiratory tract) (U.S. EPA.
5 2012b. 2009b). Species differences in absorption and respiratory tract uptake for 1,4- dioxane have not
6 been studied, thus it still represents an area of uncertainty for this compound.
4.7.3.6 Other Possible Modes of Action
7 An alternate MO A could be hypothesized that 1,4-dioxane alters DNA, either directly or
8 indirectly (Kasai et al.. 2009). which causes mutations in critical genes for tumor initiation, such as
9 oncogenes or tumor suppressor genes. Following these events, tumor growth may be promoted by a
10 number of molecular processes leading to enhanced cell proliferation or inhibition of programmed cell
11 death. The results from in vitro and in vivo assays do not provide overwhelming support for the
12 hypothesis of a genotoxic MOA for 1,4-dioxane carcinogenicity. The genotoxicity data for 1,4-dioxane
13 were reviewed in Section 4.5.1 and were summarized in Table 4-23. Negative findings were reported for
14 mutagenicity in Salmonella typhimurium, Escherichia coli, and Photobacterium phosphoreum (Mutatox
15 assay) (Morita and Hayashi. 1998; Hellmer and Bolcsfoldi. 1992; Kwanetal. 1990; Khudoley et al..
16 1987; Nestmann et al.. 1984a: Haworth et al.. 1983; Stottetal. 1981). Negative results were also
17 indicated for the induction of aneuploidy in yeast (Saccharomyces cerevisiae) and the sex-linked
18 recessive lethal test in Drosophila melanogaster (Zimmermann et al.. 1985a). In contrast, positive results
19 were reported in assays for sister chromatid exchange (Galloway et al.. 1987a). DNA damage (Kitchin
20 and Brown. 1990a). and in in vivo micronucleus formation in bone marrow (Roy et al.. 2005a: Mirkova.
21 1994a). and liver (Roy et al.. 2005a: Morita and Hayashi. 1998). Lastly, in the presence of toxicity,
22 positive results were reported for meiotic nondisjunction in drosophila (Munoz and Barnett. 2002). DNA
23 damage (Sina et al.. 1983). and cell transformation (Sheu et al.. 1988).
24 Additionally, 1,4-dioxane metabolism did not produce reactive intermediates that covalently
25 bound to DNA (Stott et al.. 1981; Woo et al.. 1977c) and DNA repair assays were generally negative
26 (Golds worthy et al.. 1991; Stottetal.. 1981). No studies were available to assess the ability of
27 1,4-dioxane or its metabolites to induce oxidative damage to DNA.
98
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 MO A by which 1,4-dioxane produces liver tumors is unknown,
2 and 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. 1974a) and a regenerative response to tissue injury was
7 demonstrated by histopathology. Tumor formation has also been observed in the absence of cytotoxicity
8 (Kano et al.. 2009; JBRC. 1998). Cell proliferation and tumor promotion have been shown to occur after
9 prolonged exposure to 1,4-dioxane (Miyagawa et al.. 1999; Uno et al.. 1994; Golds worthy et al.. 1991;
10 Lundberg et al.. 1987: Bulletal. 1986: Stottetal. 1981: Kingetal. 1973a).
1 4.7.3.7.2 Nasal cavity. The MOA for the formation of nasal cavity tumors is
2 unknown, and evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
3 inconclusive. Nasal carcinogens are generally characterized as potent genotoxins (Ashby. 1994):
4 however, other MOAs have been proposed for nasal carcinogens that induce effects through other
5 mechanisms (Rasper et al.. 2007: Green et al.. 2000). In the human studies evidence of nasal tumors were
6 not assessed, nor genotoxicity in human or animal studies following inhalation exposure to 1,4-dioxane,
7 so the role of genotoxicity cannot be ruled out. A MOA hypothesis involving nasal damage, cell
8 proliferation, and hyperplasia is possible, but data are not available to support this hypothesis. In studies
9 that examined nasal effects after exposure to 1,4-dioxane, at least one of these events is missing. More
10 specifically, nasal cavity tumors have been reported by Kasai et al. (2009) in the absence of cytotoxicity
11 and in Kano et al. (2009) in the absence of hyperplasia. Therefore, as per EPA's Cancer Guidelines (U.S.
12 EPA. 2005a). there is insufficient biological support for potential key events and to have reasonable
13 confidence in the sequence of events and how they relate to the development of nasal tumors following
14 exposure to 1,4-dioxane. Using the modified Hill criteria, exposure-response and temporal relationships
15 have not been established in support of any hypothetical mode of carcinogenic action for 1,4-dioxane.
4.7.3.8 Relevance of the Mode of Action to Humans
1 Several hypothesized MOAs for 1,4-dioxane induced tumors in laboratory animals have been
2 discussed along with the supporting evidence for each. Some mechanistic information is available to
3 inform the MOA of the liver and nasal tumors but no information exists to inform the MOA of the other
4 tumor types (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994b). Human
5 relevancy is assumed unless information indicates otherwise (U.S. EPA. 2005b).
99
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 (Watanabe et al..
14 1994; Hayashi et al.. 1991) and were considered to be possible factors in the abnormal liver function seen
15 in workers exposed to vinyl chloride (Huang etal.. 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.
100
DRAFT - DO NOT CITE OR QUOTE
-------
5 DOSE-RESPONSE ASSESSMENTS
5.1 Oral Reference Dose (RfD)
5.1.1 Choice of Principal Studies and Critical Effect with Rationale and
Justification
1 Liver and kidney toxicity were the primary noncancer health effects associated with 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: Kesten et al. 1939: Laugetal. 1939: Schrenk and Yant.
6 1936: deNavasquez. 1935: Fairley et al.. 1934a). 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.. 1994a:
8 NCI. 1978: Kocibaetal. 1974a: Argus etal. 1973a: Argus etal. 1965a) (see Table 4-25).
9 Liver toxicity in the available chronic studies was characterized by necrosis, spongiosis hepatis,
10 hyperplasia, cyst formation, clear foci, and mixed cell foci. Kociba et al. (1974a) demonstrated
11 hepatocellular degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats) or greater, as
12 well as hepatocellular regeneration as indicated by hepatocellular hyperplastic nodule formation at these
13 doses. The NOAEL for liver toxicity was 9.6 mg/kg-day and 19 mg/kg-day in male and female rats,
14 respectively. No quantitative incidence data were provided in this study. Argus et al. (1973a) described
15 early preneoplastic changes in the liver and JBRC (1998) demonstrated liver lesions that are primarily
16 associated with the carcinogenic process. Clear and mixed-cell foci in the liver are commonly considered
17 preneoplastic changes and would not be considered evidence of noncancer toxicity. In the JBRC (1998)
18 study, spongiosis hepatis was associated with other preneoplastic changes in the liver (clear and
19 mixed-cell foci) and no other lesions indicative of liver toxicity were seen. Spongiosis hepatis was
20 therefore not considered indicative of noncancer effects in this study. The activity of serum enzymes (i.e.,
21 AST, ALT, LDH, and ALP) was increased in mice and rats chronically exposed to 1,4-dioxane (JBRC.
22 1998): however, these increases were seen only at tumorigenic dose levels. Blood samples were collected
23 at study termination and elevated serum enzymes may reflect changes associated with tumor formation.
24 Histopathological evidence of liver toxicity was not seen in rats from the JBRC (1998) study. The highest
25 non-tumorigenic dose levels for this study approximated the LOAEL derived from the Kociba et al.
26 (1974a) study (94 and 148 mg/kg-day for male and female rats, respectively).
27 Kidney damage in chronic toxicity studies was characterized by degeneration of the cortical
28 tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCI. 1978: Kociba et al.. 1974a: Argus et
29 al.. 1973a: Argus etal.. 1965a: Fairley et al.. 1934a). Kociba et al. (1974a) described renal tubule
30 epithelial cell degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats) or greater, with
31 a NOAEL of 9.6 mg/kg-day. No quantitative incidence data were provided in this study (Kociba et al..
101
DRAFT - DO NOT CITE OR QUOTE
-------
1 1974a). Doses of > 430 mg/kg-day 1,4-dioxane induced marked kidney alterations (Argus et al. 1973a).
2 The observed changes included glomerulonephritis and pyelonephritis, with characteristic epithelial
3 proliferation of Bowman's capsule, periglomerular fibrosis, and distension of tubules. Quantitative
4 incidence data were not provided in this study. In the NCI (1978) study, kidney lesions in rats consisted
5 of vacuolar degeneration and/or focal tubular epithelial regeneration in the proximal cortical tubules and
6 occasional hyaline casts. Kidney toxicity was not seen in rats from the JBRC (1998) study at any dose
7 level (highest dose was 274 mg/kg-day in male rats and 429 mg/kg-day in female rats).
8 Kociba et al. (1974a) was chosen as the principal study for derivation of the RfD because the liver
9 and kidney effects in this study are considered adverse and represent the most sensitive effects identified
10 in the database (NOAEL 9.6 mg/kg-day, LOAEL 94 mg/kg-day in male rats). Kociba et al. (1974a)
11 reported degenerative effects in the liver, while liver lesions reported in other studies (JBRC. 1998; Argus
12 et al.. 1973a) appeared to be related to the carcinogenic process. Kociba et al. (1974a) also reported
13 degenerative changes in the kidney. NCI (1978) and Argus et al. (1973a) provided supporting data for this
14 endpoint; however, kidney toxicity was observed in these studies at higher doses. JBRC (1998) reported
15 nasal inflammation in rats (NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and mice (NOAEL
16 66 mg/kg-day, LOAEL 278 mg/kg-day).
17 Even though the study reported by Kociba et al. (1974a) had one noteworthy weakness, it had
18 several noted strengths, including: (1) two-year study duration; (2) use of both male and female rats and
19 three dose levels, 10-fold apart, plus a control group; (3) a sufficient number of animals per dose group
20 (60 animals/sex/dose group; and (4) the authors conducted a comprehensive evaluation of the animals
21 including body weights and clinical observations, blood samples, organ weights of all the major tissues,
22 and a complete histopathological examination of all rats. The study weakness was that the authors did not
23 report individual incidence data that would have allowed for a BMD analysis of this robust dataset.
5.1.2 Methods of Analysis—Including Models (PBPK, BMD, etc.)
24 Available human PBPK models were evaluated to determine if an adequate fit of the model to the
25 empirical model output or experimental observations could be attained using biologically plausible values
26 for the model parameters. The re-calibrated model predictions for blood 1,4-dioxane levels did not
27 adequately fit the experimental values (See Appendix B). The model structure is insufficient to capture
28 the apparent species difference in the blood 1,4-dioxane Vd between rats and humans. Differences in the
29 ability of rat and human blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this
30 is expected to be evident in very different values for rat and human blood:air partition coefficients, which
31 is not the case (Table B-l). Additionally, the models do not account for induction in metabolism, which
32 may be present in animals exposed repeatedly to 1,4-dioxane. Therefore, some other modification(s) to
33 the Reitz et al. (1990b) PBPK model structure would be necessary to correct the PBPK models for use in
34 derivation of toxicity values (See Appendix B for more details).
35 Kociba et al. (1974a) did not provide quantitative incidence or severity data for liver and kidney
36 degeneration and necrosis. Therefore, benchmark dose (BMD) modeling could not be performed for this
102
DRAFT - DO NOT CITE OR QUOTE
-------
1 study, and thus the NOAEL for liver and kidney degeneration (9.6 mg/kg-day in male rats) was used as
2 the point of departure (POD) in deriving the RfD for 1,4-dioxane.
3 An alternative POD was derived using incidence data reported for cortical tubule degeneration in
4 the kidneys in male and female rats (NCI. 1978). The incidence data for cortical tubule cell degeneration
5 in male and female rats exposed to 1,4-dioxane in the drinking water for 2 years are presented in
6 Table 5-1. Details of the BMD analysis of these data are presented in Appendix C. Male rats were more
7 sensitive to the kidney effects of 1,4-dioxane than females, and the male rat data provided the lowest POD
8 based on cortical tubule degeneration in the NCI (1978) study (BMDLio of 22.3 mg/kg-day) (Table 5-2).
9 The BMDL10 value of 22.3 mg/kg-day from the NCI (1978) study is about double the NOAEL
10 (9.6 mg/kg-day) observed by Kociba et al. (1974a).
Table 5-1 Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed to
1,4-dioxane in drinking water for 2 years
Males (mg/kg-day)
0
0/3 1a
240
20/3 1b
530
27/33b
Females (mg/kg-day)
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 the incidence 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) BMDLio (mg/kg-day)
Male rats 28.8 22.3
Female rats 596.4 452.4
Source: NCI (1978).
5.1.3 RfD Derivation - Including Application of Uncertainty Factors (UFs)
11 The RfD of 3 x 1CT2 mg/kg-day is based on liver and kidney toxicity in rats exposed to
12 1,4-dioxane in the drinking water for 2 years (Kociba et al.. 1974a). The Kociba et al. (1974a) study was
13 chosen as the principal study because it provides the most sensitive measure of adverse effects by
14 1,4-dioxane. The incidence of liver and kidney lesions was not reported for each dose group. Therefore,
15 BMD modeling could not be used to derive a POD. The RfD for 1,4-dioxane is derived by dividing the
16 NOAEL of 9.6 mg/kg-day (Kociba et al.. 1974a) by a composite UF of 300, as follows:
103
DRAFT - DO NOT CITE OR QUOTE
-------
1 RfD = NOAEL/UF
2 = 9.6 mg/kg-day / 300
3 = 0.03 or 3 x 10~2 mg/kg-day
4 The composite UF of 300 includes factors of 10 for animal-to-human extrapolation and for
5 interindividual variability, and an UF of 3 for database deficiencies.
6 A default interspecies UF of 10 (UFA) was used to account for pharmacokinetic and
7 pharmacodynamic differences between rats and humans. Existing PBPK models could not be used to
8 derive an oral RfD for 1,4-dioxane (Appendix B).
9 A default interindividual variability UF of 10 (UFH) was used to account for variation in
10 sensitivity within human populations because there is limited information on the degree to which humans
11 of varying gender, age, health status, or genetic makeup might vary in the disposition of, or response to,
12 1,4-dioxane.
13 An UF of 3 for database deficiencies was applied due to the lack of a multigeneration
14 reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
15 1,4-dioxane (Giavini et al.. 1985a). This developmental study indicated that the developing fetus may be a
16 target of toxicity.
17 An UF to extrapolate from a subchronic to a chronic (UFS) exposure duration was not necessary
18 because the RfD was derived from a study using a chronic exposure protocol.
19 An UF to extrapolate from a LOAEL to a NOAEL (UFL) was not necessary because the RfD was
20 based on a NOAEL. Kociba et al. (1974a) was a well-conducted, chronic drinking water study with an
21 adequate number of animals. Histopathological examination was performed for many organs and tissues,
22 but clinical chemistry analysis was not performed. NOAEL and LOAEL values were derived by the study
23 authors based on liver and kidney toxicity; however, quantitative incidence data were not reported.
24 Several additional oral studies (of acute/short-term, subchronic, and chronic durations) were available that
25 support liver and kidney toxicity as the critical effect (Kano et al.. 2008; JBRC. 1998; NCI. 1978; Argus
26 et al.. 1973a) (Table 4-15 and Table 4-17). Although degenerative liver and kidney toxicity was not
27 observed in rats from the JBRC (1998) study at doses at or below the LOAEL in the Kociba et al. (1974a)
28 study, other endpoints such as metaplasia and hyperplasia of the nasal epithelium, nuclear enlargement,
29 and hematological effects, were noted.
5.1.4 RfD Comparison Information
30 PODs and candidate oral RfDs based on selected studies included in Table 4-18 are arrayed in
31 Figure 5-1 to Figure 5-3, and provide perspective on the RfD supported by Kociba et al. (1974a). These
32 figures should be interpreted with caution because the PODs across studies are not necessarily
33 comparable, nor is the confidence in the data sets from which the PODs were derived the same. PODs in
34 these figures may be based on a NOAEL, LOAEL, or BMDL (as indicated), and the nature, severity, and
104
DRAFT - DO NOT CITE OR QUOTE
-------
1 incidence of effects occurring at a LOAEL are likely to vary. To some extent, the confidence associated
2 with the resulting candidate RfD is reflected in the magnitude of the total UF applied to the POD (i.e., the
3 size of the bar); however, the text of Sections 5.1.1 and 5.1.2 should be consulted for a more complete
4 understanding of the issues associated with each data set and the rationale for the selection of the critical
5 effect and principal study used to derive the candidate RfD.
6 The predominant noncancer effect of chronic oral exposure to 1,4-dioxane is degenerative effects
7 in the liver and kidney. Figure 5-1 provides a graphical display of effects that were observed in the liver
8 following chronic oral exposure to 1,4-dioxane. Information presented includes the PODs and UFs that
9 could be considered in deriving the oral RfD. As discussed in Sections 5.1.1 and 5.1.2, among those
10 studies that demonstrated liver toxicity, the study by Kociba et al. (1974a) provided the data set most
11 appropriate for deriving the RfD. For degenerative liver effects resulting from 1,4-dioxane exposure, the
12 Kociba et al. (1974a) study represents the most sensitive effect and dataset observed in a chronic bioassay
13 (Figure 5-1).
14 Kidney toxicity as evidenced by glomerulonephritis (Argus et al.. 1973a: Argus et al.. 1965a) and
15 degeneration of the cortical tubule (NCI. 1978; Kociba et al.. 1974a) has also been observed in response
16 to chronic exposure to 1,4-dioxane. As was discussed in Sections 5.1 and 5.2, degenerative effects were
17 observed in the kidney at the same dose level as effects in the liver (Kociba et al.. 1974a). A comparison
18 of the available datasets from which an RfD could potentially be derived based on this endpoint is
19 presented in Figure 5-2.
20 Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
21 dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice).
22 JBRC (1998) reported nasal inflammation in rats (NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and
23 mice (NOAEL 66 mg/kg-day, LOAEL 278 mg/kg-day). A comparison of the available datasets from
24 which an RfD could potentially be derived based on this endpoint is presented in Figure 5-3.
25 Figure 5-4 displays PODs for the major targets of toxicity associated with oral exposure to
26 1,4-dioxane. Studies in experimental animals have also found that relatively high doses of 1,4-dioxane
27 (1,000 mg/kg-day) administered during gestation can produce delayed ossification of the sternebrae and
28 reduced fetal BWs (Giavini et al.. 1985a). This graphical display (Figure 5-4) compares organ specific
29 toxicity for 1,4-dioxane, including a single developmental study. The most sensitive measures of toxicity
30 are degenerative liver and kidney effects. The sample RfDs for degenerative liver and kidney effects are
31 identical since they were derived from the same study and dataset (Kociba et al.. 1974a) and are presented
32 for completeness.
105
DRAFT - DO NOT CITE OR QUOTE
-------
100
10
Rat
Mouse
Rat
o
E.
v
g
o
0 1
0.01
• POD
H|Animal-io-human
H3LOAEL to NOAEL
DSubchronicto Chronic
^Database deficiencies
ORfD
Hepatocellulardegenerationand Increase in serum liverenzymes; Increase in serum liverenzymes;
necrosis: NOAEL: 2 yr rat drinking water NOAEL: 2 yr mouse drinking water study NOAEL: 2 yr rat drinking water study
study (KocibaetaL 1974.062929; (JBRC, 1998.196240) (JBRC. 1993.196240}
Figure 5-1 Potential points of departure (POD) based on liver toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane.
1000
100
10
Rat
Rat
Rat
0.1
0.01
• POD
nTllAnlmal-to-human
I [Human variation
EI]LOAELto NOAEL
CUSubchronicto Chronic
Moatabase deficiencies
ORfD
Glomerulonephritis: LOAEL: 13 month
rat drinkingwater study (Argus et al..
1973. 0629121
Degeneration and necrosis oftubular Cortical tubule degeneration: BMDL1D:
epithelium: NOAEL: 2 yr rat drinking 2yr rat drinking water study (NCI. 1978.
water study (Kociba et al., 1974. 062935)
062929)
Figure 5-2 Potential points of departure (POD) based on kidney toxicity with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane.
106
DRAFT - DO NOT CITE OR QUOTE
-------
Mouse
Rat
100
10
0 1
I
O
• POD
HIjAnimal-to-human
Q] Human variation
HLOAELto NOAEL
Dsubchronicto Chronic
• Database deficiencies
oRfD
Nasal inflammation: NOAEL: 2 yr mouse drinking water study
(JBRC: 1998,196240)
Nasal inflammation: NOAEL: 2 yr rat drinking water study
(JBRC, 1998,196240)
Figure 5-3 Potential points of departure (POD) based on nasal inflammation with
corresponding applied uncertainty factors and derived candidate RfDs
following chronic oral exposure to 1,4-dioxane.
1000
100
Rat
Kidney
Rat
Liver
Rat
Developmental
Mouse
Respiratory
10
ci
001
.
Degeneration and necrosis Hepatocellular degeneration
oftubularepitnelium; and necrosis; NOAEL;2yr
NOAEL; 2 yr rat drinking rat drinking water sduy
waterstudy (KocibaetaL (Kocibaetal., 1974,
1974,062929) 062929)
Delayed ossification of Nasal inflammation:
stemebrae and reduced NOAEL; 2 yr mouse
fetal body weight: NOAEL; drinking water study (JBRC.
rat study gestation days 6- 1998,196240)
15(Giavinietal., 1985
062924)
Figure 5-4 Potential points of departure (POD) based on organ-specific toxicity
endpoints with corresponding applied uncertainty factors and derived
candidate RfDs following chronic oral exposure to 1,4-dioxane.
107
DRAFT - DO NOT CITE OR QUOTE
-------
5.1.5 Previous RfD Assessment
1 An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
2 oral RfD was not developed as part of the 1988 assessment.
5.2 Inhalation Reference Concentration (RfC)
5.2.1 Choice of Principal Study and Candidate Critical Effect(s) with
Rationale and Justification
3 Two human studies of occupational exposure to 1,4-dioxane have been published (Buffler et al..
4 1978a: Thiess et al.. 1976a): however, neither study provides sufficient information and data to quantify
5 subchronic or chronic noncancer effects. In each study, findings were negative and deemed inconclusive
6 by the EPA due to the small cohort size and the limited number of reported cases (Buffler et al.. 1978a:
7 Thiess etal. 1976a).
8 Four inhalation studies in animals were identified in the literature; two 13-week subchronic
9 studies in several species of laboratory animals (Kasai et al.. 2008; Fairley et al.. 1934a) and two 2-year
10 chronic studies in rats (Kasai et al.. 2009; Torkelson et al.. 1974a).
11 In the subchronic study by Fairley et al. (1934a). rabbits, guinea pigs, rats, and mice
12 (3-6/species/group) were exposed to 1,000, 2,000, 5,000, or 10,000 ppm of 1,4-dioxane vapor for
13 1.5 hours two times a day for 5 days, 1.5 hours for one day, and no exposure on the seventh day. Animals
14 were exposed until death occurred or were sacrificed after various durations of exposure (3-202.5 hours).
15 Detailed dose-response information was not provided; however, severe kidney and liver damage and
16 acute vascular congestion of the lungs were observed at concentrations > 1,000 ppm. Kidney damage was
17 described as patchy degeneration of cortical tubules with vascular congestion and hemorrhage. Liver
18 lesions varied from cloudy hepatocyte swelling to large areas of necrosis. In this study, a LOAEL of
19 1,000 ppm for liver and kidney degeneration in rats, mice, rabbits, and guinea pigs was identified by EPA.
20 In the subchronic study by Kasai et al. (2008). male and female rats (10/group/sex) were exposed
21 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
22 weeks. This study observed a range of 1,4-dioxane-induced nonneoplastic effects across several organ
23 systems including the liver and respiratory tract (from the nose to the bronchus region) in both sexes and
24 the kidney in females. Detailed dose-response information was provided, illustrating a vapor
25 concentration-dependent increase of nuclear enlargement of nasal (respiratory and olfactory), trachea, and
26 bronchus epithelial cells (both sexes); vacuolic changes in nasal and bronchial epithelial cells (both
27 sexes), necrosis and centrilobular swelling of hepatocytes (both sexes); and hydropic change in the
28 proximal tubules of the kidney (females). The study authors determined nuclear enlargement of the nasal
29 respiratory epithelium as the most sensitive lesion and a LOAEL of 100 ppm was identified based on this
30 effect. However, it is important to note that the severity of the change (i.e., nuclear enlargement) was
108
DRAFT - DO NOT CITE OR QUOTE
-------
1 similar (i.e., slight) at the four lowest tested vapor levels (i.e., 100, 200, 400 and 800 ppm) in male and
2 female rats; with only a moderate observation of severity noted at the two highest tested vapor levels (i.e.,
3 1,600 and 3,200 ppm). Furthermore, nuclear enlargement has only been reported in studies conducted by
4 JBRC (Kano et al. 2009: Kasai et al. 2009: Kano et al. 2008: Kasai et al. 2008). and there is a lack of
5 information concerning the nature, severity, and significance of this observation elsewhere in the
6 literature. Thus, the toxicological significance of nuclear enlargement is largely unknown.
7 Torkelson et al. (1974a) performed a chronic inhalation study in which male and female Wistar
8 rats (288/sex) were exposed to 111 ppm 1,4-dioxane vapor for 7 hours/day, 5 days/week for 2 years.
9 Control rats (192/sex) were exposed to filtered air. No significant effects were observed on BWs,
10 survival, organ weights, hematology, clinical chemistry, or histopathology. A free standing NOAEL of
11 111 ppm was identified in this study by EPA.
12 Kasai et al. (2009) reported data for groups of male F344 rats (50/group) exposed to 0, 50, 250,
13 and 1,250 ppm of 1,4-dioxane for 6 hours/day, 5 days/week, for 2 years. In contrast to the subchronic
14 Kasai et al. (2008) study, this 2-year bioassay reported more nonneoplastic effects in multiple organ
15 systems. Effectsobserved included: (1) inflammation of nasal respiratory and olfactory epithelium, (2)
16 squamous cell metaplasia and hyperplasia of nasal respiratory epithelium, (3) atrophy and respiratory
17 metaplasia of olfactory epithelium, (4) hydropic change and sclerosis in the lamina propria of nasal
18 cavity, (5) nuclear enlargement in proximal tubules of the kidney, in the centrilobular region of the liver,
19 and of the respiratory and olfactory epithelium, (6) centrilobular necrosis in the liver, and (7) spongiosis
20 hepatis. Some of these histopathological lesions were significantly increased compared to controls at the
21 lowest exposure level (50 ppm), including nuclear enlargement of respiratory and olfactory epithelium;
22 and atrophy and respiratory metaplasia of olfactory epithelium. Many of these histopathological lesions
23 were increased in a concentration-dependent manner.
24 Whether spongiosis hepatis/cystic degeneration represents a preneoplastic change or a
25 nonneoplastic change has been the subject of scientific controversy (Karbe and Kerlin. 2002b: Stroebel et
26 al.. 1995: Bannasch et al.. 1982b). Spongiosis hepatis is commonly seen in aging rats, but has been shown
27 to increase in incidence following exposure to hepatocarcinogens. Spongiosis hepatis can be seen in
28 combination with preneoplastic foci in the liver or with hepatocellular adenoma or carcinoma and has
29 been considered a preneoplastic lesion (Bannasch. 2003: Stroebel et al.. 1995). In contrast, it can also be
30 associated with hepatocellular hypertrophy and liver toxicity and has been regarded as a secondary effect
31 of some liver carcinogens (Karbe and Kerlin. 2002a). Following inhalation of 1,4-dioxane, spongiosis
32 hepatis was associated with other preneoplastic (e.g., liver foci) and nonneoplastic (e.g., centrilobular
33 necrosis) changes in the liver (Kasai et al.. 2009). However, the incidence rates of spongiosis hepatis and
34 liver tumors were highly correlated; therefore, spongiosis hepatis was considered a preneoplastic lesion
35 following inhalation exposure and not considered further in the noncancer analysis.
36 The Fairley et al. (1934a) study was inadequate to characterize the inhalation risks of 1,4-dioxane
37 because control animals were not used, thus limiting the ability to perform statistical analysis;
38 additionally, no data for low-dose exposure were reported. Because Torkelson et al. (1974a) identified a
39 free-standing NOAEL only, this study was also deemed inadequate to characterize the inhalation risks of
109
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-dioxane. A route-to-route extrapolation from the oral toxicity data was not performed because
2 1,4-dioxane inhalation causes direct effects on the respiratory tract (i.e., respiratory irritation in humans,
3 pulmonary congestion in animals) (Wirth and Klimmer. 1936; Fairley et al.. 1934a: Yant et al.. 1930).
4 which would not be accounted for in a cross-route extrapolation. In addition, available kinetic models are
5 not suitable for this purpose (Appendix B).
6 Therefore, the chronic Kasai et al. (2009) study was selected as the principal study for the
7 derivation of the RfC. The Kasai et al. (2009) 2-year bioassay utilized 50 animals per exposure group, a
8 range of exposure concentrations which were based on the results of the subchronic study (Kasai et al..
9 2008). and thoroughly examined toxicity of 1,4-dioxane in multiple organ systems. Based on the
10 noncancer database for 1,4-dioxane, this study demonstrated exposure concentration-related effects for
11 histopathological lesions at a lower concentration (50 ppm) compared to the subchronic Kasai et al. study
12 (2008). The 2-year bioassay (Kasai et al.. 2009) did not observe effects in both sexes, but the use of only
13 male rats was proposed by the study authors as justified because of data illustrating the absence of
14 induced mesotheliomas in female rats following exposure to 1,4-dioxane in drinking water (Yamazaki et
15 al.. 1994a). Additionally, a similar pattern of effects was observed after oral exposure to 1,4-dioxane
16 (Kano et al.. 2009; JBRC. 1998) as was observed in the Kasai et al. (2009) 2-year inhalation study.
17 Incidences of nonneoplastic lesions from the Kasai et al. (2009) study that were statistically
18 significantly increased as compared to control were considered candidates for the critical effect. These
19 candidate endpoints included centrilobular necrosis of the liver, squamous cell metaplasia of the nasal
20 respiratory epithelium, squamous cell hyperplasia of the nasal respiratory epithelium, respiratory
21 metaplasia of the nasal olfactory epithelium, sclerosis in the lamina propria of the nasal cavity, and two
22 degenerative nasal lesions, that is, atrophy of the nasal olfactory epithelium and hydropic change in the
23 lamina propria (Table 5-). Despite statistically significant increases at the low- and mid-exposure
24 concentrations (50 and 250 ppm, respectively), incidences of nuclear enlargement of the respiratory
25 epithelium (nasal cavity), olfactory epithelium (nasal cavity), and proximal tubule (kidney) were not
26 considered candidates for the critical effect given that the toxicological significance of nuclear
27 enlargement is uncertain, as discussed previously (See Section 4.6.2 and Table 4-22).
110
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-3 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)
Liver Centrilobular necrosis
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
0/50
0/50
11/50
0/50
0/50
0/50
50
3/50
0/50
0/50
34/50a
40/50a
2/50
0/50
250
6/50
7/50b
1/50
49/50a
47/50a
36/50a
22/50a
1,250
12/50a
44/50a
10/50a
48/50a
48/50a
49/50a
40/50a
ap<0.01 by x^ test.
Source: Kasai et al. (2QQ9).
5.2.2 Methods of Analysis
1 Benchmark dose (BMD) modeling (U.S. EPA. 2012a) was used to analyze the candidate
2 endpoints identified for 1,4-dioxane. Use of BMD methods involves fitting mathematical models to the
3 observed dose-response data and provides a BMD and its 95% lower confidence limit (BMDL) associated
4 with a predetermined benchmark response (BMR). For 1,4-dioxane, the selected datasets in Table 5- were
5 considered as candidate critical effects and analyzed using BMD modeling to determine potential PODs.
6 Information regarding the degree of change in the selected endpoints that is considered biologically
7 significant was not available. Therefore, a BMR of 10% extra risk was selected under the assumption that
8 it represents a minimally biologically significant response level (U.S. EPA. 2012a).
9 The estimated BMDs and BMDLs based on incidences of centrilobular necrosis, squamous cell
10 metaplasia and hyperplasia of the respiratory epithelium, and hydropic change of lamina propria are
11 presented in Table 5-. Due to lack of fit or substantial model uncertainty, BMD modeling results were
12 deemed inadequate for the following endpoints: atrophy (olfactory epithelium), respiratory metaplasia
13 (olfactory epithelium), and sclerosis (lamina propria). Consequently, for these last three endpoints, the
14 NOAEL/LOAEL approach was used to determine potential PODs. The detailed results of the BMD
15 analysis are provided in Appendix F.
5.2.3 Exposure Duration and Dosimetric Adjustments
16 Because an RfC assumes continuous human exposure over a lifetime, data derived from
17 inhalation studies in animals need to be adjusted to account for the noncontinuous exposure protocols
18 used in these 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 PODs for liver and nasal lesions in rats were
2 calculated as follows:
hours exposed per day days exposed per week
PODADJ (ppm) = POD (ppm) x - ^^ - x - — -
3 RfCs are typically expressed in units of mg/m3; so PODADj (ppm) values were converted using
4 the chemical specific conversion factor of 1 ppm = 3.6 mg/m3 for 1,4-dioxane (Table 2-1). The following
5 calculation was used:
PODADJ (mg/m3) = PODADJ (ppm) x
6 The calculated POD^ (mg/m3) values for all considered endpoints are presented in the last
7 column of Table 5-4.
Table 5-4 Duration adjusted POD estimates for BMDLs (from best fitting BMDS models) or
NOAELs/LOAELs from chronic exposure to 1,4-dioxane
NOAEL3 LOAEL1
(ppm) (ppm)
Model
BMR
BMD
(ppm)
BMDL
(ppm)
PODADJ
(ppm)
PODADJ
(mg/m3)
Liver Effects
Centrilobular
necrosis; Liver
Dichotomous-Hill
10
220
60
10.7
38.6
Nasal Effects
Squamous cell
metaplasia;
respiratory epithelium
Log-probit
10
218
160
28.6
103
Squamous cell
hyperplasia; - - Log-probit 10 756 561 100 361
respiratory epithelium
Respiratory
metaplasia; olfactory
epithelium
Atrophy; olfactory
epithelium
Hydropic change;
lamina propria
Sclerosis; lamina „
propria
50
50
Log-logistic 10 69 47
250
8.9
8.9
8.4
8.9
32.2
32.2
30.2
32.2d
aNOAEL is identified in this assessment as the highest tested exposure dose at which there is no statistically significant effect in
the exposed group as compared to control.
bLOAEL is identified in this assessment as the lowest tested exposure dose at which there is a statistically significant effect in the
exposed group as compared to control.
°BMD modeling results are inadequate for use in deriving a POD. Therefore, the NOAEL/LOAEL approach is used to determine
a POD for these endpoints. BMD analysis for these endpoints is described in Appendix F.
dBased on the NOAEL of 50 ppm.
8 Based on a review of the data in Table 5-4, hepatic centrilobular necrosis was shown to be less
9 sensitive than the nasal effects and was not considered further as a candidate critical effect. Similarly, the
10 squamous cell metaplasia and hyperplasia of the respiratory epithelium yielded potential PODs that were
11 at least 3-fold higher than the remaining nasal effects; thus, these two effects were not considered further
12 as candidate critical effects. The PODs (adjusted for continuous exposure) for sclerosis of the lamina
112
DRAFT - DO NOT CITE OR QUOTE
-------
1 propria, atrophy of the olfactory epithelium, and respiratory metaplasia of the olfactory epithelium were
2 identical (32.2 mg/m3) and similar to the P()DADJ for hydropic change of the lamina propria (30.2 mg/m3).
3 Although the P()DADJ estimates for these four endpoints were either identical or similar, the magnitude of
4 response (i.e., increased incidence of effect) at each POD ADJ for these effects varied (i.e., 0% for
5 sclerosis, 10% for hydropic change, 59% for respiratory metaplasia, 80% for atrophy).
6 As shown in Table 5-3, atrophy and respiratory metaplasia of the olfactory epithelium were the
7 most sensitive effects based on responses of 80 and 59% at their respective PODs of 50 ppm (LOAELs).
8 Increased incidences of the other nasal effects, as well as liver effects (i.e., centrilobular necrosis), were
9 observed at exposures of 50 ppm or greater and the magnitude of the responses at these exposures were
10 lower than those observed for atrophy and respiratory metaplasia of the olfactory epithelium. Typically,
11 chemically-induced nasal effects include atrophy and/or necrosis, cell proliferation/hyperplasia, and
12 metaplasia depending on the nature of the tissue damage and level of exposure (Harkema et al.. 2006;
13 Boorman et al.. 1990; Gaskell. 1990). However, the pathological progression of these events is uncertain
14 and often accompanied by an inflammatory response. Since the data do not support a continuum of
15 pathological events associated with respiratory tract effects, both atrophy and respiratory metaplasia of
16 the olfactory epithelium were selected as co-critical effects in this assessment. Additionally, these effects
17 were the most sensitive non-cancer effects considered following inhalation of 1,4-dioxane.
18 For the derivation of a RfC based upon an animal study, the selected POD must be adjusted to
19 reflect the human equivalent concentration (HEC). The HEC was calculated by the application of a
20 dosimetric adjustment factor (DAF), in accordance with the U.S. EPA Methods for Derivation of
21 Inhalation Reference Concentrations and Application of Inhalation Dosimetry (hereafter referred to as the
22 RfC methodology) (U.S. EPA. 1994a). DAFs are ratios of animal and human physiologic parameters, and
23 are dependent on the nature of the contaminant (particle or gas) and the target site (e.g., respiratory tract
24 or remote to the portal-of-entry) (U.S. EPA. 1994b).
25 1,4-Dioxane is miscible with water and has a high blood:air partition coefficient. Typically,
26 highly water-soluble and directly reactive chemicals (i.e. Category 1 gases) partition predominantly into
27 the upper respiratory tract, induce portal-of-entry effects, and do not accumulate significantly in the
28 blood. 1,4-Dioxane induces effects at the portal-of-entry (i.e., respiratory tract), liver, and kidneys, and it
29 has been measured in the blood after inhalation exposure (Kasai et al.. 2009; Kasai et al.. 2008). The
30 observations of systemic (i.e., nonrespiratory) effects and measured blood levels resulting from
31 1,4-dioxane exposure indicate that this compound is absorbed into the bloodstream and distributed
32 throughout the body. Thus, 1,4-dioxane might be best described as a water-soluble and non-directly
33 reactive gas. Gases such as these are readily taken up into respiratory tract tissues and can also diffuse
34 into the blood (Medinsky and Bond. 2001). The effects observed in the olfactory epithelium may be the
35 result of the metabolism of 1,4-dioxane to an acid metabolite; however, for the reasons stated above, it is
36 unclear whether or not these effects are solely the result of portal-of-entry or systemic delivery. A similar
37 pattern of effects was observed after oral exposure to 1,4-dioxane (Kano et al.. 2009; JBRC. 1998).
113
DRAFT - DO NOT CITE OR QUOTE
-------
1 In consideration of the evidence described above, the human equivalent concentration (HEC) for
2 1,4-dioxane was calculated by the application of the appropriate dosimetric adjustment factor (DAF) for
3 systemic acting gases, in accordance with the U.S. EPA RfC methodology (U.S. EPA. 1994a).
4 The calculation of the HEC used in this assessment is as follows:
5 DAF = (Hb/g)A/(Hb/g)H
6 DAF = 1,861/1,666
7 DAF =1.12
8 where:
9 (Hb/g)A = the animal blood:air partition coefficient =1,861 (Sweeney et al. 2008a)
10 (Hb/g)H = the human blood:air partition coefficient =1,666 (Sweeney et al.. 2008a)
11 Given that the animal blood:air partition coefficient is higher than the human value resulting in a
12 DAF>1, a default value of 1 is substituted in accordance with the U.S. EPA RfC methodology (U.S. EPA.
13 1994a). Analysis of the existing inhalation dosimetry modeling database supports the application of a
14 DAF of 1 for a systemic acting gas (U.S. EPA. 2012b. 2009c. b). In addition, a robust computational fluid
15 dynamic (CFD) and PBPK modeling database supports the scientific rationale to apply of DAF of 1 for
16 both portal of entry and systemic effects irrespective of "gas categorization" (U.S. EPA. 2012b).
17 Application of these models to gases that have similar physicochemical properties and induce similar
18 nasal effects as 1,4-dioxane yield estimated DAFs > 1.
19 Utilizing a DAF of 1, the HEC for atrophy and respiratory metaplasia of the olfactory epithelium
20 in male F344/DuCrj rats is calculated as follows:
21 PODnEc (mg/m3) = POD^ (mg/m3) x DAF
22 = PODADj (mg/m3) x 1.0
23 = 32.2 mg/m3 x 1.0
24 = 32.2 mg/m3
25 Therefore, the PODnEc of 32.2 mg/m3 for the co-critical effects of atrophy and respiratory
26 metaplasia of the olfactory epithelium is used for the derivation of a RfC for 1,4-dioxane.
5.2.4 RfC Derivation- Including Application of Uncertainty Factors (UFs)
27 The RfC of 3 x 1CT2 mg/m3 is based on atrophy and respiratory metaplasia of the olfactory
28 epithelium in male rats exposed to 1,4-dioxane via inhalation for 2 years (Kasai et al.. 2009). The RfC for
29 1,4-dioxane is derived by dividing the PODnEc by a composite UF of 1,000.
114
DRAFT - DO NOT CITE OR QUOTE
-------
1 RfC = PODnEc / UF
2 = 32.2 mg/m3/1,000
3 = 0.0322 or 3 x 10~2 mg/m3 (rounded to 1 significant figure)
4 An interspecies UF of 3 (UFA) was used for animal-to-human extrapolation to account for
5 pharmacodynamic differences between species. This uncertainty factor is comprised of two separate areas
6 of uncertainty to account for differences in the toxicokinetics and toxicodynamics of animals and humans.
7 In this assessment, the toxicokinetic uncertainty was accounted for by the calculation of a HEC and
8 application of a dosimetric adjustment factor as outlined in the RfC methodology (U.S. EPA. 1994a). As
9 the toxicokinetic differences are thus accounted for, only the toxicodynamic uncertainties remain, and an
10 UFA of 3 is retained to account for this uncertainty.
11 A default interindividual variability UF of 10 (UFH) was used to account for variation in
12 sensitivity within human populations because there is limited information on the degree to which humans
13 of varying gender, age, health status, or genetic makeup might vary in the disposition of, or response to,
14 1,4-dioxane.
15 An UF to extrapolate from a subchronic to a chronic (UFS) exposure duration was not necessary
16 (e.g., UFS = 1) because the RfC was derived from a study using a chronic exposure protocol.
17 An UF of 10 (UFL) was used to extrapolate from a LOAEL to a NOAEL because a LOAEL was
18 used as the POD. A NOAEL for atrophy and respiratory metaplasia of the olfactory epithelium was not
19 identified in the study by Kasai et al. (2009).
20 An UF of 3 for database deficiencies (UFD) 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.. 1985a). Giavini et al. (1985b)
23 administered 1,4-dioxane by gavage in water to pregnant rats. The authors found statistically significant
24 changes in fetal body weight at the highest dose group and reduced ossification of the sternebrae;
25 however, the lack of a multigenerational reproductive study in which animals were exposed to 1,4-
26 dioxane via oral or inhalation routes warrants the use of a 3 for UFD.
5.2.5 RfC Comparison Information
27 Figure 5-5 presents PODs, applied UFs, and derived candidate RfCs based on each of the
28 endpoints from the chronic inhalation study by Kasai et al. (2009) in male rats. The PODs are based on
29 the BMDLio, NOAEL, or LOAEL, and appropriate unit conversions, duration, and dosimetric
30 adjustments were applied before applications of UFs. The predominant noncancer effects of chronic
31 inhalation exposure to 1,4-dioxane include nasal and liver effects. Figure 5-5 provides a graphical display
32 of these effects that were observed in the Kasai et al. (2009) study. The nasal effects involving the
33 olfactory epithelium represent the most sensitive effects.
115
DRAFT - DO NOT CITE OR QUOTE
-------
1000
• POD
QDAnimal-to-human
QHuman variation
LOAELtoNOAEL
DSubchronicto Chronic
• Database deficiencies
ORfC
0.01
Squamous cell Squamous call Respiratory Atrophyinthe Hydropic Sclerosis of the Centrilobular
metaplasia in hyperplasiain metaplasia in nasal olfactory change in lamina propria; necrosis in the
the respiratory the respiratory theolfactory epithelium; lamina propria; NOAEL liver; BMDL10
epithelium; epithelium; epithelium; LOAEL NOAEL
NOAEL NOAEL LOAEL
Figure 5-5 Potential points of departure (POD) for candidate endpoints with
corresponding applied uncertainty factors and derived candidate RfCs
following chronic inhalation exposure of F344 male rats to 1,4-dioxane.
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
2 The following discussion identifies the uncertainties associated with deriving the RfD and RfC
3 for 1,4-dioxane. As presented earlier in this section (see Sections 5.1.2, 5.1.3 for the RfD and Sections
4 5.2.2, and 5.2.3 for the RfC), the uncertainty factor approach (U.S. EPA. 2002b. 1994a) was used to
5 derive the RfD and RfC for 1,4-dioxane. Using this approach, the POD was divided by a set of factors to
6 account for uncertainties associated with a number of steps in the analysis, including extrapolation from
7 LOAEL to NOAEL, extrapolation from animals to humans, a diverse population of varying
8 susceptibilities, and to account for database deficiencies. Because information specific to 1,4-dioxane was
9 unavailable to fully inform these extrapolations, default factors were generally applied.
10 An adequate range of animal toxicology data are available for the hazard assessment of
11 1,4-dioxane, as described throughout the previous section (Section 4). The database of oral toxicity
116
DRAFT - DO NOT CITE OR QUOTE
-------
1 studies includes chronic drinking water studies in rats and mice, multiple subchronic drinking water
2 studies conducted in rats and mice, and a developmental study in rats. Toxicity associated with oral
3 exposure to 1,4-dioxane is observed predominately in the liver and kidney. The database of inhalation
4 toxicity studies in animals includes two subchronic bioassays in rabbits, guinea pigs, mice, and rats, and
5 two chronic inhalation bioassays in rats. Toxicity associated with inhalation exposure to 1,4-dioxane was
6 observed predominately in the liver and nasal cavity. In addition to oral and inhalation data, there are
7 PBPK models and genotoxicity studies of 1,4-dioxane. Critical data gaps have been identified and
8 uncertainties associated with data deficiencies of 1,4-dioxane are more fully discussed below.
9 Consideration of the available dose-response data led to the selection of the two-year drinking
10 water bioassay in Sherman rats (Kociba et al. 1974a) as the principal study and increased liver and
11 kidney degeneration as the critical effects for deriving the RfD for 1,4-dioxane. The dose-response
12 relationship for oral exposure to 1,4-dioxane and cortical tubule degeneration in Osborne-Mendel rats
13 (NCI. 1978) was also suitable for deriving a RfD, but it is associated with a higher POD and potential
14 RfD compared to the same values derived from Kociba et al. (1974a).
15 The RfD was derived by applying UFs to a NOAEL for degenerative liver and kidney effects.
16 The incidence data for the observed effects were not reported in the principal study (Kociba et al.. 1974a).
17 precluding BMD modeling of the dose-response. However, confidence in the NOAEL can be derived
18 from additional studies (JBRC. 1998: NCI. 1978: Argus etal.. 1973a: Argus etal.. 1965a) that observed
19 effects on the same organs at comparable dose levels and by the BMDL generated by modeling of the
20 kidney dose-response data from the chronic NCI (1978) study.
21 The RfC was derived by applying UFs to a LOAEL for atrophy and respiratory metaplasia of the
22 olfactory epithelium. The incidence data for the observed effects were not amenable to BMD modeling
23 (see Appendix F). The LOAEL for these effects was less than or equal to the LOAEL or NOAEL for
24 other effects observed in the Kasai et al. (2009) study.
25 Extrapolating from animals to humans embodies further issues and uncertainties. The effect and
26 the magnitude associated with the dose at the POD in rodents are extrapolated to human response.
27 Pharmacokinetic models are useful to examine species differences in pharmacokinetic processing;
28 however, it was determined that dosimetric adjustment using pharmacokinetic modeling to reduce
29 uncertainty following oral exposure to 1,4-dioxane was not supported. Insufficient information was
30 available to quantitatively assess toxicokinetic or toxicodynamic differences between animals and
31 humans, so a 10-fold UF was used to account for uncertainty in extrapolating from laboratory animals to
32 humans in the derivation of the RfD. A DAF was used to account for pharmacokinetic differences
33 between rodents and humans in the derivation of the RfC; however, there was no information to inform
34 pharmacodynamic differences between species, so an UF of 3 was used in derivation of the RfC to
3 5 account for these uncertainties.
36 Heterogeneity among humans is another uncertainty associated with extrapolating doses from
37 animals to humans. Uncertainty related to human variation needs consideration. In the absence of
38 1,4-dioxane-specific data on human variation, a factor of 10 was used to account for uncertainty
39 associated with human variation in the derivation of the RfD and RfC. Human variation may be larger or
117
DRAFT - DO NOT CITE OR QUOTE
-------
1 smaller; however, 1,4-dioxane-specific data to examine the potential magnitude of over- or
2 under-estimation are unavailable.
3 Uncertainties in the assessment of the health hazards of 1,4-dioxane are associated with
4 deficiencies in reproductive toxicity information. The oral and inhalation databases lack a multigeneration
5 reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
6 1,4-dioxane (Giavini et al. 1985a). This developmental study indicates that the developing fetus may be a
7 target of toxicity. No developmental studies are available following inhalation to 1,4-dioxane.
5.4 Cancer Assessment
5.4.1 Choice of Study/Data - with Rationale and Justification
5.4.1.1 Oral Study/Data
8 Three chronic drinking water bioassays provided incidence data for liver tumors in rats and mice,
9 and nasal cavity, peritoneal, and mammary gland tumors in rats only (Kano et al.. 2009; JBRC. 1998;
10 Yamazaki et al.. 1994a: NCI. 1978; Kocibaetal. 1974a). The dose-response data from each of these
11 studies are summarized in Table 5-5. With the exception of the NCI (1978) study, the incidence of nasal
12 cavity tumors was generally lower than the incidence of liver tumors in exposed rats. The Kano et al.
13 (2009) drinking water study was chosen as the principal study for derivation of an oral cancer slope factor
14 (CSF) for 1,4-dioxane. This study used three dose groups in addition to controls and characterized the
15 dose-response relationship at lower exposure levels, as compared to the high doses employed in the NCI
16 (1978) bioassay (Table 5-5). The Kociba et al. (1974a) study also used three dose groups and low
17 exposures; however, the study authors only reported the incidence of hepatocellular carcinomas, which
18 may underestimate the combined incidence of rats with adenomas or carcinomas. In addition to increased
19 incidence of liver tumors, chosen as the most sensitive target organ for tumor formation, the Kano et al.
20 (2009) study also noted increased incidence of peritoneal and mammary gland tumors, and nasal cavity
21 tumors were also seen in high-dose male and female rats.
22 In a personal communication, Dr. Yamazaki (2006) provided data that showed that the survival of
23 mice in the Kano et al. (2009) study was low in all male groups (31/50, 33/50, 25/50 and 26/50 in control,
24 low-, mid-, and high-dose groups, respectively) and particularly low in high-dose females (29/50, 29/50,
25 17/50, and 5/50 in control, low-, mid-, and high-dose groups, respectively). These deaths occurred
26 primarily during the second year of the study. Survival at 12 months in male mice was 50/50, 48/50,
27 50/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively. Female mouse survival at
28 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively
29 (Yamazaki. 2006). Furthermore, these deaths were primarily tumor related. Liver tumors were listed as
118
DRAFT - DO NOT CITE OR QUOTE
-------
1 the cause of death for 31
2 2006).
of the 45 pretermination deaths in high-dose female Crj:BDFl mice (Yamazaki.
Table 5-5 Incidence of liver, nasal cavity, peritoneal, and mammary gland tumors in rats and
mice exposed to 1,4-dioxane in drinking water for 2 years (based on survival to
12 months)
Study Species/strain/gender
Kociba et al. Sherman rats, male and
(1974a) female combined3'13
rat=;b
MPI Mn7n\
Male B6C3F-I miced
Female B6C3F-I miced
Male F344/DuCrj
ratsd'e'f'g
Female F344/DuCrj
ratsd'e'f'g
Kano et al. (2009)
Mala Pri-RDP1 mii~ad
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
39/5C?'K
3/50
1/50
6/50
48/50)'K
23/50
31/50
37/50'
40/50)'K
5/50
35/501
41/501
46/50)'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
28/50)'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
DRAFT - DO NOT CITE OR QUOTE
119
-------
5.4.1.2 Inhalation Study/Data
1 Epidemiological studies of populations exposed to 1,4-dioxane via inhalation are not adequate for
2 dose-response analysis and thus derivation of an inhalation unit risk (IUR). However, two chronic
3 inhalation studies in animals are available and were evaluated for the potential to estimate an IUR
4 (Table 5-6). The chronic inhalation study conducted by Torkelson et al. (1974a) in rats did not find any
5 treatment-related tumors; however, only a single exposure concentration was used (111 ppm 1,4-dioxane
6 vapor for 7 hours/day, 5 days/week for 2 years). A chronic bioassay of 1,4-dioxane by the inhalation route
7 reported by Kasai et al. (2009) provides data adequate for dose-response modeling and was subsequently
8 chosen as the study for the derivation of an IUR for 1,4-dioxane. In this bioassay, groups of 50 male F344
9 rats were exposed to either 0, 50, 250 or 1,250 ppm 1,4-dioxane, 6 hours/day, 5 days/week, for 2 years
10 (104-weeks). In male F344 rats, 1,4-dioxane produced a statistically significant increase in incidence
11 and/or a statistically significant dose-response trend for the following tumor types: hepatomas, nasal
12 squamous cell carcinomas, renal cell carcinomas, peritoneal mesotheliomas, mammary gland
13 fibroadenomas, Zymbal gland adenomas, and subcutis fibromas (Kasai et al.. 2009). The incidence of
14 adenomas and carcinomas were combined in this assessment in accordance with EPA's Guidelines on
15 Carcinogen Risk Assessment which notes that etiologically similar tumor types, i.e., benign and malignant
16 tumors of the same cell type, can be combined due to the possibility that benign tumors could progress to
17 the malignant form (U.S. EPA. 2005a: McConnell et al.. 1986). Consistent with the oral cancer
18 assessment (Appendix D), the incidence of hepatic adenomas and carcinomas (combined) was used to
19 calculate an IUR (See Table 5 -6).
120
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-6 Incidence of liver, nasal cavity, kidney, peritoneal, and mammary gland, Zymbal
gland, and subcutis tumors in rats exposed to 1,4-dioxane via inhalation for 2 years.
Study
Torkelson
etal.
(1974a)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
Liver0
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
1/1391
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
11/139K
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.
""Incidence 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 5-8.
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.
mTumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.05).
"Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p & 0.01).
NA = data are not available
5.4.2 Dose-Response Data
5.4.2.1 Oral Data
I Table 5-7 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 statistically significant increase in the incidence of peritoneal mesotheliomas was observed in
7 high-dose male rats only (28/50 rats, Table 5-5). The incidence of peritoneal mesotheliomas was lower
8 than the observed incidence of hepatocellular adenomas or carcinomas in male rats (Table 5-7); therefore,
9 the incidence of hepatocellular adenomas or carcinomas was used to derive an oral CSF for 1,4-dioxane.
121
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-7 Incidence of hepatocellular adenomas or carcinomas in rats and mice exposed to
1,4-dioxane in drinking water for 2 years
Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male Crj:BDF1 mice
Female Crj:BDF1 mice
USSS? incidence of Nver tumors'
0
11
55
274
0
18
83
429
0
49
191
677
0
66
278
964
3/50
4/50
7/50
39/50°'c
3/50
1/50
6/50
48/50D'c
23/50
31/50
37/50°
40/50°'c
5/50
35/50c
41/50C
46/50D'c
Incidence of either hepatocellular adenomas or carcinomas.
bp < 0.05; positive dose-related trend (Peto's test).
""Significantly different from control at p < 0.01 by Fisher's Exact test.
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, 5 days/week, for 2 years (104-weeks).
4 Statistically significant positive dose-response trends were observed for nasal cavity squamous cell
5 carcinomas, hepatomas, renal cell carcinomas, peritoneal mesotheliomas, mammary gland fibroadenomas,
6 and Zymbal gland adenomas. Following 250 ppm 1,4-dioxane exposure, statistically significantly
7 elevated tumor incidences were found in two tissue types (i.e., peritoneal mesothelioma and subcutis
8 fibroma) compared to controls. It is important to note, for observations of subcutis fibroma, the incidence
9 was increased compared to controls at all concentrations, but a decrease in incidence, compared to the
10 mid-concentration, was noted at the highest concentration (1,250 ppm). However, a statistically
11 significantly decreased survival rate was noted in this exposure group by the study authors. Interim
12 sacrifices were not performed. Tumor incidences following 1,250 ppm inhalation exposure to 1,4-dioxane
13 were statistically elevated compared to controls in three tissues (i.e., 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 5-8.
122
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-8 Incidence of tumors in F344 male rats exposed to 1,4-dioxane via inhalation for 104
weeks (6 hours/day, 5 days/week)
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
50
0/50
2/50
0/50
2/50
0/50
4/50
2/50
0/50
0/50
4/50
250
1/50
3/50
1/50
4/50
0/50
14/50C
3/50
0/50
0/50
9/50c
1,250
6/50a'°
21/50a'c
2/50
22/50a'c
4/50a
41/50a'c
5/50°
1/50
4/50a
5/50
Statistically significant trend for increased tumor incidence by Peto's test (p £ 0.01).
bTumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.05).
°Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.01).
Statistically significant trend for increased tumor incidence by Peto's test (p < 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. 2011). This was accomplished using the following equation:
/animal BW [kg]\
HED = animal dose (mg/kg) x I —-
v &/ &J Vhuman BW [kg]/
0.25
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 5-9. HEDs were also calculated for supporting studies (NCI. 1978; Kociba et al.
6 1974a) and are also shown in Table 5-9.
123
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-9 Calculated HEDs for the tumor incidence data used for dose-response modeling
Study Species/strain/gender Anim^W <9>
Male F344/DuCrj rats
Female F344/DuCrj rats
K-inn nt -ii fonnrn
Male Crj:BDF1 mice
Female Crj:BDF1 mice
Kociba et al. (1 974a) Male anj?hff^f'® (<;°mbined>
...14.
MPI M 07^
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. (1974a): and NCI (19781
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 liver tumors is unknown. Therefore, a
6 linear low-dose extrapolation approach was used to estimate human carcinogenic risk associated with
7 1,4-dioxane oral exposure.
8 However, several of the external peer review panel members for the oral assessment (see
9 Appendix A: Summary of External Peer Review and Public Comments and Disposition) recommended
10 that the mode of action data support the use of a nonlinear extrapolation approach to estimate human
11 carcinogenic risk associated with exposure to 1,4-dioxane and that such an approach should be presented
12 in the Toxicological Review. As discussed in Section 4.5.1, numerous short-term in vitro and a few in
13 vivo tests were nonpositive for 1,4-dioxane-induced genotoxicity. Results from two-stage mouse skin
14 tumor bioassays demonstrated that 1,4-dioxane does not initiate mouse skin tumors, but it is a promoter of
15 skin tumors initiated by DMBA (King et al.. 1973a). These data suggest that a potential mode of action
16 for 1,4-dioxane-induced tumors may involve proliferation of cells initiated spontaneously, or by some
124
DRAFT - DO NOT CITE OR QUOTE
-------
1 other agent, to become tumors (Miyagawa et al.. 1999; Uno et al. 1994; Golds worthy et al. 1991;
2 Lundberg et al.. 1987; Bulletal. 1986; Stott et al.. 1981; Kingetal. 1973a). However, key events related
3 to the promotion of tumor formation by 1,4-dioxane are unknown. Therefore, under the U.S. EPA
4 Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). EPA concluded that the available
5 information does not establish a plausible mode of action for 1,4-dioxane and data are insufficient to
6 establish significant biological support for a nonlinear approach. EPA determined that there are no data
7 available to inform the low-dose region of the dose response, and thus, a nonlinear approach was not
8 included.
9 Accordingly, the CSF for 1,4-dioxane was derived via a linear extrapolation from the POD
10 calculated by fitting a curve in BMDS to the experimental dose-response data. The POD is the 95% lower
11 confidence limit on the dose associated with a benchmark response (BMR) near the lower end of the
12 observed data. The BMD modeling analysis used to estimate the POD is described in detail in Appendix
13 D and is summarized below in Section 5.4.4.
14 Model estimates were derived for all available bioassays and tumor endpoints (Appendix D);
15 however, the POD used to derive the CSF is based on the most sensitive species and target organ in the
16 principal study (Kano et al.. 2009).
17 The oral CSF was calculated using the following equation:
18 CSF = BMR /
5.4.3.2 Inhalation
19 In accordance with the U.S. EPA (1994a) RfC methodology, the HEC values were calculated by
20 the application of DAFs. As discussed in Section 5.2.3. since 1,4-dioxane is miscible with water, has a
21 high partition coefficient, and induces effects throughout the body of the rat, this substance was
22 considered to be a systemic acting gas and a DAF of 1.0 was applied. The lifetime continuous inhalation
23 risk for humans is defined as the slope of the line drawn from the POD through the origin, with the POD
24 defined as the lower 95% bound on the exposure associated with a level of extra risk near the low end of
25 the data range.
26 All PODs were converted to equivalent continuous exposure levels by multiplying by [(6
27 hours)/(24 hours)] x[(5 days)/(7 days)], under the assumption of equal cumulative exposures leading to
28 equivalent outcomes.
29 Given the multiplicity of tumor sites observed in animals, basing the IUR on one tumor site may
30 underestimate the carcinogenic potential of 1,4-dioxane via inhalation. Also, simply pooling the counts of
31 animals with one or more tumors (i.e., counts of tumor bearing animals) would tend to underestimate the
32 overall risk for tumors observed at independent sites and ignores potential differences in the
33 dose-response relationships across the sites (NRC. 1994; Bogen. 1990). NRC (1994) has also noted that
125
DRAFT - DO NOT CITE OR QUOTE
-------
1 the assumption of independence across tumor types is not likely to produce substantial error in the risk
2 estimates unless tumors across multiple sites are known to be biologically dependent.
3 The U.S. EPA's BMDS (v2.2 beta) MS_Combo program was utilized as a computational
4 approach to calculating the dose associated with a specified composite risk under the assumption of
5 independence of tumors. The best fitting BMDS multistage model was determined for each individual
6 tumor type as shown in Section 5.4.4.2 and APPENDIX G. These models account for spontaneous tumor
7 generation in controls. The Guidelines for Carcinogen Risk Assessment recommend calculation of an
8 upper bound to account for uncertainty in the estimate (U.S. EPA. 2005a). Complete details of this
9 analysis are included in Appendix G. In addition, Bayesian MCMC computations were conducted as
10 described by Kopylev et al. (2009) using WinBugs (Spiegelhalter et al. 2003). For uncertainty
11 characterization, MCMC methods have the advantage of providing information about the full distribution
12 of risk and/or BMDs, which can be used in generating a confidence bound. This MCMC approach, which
13 builds on the re-sampling approach recommended by Bogen (1990). also provides a distribution of the
14 combined potency across sites. This supporting analysis was completed in addition to the MS_Combo
15 analysis and additional details are included in Appendix G.
16 Several hypothesized MOA(s) have been proposed for liver and nasal tumors, although these
17 MOA(s) are not supported by the available data (see Sections 4.7.3.3 and 4.7.3.4). Specifically, tumors
18 occur in rodent models in the absence of data to identify hypothesized key events (e.g., cytotoxicity).
19 Also, studies evaluating the kinetics of 1,4-dioxane suggest that liver carcinogenicity is related to the
20 accumulation of the parent compound following metabolic saturation; however, data are not available to
21 determine the toxic moiety (i.e., parent compound and/or metabolite(s)) (see Section 3.3 and 4.7.3.1.1).
22 For kidney, lung, peritoneal (mesotheliomas), mammary gland, Zymbal gland, and subcutis tumors, there
23 are no available data regarding any hypothesized carcinogenic MOA(s) for 1,4-dioxane.
24
25 The EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). recommend that the
26 method used to characterize and quantify cancer risk from a chemical is determined by what is known
27 about the MOA of the carcinogen and the shape of the cancer dose-response curve. The linear
28 extrapolation approach is used as a default option if the mode of carcinogenic action is not identified. A
29 nonlinear extrapolation approach can be used for cases with sufficient data to ascertain the mode of action
30 and to conclude that it is not linear at low doses. Also, nonlinear extrapolation having significant
3 1 biological support may be presented in addition to a linear approach when the available data and weight
32 of evidence support a nonlinear approach. In the case of 1,4-dioxane, there is insufficient biological
33 support to identify key events and to have reasonable confidence in the sequence of events and how they
34 relate to the development of tumors following exposure to 1,4-dioxane; thus, the data are not strong
35 enough to ascertain the mode of action applying the Agency's mode of action framework (U.S. EPA.
36 2005a). Therefore, EPA concluded that a default linear extrapolation should be utilized to estimate the
37 cancer risk estimates for inhalation and oral exposure to 1,4-dioxane.
38 IUR estimates were calculated using the following equation:
3 9 IUR = BMR /
126
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 drinking water (Kano et al.. 2009; NCI.
4 1978; Kociba et al.. 1974a) (Table 5-5). Animal doses were used for BMD modeling, and then HED
5 BMD and BMDL values were calculated using BW3/4 scaling employing animal TWA body weights
6 (Table 5-10) and a human BW of 70 kg. For all models, a BMR of 10% extra risk was employed. BMDs
7 and BMDLs from all models are reported, and the model outputs and plots corresponding to the
8 best-fitting models are shown (Appendix D). When the best-fitting model is not a multistage model, the
9 multistage model output and plot are also provided (Appendix D). A summary of the BMD modeling
10 results for the Kano et al. (2009). NCI (1978). and Kociba et al. (1974a) studies is shown in Table 5-10.
127
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-10 BMD HED and BMDLHED values from best-fit models fit to tumor incidence data for rats
and mice exposed to 1,4-dioxane in drinking water for 2 years and corresponding oral
CSFs
Study
Kano et al.
(2009)
Kociba et al.
NCI (1978)
Gender/strain/species
Male F344/DuCrj rats"
Female F344/DuCrj ratsc
Male Crj:BDF1 mice0
Female Crj:BDF1 mice0
Female Crj:BDF1 mice0'8
Female Crj:BDF1 mice0''
Female F344/DuCrj rats0
Male F344/DuCrj rats0
Male F344/DuCrj ratsb
Female F344/DuCrj ratsd
Male and female (combined)
Sherman rats0
Male and female (combined)
Sherman ratsb
Male Osborne Mendel rats0
Female Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female B6C3F-I micec
Male B6C3F-I miceh
_ . BM.DHED BMDLHED Oral CSF
i umor type (mg/kg_day) (mg/kg_day) (mg/kg-day)-1
I lep
2 |0.11) to the data for the incidence of hepatocellular adenoma or carcinoma in female mice (Appendix D).
3 The high dose was dropped for the female mouse liver tumor dataset in an attempt to achieve an adequate
4 fit; however, an adequate fit was still not achieved. Because the female mice were clearly the most
5 sensitive group tested, other BMD models were applied to the female mouse liver tumor dataset to
6 achieve an adequate fit. The log-logistic model was the only model that provided adequate fit for this data
7 set due to the steep rise in the dose-response curve (70% incidence at the low dose) followed by a plateau
8 at near maximal tumor incidence in the mid- and high-dose regions (82 and 92% incidence, respectively).
9 The predicted BMDio and BMDLio for the female mouse data are presented in Table 5-10, as well as
10 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 BMDio and BMDLio for female rat mammary
128
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 5-10.
3 A comparison of the BMD and BMDL estimates derived for rats and mice from the Kano et al.
4 (2009). NCI (1978). and Kociba et al. (1974a) studies (Table 5-10) indicates that female mice are more
5 sensitive to liver carcinogenicity induced by 1,4-dioxane compared to other species or tumor types.
6 Therefore, the BMDL50 HED for the female mouse data was chosen as the POD and the CSF of 0.10
7 (mg/kg-day)"1 was calculated as follows:
0.50
CSF = 4.95 mg/kg - day (BMDL50 HED for female mice) = °-10 (mg/kg ~ ^
8 Calculation of a CSF for 1,4-dioxane is based upon the dose-response data for the most sensitive
9 species and gender.
5.4.4.2 Inhalation Unit Risk
10 As stated in Section 5.4.2.2, multiple tumor types have been observed in rats following inhalation
11 exposure to 1,4-dioxane. These data have been used to develop IUR estimates for 1,4-dioxane. The
12 multistage cancer models available in the BMDS (version 2.1.1) were fit to the incidence data for each
13 tumor type observed in rats exposed to 1,4-dioxane via inhalation (Kasai et al.. 2009) to determine the
14 degree (e.g., 1st, 2nd, or 3rd) of the multistage model that best fit the data (details in Appendix G). In
15 contrast to the oral slope factor analysis, suitable multistage model fits were obtained for all of the
16 datasets included in the inhalation unit risk analysis. Then, the best fitting models for each endpoint were
17 used in the BMDS (version 2.2Beta) MS_Combo program to estimate a total tumor BMC and BMCLi0.
18 A Bayesian MCMC analysis was also performed using WinBUGS to calculate the total tumor risk and it
19 yielded similar results (See Appendix G). A summary of the BMDS model predictions for the Kasai et al.
20 (2009) study is shown in Table 5-11. Experimental exposure concentrations were used for BMD
21 modeling and then continuous human equivalent exposures were calculated by adjusting for duration of
22 exposure (Table 5-11) 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 BMCLio (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 G).
28 The IUR estimates are provided in Table 5-11. Human equivalent risks estimated from the
29 individual rat tumor sites ranged from 2 x 10"7 to 2 x 10"6 (jig/m3)"1 (rounded to one significant figure).
30 The highest IUR (2 x 10"6 ((ig/m3)"1) corresponded to peritoneal mesotheliomas in male rats, and the
31 lowest IUR (2 x 10"7 ((ig/m3)"1) corresponded to renal cell carcinoma and Zymbal gland adenomas in male
32 rats. The MS_Combo analysis yielded an IUR estimate of 5 x 1Q"6 (jig/m3)"1.
129
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-11 Dose-response modeling summary results for male rat tumors associated with
inhalation exposure to 1,4-dioxane for 2 years
Tumor Type3
Nasal cavity squamous cell
carcinoma
Hepatocellular adenoma or
carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland
fibroadenoma
Zymbal gland adenoma
Subcutis fibroma
BMDS MS Combo Total Tumor
Multistage
Model
Degreeb
1
1
3
1
1
3
1
Analysis'
Point of Departure0
Bioassay Exposure
Concentration (ppm)
BMCio
1107
252.8
1355
82.21
1635
1355
141.8
40.4
BMCLio
629.9
182.3
1016
64.38
703.0
1016
81.91
30.3
HEC
(mg/m3)d
BMCio
712.3
162.7
872
52.89
1052
872
91.21
26.0
BMCLio
405.3
117.3
653.7
41.42
452.4
653.7
52.70
19.5
Estirr
([jg/n
2.5 x
8.5 x
1.5x
2.4 x
2.2 x
1.5x
1.9x
5.0 x
R
iatee
n3)-1
ID'7
ID'7
10"
10'b
ID'7
10"
10'b
10'b
aTumor incidence data from Kasai et al. (2009).
Best-fitting multistage model degree (p>0.1, lowest AIC). See Appendix G for modeling details.
°BMC = Concentration at specified extra risk (benchmark dose); BMCL = 95% lower bound on concentration at specified extra risk.
iman continuous equi\
1,4-dioxane]/ 24.45.
b
c
dHuman continuous equivalent estimated by multiplying exposures by [(6 hours)/(24 hours) x (5 days)/(7 days) x molecular weight of
eThe inhalation unit risk (ug/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 (0.1 0) by the BMCLio. Thus, representing an upper bound, continuous lifetime
exposure estimate of cancer potency.
'Results in this table are from the BMDS MS_Combo model. Additionally, Bayesian analysis using WinBUGS was performed and
yielded similar results (See Appendix G).
1 Given the multiplicity of tumor sites, basing the inhalation unit risk on one tumor site may
2 underestimate the carcinogenic potential of 1,4-dioxane. Consistent with recommendations of the NRC
3 (1994) and the EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005aX the total risk and
4 upper bound risk for all tumor sites in male F344 rats was estimated. This estimate of total risk describes
5 the risk of developing any combination of the tumor types considered. As shown in Table 5-11, the
6 resulting inhalation unit risk for all tumor types in male F344 rats was 5 x 10~6 ((ig/m3)"1. Consideration of
7 all tumor sites approximately doubled the unit risk compared to the highest unit risk associated with any
8 individual tumor type, 2 x 10~6 (jig/m3)"1 for male peritoneal mesotheliomas.
9 The HEC BMCLio for the combined tumor estimate in male rats was chosen as the POD and the
10 IUR of 5 x 10'6 (ng/m3)"1 was calculated as follows:
IUR (mg/m3)-1 = °3 = 0-005 (mg/m3)-1
"1
IUR (ng/m3)-1 = 0.005 (mg/m3)-1 x ^|- = 5 x 10"6 (ng/m3)
IUR (ng/m3)'1 = 5 x ID' '
12 Based on the analysis discussed above, the recommended upper bound estimate on human extra
13 cancer risk from continuous lifetime inhalation exposure to 1,4-dioxane is 5 x 1Q"6 ((ig/m3)"1. The IUR
14 reflects the exposure-response relationships for the multiple tumor sites in male F344 rats.
130
DRAFT - DO NOT CITE OR QUOTE
-------
5.4.5 Previous Cancer Assessment
1 A previous cancer assessment was posted for 1,4-dioxane on IRIS in 1988. 1,4-Dioxane was
2 classified as a Group B2 Carcinogen (probable human carcinogen; sufficient evidence from animal
3 studies and inadequate evidence or no data from human epidemiology studies (U.S. EPA. 1986bV) based
4 on the induction of nasal cavity and liver carcinomas in multiple strains of rats, liver carcinomas in mice,
5 and gall bladder carcinomas in guinea pigs. An oral CSF of 0.011 (mg/kg-day)"1 was derived from the
6 tumor incidence data for nasal squamous cell carcinoma in male rats exposed to 1,4-dioxane in drinking
7 water for 2 years (NCI. 1978). The linearized multistage extra risk procedure was used for linear low dose
8 extrapolation. An inhalation unit risk was not previously derived.
5.5 Uncertainties in Cancer Risk Values
9 In this assessment, extrapolation of high-dose data from laboratory animals to estimate potential
10 risks to human populations from low-dose exposure to 1,4-dioxane has engendered some uncertainty in
11 the results. Several types of uncertainty may be considered quantitatively, but other important
12 uncertainties can only be considered qualitatively. Thus, an overall integrated quantitative uncertainty
13 analysis is not presented. However, the sources of uncertainty and their potential impacts on the
14 assessment are described below and in Table 5-12.
5.5.1 Sources of Uncertainty
5.5.1.1 Choice of Low-Dose Extrapolation Approach
15 The range of possibilities for the low-dose extrapolation of tumor risk from exposure to
16 1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible MOA(s)
17 for the observed tumors. The MOA is a key consideration in clarifying how risks should be estimated for
18 low-dose exposure. Exposure to 1,4-dioxane has been observed in animal models to induce multiple
19 tumor types, including liver adenomas and carcinomas, nasal carcinomas, mammary adenomas and
20 fibroadenomas, and mesotheliomas of the peritoneal cavity (Kano et al. 2009; Kasai et al. 2009; JBRC.
21 1998; NCI. 1978; Kocibaetal.. 1974a). MOA information that is available for the carcinogenicity of
22 1,4-dioxane has largely focused on liver adenomas and carcinomas, with little or no MOA information
23 available for the remaining tumor types. In Section 4.7.3, hypothesized MO As were explored for
24 1,4-dioxane. Information that would provide sufficient support for any MOA is not available. In the
25 absence of a MOA(s) for the observed tumor types, a linear low-dose extrapolation approach was used to
26 estimate human carcinogenic risk associated with 1,4-dioxane exposure.
27 It is not possible to predict how additional MOA information would impact the dose-response
28 assessment for 1,4-dioxane because of the variety of tumors observed and the lack of data on how
131
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-dioxane or its metabolite interacts with cells initiating the progression to the observed tumors. In
2 general, the Agency has preferred to use the multistage model for analyses of tumor incidence and related
3 endpoints because this model has a generic biological motivation based on long-established biologically-
4 based mathematical models such as the Moolgavkar-Venzon-Knudsen (MVK) model. The MVK model
5 does not necessarily characterize all modes of tumor formation, but it is a starting point for most
6 investigations and, much more often than not, has provided at least an adequate description of tumor
7 incidence data.
8 The multistage cancer model provided adequate fits for the tumor incidence data following a
9 2-year inhalation exposure to 1,4-dioxane by male rats (Kasai et al. 2009). In the studies evaluated for the
10 oral cancer assessment (Kano et al.. 2009; NCI. 1978; Kocibaetal. 1974a). the multistage model
11 provided good descriptions of the incidence of a few tumor types in male (nasal cavity) and female
12 (hepatocellular and nasal cavity) rats and in male mice (hepatocellular) exposed to 1,4-dioxane via
13 ingestion (Appendix D for details). The multistage model did not provide an adequate fit for the female
14 mouse liver tumor dataset based upon the following (U.S. EPA. 2012a):
15 • Goodness-of-fit/7-value was less than 0.10 indicating statistically significant lack
16 of fit;
17 • Akaike's Information Criterion (AIC) was larger than other acceptable models;
18 • Observed data deviated substantially from the fitted model, as measured by their
19 standardized %2 residuals (i.e., residuals with values greater than an absolute
20 value of one).
21 By default, the BMDS software imposes constraints on the values of certain parameters of the
22 models. When these constraints were imposed, the multistage model and most other models did not fit the
23 incidence data for female mouse liver adenomas or carcinomas, even after dropping the highest dose
24 group.
25 The log-logistic model was selected because it was the only model that provided an adequate fit
26 to the female mouse liver tumor data (Kano et al.. 2009). A BMR of 50% was used because it is
27 proximate to the response at the lowest dose tested, and the BMDL50 HED was estimated by applying
28 appropriate parameter constraints to the selected model, consistent with the BMD Technical Guidance
29 Document (U.S. EPA. 2012a).
30 The human equivalent oral CSFs estimated from tumor datasets with statistically significant
31 increases ranged from 4.2 x 10"4 to 1.0 x 10"1 per mg/kg-day (Table 5-10), a range of about three orders
32 of magnitude, with the upper and lower extremes coming from the combined male and female rat data for
33 hepatocellular carcinomas (Kocibaetal.. 1974a) and the female mouse combined liver adenoma and
34 carcinomas (Kano et al.. 2009).
132
DRAFT - DO NOT CITE OR QUOTE
-------
5.5.1.2 Dose Metric
1 1,4-Dioxane is known to be metabolized in vivo. However, it is unknown whether a metabolite or
2 the parent compound, or some combination of parent compound and metabolites, is responsible for the
3 observed carcinogenicity. If the actual carcinogenic moiety is proportional to administered exposure, then
4 use of administered exposure as the dose metric is the least biased choice. On the other hand, if this is not
5 the correct dose metric, then the impact on the CSF and IUR is unknown.
5.5.1.3 Cross-Species Scaling
6 For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied(U.S.
7 EPA. 2011) to address toxicological equivalence of internal doses between each rodent species and
8 humans, consistent with the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). It is assumed
9 that equal risks result from equivalent constant lifetime exposures.
10 Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
11 in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
12 However, since similar cell types are prevalent throughout the respiratory tract of both rats and humans,
13 the tumors are considered biologically plausible and relevant to humans.
5.5.1.4 Statistical Uncertainty at the POD
14 Parameter uncertainty can be assessed through confidence intervals. Each description of
15 parameter uncertainty assumes that the underlying model and associated assumptions are valid. For the
16 log-logistic model applied to the female mouse data following oral exposure, there is a reasonably small
17 degree of uncertainty at the 50% excess incidence level (the POD for linear low-dose extrapolation), as
18 indicated by the proximity of the BMDLnED (4.95 mg/kg-day) to the BMDnED (7.51 mg/kg-day). For the
19 multistage model applied for the male rat inhalation dataset, there is a reasonably small degree of
20 uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
5.5.1.5 Bioassay Selection
21 The study by Kano et al. (2009) was used for development of an oral CSF. This was a
22 well-designed study, conducted in both sexes in two species (rats and mice) with a sufficient number
23 (N=50) of animals per dose group. The number of test animals allocated among three dose levels and an
24 untreated control group was adequate, with examination of appropriate toxicological endpoints in both
25 sexes of rats and mice. Alternative bioassays (NCI. 1978; Kocibaetal.. 1974a) were available and were
26 fully considered for the derivation of the oral CSF.
133
DRAFT - DO NOT CITE OR QUOTE
-------
1 The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
2 well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
3 Three dose levels plus an untreated control group were examined following exposure to 1,4-dioxane via
4 inhalation for 2 years.
5.5.1.6 Choice of Species/Gender
5 The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the female
6 mouse, which was shown to be more sensitive than male mice or either sex of rats to the carcinogenicity
7 of 1,4-dioxane. While all data, both species and sexes reported from the Kano et al. (2009) study, were
8 suitable for deriving an oral CSF, the female mouse data represented the most sensitive indicator of
9 carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day or 10 mg/kg-day [HED])
10 resulted in a considerable and significant increase in combined liver adenomas and carcinomas observed.
11 Additional testing of doses within the range of control and the lowest dose (66 mg/kg-day or
12 10 mg/kg-day [HED]) could refine and reduce uncertainty for the oral CSF.
13 A personal communication from Dr. Yamazaki (2006) provided that the survival of mice in the
14 Kano et al. (2009) study was particularly low in high-dose females (29/50, 29/50, 17/50, and 5/50 in
15 control, low-, mid-, and high-dose groups, respectively). These deaths occurred primarily during the
16 second year of the study. Female mouse survival at 12 months was 50/50, 50/50, 48/50, and 48/50 in
17 control, low-, mid-, and high-dose groups, respectively (Yamazaki. 2006). Furthermore, these deaths were
18 primarily tumor related. Liver tumors were listed as the cause of death for 1/21, 2/21, 8/33, and 31/45 of
19 the pretermination deaths in control, low-, mid- and, high-dose female Crj:BDFl mice (Yamazaki. 2006).
20 Therefore, because a number of the deaths in female mice were attributed to liver tumors, this endpoint
21 and species was still considered to be relevant for this analysis; however, the high mortality rate does
22 contribute uncertainty. Additionally, the OSF may actually be larger if the survival adjusted tumor data
23 were available.
24 Additionally, the incidence of hepatocellular adenomas and carcinomas in historical controls was
25 evaluated with the data from Kano et al. (2009). Katagiri et al. (1998) summarized the incidence of
26 hepatocellular adenomas and carcinomas in control male and female BDF1 mice from ten 2-year
27 bioassays at the JBRC. For female mice, out of 499 control mice, the incidence rates were 4.4% for
28 hepatocellular adenomas and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported a 10%
29 incidence rate for hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in
30 control female BDF1. These incidence rates are near the historical control values, and thus are appropriate
31 for consideration in this assessment.
32 Male F344 rat data were used to estimate risk following inhalation of 1,4-dioxane. Kano et al.
33 (2009) showed that male rats were more sensitive than female rats to the effects of 1,4-dioxane following
34 oral administration; therefore, male rats were chosen to be studies in the 2-year bioassay conducted by the
35 same laboratory (Kasai et al.. 2009). The sensitivity and tumorigenic response of female rats or male or
36 female mice following inhalation of 1,4-dioxane is unknown. Since female mice were the most sensitive
134
DRAFT - DO NOT CITE OR QUOTE
-------
1 gender and species examined in the Kano et al. (2009) oral study, female mice may also be more sensitive
2 to the inhalation of 1,4-dioxane, which would result in a greater risk.
5.5.1.7 Relevance to Humans
3 The derivation of the oral CSF is derived using the tumor incidence in the liver of female mice. A
4 thorough review of the available toxicological data available for 1,4-dioxane provides no scientific
5 justification to propose that the liver adenomas and carcinomas observed in animal models due to
6 exposure to 1,4-dioxane are not relevant to humans. As such, liver adenomas and carcinomas were
7 considered relevant to humans due to exposure to 1,4-dioxane.
8 The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
9 rats. There is no information on 1,4-dioxane to indicate that the observed rodent tumors are not relevant to
10 humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
11 rodents and humans. In the absence of information to indicate otherwise and considering similar cell types
12 are prevalent throughout the respiratory tract of rats and humans, the nasal, liver, renal, peritoneal,
13 mammary gland, Zymbal gland and subcutis tumors were considered relevant to humans.
5.5.1.8 Human Population Variability
14 The extent of inter-individual variability in 1,4-dioxane metabolism has not been characterized. A
15 separate issue is that the human variability in response to 1,4-dioxane is also unknown. Data exploring
16 whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life stages are unavailable.
17 This lack of understanding about potential differences in metabolism and susceptibility across exposed
18 human populations thus represents a source of uncertainty. Also, the lack of information linking a MOA
19 for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.
135
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-12 Summary of uncertainty in the 1,4-dioxane cancer risk estimation
Consideration/
approach
Potential Impact
Decision
Justification
Low-dose
extrapolation
procedure
Departure from
EPA's Guidelines for
Carcinogen Risk
Assessment POD
paradigm, if justified,
could 1 or t unit risk
an unknown extent
Log-logistic model
to determine POD,
for CSF; Combined
tumor modeling for
IUR; linear
low-dose
extrapolation from
POD
A linear low-dose extrapolation approach was used to
estimate human carcinogenic risk associated with
1,4-dioxane exposure. Where data are insufficient to
ascertain the MOA, EPA's 2005 Guidelines for
Carcinogen Risk Assessment recommend application
of a linear low-dose extrapolation approach.
Dose metric
Cross-species
scaling
Bioassay
Alternatives could f
or I CSF by an
unknown extent
Alternatives could J,
or t CSF [e.g.,
3.5-fold | (scaling by
BW) or t twofold
(scaling by BW067 )]
Alternatives could f
or | cancer potency
by an unknown
extent
Used administered
exposure
BW075 (default
approach)
CSF (Kano et al.,
2009); IUR (Kasai
etal.,2009)
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 '3
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.
Species /gender
combination
Human risk could J, or
t, depending on
relative sensitivity
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
Female mouse gender/species tested; true correspondence is
(CSF); male rat unknown. Calculation of the CSF for 1,4-dioxane was
(IUR) based on dose-response data from the most sensitive
species and gender. The carcinogenic response
occurs across species. No female mouse data were
available for derivation of the IUR.
Human
relevance of
mouse tumor
data
If rodent tumors
proved not to be
relevant to humans,
unit risk would not
apply i.e., could J,
CSF
Mouse liver
adenomas and
carcinomas are
relevant to humans
(basis for CSF). Rat
tumors at multiple
sites are relevant to
humans (basis for
IUR)
1,4-dioxane is a multi-site carcinogen in rodents and
the MOA(s) is unknown; carcinogenicity observed in
the rodent studies is considered relevant to human
exposure.
Human
population
variability in
metabolism and
response/
sensitive
subpopulations
Risk t or | to an
unknown extent
Considered
qualitatively
No data to support range of human
variability/sensitivity, including whether children are
more sensitive.
136
DRAFT - DO NOT CITE OR QUOTE
-------
6 MAJOR CONCLUSIONS IN THE
CHARACTERIZATION OF HAZARD AND DOSE
RESPONSE
6.1 Human Hazard Potential
1 1,4-Dioxane is absorbed rapidly following oral and inhalation exposure, with much less
2 absorption occurring from the dermal route. 1,4-Dioxane is primarily metabolized to HEAA, which is
3 excreted in the urine. Liver, kidney, and nasaljoxicity 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.. 1934aV
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.. 1974a) 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: NCI. 1978: Argus etal. 1973aV Liver
17 and kidney toxicity appear to be related to saturation of clearance pathways and an increase in the
18 1,4-dioxane concentration in the blood (Kociba et al.. 1974b). Kidney damage was characterized by
19 degeneration of the cortical tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCI. 1978:
20 Kociba etal.. 1974b: Argus etal.. 1973b: Argus etal.. 1965b: Fairlev et al.. 1934b). In chronic inhalation
21 studies conducted in rats, nasal and liver toxicity were the primary noncancer health effects. Degeneration
22 of nasal tissue (i.e. metaplasia, hyperplasia, atrophy, hydropic change, and vacuolic change) and
23 preneoplastic cell proliferation were observed in the nasal cavity following inhalation exposure to 1,4-
24 dioxane for 2 years (Kasai et al.. 2009). Liver toxicity was described as necrosis of the centrilobular
25 region and preneoplastic changes were noted as well.
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: NCI. 1978: Kociba etal.. 1974b: Torkelson et al..
28 1974b: Argus etal.. 1973b: Hoch-Ligeti and Argus. 1970b: Hoch-Ligeti et al.. 1970b: Argus et al..
29 1965b). 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, Zymbal gland, and
32 subcutaneous tumors were also observed in rats, but were not seen in mice. With the exception of the NCI
137
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. 2005b). 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 (Kano et al.. 2009;
6 Kasai et al. 2009: JBRC. 1998: NCI. 1978: Kocibaetal. 1974b: Argus etal. 1973b: Hoch-Ligeti and
7 Argus. 1970b: Hoch-Ligeti et al.. 1970b: Argus et al.. 1965b). 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 et al.. 1978b: Thiess et al.. 1976b).
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 (Roy et al.. 2005b: Morita and Hayashi.
14 1998: Mirkova. 1994b: Kitchin and Brown. 1990b: Galloway et al.. 1987b). A MOA hypothesis for liver
15 tumorsjnvolving sustained proliferation of spontaneously transformed liver cells has some support by
16 evidence that suggests 1,4-dioxane is a tumor promoter in mouse skin and rat liver bioassays (Lundberg et
17 al.. 1987: King et al.. 1973b). Some dose-response and temporal evidence support the occurrence of cell
18 proliferation prior to the development of liver tumors (JBRC. 1998: Kocibaetal.. 1974b). However, the
19 dose-response relationship for the induction of hepatic cell proliferation has not been characterized, and it
20 is unknown if it would reflect the dose-response relationship for liver tumors in the 2-year rat and mouse
21 studies. Data from rat and mouse bioassays (JBRC. 1998: Kocibaetal.. 1974b) suggest that cytotoxicity
22 is not a required precursor event for 1,4-dioxane-induced cell proliferation. Liver tumors were observed
23 in female rats and female mice in the absence of lesions indicative of cytotoxicity (Kano et al.. 2009:
24 JBRC. 1998: NCI. 1978). Data regarding a plausible dose response and temporal progression from
25 cytotoxicity to cell proliferation and eventual liver tumor formation are not available. Hypothesized
26 MOAs by which 1,4-dioxane induces tumors in other organ systems such as the respiratory system lack
27 supporting data (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 (Kocibaetal.. 1974b). 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
138
DRAFT - DO NOT CITE OR QUOTE
-------
1 variability in human susceptibility; thus, the interspecies and intraspecies uncertainty factors of 10 were
2 applied. In addition, a threefold database uncertainty factor was applied due to the lack of information
3 addressing the potential reproductive toxicity associated with 1,4-dioxane.
4 The overall confidence in the RfD is medium. Confidence in the principal study (Kociba et al..
5 1974b) is medium. Confidence in the database is medium due to the lack of a multigeneration
6 reproductive toxicity study. Reflecting medium confidence in the principal study and medium confidence
7 in the database, confidence in the RfD is medium.
6.2.2 Noncancer/lnhalation
8 The RfC of 3 x 10~2 mg/m3 was derived based on co-critical effects of olfactory epithelium
9 atrophy and respiratory metaplasia in rats exposed for 2 years to 1,4-dioxane via inhalation (Kasai et al..
10 2009). This study was chosen as the principal study because it provides an adequate study design and the
11 most sensitive measure of adverse effects by 1,4-dioxane. The POD was derived using the LOAEL for
12 olfactory epithelium atrophy and respiratory metaplasia in male rats (Kasai et al.. 2009). A composite UF
13 of 1,000 was applied, consisting of factors of 10 for a LOAEL-to NOAEL extrapolation, 10 for
14 interindividual variability, 3 for animal-to-human extrapolation, and 3 for database deficiencies.
15 The overall confidence in the RfC is medium. Confidence in the principal study (Kasai et al..
16 2009) is medium. Confidence in the database is medium due to the lack of supporting studies and a
17 multigeneration reproductive toxicity study. Reflecting medium confidence in the principal study and
18 medium confidence in the database, the confidence in the RfC is medium.
6.2.3 Cancer
19 Under EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005b). 1,4-dioxane is
20 "likely to be carcinogenic to humans" by all routes of exposure. This descriptor is based on evidence of
21 carcinogenicity from animal studies.
6.2.3.1 Oral
22 An oral CSF for 1,4-dioxane of 0.10 (mg/kg-day)"1 was based on liver tumors in female mice
23 from a chronic study (Kano et al.. 2009). The available data indicate that the MOA(s) by which
24 1,4-dioxane induces peritoneal, mammary, or nasal tumors in rats and liver tumors in rats and mice is
25 unknown (see Section 4.7.3 for a more detailed discussion of 1,4-dioxane's hypothesized MOAs).
26 Therefore, based on the U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005b). a
27 linear low dose extrapolation was used. The POD was calculated by curve fitting the animal experimental
28 dose-response data from the range of observation and converting it to a HED (BMDL50 HED of
29 4.95 mg/kg-day).
30 The uncertainties associated with the quantitation of the oral CSF are discussed below.
139
DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.2 Inhalation
1 The IUR for 1,4-dioxane of 5 x 10~6 ((ig/m3)"1 was based on a chronic inhalation study conducted
2 by Kasai et al. (2009). Statistically significant increases in tumor incidence and positive dose-response
3 trends were observed at multiple sites in the male rat including the nasal cavity (squamous cell
4 carcinoma), liver (adenoma), peritoneal (mesothelioma), and the subcutis (fibroma). Statistically
5 significant dose-response trends were also observed in the kidney (carcinoma), mammary gland
6 (fibroadenoma), and the Zymbal gland (adenoma). The available data indicate that the MOA(s) by which
7 1,4-dioxane induces tumors in rats is unknown (see Section 4.7.3 for a more detailed discussion of
8 1,4-dioxane's hypothesized MOAs). Therefore, based on the EPA's Guidelines for Carcinogen Risk
9 Assessment (U.S. EPA. 2005b). a linear low dose extrapolation was used. A combined tumor BMD
10 approach (see Section 5.4.3.2 and Appendix G for details) was used to calculate the POD for the total
11 tumor risk following inhalation of 1,4-dioxane. The POD was calculated by curve fitting the animal
12 experimental dose-response data from the range of observation and converting it to a continuous human
13 equivalent exposure.
14 The uncertainties associated with the quantitation of the IUR are discussed below.
6.2.3.3 Choice of Low-Dose Extrapolation Approach
15 The range of possibilities for the low-dose extrapolation of tumor risk from exposure to
16 1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible MOA(s)
17 for the observed tumors. The MOA is a key consideration in clarifying how risks should be estimated for
18 low-dose exposure. Exposure to 1,4-dioxane has been observed in animal models to induce multiple
19 tumor types, including liver adenomas and carcinomas, nasal carcinomas, mammary adenomas and
20 fibroadenomas, and mesotheliomas of the peritoneal cavity (Kano et al.. 2009). MOA information that is
21 available for the carcinogenicity of 1,4-dioxane has largely focused on liver adenomas and carcinomas,
22 with little or no MOA information available for the remaining tumor types. In Section 4.7.3,
23 hypothesized MOAs were explored for 1,4-dioxane. Data are not available to support a carcinogenic
24 MOA for 1,4-dioxane. In the absence of a MOA(s) for the observed tumor types associated with
25 exposure to 1,4-dioxane, a linear low-dose extrapolation approach was used to estimate human
26 carcinogenic risk associated with 1,4-dioxane exposure.
27 In general, the Agency has preferred to use the multistage model for analyses of tumor incidence
28 and related endpoints because they have a generic biological motivation based on long-established
29 mathematical models such as the MVK model. The MVK model does not necessarily characterize all
30 modes of tumor formation, but it is a starting point for most investigations and, much more often than not,
31 has provided at least an adequate description of tumor incidence data.
32 The multistage cancer model provided adequate fits for the tumor incidence data following a 2-
33 year inhalation exposure to 1,4-dioxane by male rats (Kasai et al.. 2009). However, in the studies
34 evaluated for the oral cancer assessment (Kano et al.. 2009; NCI. 1978; Kocibaetal.. 1974b) the
35 multistage model provided good descriptions of the incidence of a few tumor types in male (nasal cavity)
36 and female (hepatocellular and nasal cavity) rats and in male mice (hepatocellular) exposed to
140
DRAFT - DO NOT CITE OR QUOTE
-------
1 1,4-dioxane via ingestion (see Appendix D for details). However, the multistage model did not provide
2 an adequate fit for female mouse liver tumor dataset based upon the following (U.S. EPA. 2012a):
• Goodness-of-fit/7-value was less than 0.10 indicating statistically significant lack of fit;
• AIC was larger than other acceptable models;
• Observed data deviated substantially from the fitted model, as measured by their standardized %2
residuals (i.e., residuals with values greater than an absolute value of one).
3 By default, the BMDS software imposes constraints on the values of certain parameters of the
4 models. When these constraints were imposed, the multistage model and most other models did not fit
5 the incidence data for female mouse liver adenomas or carcinomas, even after dropping the highest dose
6 group.
7 The log-logistic model was selected because it was the only model that provided an adequate fit
8 to the female mouse liver tumor data (Kano et al.. 2009). A BMR of 50% was used because it is
9 proximate to the response at the lowest dose tested and the BMDL50 was derived by applying appropriate
10 parameter constraints, consistent with recommended use of BMDS in the BMD Technical Guidance
11 Document (U.S. EPA. 2012a).
12 The human equivalent oral CSF estimated from liver tumor datasets with statistically significant
13 increases ranged from 4.2 x 10"4 to 1.0 x 10"1 per mg/kg-day, a range of about three orders of magnitude,
14 with the upper and lower extremes coming from the combined male and female data for hepatocellular
15 carcinomas (Kociba et al.. 1974b) and the female mouse liver adenoma and carcinoma dataset (Kano et
16 al.. 2009V
6.2.3.4 Dose Metric
17 1,4-Dioxane is known to be metabolized in vivo. However, evidence does not exist to determine
18 whether the parent compound, metabolite(s), or a combination of the parent compound and metabolites is
19 responsible for the observed toxicity following exposure to 1,4-dioxane. If the actual carcinogenic moiety
20 is proportional to administered exposure, then use of administered exposure as the dose metric is the least
21 biased choice. On the other hand, if this is not the correct dose metric, then the impact on the CSF is
22 unknown.
6.2.3.5 Cross-Species Scaling
23 For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied to
24 address toxicological equivalence of internal doses between each rodent species and humans, consistent
25 with the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005b). It is assumed that equal risks
26 result from equivalent constant lifetime exposures.
27 Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
28 in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
141
DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.6 Statistical Uncertainty at the POD
1 Parameter uncertainty can be assessed through confidence intervals. Each description of
2 parameter uncertainty assumes that the underlying model and associated assumptions are valid. For the
3 log-logistic model applied to the female mouse data following oral exposure, there is a reasonably small
4 degree of uncertainty at the 50% excess incidence level (the POD for linear low-dose extrapolation), as
5 indicated by the proximity of the BMDLnED (4.95 mg/kg-day) to the BMDnED (7.51 mg/kg-day) . For the
6 multistage model applied for the male rat inhalation dataset, there is a reasonably small degree of
7 uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
6.2.3.7 Bioassay Selection
8 The study by Kano et al. (2009) was used for development of an oral CSF. This was a well-
9 designed study, conducted in both sexes in two species (rats and mice) with a sufficient number (N=50)
10 of animals per dose group. The number of test animals allocated among three dose levels and an
11 untreated control group was adequate, with examination of appropriate toxicological endpoints in both
12 sexes of rats and mice. Alternative bioassays (NCI. 1978; Kocibaetal.. 1974b) were available and were
13 fully considered for the derivation of the oral CSF.
14 The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
15 well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
16 Three dose levels plus an untreated control group were examined following exposure to 1,4-dioxane via
17 inhalation for 2 years.
6.2.3.8 Choice of Species/Gender
18 The oral CSF for 1,4-dioxane was derived using the tumor incidence data for the female mouse,
19 which was thought to be more sensitive than male mice or either sex of rats to the carcinogenicity of
20 1,4-dioxane. While all data, from both species and sexes reported from the Kano et al. (2009) study, were
21 suitable for deriving an oral CSF, the female mouse data represented the most sensitive indicator of
22 carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day [animal dose] or
23 10 mg/kg-day [HED]) observed a considerable and significant increase in combined liver adenomas and
24 carcinomas. Additional testing of doses within the range of control and the lowest dose (66 mg/kg-day
25 [animal dose] or 10 mg/kg-day [HED]) could refine and reduce uncertainty for the oral CSF.
26 Male F344 rat data were used to estimate risk following inhalation of 1,4-dioxane. Kano et al.
27 (2009) showed that male rats were more sensitive than female rats to the effects of 1,4-dioxane following
28 oral administration; therefore, male rats were studied in the 2-year bioassay conducted by the same
29 laboratory (Kasai et al.. 2009). The sensitivity and tumorigenic response of female rats or male or female
30 mice following inhalation of 1,4-dioxane is unknown. Since female mice were the most sensitive gender
31 and species examined in the Kano et al. (2009) study, female mice may also be more sensitive to the
32 inhalation of 1,4-dioxane which would result in a greater risk.
142
DRAFT - DO NOT CITE OR QUOTE
-------
6.2.3.9 Relevance to Humans
1 The oral CSF was derived using the tumor incidence in the liver of female mice. A thorough
2 review of the available toxicological data available for 1,4-dioxane provides no scientific justification to
3 propose that the liver adenomas and carcinomas observed in animal models following exposure to
4 1,4-dioxane are not plausible in humans. Liver adenomas and carcinomas were considered plausible
5 outcomes in humans due to exposure to 1,4-dioxane.
6 The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
7 rats. There is no information on 1,4-dioxane to indicate that the observed rodent tumors are not relevant to
8 humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
9 rodents and humans.
6.2.3.10 Human Population Variability
10 The extent of inter-individual variability in 1,4-dioxane metabolism has not been characterized.
11 A separate issue is that the human variability in response to 1,4-dioxane is also unknown. Data exploring
12 whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life stages is unavailable.
13 This lack of understanding about potential differences in metabolism and susceptibility across exposed
14 human populations thus represents a source of uncertainty. Also, the lack of information linking a MOA
15 for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.
143
DRAFT - DO NOT CITE OR QUOTE
-------
REFERENCES
1 Agrawal A. and Shapiro. B. (2000) Differential expression of gender-dependent hepatic isoforms of cytochrome P-
2 450 by pulse signals in the circulating masculine episodic growth hormone profile of the rat. J Pharmacol
3 ExpTher 292: 228-237.
4 Amendments to the Clean Air Act. Sec. 604. Phase-out of production and consumption of class I substances. 42.
5 U.S.C. §76710(1990).
6 Andersen. M: Clewell. H.. Ill: Gargas. M: Smith. F.: Reitz. R. (1987) Physiologically based pharmacokinetics and
7 the risk assessment process for methylene chloride. Toxicol Appl Pharmacol 87: 185-205.
8 http://dx.doi.org/10.1016/0041-008X(87)90281-X.
9 Argus. M.: Arcos. J.: Hoch-Ligeti. C. (1965a) Studies on the carcinogenic activity of protein-denaturing agents:
10 Hepatocarcinogenicity of dioxane. J Natl Cancer Inst 35: 949-958.
11 Argus. M.: Sohal. R.: Bryant. G.: Hoch-Ligeti. C.: Arcos. J. (1973a) Dose-response and ultrastructural alterations in
12 dioxane carcinogenesis. Influence of methylcholanthrene on acute toxicity. Eur J Cancer 9: 237-243.
13 http://dx.doi.org/10.1016/0014-2964(73)90088-1.
14 Argus. M. F.: Arcos. J. C.: Hoch-Ligeti. C. (1965b) Studies on the carcinogenic activity of protein-denaturing
15 agents: Hepatocarcinogenicity of dioxane. J Natl Cancer Inst 35: 949-958.
16 Argus. M. F.: Sohal. R. S.: Bryant. G. M.: Hoch-Ligeti. C.: Arcos. J. C. (1973b) Dose-response and ultrastructural
17 alterations in dioxane carcinogenesis. Influence of methylcholanthrene on acute toxicity. Eur J Cancer 9:
18 237-243. http://dx.doi.org/10.1016/0014-2964(73)90088-1.
19 Ashbv. J. (1994) The genotoxicity of 1,4-dioxane. Mutat Res 322: 141-142. http://dx.doi.org/10.1016/0165-
20 1218(94)00022-0.
21 Atkinson. R. (1989) Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic
22 compounds. J Phys Chem Ref Data 1: 1 -246.
23 ATSDR. (Agency for Toxic Substances and Disease Registry). (2007) Toxicological profile for 1,4 dioxane. Draft
24 for public comment. Atlanta, GA: U.S. Department of Health and Human Services, Public Health Service.
25 http://www.atsdr.cdc.gov/toxprofiles/tpl87.pdf.
26 Bannasch. P.: Moore. M.: Klimek. F.: Zerban. H. (1982a) Biological markers of preneoplastic foci and neoplastic
27 nodules in rodent liver. Toxicol Pathol 10: 19-34. http://dx.doi.org/10.1177/019262338201000204.
28 Bannasch. P.: Moore. M. A.: Klimek. F.: Zerban. H. (1982b) Biological markers of preneoplastic foci and neoplastic
29 nodules in rodent liver. Toxicol Pathol 10: 19-34. http://dx.doi.org/10.1177/019262338201000204.
30 Bannasch. P. (2003) Comments on R. Karbe and R.L. Kerlin (2002) Cystic degeneration/spongiosis hepatis (Toxicol
31 Pathol 30 (2), 216-227). Toxicol Pathol 31: 566-570. http://dx.doi.org/10.1080/01926230390224700.
32 Barber. H. (1934) Haemorrhagic nephritis and necrosis of the liver from dioxan poisoning. Guy's Hosp Rep 84: 267-
33 280.
34 Bogen. K. (1990) Uncertainty in environmental health risk assessment. New York: Garlend Publishing Company.
35 Boorman. G.: Morgan. K.: Uriah. L. (1990). Nose, larynx and trachea Pathology of the Fischer Rat Reference and
36 Atlas (pp. 315-337). San Diego, CA: Academic Press, San Diego.
37 Braun. W. and Young. J. (1977) Identification of beta-hydroxyethoxyacetic acid as the major urinary metabolite of
38 1,4-dioxane in the rat. Toxicol Appl Pharmacol 39: 33-38. http://dx.doi.org/10.1016/0041-008X(77)90174-
39 0.
40 Bronaugh. R. (1982a). Percutaneous absorption of cosmetic ingredients. In P. Frost & S. N. Horwitz (Eds.),
41 Principles of cosmetics for the dermatologist (pp. 277-284). St. Louis, MO: C.V. Mosby.
42 Bronaugh. R. L. (Ed.). (1982b). Percutaneous absorption of cosmetic ingredients. St. Louis, MO: C.V. Mosby.
43 Brown. R.: Delp. M.: Lindstedt. S.: Rhomberg. L.: Beliles. R. (1997) Physiological parameter values for
44 physiologically based pharmacokinetic models. Toxicol Ind Health 13: 407-484.
45 http://dx.doi.org/10.1177/074823379701300401.
46 Buffler. P.: Wood. S.: Suarez. L.: Kilian. D. (1978a) Mortality follow-up of workers exposed to 1,4-dioxane. J
47 Occup Environ Med 20: 255-259.
48 Buffler. P. A.: Wood. S. M.: Suarez. L.: Kilian. D. J. (1978b) Mortality follow-up of workers exposed to 1,4-
49 dioxane. J Occup Environ Med 20: 255-259.
50 Bull. R.: Robinson. M.: Laurie. R. (1986) Association of carcinoma yield with early papilloma development in
51 SENCAR mice. Environ Health Perspect 68: 11-17.
52 Burmistrov. S.: Arutvunyan. A.: Stepanov. M.: Oparina. T.: Prokopenko. V. (2001) Effect of chronic inhalation of
53 toluene and dioxane on activity of free radical processes in rat ovaries and brain. Bull Exp Biol Med 132:
54 832-836.
144
DRAFT - DO NOT CITE OR QUOTE
-------
1 Carpenter. S.: Lasker. I: Raucv. J. (1996) Expression, induction, and catalytic activity of the ethanol-inducible
2 cytochrome P450 (CYP2E1) in human fetal liver and hepatocytes. Mol Pharmacol 49: 260-268.
3 Clark. B.: Furlong. J.: Ladner. A.: Slovak. A. (1984) Dermal toxicity of dimethyl acetylene dicarboxylate, N-methyl
4 pyrrolidone, triethylene glycol dimethyl ether, dioxane and tetralin in the rat. IRCS Med Sci 12: 296-297.
5 David. H. (1964) Electron-microscopic findings in dioxan-dependent nephrosis in rat kidneys. Beitr Pathol Anat
6 130: 187-212.
7 de Navasquez. S. (1935) Experimental tubular necrosis of the kidneys accompanied by liver changes due to dioxane
8 poisoning. J Hyg 35: 540-548.
9 Derosa. C.: Wilbur. S.: Holler. J.: Richter. P.: Stevens. Y. (1996) Health evaluation of 1,4-dioxane. Toxicol Ind
10 Health 12: 1-43. http://dx.doi.org/10.1177/074823379601200101.
11 Drew. R.: Patel J.: Lin. F. (1978) Changes in serum enzymes in rats after inhalation of organic solvents singly and
12 in combination. Toxicol Appl Pharmacol 45: 809-819. http://dx.doi.org/10.1016/0041-008X(78)90172-2.
13 Ernstgard. L.: Iregren. A.: Sjogren. B.: Johanson. G. (2006) Acute effects of exposure to vapours of dioxane in
14 humans. Hum Exp Toxicol 25: 723-729. http://dx.doi.org/10.1177/0960327106073805.
15 Fairlev. A.: Linton. E.: Ford-Moore. A. (1934a) The toxicity to animals of 1:4 dioxan. J Hyg 34: 486-501.
16 http://dx.doi.org/10.1017/S0022172400043266.
17 Fairlev. A.: Linton. E. C.: Ford-Moore. A. H. (1934b) The toxicity to animals of 1:4 dioxan. J Hyg 34: 486-501.
18 http://dx.doi.org/10.1017/S0022172400043266.
19 Fisher. J.: Mahle. P.: Bankston. L.: Greene. R.: Gearhart. J. (1997) Lactational transfer of volatile chemicals in
20 breast milk. Am Ind Hyg Assoc J 58: 425-431. http://dx.doi.org/10.1080/15428119791012667.
21 Franke. C.: Studinger. G.: Berger. G.: Bohling. S.: Bruckmann. U.: Cohors-Fresenborg. P.: Johncke. U. (1994) The
22 assessment of bioaccumulation. Chemosphere 29: 1501-1514. http://dx.doi.org/10.1016/0045-
23 6535(94)90281-X.
24 Frantik. E.: Hornychova. M: Horvath. M. (1994) Relative acute neurotoxicity of solvents: Isoeffective air
25 concentrations of 48 compounds evaluated in rats and mice. Environ Res 66: 173-185.
26 http://dx.doi.org/10.1006/enrs. 1994.1053.
27 Galloway. S.: Armstrong. M.: Reuben. C.: Colman. S.: Brown. B.: Cannon. C Zeiger. E. (1987a) Chromosome
28 aberrations and sister chromatid exchanges in Chinese hamster ovary cells: Evaluations of 108 chemicals.
29 Environ Mol Mutagen 10: 1-175. http://dx.doi.org/10.1002/em.2850100502.
30 Galloway. S. M.: Armstrong. M. J.: Reuben. C.: Colman. S.: Brown. B.: Cannon. C Zeiger. E. (1987b)
31 Chromosome aberrations and sister chromatid exchanges in Chinese hamster ovary cells: Evaluations of
32 108 chemicals. Environ Mol Mutagen 10: 1-175. http://dx.doi.org/10.1002/em.2850100502.
33 Gaskell B. (1990) Nonneoplastic changes in the olfactory epithelium-- experimental studies. Environ Health
34 Perspect 85: 275-289.
35 Giavini. E.: Vismara. C.: Broccia. M. (1985a) Teratogenesis study of dioxane in rats. Toxicol Lett 26: 85-88.
36 http://dx.doi.org/10.1016/0378-4274(85)90189-4.
37 Giavini. E.: Vismara. C.: Broccia. M. L. (1985b) Teratogenesis study of dioxane in rats. Toxicol Lett 26: 85-88.
38 http://dx.doi.org/10.1016/0378-4274(85)90189-4.
3 9 Goldberg. M.: Johnson. H.: Pozzani. U.: Smyth. H. J. (1964) Effect of repeated inhalation of vapors of industrial
40 solvents on animal behavior: I. Evaluation of nine solvent vapors on pole-climb performance in rats. Am
41 Ind Hyg Assoc J 25: 369-375.
42 Goldsworthy. T.: Monticello. T.: Morgan. K.: Bermudez. E.: Wilson. P.: Ja'ckh. R.: BE. B. (1991) Examination of
43 potential mechanisms of carcinogenicity of 1,4-dioxane in rat nasal epithelial cells and hepatocytes. Arch
44 Toxicol 65: 1-9. http://dx.doi.org/10.1007/BF01973495.
45 Green. T.: Lee. R.: Moore. R.: Ashbv. J.: Willis. G.: Lund. V.: MJL. C. (2000) Acetochlor-induced rat nasal tumors:
46 Further studies on the mode of action and relevance to humans. Regul Toxicol Pharmacol 32: 127-133.
47 http://dx.doi.org/10.1006/rtph.2000.1413.
48 Grosjean. D. (1990) Atmospheric chemistry of toxic contaminants. 2. Saturated aliphatics: Acetaldehyde, dioxane,
49 ethylene glycol ethers, propylene oxide. J Air Waste Manag Assoc 40: 1522-1531.
50 Guilmette. R. A.: Cheng. Y. S.: Griffith. W. C. (1997) Characterising the variability in adult human nasal airway
51 dimensions. Ann Occup Hyg 41: 491 -496.
52 Hall W. C. (Ed.). (1990). Peritoneum, retroperitoneum, mesentery and abdominal cavity. San Diego, CA: Academic
53 Press.
54 Hansch. C.: Leo. A.: Hoekman. D. (1995) Exploring QSAR: Hydrophobic, electronic, and steric constants.
55 Washington, DC: American Chemical Society.
56 Harkema. J.: Carey. S.: Wagner. J. (2006) The nose revisited: A brief review of the comparative structure, function,
57 and toxicologic pathology of the nasal epithelium. Toxicol Pathol 34: 252-269.
58 Haseman. J.: Huff. J.: Boorman. G. (1984) Use of historical control data in carcinogenicity studies in rodents.
59 Toxicol Pathol 12: 126-135. http://dx.doi.org/10.1177/019262338401200203.
145
DRAFT - DO NOT CITE OR QUOTE
-------
1 Haseman. J. and Hailev. J. (1997a) An update of the National Toxicology Program database on nasal carcinogens.
2 MutatRes 380: 3-11. http://dx.doi.org/10.1016/S0027-5107(97)00121-8.
3 Haseman. J. K. and Hailev. J. R. (1997b) An update of the National Toxicology Program database on nasal
4 carcinogens. Mutat Res 380: 3-11. http://dx.doi.org/10.1016/S0027-5107(97)00121-8.
5 Haseman. J. K.: Hailev. J. R.: Morris. R. W. (1998) Spontaneous neoplasm incidences in Fischer 344 rats and
6 B6C3F1 mice in two-year carcinogenicity studies: A National Toxicology Program update. Toxicol Pathol
7 26: 428-441. http://dx.doi.org/10.1177/019262339802600318.
8 Hawlev. G. and Lewis. R.. Sr (Eds.). (2001). Hawley's Condensed Chemical Dictionary (14 ed.). New York, NY:
9 John Wiley & Sons, Inc.
10 Haworth. S.: Lawlor. T.: Mortelmans. K.: Speck. W.: Zeiger. E. (1983) Salmonella mutagenicity test results for 250
11 chemicals. Environ Mutagen 5: 3-142. http://dx.doi.org/10.1002/em.2860050703.
12 Havashi. S.: Watanabe. J.: Kawajiri. K. (1991) Genetic polymorphisms in the 5'-flanking region change
13 transcriptional regulation of the human cytochrome P450IIE1 gene. J Biochem 110: 559-565.
14 Hellmer. L. and Bolcsfoldi. G. (1992) An evaluation of the E. coli K-12 uvrB/recA DNA repair host-mediated assay.
15 I. In vitro sensitivity of the bacteria to 61 compounds. Mutat Res 272: 145-160.
16 http://dx.doi.org/10.1016/0165-1161(92)90043-L.
17 HF Jr. S.: Seaton. J.: Fischer. L. (1941) The single dose toxicity of some glycols and derivatives. J Ind Hyg Toxicol
18 23:259-268.
19 Hoch-Ligeti. C. and Argus. M. (1970a). Effect of carcinogens on the lung of guinea pigs. In P. Nettlesheim, M. G.
20 Hanna Jr & J. W. Deatherage Jr (Eds.), Morphology of Experimental Respiratory Carcinogenesis:
21 Proceedings of a Biology Division, Oak Ridge National Laboratory, Conference held in Gatlinburg,
22 Tennessee, May 13-16, 1970 (Vol. 21, pp. 267-279). Oak Ridge, TN: United States Atomic Energy
23 Comission, Division of Technical Information.
24 Hoch-Ligeti. C.: Argus. M.: Arcos. J. (1970a) Induction of carcinomas in the nasal cavity of rats by dioxane. Br J
25 Cancer 24: 164-167.
26 Hoch-Ligeti. C. and Argus. M. F. (1970b). Effect of carcinogens on the lung of guinea pigs. In P. Nettlesheim, M.
27 G. Hanna Jr & J. W. Deatherage Jr (Eds.), Morphology of Experimental Respiratory Carcinogenesis:
28 Proceedings of a Biology Division, Oak Ridge National Laboratory, Conference held in Gatlinburg,
29 Tennessee, May 13-16, 1970 (Vol. 21, pp. 267-279). Oak Ridge, TN: United States Atomic Energy
30 Comission, Division of Technical Information.
31 Hoch-Ligeti. C.: Argus. M. F.: Arcos. J. C. (1970b) Induction of carcinomas in the nasal cavity of rats by dioxane.
32 BrJ Cancer 24: 164-167.
33 HSDB. (Hazardous Substances Data Bank). (2007) 1,4-Dioxane. Bethesda, Maryland: National Library of
34 Medicine, National Toxicology Program, Hazardous Substances Data Bank.
35 Huang. C.-Y.: Huang. K.-L.: Cheng. T.-J.: Wang. J.-D.: Hsieh. L.-L. (1997) The GST Tl and CYP2E1 genotypes
36 are possible factors causing vinyl chloride induced abnormal liver function. Arch Toxicol 71: 482-488.
37 http://dx.doi.org/10.1007/s002040050416.
38 IARC. (International Agency for Research on Cancer). (1999) 1,4-Dioxane. Lyon, France: World Health
39 Organization. http://monographs.iarc.fr/ENG/Monographs/vol71/mono71-25.pdf.
40 ICRP. (International Commission on Radiological Protection). (1975) Report of the task group on reference man:
41 ICRP publication 23 (Vol. 23). New York, NY: International Commission of Radiological Protection,
42 Pergamon Press.
43 JBRC. (Japan Bioassay Research Center). (1998) Two-year studies of 1,4-dioxane in F344 rats and BDF1 mice
44 (drinking water). Kanagawa, Japan.
45 Johnstone. R. (1959) Death due to dioxane? AMA Arch Ind Health 20: 445-447.
46 Kanada. M.: Miyagawa. M.: Sato. M.: Hasegawa. H.: Honma. T. (1994) Neurochemical profile of effects of 28
47 neurotoxic chemicals on the central nervous system in rats (1) Effects of oral administration on brain
48 contents of biogenic amines and metabolites. Ind Health 32: 145-164.
49 http://dx.doi.org/10.2486/indhealth.32.145.
50 Kano. H.: Umeda. Y: Saito. M.: Senoh. H.: Ohbavashi. H.: Aiso. S Fukushima. S. (2008) Thirteen-week oral
51 toxicity of 1,4-dioxane in rats and mice. J Toxicol Sci 33: 141-153. http://dx.doi.org/10.2131/jts.33.141.
52 Kano. H.: Umeda. Y.: Kasai T.: Sasaki. T.: Matsumoto. M.: Yamazaki. K Fukushima. S. (2009)
53 Carcinogenicity studies of 1,4-dioxane administered in drinking-water to rats and mice for 2 years. Food
54 Chem Toxicol 47: 2776-2784. http://dx.doi.0rg/10.1016/j.fct.2009.08.012.
55 Karbe. E. and Kerlin. R. (2002a) Cystic degeneration/spongiosis hepatis in rats. Toxicol Pathol 30: 216-227.
56 http://dx.doi.org/10.1080/019262302753559551.
57 Karbe. E. and Kerlin. R. L. (2002b) Cystic degeneration/spongiosis hepatis in rats. Toxicol Pathol 30: 216-227.
58 http://dx.doi.org/10.1080/019262302753559551.
59 Kasai. T. (2008) [1,4-Dioxane toxicity studies].
146
DRAFT - DO NOT CITE OR QUOTE
-------
1 Kasai T.: Saito. M.: Senoh. H.: Umeda. Y.: Also. S.: Ohbavashi. H Fukushima. S. (2008) Thirteen-week
2 inhalation toxicity of 1,4-dioxane in rats. Inhal Toxicol 20: 961-971.
3 http://dx.doi.org/10.1080/08958370802105397.
4 Kasai. T.: Kano. H.: Umeda. Y.: Sasaki. T.: Ikawa. N.: Nishizawa. T Fukushima. S. (2009) Two-year inhalation
5 study of carcinogenicity and chronic toxicity of 1,4-dioxane in male rats. Inhal Toxicol 21: 889-897.
6 http://dx.doi.org/10.1080/08958370802629610.
7 Kasper. P.: Uno. Y.: Mauthe. R.: Asano. N.: Douglas. G.: Matthews. E Speit G. (2007) Follow-up testing of
8 rodent carcinogens not positive in the standard genotoxicity testing battery: IWGT workgroup report. Mutat
9 Res 627: 106-116. http://dx.doi.0rg/10.1016/j.mrgentox.2006.10.007.
10 Katagirl T.: Nagano. K.: Aiso. S.: Senoh. H.: Sakura. Y.: Takeuchj T.: Okudaira. M. (1998) A pathological study
11 on spontaneous hepatic neoplasms in BDF1 mice. J Toxicol Pathol 11: 21-25.
12 http://dx.doi.org/10.1293/tox. 11.21.
13 Kesten. H.: Mulinos. M.: Pomerantz. L. (1939) Pathologic effects of certain glycols and related compounds. Arch
14 Pathol 27: 447-465.
15 Khudolev. V.: Mizgireuv. L: Pliss. G. (1987) The study of mutagenic activity of carcinogens and other chemical
16 agents with Salmonella typhimurium assays: Testing of 126 compounds. Arch Geschwulstforsch 57: 453-
17 462.
18 King. M.: Shefner. A.: Bates. R. (1973a) Carcinogenesis bioassay of chlorinated dibenzodioxins and related
19 chemicals. Environ Health Perspect 5: 163-170.
20 King. M. E.: Shefner. A. M.: Bates. R. R. (1973b) Carcinogenesis bioassay of chlorinated dibenzodioxins and
21 related chemicals. Environ Health Perspect 5: 163-170.
22 Kitchin. K. and Brown. J. (1990a) Is 1,4-dioxane a genotoxic carcinogen? Cancer Lett 53: 67-71.
23 http://dx.doi.org/10.1016/0304-3835(90)90012-M.
24 Kitchin. K. T. and Brown. J. L. (1990b) Is 1,4-dioxane a genotoxic carcinogen? Cancer Lett 53: 67-71.
25 http://dx.doi.org/10.1016/0304-3835(90)90012-M.
26 Knoefel. P. (1935) Narcotic potency of some cyclic acetals. J Pharmacol Exp Ther 53: 440-444.
27 Kociba. R.: McCollister. S.: Park. C.: Torkelson. T.: Gehring. P. (1974a) 1,4-dioxane. I. Results of a 2-year
28 ingestion study in rats. Toxicol Appl Pharmacol 30: 275-286. http://dx.doi.org/10.1016/0041-
29 008X(74)90099-4.
30 Kociba. R.: Torkelson. T.: Young. J.: Gehring. P. (1975a). 1,4-Dioxane: Correlation of the results of chronic
31 ingestion and inhalation studies with its dose-dependent fate in rats Proceedings of the 6th Annual
32 Conference on Environmental Toxicology (pp. 345-354). Wright-Patterson Air Force Base, OH: Wright-
33 Patterson Air Force Base, Air Force Systems Command, Aerospace Medical Division, Aerospace Medical
34 Research Laboratory.
35 Kociba. R. J.: McCollister. S. B.: Park. C.: Torkelson. T. R.: Gehring. P. J. (1974b) 1,4-dioxane. I. Results of a 2-
36 year ingestion study in rats. Toxicol Appl Pharmacol 30: 275-286. http://dx.doi.org/10.1016/0041-
37 008X(74)90099-4.
38 Kociba. R. J.: Torkelson. T. R.: Young. J. P.: Gehring. P. J. (1975b) 1,4-Dioxane: Correlation of the results of
39 chronic ingestion and inhalation studies with its dose-dependent fate in rats. Wright-Patterson Air Force
40 Base, OH: Wright-Patterson Air Force Base, Air Force Systems Command, Aerospace Medical Division,
41 Aerospace Medical Research Laboratory.
42 Kopylev. L.: John Fox. J.: Chen. C. (2009). Combining risks from several tumors using Markov Chain Monte Carlo.
43 In R. M. Cooke (Ed.), Uncertainty Modeling in Dose Response (1 ed., pp. 197-205). Hoboken, NJ: John
44 Wiley & Sons.
45 Kurl. R.: Poellinger. L.: Lund. J.: Gustafsson. J.-A. (1981) Effects of dioxane on RNA synthesis in the rat liver.
46 Arch Toxicol 49: 29-33. http://dx.doi.org/10.1007/BF00352068.
47 Kwan. K.: Dutka. B.: Rao. S.: Liu. D. (1990) Mutatox test: A new test for monitoring environmental genotoxic
48 agents. Environ Pollut 65: 323-332. http://dx.doi.org/10.1016/0269-7491(90)90124-U.
49 Laug. E.: Calvery. H.: Morris. H.: Woodard. G. (1939) The toxicology of some glycols and derivatives. J Ind Hyg
50 Toxicol 21: 173-201.
51 Lesage. S.: Jackson. R.: Priddle. M.: Riemann. P. (1990) Occurrence and fate of organic solvent residues in anoxic
52 groundwater at the Gloucester landfill, Canada. Environ Sci Technol 24: 559-566.
53 http://dx.doi.org/10.1021/es00074a016.
54 Leung. H.-W. and Paustenbach. D. (1990a) Cancer risk assessment for dioxane based upon a physiologically-based
55 pharmacokinetic approach. Toxicol Lett 51: 147-162.
56 Leung. H. W. and Paustenbach. D. J. (1990b) Cancer risk assessment for dioxane based upon a physiologically-
57 based pharmacokinetic approach. Toxicol Lett 51: 147-162.
147
DRAFT - DO NOT CITE OR QUOTE
-------
1 Lewandowski. T. and Rhomberg. L. (2005) A proposed methodology for selecting a trichloroethylene inhalation
2 unit risk value for use in risk assessment. Regul Toxicol Pharmacol 41: 39-54.
3 http://dx.doi.org/10.1016/j.yrtph.2004.09.003.
4 Lewis. R.. Sr. (2000) Sax's Dangerous Properties of Industrial Materials (10 ed.). New York, NY: John Wiley &
5 Sons, Inc.
6 Lide. D. R.. (Ed.). (2000). Handbook of Chemistry and Physics (81 ed.). Boca Raton, FL: CRC Press LLC.
7 Lundberg. L: Ekdahl. M: Kronevi T.: Lidums. V.: Lundberg. S. (1986) Relative hepatotoxicity of some industrial
8 solvents after intraperitoneal injection or inhalation exposure in rats. Environ Res 40: 411-420.
9 http://dx.doi.org/10.1016/S0013-9351(86)80116-5.
10 Lundberg. L: Hogberg. I: Kronevi. T.: Holmberg. B. (1987) Three industrial solvents investigated for tumor
11 promoting activity in the rat liver. Cancer Lett 36: 29-33. http://dx.doi.org/10.1016/0304-3835(87)90099-1.
12 Lyman. W.: Reehl. W.: DH Rosenblatt (Eds.). (American Chemical Society). (1990). Handbook of Chemical
13 Property Estimation Methods: Environmental Behavior of Organic Compounds. Washington, DC:
14 American Chemical Society.
15 Marzulli. F.: Anjo. P.: Maibach. H. (1981) In vivo skin penetration studies of 2,4-toluenediamine, 2,4-
16 diaminoanisole, 2-nitro-p-phenylenediamine, p-dioxane and N-nitrosodiethanolamine in cosmetics. Food
17 Cosmet Toxicol 19: 743-747. http://dx.doi.org/10.1016/0015-6264(81)90530-7.
18 McConnell E.: Solleveld. H.: Swenberg. J.: Boorman. G. (1986) Guidelines for combining neoplasms for evaluation
19 of rodent carcinogenesis studies. J Natl Cancer Inst 76: 283-289.
20 McFee. A.: Abbott. M.: Gulati. P.: Shelby. M. (1994) Results of mouse bone marrow micronucleus studies on 1,4-
21 dioxane. Mutat Res 322: 145-148.
22 McGregor. P.: Brown. A.: Howgate. S.: McBride. P.: Riach. C.: Caspary. W. (1991a) Responses of the L5178Y
23 mouse lymphoma cell forward mutation assay. V: 27 coded chemicals. Environ Mol Mutagen 17: 196-219.
24 http://dx.doi.org/10.1002/em.2850170309.
25 McGregor. D. B.: Brown. A. G.: Howgate. S.: McBride. P.: Riach. C.: Caspary. W. J. (1991b) Responses of the
26 L5178Y mouse lymphoma cell forward mutation assay. V: 27 coded chemicals. Environ Mol Mutagen 17:
27 196-219. http://dx.doi.org/10.1002/em.2850170309.
28 Medinsky. M. and Bond. J. (2001) Sites and mechanisms for uptake of gases and vapors in the respiratory tract.
29 Toxicology 160: 165-172.
30 Meylan. W.: Howard. P.: Boethling. R.: Aronson. P.: Printup. H.: Gouchie. S. (1999) Improved method for
31 estimating bioconcentration/bioaccumulation factor from octanol/water partition coefficient. Environ
32 Toxicol Chem 18: 664-672. http://dx.doi.org/10.1002/etc.5620180412.
33 Mikheev. M.: Gorlinskava Ye. P.: Solowova. T. (1990) The body distribution and biological action of xenobiotics. J
34 Hyg Epidemiol Microbiol Immunol 34: 329-336.
35 Mirkova. E. (1994a) Activity of the rodent carcinogen 1,4-dioxane in the mouse bone marrow micronucleus assay.
36 Mutat Res 322: 142-144.
37 Mirkova. E. T. (1994b) Activity of the rodent carcinogen 1,4-dioxane in the mouse bone marrow micronucleus
38 assay. Mutat Res 322: 142-144.
39 Miyagawa. M.: Shirotori. T.: Tsuchitani. M.: Yoshikawa. K. (1999) Repeat-assessment of 1,4-dioxane in a rat-
40 hepatocyte replicative DNA synthesis (RDS) test: Evidence for stimulus of hepatocyte proliferation. Exp
41 Toxicol Pathol 51: 555-558.
42 Morgan. K.: Patterson. P.: Gross. E. (1986) Responses of the nasal mucociliary apparatus of F-344 rats to
43 formaldehyde gas. Toxicol Appl Pharmacol 82: 1-13. http://dx.doi.org/10.1016/0041-008X(86)90431-X.
44 Morita. T. (1994) No clastogenicity of 1,4 dioxane as examined in the mouse peripheral blood micronucleus test.
45 Mammalian Mutagenicity Study Group Communications 2: 7-8.
46 Morita. T. and Havashi. M. (1998) 1,4-Dioxane is not mutagenic in five in vitro assays and mouse peripheral blood
47 micronucleus assay, but is in mouse liver micronucleus assay. Environ Mol Mutagen 32: 269-280.
48 http://dx.doi.org/10.1002/(SICI)1098-2280(1998)32:3<269::AID-EM10>3.0.CO;2-8.
49 Mungikar. A. and Pawar. S. (1978) Induction of the hepatic microsomal mixed function oxidase system in mice by
50 p-dioxane. Bull Environ Contain Toxicol 20: 797-804. http://dx.doi.org/10.1007/BFO1683603.
51 Munoz. E. and Barnett B. (2002) The rodent carcinogens 1,4-dioxane and thiourea induce meiotic non-disjunction
52 inDrosophilamelanogasterfemales. MutatRes 517: 231-238. http://dx.doi.org/10.1016/S1383-
53 5718(02)00083-9.
54 Nannelli. A.: De Rubertis. A.: Longo. V.: Gervasi P. (2005a) Effects of dioxane on cytochrome P450 enzymes in
55 liver, kidney, lung and nasal mucosa of rat. Arch Toxicol 79: 74-82. http://dx.doi.org/10.1007/s00204-004-
56 0590-z.
57 Nannelli. A.: De Rubertis. A.: Longo. V.: Gervasi. P. G. (2005b) Effects of dioxane on cytochrome P450 enzymes in
58 liver, kidney, lung and nasal mucosa of rat. Arch Toxicol 79: 74-82. http://dx.doi.org/10.1007/s00204-004-
59 0590-z.
148
DRAFT - DO NOT CITE OR QUOTE
-------
1 NCI. (National Cancer Institute). (1978) Bioassay of 1,4-dioxane for possible carcinogenicity. (78-1330 NCICGTR-
2 80). Bethesda, MD. http://ntp.niehs.nih.gov/ntp/htdocs/LT_rpts/tr080.pdf.
3 Nelson. N. (1951) Solvent toxicity with particular reference to certain octyl alcohols and dioxanes. Med Bull 11:
4 226-238.
5 Nestmann. E.: Otson. R.: Kowbel. P.: Bothwell. P.: Harrington. T. (1984a) Mutagenicity in a modified Salmonella
6 assay of fabric-protecting products containing 1,1,1-trichloroethane. Environ Mol Mutagen 6: 71-80.
7 http://dx.doi.org/10.1002/em.2860060109.
8 Nestmann. E. R.: Otson. R.: Kowbel D. I: Bothwell P. P.: Harrington. T. R. (1984b) Mutagenicity in a modified
9 Salmonella assay of fabric-protecting products containing 1,1,1-trichloroethane. Environ Mol Mutagen 6:
10 71-80. http://dx.doi.org/10.1002/em.2860060109.
11 NRC. (National Research Council). (1994) Science and judgment in risk assessment. Washington, DC: National
12 Academy Press.
13 NRC. (National Research Council). (2009) Science and decisions: Advancing risk assessment. Washington, DC:
14 National Academies Press.
15 O'Neil M. I: Smith. A.: Heckelman. P. E.: Obenchain. J. R.: Gallipeau. J. R.: D'Arecca. M. A. (Eds.). (2001). The
16 Merck index: An encyclopedia of chemicals, drugs, and biologicals (13th ed.). Whitehouse Station, NJ:
17 Merck & Co., Inc.
18 Overton. J. H.: Kimbell J. S.: Miller. F. J. (2001) Dosimetry modeling of inhaled formaldehyde: The human
19 respiratory tract. Toxicol Sci 64: 122-134.
20 Park. J.: Hussam. A.: Couasnon. P.: Fritz. P.: Carr. P. (1987) Experimental reexamination of selected partition
21 coefficients from Rohrschneider's data set. Anal Chem 59: 1970-1976.
22 http://dx.doi.org/10.1021/ac00142a016.
23 Platz. J.: Sehested. J.: Mogelberg. T.: Nielsen. P.: Wallington. T. (1997) Atmospheric chemistry of 1,4-dioxane.
24 Faraday Trans 1 93: 2855-2863. http://dx.doi.org/10.1039/a700598i.
25 Pozzani. U.: Weil C.: Carpenter. C. (1959) The toxicological basis of threshold limit values. 5: The experimental
26 inhalation of vapor mixtures by rats, with notes upon the relationship between single dose inhalation and
27 single dose oral data. Am Ind Hyg Assoc J 20: 364-369. http://dx.doi.org/10.1080/00028895909343733.
28 Ramsey. J. and Andersen. M. (1984) A physiologically based description of the inhalation pharmacokinetics of
29 styrene in rats and humans. Toxicol Appl Pharmacol 73: 159-175. http://dx.doi.org/10.1016/0041-
30 008X(84)90064-4.
31 Reitz. R.: McCroskev. P.: Park. C.: Andersen. M.: Gargas. M. (1990a) Development of a physiologically based
32 pharmacokinetic model for risk assessment with 1,4-dioxane. Toxicol Appl Pharmacol 105: 37-54.
33 http://dx.doi.org/10.1016/0041-008X(90)90357-Z.
34 Reitz. R. H.: McCroskev. P. S.: Park. C. N.: Andersen. M. E.: Gargas. M. L. (1990b) Development of a
35 physiologically based pharmacokinetic model for risk assessment with 1,4-dioxane. Toxicol Appl
36 Pharmacol 105: 37-54. http://dx.doi.org/10.1016/0041-008X(90)90357-Z.
37 Rosenkranz. H. and Klopman. G. (1992) 1,4-dioxane: Prediction of in vivo clastogenicity. Mutat Res 280: 245-251.
38 http://dx.doi.org/10.1016/0165-1218(92)90054-4.
39 Roy. S.: Thilagar. A.: Eastmond. D. (2005a) Chromosome breakage is primarily responsible for the micronuclei
40 induced by 1,4-dioxane in the bone marrow and liver of young CD-I mice. Mutat Res 586: 28-37.
41 http://dx.doi.0rg/10.1016/j.mrgentox.2005.05.007.
42 Roy. S. K.: Thilagar. A. K.: Eastmond. D. A. (2005b) Chromosome breakage is primarily responsible for the
43 micronuclei induced by 1,4-dioxane in the bone marrow and liver of young CD-I mice. Mutat Res 586: 28-
44 37. http://dx.doi.0rg/10.1016/j.mrgentox.2005.05.007.
45 Sato. K. (1989) Glutathione transferases as markers of preneoplasia and neoplasia. Adv Cancer Res 52: 205-255.
46 Schrenk. H. and Yant W. (1936) Toxicity of dioxan. J Ind Hyg Toxicol 18: 448-460.
47 Sheu. C.: Moreland. F.: Lee. J.: Dunkel. V. (1988) In vitro BALB/3T3 cell transformation assay of nonoxynol-9 and
48 1,4-dioxane. Environ Mol Mutagen 11: 41-48. http://dx.doi.org/10.1002/em.2850110106.
49 Silverman. L.: Schulte. H.: First M. (1946) Further studies on sensory response to certain industrial solvent vapors.
50 J Ind Hyg Toxicol 28: 262-266.
51 Sina. J.: Bean. C.: Dysart. G.: Taylor. V.: Bradley. M. (1983) Evaluation of the alkaline elution/rat hepatocyte assay
52 as a predictor of carcinogenic/mutagenic potential. Mutat Res 113:357-391.
53 http://dx.doi.org/10.1016/0165-1161(83)90228-5.
54 Spiegelhalter. P.: Thomas. A.: Best. N.: Lunn. D. (2003) WinBugs version 1.4 user manual. Cambridge, UK: MRC
55 Biostatistics Unit.
56 Sticknev. J.: Sager. S.: Clarkson. I: Smith. L.: Locev. B.: Bock. M Olp. S. (2003) An updated evaluation of the
57 carcinogenic potential of 1,4-dioxane. Regul Toxicol Pharmacol 38: 183-195.
58 http://dx.doi.org/10.1016/80273-2300(03)00090-4.
149
DRAFT - DO NOT CITE OR QUOTE
-------
1 Stoner. G.: Conran. P.: Greisiger. E.: Stober. J.: Morgan. M: Pereira. M. (1986) Comparison of two routes of
2 chemical administration on the lung adenoma response in strain A/J mice. Toxicol Appl Pharmacol 82: 19-
3 31. http://dx.doi.org/10.1016/0041-008X(86)90433-3.
4 Stott W.: Quast J.: Watanabe. P. (1981) Differentiation of the mechanisms of oncogenicity of 1,4-dioxane and 1,3-
5 hexachlorobutadiene in the rat. Toxicol Appl Pharmacol 60: 287-300. http://dx.doi.org/10.1016/0041-
6 008X(91)90232-4.
7 Stroebel P.: Mayer. F.: Zerban. H.: Bannasch. P. (1995) Spongiotic pericytoma: A benign neoplasm deriving from
8 the perisinusoidal (Ito) cells in rat liver. Am J Pathol 146:903-913.
9 Surprenant K. (2002). Dioxane Ullmann's Encyclopedia of Industrial Chemistry (6th ed., pp. 543). Weinheim,
10 Germany: Wiley-VCH Verlag.
11 Sweeney. L.: Thrall K.: Poet. T.: Corlev. R.: Weber. T.: Locev. B Gargas. M. (2008a) Physiologically based
12 pharmacokinetic modeling of 1,4-dioxane in rats, mice, and humans. Toxicol Sci 101: 32-50.
13 http://dx.doi.org/10.1093/toxsci/kfm251.
14 Sweeney. L. M.: Thrall K. P.: Poet T. S.: Corlev. R. A.: Weber. T. J.: Locev. B. J Gargas. M. L. (2008b)
15 Physiologically based pharmacokinetic modeling of 1,4-dioxane in rats, mice, and humans. Toxicol Sci
16 101: 32-50. http://dx.doi.org/10.1093/toxsci/kfm251.
17 Takano. R.: Muravama. N.: Horiuchi. K.: Kitajima. M.: Shono. F.: Yamazaki. H. (2010) Blood concentrations of
18 1,4-dioxane in humans after oral administration extrapolated from in vivo rat pharmacokinetics, in vitro
19 human metabolism, and physiologically based pharmacokinetic modeling. J Health Sci 56: 557-565.
20 http://dx.doi.org/10.1248/jhs.56.557.
21 Thiess. A.: Tress. E.: Fleig. I. (1976a) Arbeitsmedizinische Untersuchungsergebnisse von Dioxan-exponierten
22 Mitarbeitern [Industrial-medical investigation results in the case of workers exposed to dioxane].
23 Arbeitsmedizin, Sozialmedizin, Umweltmedizin 11: 35-46.
24 Thiess. A. M.: Tress. E.: Fleig. I. (1976b) Arbeitsmedizinische Untersuchungsergebnisse von Dioxan-exponierten
25 Mitarbeitern [Industrial-medical investigation results in the case of workers exposed to dioxane].
26 Arbeitsmedizin, Sozialmedizin, Umweltmedizin 11: 35-46.
27 Thurman. G.: Simms. B.: Goldstein. A.: Kilian. D. (1978) The effects of organic compounds used in the
28 manufacture of plastics on the responsivity of murine and human lymphocytes. Toxicol Appl Pharmacol
29 44: 617-641. http://dx.doi.org/10.1016/0041-008X(78)90269-7.
30 Tinwell H. and Ashbv. J. (1994) Activity of 1,4-dioxane in mouse bone marrow micronucleus assays. Mutat Res
31 322: 148-150.
32 Torkelson. T.: Leong. B.: Kociba. R.: Richter. W.: Gehring. P. (1974a) 1,4-Dioxane. II. Results of a 2-year
33 inhalation study in rats. Toxicol Appl Pharmacol 30: 287-298. http://dx.doi.org/10.1016/0041-
34 008X(74)90100-8.
35 Torkelson. T. R.: Leong. B. K. J.: Kociba. R. J.: Richter. W. A.: Gehring. P. J. (1974b) 1,4-Dioxane. II. Results of a
36 2-year inhalation study in rats. Toxicol Appl Pharmacol 30: 287-298. http://dx.doi.org/10.1016/0041-
37 008X(74)90100-8.
38 U.N. Environment Programme. (United Nations Environment Programme). (2000) The Montreal Protocol on
39 substances that deplete the ozone layer. Nairobi, Kenya: United Nations Environment Programme, Ozone
40 Secretariat.
41 http://www.google.com/url?sa=t&source=web&cd= l&ved=OCBIQFj AA&url=http%3 A%2F%2Fwww.une
42 p.org%2Fozone%2Fpdfs%2Fmontreal-protocol2000.pdf&ei=-c89TPXON9PRngf-i-
43 jdDg&usg=AFOjCNH4OH15inPn5XFcYTvblPPRDZu-fQ&sig2=qqSaM_nuQlXlHc409kBvgw.
44 U.S. EPA. (U.S. Environmental Protection Agency). (1986a) Guidelines for mutagenicity risk assessment.
45 (EPA/630/R-98/003). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
46 http://www.epa.gov/iris/backgrd.html.
47 U.S. EPA. (U.S. Environmental Protection Agency). (1986b) Guidelines for carcinogen risk assessment.
48 (EPA/630/R-00/004). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
49 http://epa.gov/raf/publications/pdfs/CA%20GUIDELINES_1986.PDF.
50 U.S. EPA. (U.S. Environmental Protection Agency). (1986c) Guidelines for the health risk assessment of chemical
51 mixtures. (EPA/630/R-98/002). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment
52 Forum. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=22567.
53 U.S. EPA. (U.S. Environmental Protection Agency). (1988) Recommendations for and documentation of biological
54 values for use in risk assessment. (EPA/600/6-87/008). Cincinnati, OH: U.S. Environmental Protection
55 Agency, Environmental Criteria and Assessment Office.
56 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=34855.
57 U.S. EPA. (U.S. Environmental Protection Agency). (1991) Guidelines for developmental toxicity risk assessment.
58 (EPA/600/FR-91/001). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
59 http://www.epa.gov/iris/backgrd.html.
150
DRAFT - DO NOT CITE OR QUOTE
-------
1 U.S. EPA. (U.S. Environmental Protection Agency). (1994a) Methods for derivation of inhalation reference
2 concentrations and application of inhalation dosimetry. (EPA/600/8-90/066F). Research Triangle Park, NC:
3 U.S. Environmental Protection Agency, Office of Research and Development, Office of Health and
4 Environmental Assessment, Environmental Criteria and Assessment Office.
5 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=71993.
6 U.S. EPA. (U.S. Environmental Protection Agency). (1994b) Methods for derivation of inhalation reference
7 concentrations and application of inhalation dosimetry. (EPA/600/8-90/066F). Research Triangle Park, NC.
8 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=71993.
9 U.S. EPA. (U.S. Environmental Protection Agency). (1994c) Interim policy for particle size and limit concentration
10 issues in inhalation toxicity studies. Washington, DC: U.S. Environmental Protection Agency, Health
11 Effects Division, Office of Pesticide Products.
12 http://cfpub.epa. gov/ncea/cfm/recordisplay. cfm?deid= 186068.
13 U.S. EPA. (U.S. Environmental Protection Agency). (1995) The use of the benchmark dose approach in health risk
14 assessment. (EPA/630/R-94/007). Washington, DC: U.S. Environmental Protection Agency, Risk
15 Assessment Forum, http://www.epa.gov/raf/publications/useof-bda-healthrisk.htm.
16 U.S. EPA. (U.S. Environmental Protection Agency). (1996) Guidelines for reproductive toxicity risk assessment.
17 (EPA/630/R-96/009). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
18 http://www.epa.gov/raf/publications/pdfs/REPRO51.PDF.
19 U.S. EPA. (U.S. Environmental Protection Agency). (1998) Guidelines for neurotoxicity risk assessment.
20 (EPA/630/R-95/001F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
21 http://www.epa.gov/raf/publications/pdfs/NEUROTOX.PDF.
22 U.S. EPA. (U.S. Environmental Protection Agency). (2000a) Benchmark dose technical guidance document
23 [external review draft]. (EPA/630/R-00/001). Washington, DC: U.S. Environmental Protection Agency,
24 Risk Assessment Forum, http://www.epa.gov/raf/publications^enchmark-dose-doc-draft.htm.
25 U.S. EPA. (U.S. Environmental Protection Agency). (2000b) Science policy council handbook: Risk
26 characterization. (EPA 100-B-00-002). Washington, D.C.: U.S. Environmental Protection Agency, Office
27 of Research and Development, Office of Science Policy, http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf.
28 U.S. EPA. (U.S. Environmental Protection Agency). (2000c) Supplementary guidance for conducting health risk
29 assessment of chemical mixtures. (EPA/630/R-00/002). Washington, DC: U.S. Environmental Protection
30 Agency, Risk Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=20533.
31 U.S. EPA. (U.S. Environmental Protection Agency). (2000d) Benchmark dose technical guidance document
32 [external review draft]. (EPA/630/R-00/001). Washington, DC.
33 http://www.epa.gov/raf/publications^enchmark-dose-doc-draft.htm.
34 U.S. EPA. (U.S. Environmental Protection Agency). (2002a). Toxic Substances Control Act (TSCA) Inventory
35 Update Database, from http://www.epa.gov/iur/
36 U.S. EPA. (U.S. Environmental Protection Agency). (2002b) A review of the reference dose and reference
37 concentration processes. (EPA/630/P-02/0002F). Washington, DC: U.S. Environmental Protection Agency,
38 Risk Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=51717.
39 U.S. EPA. (U.S. Environmental Protection Agency). (2002c) A review of the reference dose and reference
40 concentration processes. (EPA/630/P-02/002F). Washington, DC: Risk Assessment Forum, U.S.
41 Environmental Protection Agency. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=51717.
42 U.S. EPA. (U.S. Environmental Protection Agency). (2005a) Guidelines for carcinogen risk assessment.
43 (EPA/630/P-03/001F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
44 http://www.epa.gov/cancerguidelines/.
45 U.S. EPA. (U.S. Environmental Protection Agency). (2005b) Guidelines for carcinogen risk assessment, final report.
46 (EPA/630/P-03/001F). Washington, DC: U.S. Environmental Protection Agency, Risk Assessment Forum.
47 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid= 116283.
48 U.S. EPA. (U.S. Environmental Protection Agency). (2005c) Supplemental guidance for assessing susceptibility
49 from early-life exposure to carcinogens. (EPA/630/R-03/003F). Washington, DC: U.S. Environmental
50 Protection Agency, Risk Assessment Forum, http://www.epa.gov/cancerguidelines/guidelines-carcinogen-
51 supplementhtm.
52 U.S. EPA. (U.S. Environmental Protection Agency). (2006a) Peer review handbook (3rd edition). (EPA/100/B-
53 06/002). Washington, DC. http://www.epa.gov/peerreview/.
54 U.S. EPA. (U.S. Environmental Protection Agency). (2006b) Peer review handbook (3rd edition). (EPA/100/B-
55 06/002). Washington, DC: U.S. Environmental Protection Agency, Science Policy Council.
56 http://www.epa.gov/peerreview/pdfs/peer_review_handbook_2006.pdf.
151
DRAFT - DO NOT CITE OR QUOTE
-------
1 U.S. EPA. (U.S. Environmental Protection Agency). (2006c) A framework for assessing health risk of
2 environmental exposures to children. (EPA/600/R-05/093F). Washington, DC: U.S. Environmental
3 Protection Agency, National Center for Environmental Assessment.
4 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=158363.
5 U.S. EPA. (U.S. Environmental Protection Agency). (2009a) Toxicological review of 1,4-dioxane (CAS No. 123-
6 91-1) in support of summary information on the Intergrated Risk Information System (IRIS) [External
7 Review Draft]. (EPA/635/R-09/005). Washington, DC.
8 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=199330.
9 U.S. EPA. (U.S. Environmental Protection Agency). (2009b) Status report: Advances in inhalation dosimetry of
10 gases and vapors with portal of entry effects in the upper respiratory tract. (EPA/600/R-09/072). Research
11 Triangle Park, NC. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=212131.
12 U.S. EPA. (U.S. Environmental Protection Agency). (2009c) Status report: Advances in inhalation dosimetry of
13 gases and vapors with portal of entry effects in the upper respiratory tract. (EPA/600/R-09/072). Research
14 Triangle Park, NC: U.S. Environmental Protection Agency, National Center for Environmental
15 Assessment. http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=212131.
16 U.S. EPA. (U.S. Environmental Protection Agency). (2010a) Toxicological review of 1,4-Dioxane (CAS No. 123-
17 91-1) in support of summary information on the Integrated Risk Information System (IRIS). (EPA-635/R-
18 09-005-F). Washington, DC. http://www.epa.gov/iris/toxreviews/0326tr.pdf.
19 U.S. EPA. (U.S. Environmental Protection Agency). (2010b) Toxicological review of 1,4-Dioxane (CAS No. 123-
20 91-1) in support of summary information on the Integrated Risk Information System (IRIS). (EPA-635/R-
21 09-005-F). Washington, DC. www.epa.gov/iris/toxreviews/0326tr.pdf.
22 U.S. EPA. (U.S. Environmental Protection Agency). (2011) Recommended use of body weight 3/4 as the default
23 method in derivation of the oral reference dose. (EPA/100/R11/0001). Washington, DC.
24 http://www.epa.gov/raf/publications/interspecies-extrapolation.htm.
25 U.S. EPA. (U.S. Environmental Protection Agency). (2012a) Benchmark dose technical guidance. (EPA/100/R-
26 12/001). Washington, DC. http://www.epa.gov/raf/publications/pdfs^enchmark_dose_guidance.pdf.
27 U.S. EPA. (U.S. Environmental Protection Agency). (2012b) Advances in inhalation gas dosimetry for derivation of
28 a reference concentration (rfc) and use in risk assessment. (EPA/600/R-12/044). Washington, DC.
29 http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=244650.
30 Uno. Y.: Takasawa. H.: Miyagawa. M: Inoue. Y.: Murata. T.: Yoshikawa. K. (1994) An in vivo-in vitro replicative
31 DNA synthesis (RDS) test using rat hepatocytes as an early prediction assay for nongenotoxic
32 hepatocarcinogens screening of 22 known positives and 25 noncarcinogens. Mutat Res 320: 189-205.
33 http://dx.doi.org/10.1016/0165-1218(94)90046-9.
34 Valcke. M. and Krishnan. K. (2011) Assessing the impact of the duration and intensity of inhalation exposure on the
35 magnitude of the variability of internal dose metrics in children and adults. Inhal Toxicol 23: 863-877.
36 http://dx.doi.org/10.3109/08958378.2011.609918.
37 van Delft. J.: van Agen. E.: van Breda. S.: Herwijnen. M.: Staal Y.: Kleinjans. J. (2004) Discrimination of
38 genotoxic from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25: 1265-1276.
39 http://dx.doi.org/10.1093/carcin/bghl08.
40 Vieira. I.: Sonnier. M.: Cresteil T. (1996) Developmental expression of CYP2E1 in the human liver:
41 Hypermethylation control of gene expression during the neonatal period. Eur J Biochem 238: 476-483.
42 http://dx.doi.org/10.1111/j. 1432-1033.1996.0476z.x.
43 Ward. J. M.: Uno. H.: Kurata. Y.: Weghorst C. M.: Jang. J. J. (1993) Cell-proliferation not associated with
44 carcinogenesis in rodents and humans. Environ Health Perspect 101: 125-135.
45 http://dx.doi.org/10.2307/3431855.
46 Watanabe. J.: Havashi. S.: Kawajiri. K. (1994) Different regulation and expression of the human CYP2E1 gene due
47 to the Rsal polymorphism in the 5'-flanking region. J Biochem 116: 321-326.
48 Waxman. P.: Pampori. N: Ram. P.: Agrawal A.: Shapiro. B. (1991) Interpulse interval in circulating growth
49 hormone patterns regulates sexually dimorphic expression of hepatic cytochrome P450. PNAS 88: 6868-
50 6872.
51 Wirth. W. and Klimmer. O. (1936) [On the toxicology of organic solvents. 1,4 dioxane (diethylene dioxide)]. Archiv
52 fuer Gewerbepathologie und Gewerbehygiene 17: 192-206.
53 Wolfe. N. and Jeffers. P. (2000). Hydrolysis. In R. S. Boethling & D. Mackay (Eds.), Handbook of Property
54 Estimation Methods for Chemicals: Environmental and Health Sciences (pp. 311-333). Boca Raton, FL:
55 Lewis Publishers.
56 Wolford. S.: Schroer. R.: Gohs. F.: Gallo. P.: Brodeck. M.: Falk. H.: Ruhren. R. (1986) Reference range data base
57 for serum chemistry and hematology values in laboratory animals. J Toxicol Environ Health A 18: 161-
58 188. http://dx.doi.org/10.1080/15287398609530859.
152
DRAFT - DO NOT CITE OR QUOTE
-------
1 Woo. Y.-T.: Arcos. J.: Argus. M: Griffin. G.: K. N. (1977a) Structural identification of p-dioxane-2-one as the
2 major urinary metabolite of p-dioxane. Naunyn Schmiedebergs Arch Pharmacol 299: 283-287.
3 http://dx.doi.org/10.1007/BF00500322.
4 Woo. Y.-T.: Argus. M.: Arcos. J. (1977b) Metabolism in vivo of dioxane: Effect of inducers and inhibitors of
5 hepatic mixed-function oxidases. Biochem Pharmacol 26: 1539-1542. http://dx.doi.org/10.1016/0006-
6 2952(77)90431-2.
7 Woo. Y.-T.: Argus. M.: Arcos. J. (1977c) Tissue and subcellular distribution of 3H-dioxane in the rat and apparent
8 lack of microsome-catalyzed covalent binding in the target tissue. Life Sci21: 1447-1456.
9 http://dx.doi.org/10.1016/0024-3205(77)90199-0.
10 Woo. Y.-T.: Argus. M.: Arcos. J. (1978) Effect of mixed-function oxidase modifiers on metabolism and toxicity of
11 the oncogen dioxane. Cancer Res 38: 1621-1625.
12 Yamamoto. S.: Urano. K.: Koizumi. H.: Wakana. S.: Hioki. K.: Mitsumori. K T. N. (1998a) Validation of
13 transgenic mice carrying the human prototype c-Ha-ras gene as a bioassay model for rapid carcinogenicity
14 testing. Environ Health Perspect 106: 57-69.
15 Yamamoto. S.: Urano. K.: Nomura. T. (1998b) Validation of transgenic mice harboring the human prototype c-Ha-
16 ras gene as a bioassay model for rapid carcinogenicity testing. Toxicol Lett 102-103: 473-478.
17 http://dx.doi.org/10.1016/80378-4274(98)00341-5.
18 Yamamoto. S.: Ohsawa. M.: Nishizawa. T.: Saito. A.: Kasai. T.: Noguchi. T Matsushima. T. (2000) Long-term
19 toxicology study of 1,4-Dioxane in the F344 rats by multiple-route exposure (drinking water and
20 inhalation). J Toxicol Sci 25: 347.
21 Yamazakj K.: Ohno. H.: Asakura. M.: Narumi. A.: Ohbavashi. H.: Fujita. H Sakabe. H. (1994a). Two-year
22 toxicological and carcinogenesis studies of 1,4-dioxane in F344 rats and BDF1 mice. In K. Sumino, S. Sato
23 & N. G. Shinkokai (Eds.), Proceedings: Second Asia-Pacific Symposium on Environmental and
24 Occupational Health 22-24 July, 1993: Kobe (pp. 193-198). Kobe, Japan: Kobe University School of
25 Medicine, International Center for Medical Research.
26 Yamazakj K. (2006) [Correspondence between Kazunori Yamazaki and Julie Stickney].
27 Yamazaki. K.: Ohno. H.: Asakura. M.: Narumi. A.: Ohbavashi. H.: Fujita. H Sakabe. H. (Eds.). (1994b). Two-
28 year toxicological and carcinogenesis studies of 1,4-dioxane in F344 rats and BDF1 mice. Kobe, Japan:
29 Kobe University School of Medicine, International Center for Medical Research.
30 Yant W.: Schrenk. H.: Waite. C.: Patty. F. (1930) Acute response of guinea pigs to vapors of some new commercial
31 organic compounds: VI. Dioxan. Public Health Rep 45: 2023-2032.
32 Yasuhara. A.: Shiraishi. H.: Nishikawa. M.: Yamamoto. T.: Uehiro. T.: Nakasugi O Noma. Y. (1997)
33 Determination of organic components in leachates from hazardous waste disposal sites in Japan by gas
34 chromatography-mass spectrometry. J Chromatogr A 774: 321-332. http://dx.doi.org/10.1016/S0021-
35 9673(97)00078-2.
36 Yasuhara. A.: Tanaka. Y.: Tanabe. A.: Kawata. K.: Katami. T. (2003) Elution of 1,4-dioxane from waste landfill
37 sites. Bull Environ Contain Toxicol 71: 641-647. http://dx.doi.org/10.1007/s00128-003-8917-7.
38 Yoon. J.: Mason. J.: Valencia. R.: Woodruff. R.: Zimmering. S. (1985) Chemical mutagenesis testing in Drosophila.
39 IV. Results of 45 coded compounds tested for the National Toxicology Program. Environ Mutagen 7: 349-
40 367. http://dx.doi.org/10.1002/em.2860070310.
41 Young. J.: Braun. W.: Gehring. P.: Horvath. B.: Daniel R. (1976a) 1,4-Dioxane andbeta-hydroxyethoxyacetic acid
42 excretion in urine of humans exposed to dioxane vapors. Toxicol Appl Pharmacol 38: 643-646.
43 http://dx.doi.org/10.1016/0041-008X(76)90195-2.
44 Young. J.: Braun. W.: Rampy. L.: Chenoweth. M.: Blau. G. (1977a) Pharmacokinetics of 1,4-dioxane in humans. J
45 Toxicol Environ Health 3: 507-520. http://dx.doi.org/10.1080/15287397709529583.
46 Young. J.: Braun. W.: Gehring. P. (1978a) Dose—dependent fate of 1,4-dioxane in rats(b). J Toxicol Environ
47 Health A 4: 709-726. http://dx.doi.org/10.1080/15287397809529693.
48 Young. J.: Braun. W.: Gehring. P. (1978b) The dose-dependent fate of 1,4-dioxane in rats. J Environ Pathol Toxicol
49 2: 263-282. http://dx.doi.org/10.1080/15287397809529693.
50 Young. J. P.: Braun. W. H.: Gehring. P. J.: Horvath. B. S.: Daniel. R. L. (1976b) 1,4-Dioxane andbeta-
51 hydroxyethoxyacetic acid excretion in urine of humans exposed to dioxane vapors. Toxicol Appl
52 Pharmacol 38: 643-646. http://dx.doi.org/10.1016/0041-008X(76)90195-2.
53 Young. J. P.: Braun. W. H.: Rampy. L. W.: Chenoweth. M. B.: Blau. G. E. (1977b) Pharmacokinetics of 1,4-
54 dioxane in humans. J Toxicol Environ Health 3: 507-520. http://dx.doi.org/10.1080/15287397709529583.
55 Young. J. P.: Braun. W. H.: Gehring. P. J. (1978c) The dose-dependent fate of 1,4-dioxane in rats. J Environ Pathol
56 Toxicol 2: 263-282. http://dx.doi.org/10.1080/15287397809529693.
57 Young. J. P.: Braun. W. H.: Gehring. P. J. (1978d) Dose-dependentfate of 1,4-dioxane in rats(b). J Toxicol Environ
58 Health A 4: 709-726. http://dx.doi.org/10.1080/15287397809529693.
153
DRAFT - DO NOT CITE OR QUOTE
-------
1 Zimmermann. F.: Mayer. V.: Scheel. I.: Resnick. M. (1985a) Acetone, methyl ethyl ketone, ethyl acetate,
2 acetonitrile and other polar aprotic solvents are strong inducers of aneuploidy in Saccharomyces cerevisiae.
3 MutatRes 149: 339-351. http://dx.doi.org/10.1016/0027-5107(85)90150-2.
4 Zimmermann. F. K.: Mayer. V. W.: Scheel. I.: Resnick. M. A. (1985b) Acetone, methyl ethyl ketone, ethyl acetate,
5 acetonitrile and other polar aprotic solvents are strong inducers of aneuploidy in Saccharomyces cerevisiae.
6 MutatRes 149: 339-351. http://dx.doi.org/10.1016/0027-5107(85)90150-2.
154
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX A. SUMMARY OF EXTERNAL PEER
REVIEW AND PUBLIC COMMENTS AND
DISPOSITION
1 The Toxicological Review of 1,4-Dioxane has undergone two formal external peer reviews
2 performed by scientists in accordance with EPA guidance on peer review (U.S. EPA. 2006a. 2000b). The
3 first peer review focused on the toxicity following oral exposure to 1,4-dioxane. For completeness, the
4 inhalation data were added to the assessment and the combined document was submitted for a second
5 peer review and public comment - with a request for reviewers to focus on the inhalation portion of the
6 assessment.
7 The external peer reviewers were tasked with providing written answers to general questions on
8 the overall assessment and on chemical-specific questions in areas of scientific controversy or
9 uncertainty. A summary of significant comments made by the external reviewers and EPA's responses to
10 these comments arranged by charge question follow for both the oral assessment and inhalation update. In
11 many cases the comments of the individual reviewers have been synthesized and paraphrased for
12 development of Appendix A. The majority of the specific observations (in addition to EPA's charge
13 questions) made by the peer reviewers were incorporated into the document and are not discussed further
14 in this Appendix. Public comments that were received are summarized and addressed following the
15 peer-reviewers' comments and disposition for both the oral assessment and inhalation update.
A.1 External Peer Review Panel Comments -- Oral Assessment
16 The reviewers made several editorial suggestions to clarify portions of the text. These changes
17 were incorporated in the document as appropriate and are not discussed further.
18 In addition, the external peer reviewers commented on decisions and analyses in the
19 Toxicological Review of 1,4-Dioxane under multiple charge questions, and these comments were
20 organized and summarized under the most appropriate charge question.
A.1.1 General Charge Questions
21 1. Is the Toxicological Review logical, clear and concise? Has EPA accurately, clearly and objectively
22 represented and synthesized the scientific evidence for noncancer and cancer hazards?
23 Comment. All reviewers found the Toxicological Review to be logical, clear, and concise. One
24 reviewer remarked that it was an accurate, open-minded and balanced analysis of the literature.
25 Most reviewers found that the scientific evidence was presented objectively and transparently;
26 however, one reviewer suggested two things to improve the objectivity and transparency (1)
27 provide a clear description of the mode of action and how it feeds into the choice of the
A-l
DRAFT - DO NOT CITE OR QUOTE
-------
1 extrapolation for the cancer endpoint and (2) provide a presentation of the outcome if internal
2 dose was used in the cancer and noncancer assessments.
3 One reviewer commented that conclusions could not be evaluated in a few places where dose
4 information was not provided (Sections 3.2, 3.3 and 4.5.2.2). The same reviewer found the MOA
5 schematics, key event temporal sequence/dose-response table, and the POD plots to be very
6 helpful in following the logic employed in the assessment.
7 Response. The mode of action analysis and how conclusions from that analysis fed into the
8 choice of extrapolation method for the cancer assessment are discussed further under charge
9 questions C2 and C5. Because of the decision not to utilize the PBPK models, internal doses were
10 not calculated and thus were not included as alternatives to using the external dose as the POD for
11 the cancer and noncancer assessments.
12 In the sections noted by the reviewer (3.2, 3.3 and 4.5.2.2) dose information was added as
13 available. In Section 3.2, Mikheev et al. (1990) did not report actual doses, which is noted in this
14 section. All other dose information in this section was found to be present after further review by
15 the Agency. In Section 3.3, dose information for Woo et al. (1978. 1977b) was added to the
16 paragraph. In Section 4.5.2.2, study details for Nannelli et al. (2005a) were provided earlier in
17 Section 3.3 and a statement referring the reader to this section was added.
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.
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.
8 Response. The references (a-b) above will be evaluated for derivation of an RfC and IUR, which
9 will follow as an update to this oral assessment. References (c) and (e) noted above were
10 considered during development of this assessment as to the value they added to the cancer and
11 noncancer analyses. Reference (g) listed above is an abstract from conference proceedings from
12 the 27th Annual Meeting of the Japanese Society of Toxicology; abstracts are not generally
13 considered in the development of an IRIS assessment. Reference (d) reviews EPA's current risk
14 assessment procedures and provides no specific information regarding 1,4-dioxane. The Stickney
15 et al. (2003) reference was a review article and no new data were presented, thus it was not
16 referenced in this Toxicological Review but the data were considered during the development of
17 this assessment.
18 Following external peer review (as noted above) Kano et al. (2009) was added to the assessment,
19 which was an update and peer-reviewed published manuscript of the JBRC (1998) report.
20 3. Please discuss research that you think would be likely to increase confidence in the database for
21 future assessments of 1,4-dioxane.
22 Comment. All reviewers provided suggestions for additional research that would strengthen the
23 assessment and reduce uncertainty in several areas. The following is a brief list of questions that
24 were identified that could benefit from further research. What are the mechanisms responsible for
25 the acute and chronic nephrotoxicity? Is the acute kidney injury (AKI) multifactorial? Are there
26 both tubular and glomerular/vascular toxicities that result in cortical tubule degeneration and
27 evidence for glomerulonephritis? What are the functional correlates of the histologic changes in
28 terms of assessment of renal function? What is the exposure in utero and risk to the fetus and
29 newborn? What are the concentrations in breast milk following maternal exposure to
30 1,4-dioxane? What is the risk for use of contaminated drinking water to reconstitute infant
31 formula? What are the exposures during early human development? What is the pharmacokinetic
32 and metabolic profile of 1,4-dioxane during development? What are the susceptible populations
33 (e-g-, individuals with decreased renal function or chronic renal disease, obese individuals,
34 gender, age)?
35 Additional suggestions for future research include: evaluation of potential epigenetic mechanisms
36 of carcinogenicity, additional information on sources of exposure and biological concentrations as
37 well as human toxicokinetic data for derivation of parameter to refine PBPK model, studies to
A-3
DRAFT - DO NOT CITE OR QUOTE
-------
1 determine toxic moiety, focused studies to inform mode of action, additional inhalation studies
2 and a multigeneration reproductive toxicity study.
3 One reviewer suggested additional analyses of the existing data including a combined analysis of
4 the multiple datasets and outcomes for cancer and non-cancer endpoints, evaluation of the dose
5 metrics relevant to the MOA to improve confidence in extrapolation approach and uncertainty
6 factors, and complete a Bayesian analysis of human pharmacokinetic data to estimate human
7 variability in key determinants of toxicity (e.g., metabolic rates and partition coefficients).
8 Response. A number of research suggestions were provided for further research that may enhance
9 future health assessments of 1,4-dioxane. Regarding the suggested additional analyses for the
10 existing data, EPA did not identify a MOA in this assessment, thus combined analysis of the
11 cancer and non-cancer endpoints as well as application of various dose metrics to a MOA is not
12 applicable. Because the human PBPK model was not implemented in this assessment for oral
13 exposure to 1,4-dioxane a Bayesian analysis was not completed. No additional changes to the
14 Toxicological Review of 1,4-Dioxane were made in response to these research recommendations.
15 4. Please comment on the identification and characterization of sources of uncertainty in Sections 5 and
16 6 of the assessment document. Please comment on whether the key sources of uncertainty have been
17 adequately discussed. Have the choices and assumptions made in the discussion of uncertainty been
18 transparently and objectively described? Has the impact of the uncertainty on the assessment been
19 transparently and objectively described?
20 Comment. Six reviewers stated Sections 5 and 6 adequately discussed and characterized
21 uncertainty, in a succinct, and transparent manner. One reviewer suggested adding additional
22 discussion of uncertainty relating to the critical study used in the cancer assessment and another
23 reviewer suggested adding more discussion around the uncertainty of the toxic moiety.
24 One reviewer made specific comments on uncertainty surrounding the Kociba et al. (1974a) study
25 as used for derivation of the RfD, choice of the non-cancer dose metric, and use of a 10%BMR as
26 the basis for the CSF derivation. These comments and responses are summarized below under
27 their appropriate charge question.
28 Response. The majority of the reviewers thought the amount of uncertainty discussion was
29 appropriate. Since the external review, Kano et al. (2009) was published and this assessment was
30 updated accordingly (previously JBRC (1998). It is assumed the uncertainty referred to by the
31 reviewer was addressed by the published Kano et al. (2009) paper.
32 Clarification regarding the uncertainty surrounding the identification of the toxic moiety was
33 added to Section 4.6.2.1 stating that the mechanism by which 1,4-dioxane induces tissue damage
34 is not known, nor is it known whether the toxic moiety is 1,4-dioxane or a metabolite of
35 1,4-dioxane. Additional text was added to Section 4.7.3 clarifying that available data also do not
36 clearly identify whether 1,4-dioxane or one of its metabolites is responsible for the observed
A-4
DRAFT - DO NOT CITE OR QUOTE
-------
1 effects. The impact of the lack of evidence to clearly identify a toxic moiety related to
2 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
3 1. A chronic RfD for 1,4-dioxane has been derived from a 2-year drinking water study (Kociba et al..
4 1974a) in rats and mice. Please comment on whether the selection of this study as the principal study
5 has been scientifically justified. Has the selection of this study been transparently and objectively
6 described in the document? Are the criteria and rationale for this selection transparently and
7 objectively described in the document? Please identify and provide the rationale for any other studies
8 that should be selected as the principal study.
9 Comment. Seven of the reviewers agreed that the use of the Kociba et al. (1974a) study was the
10 best choice for the principal study.
11 One reviewer stated that Kociba et al. (1974a) was not the best choice because it reported only
12 NOAEL and LOAELs without providing incidence data for the endpoints. This reviewer also
13 stated that the study should not have been selected based on sensitivity of the endpoints, but
14 rather study design and adequacy of reporting of the study results. Additionally, this reviewer
15 suggested a better principal study would be either the NCI (1978) or JBRC (1998) study.
16 Response. The reviewer is correct that Kociba et al. (1974a) did not provide incidence data;
17 however, Kociba et al. (1974a) identified a NOAEL (9.6 mg/kg-day) and LOAEL (94 mg/kg-day)
18 within the text of the manuscript. Kociba et al. (1974a) was a well conducted chronic bioassay
19 (four dose levels, including controls, with 60 rats/sex/group) and seven of the peer reviewers
20 found this study to be appropriate as the basis for the RfD. Further support for the selection of the
21 Kociba et al. (1974a) as the principal study comes from comparison of the liver and kidney
22 toxicity data reported by JBRC (1998) and NCI (1978). which was presented in Section 5.1. The
23 effects reported by JBRC (1998) and NCI (1978) were consistent with what was observed by
24 Kociba et al. (1974a) and within a similar dose range. Derivation of an RfD from these datasets
25 resulted in a similar value (Section 5.1.).
26 2. Degenerative liver and kidney effects were selected as the critical effect. Please comment on whether
27 the rationale for the selection of this critical effect has been scientifically justified. Are the criteria and
28 rationale for this selection transparently and objectively described in the document? Please provide a
29 detailed explanation. Please comment on whether EPA's rationale regarding adversity of the critical
30 effect for the RfD has been adequately and transparently described and is scientifically supported by
31 the available data. Please identify and provide the rationale for any other endpoints that should be
32 considered in the selection of the critical effect.
33 Comment. Five of the reviewers agreed with the selection of liver and kidney effects as the
34 critical effect. One of these reviewers suggested analyzing all datasets following dose adjustment
A-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 (e.g., body weight scaling or PBPK model based) to provide a better rationale for selection of a
2 critical effect.
3 One reviewer stated that 1,4-dioxane causing liver and kidney organ specific effects is logical;
4 however, with regards to nephrotoxicity, the models and limited human data have not addressed
5 the mechanisms of injury or the clinical correlates to the histologic data. Also, advances in the
6 field of biomarkers have not yet been used for the study of 1,4-dioxane.
7 One reviewer found the selection of these endpoints to be 'without merit' because of the lack of
8 incidence data to justify the NOAEL and LOAEL values identified in the study. This reviewer
9 suggested selecting the most sensitive endpoint(s) from the NCI (NCI. 1978) or JBRC (1998)
10 studies for the basis of the RfD, but did not provide a suggestion as to what effect should be
11 selected.
12 Response. The liver and kidney effects from Kociba et al. (1974a) was supported as the critical
13 effect by most of the reviewers. PBPK model adjustment was not performed because the PBPK
14 model was found to be inadequate for use in the assessment. EPA acknowledges that neither the
15 mechanisms of injury nor the clinical correlates to histologic data exist for 1,4-dioxane. This type
16 of information could improve future health assessments of 1,4-dioxane.
17 As stated above, Kociba et al. (1974a) identified a NOAEL (9.6 mg/kg-day) and LOAEL
18 (94 mg/kg-day) within the text of the manuscript and was a well conducted chronic bioassay (four
19 dose levels, including controls, with 60 rats/sex/group).
20 3. Kociba et al. (1974a) derived a NOAEL based upon the observation of degenerative liver and kidney
21 effects and these data were utilized to derive the point of departure (POD) for the RfD. Please provide
22 comments with regard to whether the NOAEL approach is the best approach for determining the
23 POD. Has the approach been appropriately conducted and objectively and transparently described?
24 Please identify and provide rationales for any alternative approaches for the determination of the POD
25 and discuss whether such approaches are preferred to EPA's approach.
26 Comment. Seven reviewers agreed with the NOAEL approach described in the document. One of
27 these reviewers also questioned whether any attempt was made to "semi-qualitatively represent
28 the histopathological observations to facilitate a quantitative analysis".
29 One reviewer stated that data were not used to derive the POD, but rather a claim by the authors
30 of Kociba et al. (1974a) of the NOAEL and LOAEL for the endpoints. This reviewer preferred
31 the use of a BMD approach for which data include the reported incidence rather than a study
32 reported NOAEL or LOAEL.
33 Response. The suggestion to "semi-qualitatively represent the histopathological observations to
34 facilitate a quantitative analysis" was not incorporated into the document because it is unclear
35 how this would be conducted since Kociba et al. (1974a) did not provide incidence data and the
36 reviewer did not illustrate their suggested approach. See responses to Bl and B2 regarding the
37 NOAEL and LOAEL approach. The Agency agrees that a Benchmark Dose approach is preferred
A-6
DRAFT - DO NOT CITE OR QUOTE
-------
1 over the use of a NOAEL or LOAEL for the POD if suitable data (e.g., reflecting the most
2 sensitive sex, species, and endpoint identified) are available for modeling and, if suitable data are
3 not available, then NOAEL and LOAEL values are utilized. In this case, the data were not
4 suitable for BMD modeling and the LOAEL or NOAEL approach was used.
5 4. EPA evaluated the PBPK and empirical models available to describe kinetics following inhalation of
6 1,4-dioxane (Reitzetal.. 1990a: Young etal.. 1978a: Young etal.. 1978b: Young etal.. 1977a). EPA
7 concluded that the use of existing, revised, and recalibrated PBPK models for 1,4-dioxane were not
8 superior to default approaches for the dose-extrapolation between species. Please comment on
9 whether EPA's rationale regarding the decision to not utilize existing or revised PBPK models has
10 been adequately and transparently described and is supported by the available data. Please identify
11 and provide the rationale for any alternative approaches that should be considered or preferred to the
12 approach presented in the toxicological review.
13 Comment. Six reviewers found the decision not to utilize the available PBPK models to be
14 appropriate and supported by available data. One of these reviewers suggested presenting as part
15 of the uncertainty evaluation an adjustment of the experimental doses based on metabolic
16 saturation. Another reviewer stated Appendix B was hard to follow and that the main document
17 should include a more complete description of the model refinement effort performed by
18 Sweeney et al. (2008a).
19 Two reviewers noted a complete evaluation of the models was evident; one of the reviewers
20 questioned the decision not to use the models on the basis that they were unable to fit the human
21 blood PK data for 1,4-dioxane. This reviewer suggested the rat model might fit the human blood
22 PK data, thus raising concern in the reliance on the human blood PK data to evaluate the PBPK
23 model for 1,4-dioxane. Instead, the reviewer suggested the human urinary metabolite data may be
24 sufficient to give confidence in the model. One other reviewer also questioned the accuracy of the
25 available human data. One reviewer commented that the rationale for not using the PBPK model
26 to extrapolate from high to low dose was questioned. In addition, the reviewer suggested that two
27 aspects of the model code for Reitz et al. (1990a) need to be verified:
28 a. In the document, KLC is defined as a first-order rate constant and is scaled by BW°7.
29 This is inconsistent when multiplied by concentration does not result in units
30 of mg/hr. However, if the parameter is actually considered a clearance constant
31 (zero-order rate constant) then the scaling rule used, as well as the interpretations
32 provided, would be acceptable.
33 b. It is unclear as to why AM is calculated on the basis of RAM and not RMEX. RMEX
34 seems to represent the amount metabolized per unit time.
35 Response. The U.S. EPA performed a rigorous evaluation of the PBPK models available for
36 1,4-dioxane. This effort was extensively described in Section 3.5 and in Appendix B. In short,
37 several procedures were applied to the human PBPK model to determine if an adequate fit of the
38 model to the empirical model output or experimental observations could be attained using
A-7
DRAFT - DO NOT CITE OR QUOTE
-------
1 biologically plausible values for the model parameters. The re-calibrated model predictions for
2 blood 1,4-dioxane levels did not come within 10-fold of the experimental values using measured
3 tissue:air partition coefficients of (Leung and Paustenbach. 1990a) or (Sweeney et al.. 2008a) (
4 and Figure B-6). The utilization of a slowly perfused tissue:air partition coefficient 10-fold lower
5 than measured values produces exposure-phase predictions that are much closer to observations,
6 but does not replicate the elimination kinetics (Figure B-7). Re-calibration of the model with
7 upper bounds on the tissue:air partition coefficients results in predictions that are still six- to
8 sevenfold lower than empirical model prediction or observations (Figure B-9 and Figure B-10).
9 Exploration of the model space using an assumption of first-order metabolism (valid for the
10 50 ppm inhalation exposure) showed that an adequate fit to the exposure and elimination data can
11 be achieved only when unrealistically low values are assumed for the slowly perfused tissue:air
12 partition coefficient (Figure B-13). Artificially low values for the other tissue:air partition
13 coefficients are not expected to improve the model fit, as these parameters are shown in the
14 sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km. In the
15 absence of actual measurements for the human slowly perfused tissue:air partition coefficient,
16 high uncertainty exists for this model parameter value. Differences in the ability of rat and human
17 blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this is expected to be
18 evident in very different values for rat and human blood:air partition coefficients, which is not the
19 case (). Therefore, some other, as yet unknown, modification to model structure may be
20 necessary.
21 The results of U.S. EPA model evaluation were confirmed by other investigators (Sweeney et al..
22 2008a). Sweeney et al. (2008a) concluded that the available PBPK model with refinements
23 resulted in an under-prediction of human blood levels for 1,4-dioxane by six- to seven fold. It is
24 anticipated that the high uncertainty in predictions of the PBPK model for 1,4-dioxane would not
25 result in a more accurate derivation of human health toxicity values.
26 Because it is unknown whether the parent or the metabolite is the toxic moiety, analyses were not
27 conducted to adjust the experimental doses on the basis of metabolic saturation.
28 The discussion of Sweeney et al. (2008a) was expanded in the main document in Section 3.5.3. In
29 the absence of evidence to the contrary, the Agency cannot discount the human blood kinetic data
30 published by Young et al. (1977a). Even though the PBPK model provided satisfactory fits to the
31 rodent kinetic data, it was not used to extrapolate from high dose to low dose in the animal
32 because an internal dose metric was not identified and external doses were utilized in derivation
33 of the toxicity values.
34 KLC was implemented by the U.S. EPA during the evaluation of the model and should have been
35 described as a clearance constant (first-order rate constant) with units of L/hr/kg070. These
36 corrections have been made in the document; however, this does not impact the model predictions
37 because it was in reference to the terminology used to describe this constant.
38 The reviewer is correct that RMEX is the rate of metabolism of 1,4-dioxane per unit time;
39 however an amount of 1,4-dioxane metabolized was not calculated in the Reitz et al. (1990a)
A-8
DRAFT - DO NOT CITE OR QUOTE
-------
1 model code. Thus, AM is the amount of the metabolite (i.e., HEAA) in the body rather than the
2 amount metabolized of 1,4-dioxane. RAM was published by Reitz et al. (1990a) as equation 2 for
3 the change in the amount of metabolite in the body per unit time. AMEX is the amount of the
4 metabolite excreted in the urine. While the variables used are confusing, the code describes the
5 metabolism of 1,4-dioxane as published in the manuscripts. The comments in the model code
6 were updated to make this description more clear (Appendix B).
7 5. Please comment on the selection of the uncertainty factors applied to the POD for the derivation of
8 the RfD. For instance, are they scientifically justified and transparently and objectively described in
9 the document? If changes to the selected uncertainty factors are proposed, please identify and provide
10 a rationale(s). Please comment specifically on the following uncertainty factors:
11 • An interspecies uncertainty factor of 10 was used to account for uncertainties in extrapolating
12 from laboratory animals to humans because a PBPK model to support interspecies extrapolation
13 was not suitable.
14 • An intraspecies (human variability) uncertainty factor of 10 was applied in deriving the RfD
15 because the available information on the variability in human response to 1,4-dioxane is
16 considered insufficient to move away from the default uncertainty factor of 10.
17 • A database uncertainty factor of 3 was used to account for lack of adequate reproductive
18 toxicity data for 1,4-dioxane, and in particular absence of a multigeneration reproductive
19 toxicity study. Has the rationale for the selection of these uncertainty factors been transparently
20 and objectively described in the document? Please comment on whether the application of these
21 uncertainty factors has been scientifically justified.
22 Comment. One reviewer noted the uncertainty factors appear to be the standard default choices
23 and had no alternatives to suggest.
24 o Five reviewers agreed that the use of an uncertainty factor of 10 for the interspecies
25 extrapolation is fully supportable. One reviewer suggested using BW3/4 scaling rather than
26 an uncertainty factor of 10 for animal to human extrapolation. Along the same lines, one
27 reviewer suggested a steady-state quantitative analysis to determine the importance of
28 pulmonary clearance and hepatic clearance and stated that if hepatic clearance scales to
29 body surface and pulmonary clearance is negligible, then an adjusted uncertainty factor
30 based on body surface scaling would be more appropriate.
31 o Seven reviewers stated that the uncertainty factor of 10 for interindividual variability
32 (intraspecies) is fully supportable.
33 o Six reviewers commented that the uncertainty factor of 3 for database deficiencies is fully
34 justifiable. One reviewer suggested adding text to clearly articulate the science policy for
35 the use of a factor of 3 for database deficiencies.
A-9
DRAFT - DO NOT CITE OR QUOTE
-------
1 Response. The preferred approach to interspecies scaling is the use of a PBPK model; however,
2 the PBPK models available for 1,4-dioxane are not suitable for use in this health assessment as
3 outlined elsewhere. Another approach that has been commonly implemented in the cancer
4 assessments is the use of body weight scaling based on body surface area (BW3/4 scaling). It is not
5 standard practice to apply BW3/4 scaling in noncancer assessments at this time. The current
6 default approach used by the Agency when PBPK models are not available for extrapolation is
7 the application of an UFA of 10, which was implemented in this assessment.
8 The absence of a multigenerational reproductive study is why the uncertainty factor for database
9 deficiencies (UFD) was retained; however, it was reduced from 10 to 3. In the text in Section
10 5.1.3 text was included to clearly state that because of the absence of a multigenerational
11 reproductive study for 1,4-dioxane an uncertainty factor of 3 was used for database deficiencies.
12 No other changes regarding the use of the uncertainty factors were made to the document.
A.1.3 Carcinogenicity of 1,4-dioxane and derivation of an oral slope factor
13 1. Under the EPA's 2005 Guidelines for Carcinogen Risk Assessment (www.epa.gov/iris/backgr-d.htm),
14 the Agency concluded that 1,4-dioxane is likely to be carcinogenic to humans. Please comment on the
15 cancer weight of evidence characterization. Has the scientific justification for the weight of evidence
16 descriptor been sufficiently, transparently and objectively described? Do the available data for both
17 liver tumors in rats and mice and nasal, mammary, and peritoneal tumors in rats support the
18 conclusion that 1,4-dioxane is a likely human carcinogen?
19 Comment. All reviewers agreed with the Agency's conclusion that 1,4-dioxane is "likely to be
20 carcinogenic to humans". However, two reviewers also thought 1,4-dioxane could be categorized
21 as a potential human carcinogen, since low-dose environmental exposures would be unlikely to
22 result in cancer. One reviewer also suggested providing a brief recapitulation of the guidance
23 provided by the 2005 Guidelines for Carcinogen Risk Assessment regarding classification of a
24 compound as likely to be carcinogenic to humans and how a chemical falls into this category.
25 Response. The document includes a weight-of-evidence approach to categorize the carcinogenic
26 potential of 1,4-dioxane. This was included in Section 4.7.1 based upon U.S. EPA's Guidelines
27 for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-Dioxane can be described as likely to be
28 carcinogenic to humans based on evidence of liver carcinogenicity in several 2-year bioassays
29 conducted in three strains of rats, two strains of mice, and in guinea pigs. Additionally, tumors in
30 other organs and tissues have been observed in rats due to exposure to 1,4-dioxane.
31 2. Evidence indicating the mode of action of carcinogenicity of 1,4-dioxane was considered. Several
32 hypothesized MOAs were evaluated within the Toxicological Review and EPA reached the
33 conclusion that a MOA(s) could not be supported for any tumor types observed in animal models.
34 Please comment on whether the weight of the scientific evidence supports this conclusion. Please
35 comment on whether the rationale for this conclusion has been transparently and objectively
A-10
DRAFT - DO NOT CITE OR QUOTE
-------
1 described. Please comment on data available for 1,4-dioxane that may provide significant biological
2 support for a MOA beyond what has been described in the Toxicological Review. Considerations
3 should include the scientific support regarding the plausibility for the hypothesized MOA(s), and the
4 characterization of uncertainty regarding the MOA(s).
5 Comment. Three reviewers commented that the weight of evidence clearly supported the
6 conclusion that a mode of action could not be identified for any of the tumor sites. One reviewer
7 commented that there is inadequate evidence to support a specific MOA with any confidence and
8 low-dose linear extrapolation is necessary; this reviewer also pointed out that EPA should not rule
9 out a metabolite as the toxic moiety.
10 One reviewer stated this was outside of his/her area of expertise but indicated that the discussion
11 was too superficial and suggested adding statements as to what the Agency would consider
12 essential information to make a determination about a MOA.
13 Two reviewers commented that even though the MOA for 1,4-dioxane is not clear there is
14 substantial evidence that the MOA is non-genotoxic. One of these reviewers also suggested that a
15 nonlinear cancer risk assessment model should be utilized.
16 One reviewer suggested adding more text to the summary statement to fully reflect the available
17 MOA information which should be tied to the conclusion and choice of an extrapolation model.
18 Response. The Agency agrees with the reviewer not to rule out a toxic metabolite as the toxic
19 moiety. In Section 5.5.1.2 text is included relating that there is not enough information to
20 determine whether the parent compound, its metabolite(s), or a combination is responsible for the
21 observed toxicities following exposure to 1,4-dioxane.
22 It is not feasible to describe the exact data that would be necessary to conclude that a particular
23 MOA was operating to induce the tumors observed following 1,4-dioxane exposure. In general,
24 the data would fit the general criteria described in the U.S. EPA's Guidelines for Carcinogen Risk
25 Assessment (U.S. EPA. 2005a). For 1,4-dioxane, several MOA hypotheses have been proposed
26 and are explored for the observed liver tumors in Section 4.7.3. This analysis represents the extent
27 to which data could provide support for any particular MOA.
28 One reviewer suggested that the evidence indicating that 1,4-dioxane is not genotoxic supports a
29 nonlinear approach to low-dose extrapolation. In accordance with the U.S. EPA's Guidelines for
30 Carcinogen Risk Assessment (U.S. EPA. 2005a). the absence of evidence for genotoxicity does
31 not invoke the use of nonlinear low-dose extrapolation, nor does it define a MOA. A nonlinear
32 low-dose extrapolation can be utilized when a MOA supporting a nonlinear dose response is
33 identified. For 1,4-dioxane this is not the case; a cancer MOA for any of the tumor types observed
34 in animal models has not been elucidated. Therefore, as concluded in the Toxicological Review,
35 the application of a nonlinear low-dose extrapolation approach was not supported.
36 Additional text has been added to Section 5.4.3.2 to relay the fact that several reviewers
37 recommended that the MOA data support the use of a nonlinear extrapolation approach to
A-ll
DRAFT - DO NOT CITE OR QUOTE
-------
1 estimate human carcinogenic risk associated with exposure to 1,4-dioxane and that such an
2 approach should be presented in the Toxicological Review. Additional text has also been added to
3 the summary statement in Section 6.2.3 stating that the weight of evidence is inadequate to
4 establish a MOA(s) by which 1,4-dioxane induces peritoneal, mammary, or nasal tumors in rats
5 and liver tumors in rats and mice (see Section 4.7.3 for a more detailed discussion of
6 1,4-dioxane's hypothesized MOAs).
7 3. A two-year drinking water cancer bioassay (JBRC. 1998) was selected as the principal study for the
8 development of an oral slope factor (OSF). Please comment on the appropriateness of the selection of
9 the principal study. Has the rationale for this choice been transparently and objectively described?
10 Comment. Seven reviewers agreed with the choice of the JBRC (1998) study as the principal
11 study for the development of an OSF. However, two reviewers that agreed with the choice of
12 JBRC (1998) also commented on the description and evaluation of the study. One reviewer
13 commented the evaluation of the study should be separated from the evaluation/selection of
14 endpoints within the study. The other reviewer suggested that details on the following aspects
15 should be added to improve transparency of the study: (1) rationale for selection of doses; (2)
16 temporal information on body weight for individual treatment groups; (3) temporal information
17 on mortality rates; and (4) dosing details.
18 One reviewer thought that the complete rationale for selection of the JBRC (1998) study was not
19 provided because there was no indication of whether the study was conducted under GLP
20 conditions, and the study was not peer reviewed or published. This reviewer noted the NCI
21 (1978) study was not appropriate for use, but that the Kociba et al. (1974a) study may have
22 resulted in a lower POD had they employed both sexes of mice and combined benign and
23 malignant tumors.
24 Response. Since the External Peer Review draft of the Toxicological Review of 1,4-Dioxane was
25 released (U.S. EPA. 2009a). the cancer portion of the study conducted by the JBRC laboratory
26 was published in the peer-reviewed literature as Kano et al. (2009). This manuscript was
27 reviewed by EPA. EPA determined that the data published by Kano et al. (2009) should be
28 included in the assessment of 1,4-dioxane for several reasons: (1) while the JBRC (1998) was a
29 detailed laboratory report, it was not peer-reviewed; (2) the JBRC improved the diagnosis of pre-
30 and neoplastic lesions in the liver according to the current diagnostic criteria and submitted the
31 manuscript based on this updated data; (3) the Kano et al. (2009) peer-reviewed manuscript
32 included additional information such as body weight growth curves and means and standard
33 deviations of estimated dose for both rats and mice of both sexes. Thus, the Toxicological Review
34 was updated to reflect the inclusion of the data from Kano et al. (2009). and Appendix E was
35 added for a clear and transparent display of the data included in the multiple reports.
36 In response to the peer reviewers, dose information was updated throughout the assessment and
37 are also provided in detail in Section 4.2.1.2.6, along with temporal information on body weights
38 and mortality. Text was also added to Section 4.2.1.2.6 regarding the choice of high dose
39 selection as included in the Kano et al. (2009) manuscript. Additional discussion regarding the
A-12
DRAFT - DO NOT CITE OR QUOTE
-------
1 mortality rates was also added to Section 5.4.1 in selection of the critical study for the oral cancer
2 assessment. Documentation that the study was conducted in accordance with Organization for
3 Economic Co-operation and Development (OECD) Principles of Good Laboratory Practice
4 (GLP) is provided in the manuscript (Kano et al. 2009) and this was also added to the text in
5 Section 4.2.1.2.6.
6 4. Combined liver tumors (adenomas and carcinomas) in female CjrBDFl mice from the JBRC (1998)
7 study were chosen as the most sensitive species and gender for the derivation of the final OSF. Please
8 comment on the appropriateness of the selections of species and gender. Please comment on whether
9 the rationale for these selections is scientifically justified. Has the rationale for these choices been
10 transparently and objectively described?
11 Comment. Six reviewers agreed the female CjrBDFl mice should be used for the derivation of
12 the OSF. Five of these reviewers agreed with the rationale for the selection of the female
13 CjrBDFl mouse as the most sensitive gender and species. However, one reviewer suggested that
14 the specific rationale (i.e., that the final OSF is determined by selecting the gender/species that
15 gives the greatest OSF value) be stated clearly in a paragraph separate from the other
16 considerations of study selection.
17 One reviewer was unsure of both the scientific justification for combining benign and malignant
18 liver tumors, as well as the background incidence of the observed liver tumors in historical
19 control CjrBDFl male and female mice.
20 One reviewer commented that the scientific basis for the selection of female CjrBDFl mice was
21 unclear. This reviewer thought that the rationale for the choice of this strain/sex compared to all
22 others was not clearly articulated.
23 Response. Using the approach described in the Guidelines for Carcinogen Risk Assessment (U.S.
24 EPA. 2005a) studies were first evaluated based on their quality and suitability for inclusion in the
25 assessment. Once the studies were found to be of sufficient quality for inclusion in the
26 assessment, the dose-response analysis was performed with the goal of determining the most
27 appropriate endpoint and species for use in the derivation of an OSF. These topics are discussed
28 in detail in Section 4.7 and 5.4.
29 Benign and malignant tumors that arise from the same cell type (e.g., hepatocellular) may be
30 combined to more clearly identify the weight of evidence for a chemical. This is in accordance
31 with the U.S. EPA 2005 Guidelines for Carcinogen Risk Assessment as referenced in the
32 Toxicological Review. In the absence of a MOA (MOA analysis described in detail in Section
33 4.7.) for 1,4-dioxane carcinogenicity, it is not possible to determine which species may more
34 closely resemble humans. Text in Section 5.4.4 indicates that the calculation of an OSF for
35 1,4-dioxane is based upon the dose-response data for the most sensitive species and gender.
36 5. Has the scientific justification for deriving a quantitative cancer assessment been transparently and
37 objectively described? Regarding liver cancer, a linear low-dose extrapolation approach was utilized
A-13
DRAFT - DO NOT CITE OR QUOTE
-------
1 to derive the OSF. Please provide detailed comments on whether this approach to dose-response
2 assessment is scientifically sound, appropriately conducted, and objectively and transparently
3 described in the document. Please identify and provide the rationale for any alternative approaches for
4 the determination of the OSF and discuss whether such approaches are preferred to EPA's approach.
5 Comment. Four reviewers agreed with the approach for the dose-response assessment. One
6 reviewer commented that even if a nongenotoxic MOA were identified for 1,4-dioxane it may not
7 be best evaluated by threshold modeling. One reviewer commented the use of the female mouse
8 data provided an appropriate health protective and scientifically valid approach.
9 One reviewer commented that the basic adjustments and extrapolation method for derivation of
10 the OSF were clearly and adequately described, but disagreed with the linear low-dose
11 extrapolation. This reviewer suggested that the lack of certainty regarding the MOA was not a
12 sufficient cause to default to a linear extrapolation. Another reviewer commented that the
13 rationale for a linear low-dose extrapolation to derive the OSF was not clear, but may be in
14 accordance with current Agency policy in the absence of a known MOA. This reviewer also
15 commented that 1,4-dioxane appears to be non-genotoxic and nonlinear models should be tested
16 on the available data to determine if they provide a better fit and are more appropriate.
17 One reviewer thought that the justification for a linear extrapolation was not clearly provided and
18 that a disconnect between the MOA summary and the choice of a linear extrapolation model
19 existed. In addition, this reviewer commented that the pharmacokinetic information did not
20 support the use of a linear extrapolation approach, but rather use of animal PBPK models to
21 extrapolate from high to low dose that would result in a mixture of linear and nonlinear
22 extrapolation models was warranted.
23 One reviewer suggested consideration of an integrated assessment of the cancer and noncancer
24 endpoints; however, if linear low-dose extrapolation remains the approach of choice by the
25 Agency, then the effect of choosing BMRs other than 10% was recommended to at least be
26 included in the uncertainty discussion. Using BMRs lower than 10% may allow for the
27 identification of a risk level for which the low-dose slope is 'best' estimated.
28 Response. The EPA conducted a cancer MOA analysis evaluating all of the available data for
29 1,4-dioxane. Application of the framework in the U.S. EPA Guidelines for Carcinogen Risk
30 Assessment (2005a) demonstrates that the available evidence to support any hypothesized MOA
31 for 1,4-dioxane-induced tumors does not exist. In the absence of a MOA, the U.S. EPA
32 Guidelines for Carcinogen Risk Assessment (2005a) indicate that a low dose linear extrapolation
33 should be utilized for dose response analysis (see Section 5.4). Some of the potential uncertainty
34 associated with this conclusion was characterized in Section 5.5. Note that there is no scientific
35 basis to indicate that in the absence of evidence for genotoxicity a nonlinear low-dose
36 extrapolation should be used. As concluded in the Toxicological Review, the application of a
37 nonlinear low-dose extrapolation approach was not supported.
A-14
DRAFT - DO NOT CITE OR QUOTE
-------
1 With regards to the PBPK model available for 1,4-dioxane, it is clear that there currently exist
2 deficiencies within the model and as such, the model was not utilized for interspecies
3 extrapolation. Given the deficiencies and uncertainty in the 1,4-dioxane model it also does not
4 provide support for a MOA.
5 Lastly, in the absence of a MOA for 1,4-dioxane carcinogenicity it is not possible to harmonize
6 the cancer and noncancer effects to assess the risk of health effects due to exposure. However, the
7 choice of the BMDLi0,which was more than 15-fold lower than the response at the lowest dose
8 (66 mg/kg-day), was reconsidered in response to a public comment. BMDs and BMDLs were
9 calculated using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and BMDL50). A
10 BMR of 50% was used as it resulted in a BMDL closest to the response level at the lowest dose
11 tested in the bioassay.
A.2 Public Comments - Oral Assessment
12 Comments on the Toxicological Review of 1,4-Dioxane submitted by the public for the external
13 peer review of the oral toxicity values are summarized below in the following categories: Oral
14 reference dose for 1,4-dioxane, carcinogenicity of 1,4-dioxane, PBPK modeling, and other
15 comments.
A.2.1 Oral reference dose (RfD) for 1,4-dioxane
16 Comment: An UF for database deficiencies is not necessary because of considerable evidence
17 showing no reproductive or developmental effects from 1,4-dioxane exposure.
18 Response: Due to the lack of a multigenerational reproductive study for 1,4-dioxane an UF of 3
19 was retained for database deficiencies. Without clear evidence showing a lack of reproductive or
20 developmental effects in a multigenerational reproductive study, there is still uncertainty in this
21 area.
A.2.2 Carcinogenicity of 1,4-dioxane
22 Comment: Using liver tumors as the basis for the oral CSF is more appropriate than nasal tumors
23 (1988 IRIS assessment of 1,4-dioxane); however, the use of mouse liver tumor data is
24 inappropriate because it is inconsistent with other liver models both quantitatively and in the
25 dose-response pattern. High mortality rates in the study are also a limitation. Liver tumor data
26 from rats should be used instead, which represents a better animal model for 1,4-dioxane
27 carcinogenicity assessment.
A-15
DRAFT - DO NOT CITE OR QUOTE
-------
1 Response: Even though the dose-response is different for mice and rats, the female mice were
2 considered to be appropriate for the carcinogenicity assessment for several reasons. The female
3 mouse liver tumors from the Kano et al. (2009) report were found to be the most sensitive species
4 and endpoint. Section 4.2.1.2.6 was updated to include additional information on mortality rates.
5 The majority of the animals lived past 52 weeks (only 4 females died prior to 52 weeks, 2 in each
6 the mid- and high-dose groups). The cause of death in the female mice that died between 1 and 2
7 years was attributed to liver tumors.
8 Comment: The OSF was based on the most sensitive group, Crj:BDFl mice; however BDF1
9 mice have a high background rate of liver tumors. The incidence of liver tumors in historical
10 controls for this gender/species should be considered in the assessment. Sensitivity of the test
11 species/gender as well as other criteria should be considered in the selection of the appropriate
12 study, including internal and external validity as outlined in Lewandowski and Rhomberg (2005).
13 The female Crj:BDFl mice had a low survival rate that should be considered in the selection of
14 the animal model for 1,4-dioxane carcinogenicity.
15 Response. Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and
16 carcinomas in control male and female BDF1 mice from ten 2-year bioassays at the JBRC. For
17 female mice, out of 499 control mice, the incidence rates were 4.4% for hepatocellular adenomas
18 and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported a 10% incidence rate for
19 hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in control female
20 BDF1. These incidence rates are near the historical control values and thus are appropriate for
21 consideration in this assessment. Additional text regarding these historical controls was added to
22 the study description in Section 4.2.1.2.6.
23 Comment: Low-dose linear extrapolation for the oral CSF is not appropriate nor justified by the
24 data. The weight of evidence supports a threshold (nonlinear) MOA when metabolic pathway is
25 saturated at high doses. Nonlinear extrapolations should be evaluated and presented for
26 1,4-dioxane. Oral CSFs should be derived and presented using both the BW3/4 scaling as well as
27 available PBPK models to extrapolate across species.
28 Response: The absence of evidence for genotoxicity/mutagenicity does not indicate the use of
29 nonlinear low-dose extrapolation. For 1,4-dioxane, a MOA to explain the induction of tumors
30 does not exist so the nature of the low-dose region of the dose-response is unknown. The oral
31 CSF for 1,4-dioxane was derived using BW3/4 scaling for interspecies extrapolation. The PBPK
32 and empirical models available for 1,4-dioxane were evaluated and found not to be adequate for
33 use in this assessment, described in detail in Appendix B.
34 Comment: The POD for the BDF1 female mouse is 15-fold lower than the lowest dose in the
35 bioassay, thus the POD is far below the lower limit of the data and does not follow the U.S.
36 EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a).
37 Response. The comment is correct that the animal BMDLio was more than 15-fold lower than the
38 response at the lowest dose (66 mg/kg-day) in the bioassay. BMDs and BMDLs were calculated
A-16
DRAFT - DO NOT CITE OR QUOTE
-------
1 using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and BMDL50). A BMR of
2 50% was chosen as it resulted in a BMDL closest to the response level at the lowest dose tested in
3 the bioassay.
4 Comment. The geometric mean of the oral cancer slope factors (as done with B[a]P & DDT)
5 should have been used instead of relying on the female BDF1 mouse data, since a MOA could not
6 be determined for 1,4-dioxane.
7 Response. In accordance with the BMD Technical Guidance Document (U.S. EPA. 2012a)
8 averaging tumor incidence is not a standard or default approach. Averaging the tumor incidence
9 response diminishes the effect seen in the sensitive species/gender.
10 Comment. EPA should critically reexamine the choice of JBRC (1998) as the principal study
11 since it has not been published or peer-reviewed. A transcript of e-mail correspondence should be
12 provided.
13 Response. JBRC (1998) was published as conference proceedings as Yamazaki et al. (1994a) and
14 recently in the peer-reviewed literature as Kano et al. (2009). Additional study information was
15 also gathered from the authors (Yamazaki. 2006) and is available upon request from the IRIS
16 Hotline. The peer-reviewed and published data from Kano et al. (2009) was incorporated into the
17 final version of the Toxicological Review of 1,4-Dioxane.
18 Comment. The WOE does not support a cancer descriptor of likely to be carcinogenic to humans
19 determination, but rather suggestive human carcinogen at the high dose levels used in rodent
20 studies seems more appropriate for the following reasons: 1) lack of conclusive human
21 epidemiological data; 2) 1,4-dioxane is not mutagenic; and 3) evidence at high doses it would act
22 via cell proliferation MOA.
23 Response: A cancer classification of "likely, " based on evidence of liver carcinogenicity in
24 several two-year bioassays conducted in three strains of rats, two strains of mice, and in guinea
25 pigs was chosen. Also, mesotheliomas of the peritoneum, mammary, and nasal tumors have been
26 observed in rats. The Agency agrees that human epidemiological studies are inconclusive. The
27 evidence at any dose is insufficient to determine a MOA.
A.2.3 PBPK Modeling
28 Comment. EPA should have used and considered PBPK models to derive the oral toxicity values
29 (rat to human extrapolation) rather than relying on a default method. The draft did not consider
30 the Sweeney et al. (2008a) model. The PBPK model should be used for both noncancer and
31 cancer dose extrapolation.
32 Response: The Agency evaluated the Sweeney et al. (2008a) publication and this was included in
33 Appendix B of the document. Text was added to the main document in Section 3.5.2.4 and 3.5.3
34 regarding the evaluation of Sweeney et al. (2008a). This model was determined not to be
A-17
DRAFT - DO NOT CITE OR QUOTE
-------
1 appropriate for interspecies extrapolation. Additionally, see response to the external peer review
2 panel comment B4.
3 Comment: EPA should use the modified inhalation inputs used in the Reitz et al. (1990a) model
4 and the updated input parameters provided in Sweeney et al. (2008a) and add a compartment for
5 the kidney
6 Response: See response to previous comment regarding evaluation of Sweeney et al. (2008a).
7 Modification of the model to add a kidney compartment is not within the scope of this
8 assessment.
A.2.4 Other Comments
9 Comment: EPA should consider the Kasai et al. (2009; 2008) studies for inhalation and MO A
10 relevance.
11 Response: The 13 week and 2-year inhalation studies by Kasai et al. (2009; 2008) were published
12 late in the development stage of this assessment. The IRIS Program will evaluate these recently
13 published 1,4-dioxane inhalation data for the potential to derive an RfC in a separate assessment.
14 Comment: 1,4-Dioxane is not intentionally added to cosmetics and personal care products -
15 correct sentence on page 4.
16 Response: This oversight was corrected in the document.
A.3 External Peer Review Panel Comments - Inhalation Update
17 The reviewers made several editorial suggestions to clarify portions of the text. These changes
18 were incorporated in the document as appropriate and are not discussed further.
19 In addition, the external peer reviewers commented on decisions and analyses in the
20 Toxicological Review of 1,4-Dioxane under multiple charge questions, and these comments were
21 organized and summarized under the most appropriate charge question. In cases where comments were
22 made regarding the oral assessment for 1,4-dioxane, those comments are noted, considered, and changes
23 were made to the oral assessment as appropriate; however this was not intended to be a second peer
24 review of the oral assessment finalized in 2010 (U.S. EPA. 2010a).
A-18
DRAFT - DO NOT CITE OR QUOTE
-------
A.3.1 General Charge Questions
1 1. Is the Toxicological Review logical, clear and concise? Has EPA clearly presented and synthesized
2 the scientific evidence for noncancer and cancer health effects from exposure to 1,4-dioxane
3 viainhalation?
4 Comment. Four reviewers agreed that the Toxicological Review of 1,4-dioxane was logical,
5 clear, and concise. Two reviewers commented that the majority of the Toxicological Review was
6 logical, clear, and concise, but provided several recommendations to improve the document. The
7 specific recommendations included: (1) documentation of literature search terms, (2) description
8 of the severity of the lesions observed by Kasai et al. (2008) should be included in the main body
9 of the text, (3) clarification of the toxicological significance of nuclear enlargement with clear
10 differentiation between study author and EPA's conclusions regarding this endpoint, (4)
11 improvement of Tables 4-27 and 4-28 as they do not readily demonstrate temporal relationships
12 of interest, (5) removal of repetitive text, (6) reduction of unnecessary text in the mode of action
13 analysis, (7) correction of inconsistencies between oral and inhalation approaches to derive the
14 reference values, (8) the addition of information on ambient exposures to 1,4-dioxane, and (9)
15 improve the writing of the text of section 4.6.2 and expand Section 4.6.2.1 to focus on the
16 possibility that the parent compound is the toxic moiety.
17 Additionally, one reviewer made reference to a public comment noting an error in the PBPK
18 model code in the description of the slowly perfused tissue. This reviewer suggested the code be
19 corrected and provided in the assessment. However, the reviewer did agree with the conclusion
20 that the existing PBPK models are inadequate to perform route-to-route and cross-species
21 extrapolation of animal studies.
22 Response. (1) Additional information was provided in Section 1.1 regarding the literature search
23 strategy employed for 1,4-dioxane. (2) The severity of the nasal lesions observed by Kasai et al.
24 (2008) was included in Table 4-17; no additional language was added to the text as the data is
25 presented clearly in tabular format. (3) With regards to nuclear enlargement, consultation with an
26 Agency pathologist and a search of published literature did not provide any information
27 concerning the nature, severity, and significance of this observed response. Reports of nuclear
28 enlargement have only been found in JBRC published studies and have not been reported
29 elsewhere in literature. Thus, consideration and selection of this response as a critical endpoint
30 would add unnecessary uncertainty to the assessment of the effects induced by 1,4-dioxane.
31 Clarifying text was added to the document regarding the uncertainty surrounding this reported
32 observation to Sections 4.2.1.1.3, 4.2.1.2.6, 4.2.2.1.2, 4.2.2.2.2, and 5.2.1. (4) Tables 4-27 and 4-
33 28 were described in more depth in their accompanying sections to describe their content and the
34 temporal nature. (5)/(6) The Agency continues to evaluate and incorporate recommendations
35 made by the NAS that should streamline (i.e. reduce redundancy), strengthen and improve
36 transparency within the IRIS documents. (7) There are necessary differences in the derivation of
37 oral and inhalation reference values, discussed in Section 5.4.4, and clarified in Section 5.4.4.2.
38 For instance, the oral slope factor derivation does not use the multistage model, whereas the
A-19
DRAFT - DO NOT CITE OR QUOTE
-------
1 inhalation unit risk derivation does. This is due to a lack of a suitable multistage model being
2 identified for the female mouse liver tumor data used to derive the oral slope factor, whereas
3 appropriate multistage model fits were obtained for the tumor data used to derive the inhalation
4 unit risk. This departure resulted in a necessary and significant difference in approaches. (8)
5 While it is important for risk assessors to understand ambient exposure levels in utilization of
6 IRIS reference values, ambient exposure levels are dependent upon location and media and thus
7 are not included in IRIS assessments. In the context of the overall risk assessment paradigm,
8 IRIS documents provide the hazard identification information and the dose-response analysis in
9 support of the derivation of reference values for the chemical of interest. (9) The suggestions
10 made by the reviewer to improve the writing and summaries in 4.6.2 were all incorporated. The
11 mechanism by which 1,4-dioxane induces tissue damage is not known, nor is it known whether
12 the toxic moiety is 1,4-dioxane or a transient or terminal metabolite. As the reviewer notes, and
13 is already stated in the toxicological review, it is possible that the parent compound is the toxic
14 moiety; however, the section was not rewritten with a focus on the parent compound. Regarding
15 the PBPK model, the code errors identified by a public commenter and referenced by a member
16 of the peer review panel were corrected (discussed further in response to public comments,
17 below). Additionally, the model equations have been available in Appendix B of previous
18 version of the toxicological review released. In this final version, however, the model code is not
19 provided in the text, but is available electronically via HERO, along with the executable .m script
20 files (provide HERO link).
21 2. Please identify any additional peer-reviewed studies from the primary literature that should be
22 considered in the assessment of noncancer and cancer health effects from exposure to 1,4-dioxane via
23 inhalation.
24 Comment: Four reviewers stated they were unaware of any additional studies available to add to
25 the inhalation toxicity evaluation of 1,4-dioxane. One reviewer provided additional general
26 references pertaining to dose extrapolation for the derivation of the RfC specifically regarding the
27 default values used for the human extrathoracic surface area and minute ventilation. Another
28 reviewer provided some general references related to evaluation of tumors and mode of action,
29 along with a few 1,4-dioxane specific papers. The 1,4-dioxane specific papers suggested for
30 consideration were:
31 a. Takano, T, Murayama, N, Horiuchi, K, Kitajima, M, Shono, F. (2010). Blood
32 concentrations of 1,4-dioxane in humans after oral administration extrapolated from in
33 vivo rat pharmacokinetics, in vitro human metabolism, and physiologically based
34 pharmacokinetic modeling. JHealth Sci 56: 557-565. (Note: The reviewer noted that this
35 paper is not likely to be useful in the assessment; however, a short summary should be
36 added to the appropriate section in the toxicological review)
37 b. U.S. Army Public Health Command (2010). Studies on Metabolism of 1,4-Dioxane,
38 Toxicology Report No. 87-08 WR-09, Aberdeen Proving Ground, MD.
A-20
DRAFT - DO NOT CITE OR QUOTE
-------
1 c. WHO (World Health Organization). (2005). 1,4-Dioxane in Drinking Water,
2 WHO/SDE/WSH/05.08/120, Geneva.
3 Response: Reference (a) above was evaluated for the utility of the described PBPK model in
4 predicting toxicokinetics of 1,4-dioxane in rats and humans. A summary of Takano et al. (2010)
5 and an evaluation of the model was added to Section 3.5.2.5. Reference (b) above is a report that
6 has not undergone formal peer-review and thus, is generally not considered in the development of
7 an IRIS assessment. Reference (c) listed is a report produced by an organization other than the
8 U.S. EPA and was considered during development of this assessment; however, the Agency
9 performed an independent analysis of the scientific informa available for 1,4-dioxane and did not
10 cite this document.
11 The additional general references pertaining to dose extrapolation for the derivation of
12 the RfC specifically regarding the default values used for the human extrathoracic surface area
13 and minute ventilation were related to the inclusion of the alternative RfC calculation in
14 Appendix G. This appendix was removed following external peer review. See response to charge
15 question 4, below, relating to the RfC for more details.
A.3.2 Inhalation reference concentration (RfC) for 1,4-dioxane
16 1. A 2-year inhalation bioassay in male rats (Kasai et al.. 2009) was selected as the basis for the
17 derivation of the RfC. Please comment on whether the selection of this study is scientifically
18 supported and clearly described. If a different study is recommended as the basis for the RfC, please
19 identify this study and provide scientific support for this choice.
20 Comment: Four reviewers agreed that the selection of the 2-year bioassay in male rats (Kasai et
21 al. 2009) as the critical study used for the derivation of the RfC was scientifically justified. Two
22 reviewers also agreed with the aforementioned, but stated that decision not to collect female rat
23 data for the 2-year bioassay was not scientifically supported by the study authors (Kasai et al..
24 2009). especially given that the 13-week bioassay (Kasai et al.. 2008) showed female rats more
25 responsive than male rats following inhalation exposure. More specifically, the two reviewers
26 highlighted that one of the selected critical effects (atrophy of the olfactory epithelium) was
27 observed in female rats and not male rats following 13 weeks of exposure to 1,4-dioxane vapors,
28 thus making the female rat more responsive to 1,4-dioxane following inhalation exposure.
29 Response: The Agency did not conclude that the available data supports the female rats as
30 definitively more responsive than male rats following 13 weeks of exposure to 1,4-dioxane
31 vapors. BMD analysis of the incidence of olfactory atrophy in female rats from the Kasai et al.
32 (2008) study provides a BMCLio of 65 ppm (fit with the Dichotomous Hill model). Application
33 of a total UF of 1,000 would yield an RfC of 0.065 ppm compared to an RfC of 0.05 ppm
34 calculated from the 2 year bioassay. A review of the pathological observations also does not
35 indicate that females are definitively more responsive to 1,4-dioxane exposure. Of the lesions
A-21
DRAFT - DO NOT CITE OR QUOTE
-------
1 noted, most were considered to be of the lowest severity grade. Of these lesions, equivalent
2 responses were observed between males and females and in some cases greater in females and in
3 others greater in males. Thus, information to suggest that females are more responsive than males
4 is currently lacking. Additionally, in accordance with the weight-of-evidence framework
5 described in the Methods for Derivation of Inhalation Reference Concentrations and Application
6 of Inhalation Dosimetry (U.S. EPA. 1994a). the selection of the 2-year bioassay in male rats as
7 the critical study is justified. Furthermore, an uncertainty factor of 3 for an incomplete database
8 was applied. This uncertainty factor is intended to account for the inability of any single
9 laboratory animal study to adequately address all possible adverse outcomes in humans.
10 Therefore, in consideration of the data presented in each of the studies as well as the difference in
11 the study durations (13 vs. 104 wks), the selection of the 2-year bioassay in male rats as the
12 critical study is justified.
13 2. Atrophy and respiratory metaplasia of the olfactory epithelium in male rats were concluded by EPA
14 to be adverse effects and were selected as co-critical effects for the derivation of the RfC. Please
15 comment on whether the selection of these co-critical effects and their characterization is
16 scientifically supported and clearly described. If a different health endpoint is recommended as the
17 critical effect for deriving the RfC, please identify this effect and provide scientific support for this
18 choice.
19 Comment: Four reviewers agreed with the selection of co-critical effects in the derivation of the
20 RfC and stated that the selection was scientifically supported and clearly described. The
21 remaining two reviewers also agreed with the selection of co-critical effects in the derivations of
22 the RfC; however, they provided suggestions on how to strengthen the justification for EPA's
23 decision or improve clarity. These reviewers suggested EPA (1) provide further justification for
24 why nuclear enlargement was not considered as a critical effect and (2) clearly state the criteria
25 for selection of the critical effect. One reviewer also noted inconsistency between the oral and
26 inhalation assessments regarding the consideration of spongiosis hepatis as a nonneoplastic lesion
27 and potential critical effect.
28 Response. In response to reviewer comments, EPA further investigated nuclear enlargement. As
29 stated in response to inhalation assessment general charge question 1 (Section A.3.1), a search of
30 the literature demonstrated a lack of clear information concerning the nature, severity, and
31 significance of nuclear enlargement. Thus, the etiology of nuclear enlargement or its downstream
32 effects have not been elucidated. Therefore, consideration and selection of this response as a
33 critical endpoint would not be supported by the available scientific information. Clarifying text
34 was added to the document regarding nuclear enlargement as noted in response to charge question
35 Al, and specifically in Section 5.2.1 as to why it was not considered as a critical effect.
36 Additional clarifying text was added to Section 5.2.3 regarding the use of respiratory metaplasia
37 and atrophy of the olfactory epithelium as co-critical effects, noting that they were the most
38 sensitive effects considered following inhalation of exposure to 1,4-dioxane. EPA agrees there
39 was inconsistency in way spongiosis hepatis was considered between the oral and inhalation
A-22
DRAFT - DO NOT CITE OR QUOTE
-------
1 assessments. Spongiosis hepatis was removed from the list of candidate critical effects in the
2 inhalation assessment. However, whether spongiosis hepatis/cystic degeneration represents a
3 preneoplastic change or a nonneoplastic change has been the subject of scientific controversy
4 (Karbe and Kerlin. 2002b: Stroebel et al. 1995; Bannasch et al. 1982b). Spongiosis hepatis is
5 commonly seen in aging rats, but has been shown to increase in incidence following exposure to
6 hepatocarcinogens. Spongiosis hepatis can be seen in combination with preneoplastic foci in the
7 liver or with hepatocellular adenoma or carcinoma and has been considered a preneoplastic lesion
8 (Bannasch. 2003; Stroebel et al.. 1995). In contrast, it can also be associated with hepatocellular
9 hypertrophy and liver toxicity and has been regarded as a secondary effect of some liver
10 carcinogens (Karbe and Kerlin. 2002a). Following inhalation of 1,4-dioxane, spongiosis hepatis
11 was associated with other preneoplastic (e.g., liver foci) and nonneoplastic (e.g., centrilobular
12 necrosis) changes in the liver (Kasai et al.. 2009). Additionally, the incidence rates of spongiosis
13 hepatis and liver tumors were highly correlated; therefore, spongiosis hepatis was considered a
14 preneoplastic lesion following inhalation exposure and not considered further in the noncancer
15 analysis. This justification was added to the document in Section 5.2.1.
16
17 3. Benchmark dose (BMD) modeling methodology (U.S. EPA. 2000d) was used to analyze the
18 candidate endpoints identified for 1,4-dioxane. However, due to poor fit or substantial model
19 uncertainty, BMD model results were inadequate for the following nasal lesions: atrophy (olfactory
20 epithelium), respiratory metaplasia (olfactory epithelium), and sclerosis (lamina propria).
21 Consequently, the NOAEL/LOAEL approach was used to identify the POD for derivation of the RfC.
22 Please comment on whether this approach is scientifically supported and clearly described.
23 Comment: Six reviewers agreed that the use of the NOAEL/LOAEL approach in the derivation
24 of the RfC is scientifically supported and clearly described.
25 Response. EPA agrees with the reviewers regarding the use of the NOAEL/LOAEL approach in
26 the derivation of the RfC, no changes were made to the document.
27 4. The human equivalent concentration (HEC) for 1,4-dioxane was calculated by the application of the
28 dosimetric adjustment factor (DAF) for systemic acting gases (i.e. Category 3 gases), in accordance
29 with the U.S. EPA RfC methodology (U.S. EPA. 1994b). This conclusion was based upon a number
30 of factors, including the low reactivity of 1,4-dioxane, and the occurrence of systemic effects
31 following oral and inhalation exposure to 1,4-dioxane. However, since 1.4-dioxane is water soluble
32 and induces effects in portal-of-entry tissues, an alternative calculation of the HEC for 1,4-dioxnae
33 based on the application of the corresponding DAF for the portal-of entry acting gases (i.e., Categrory
34 1) is provided in Appendix G. Please comment on EPA's conclusion that 1.4-dioxane is a Category 3
35 gas, and the resulting application of the corresponding dosimetric adjustment factor (DAF) in
36 deriving the RfC. If a different approach is recommended in the derivation of the RfC, please identify
37 this approach and provide scientific support for the proposed changes.
38 Comment: All of the reviewers thought the approach used in the main body of the document was
39 reasonable and consistent with the Agency's current definitions and approaches, as well as the
A-23
DRAFT - DO NOT CITE OR QUOTE
-------
1 effects observed. Two reviewers thought the inclusions of an alternative approach in Appendix G
2 was reasonable. Two other reviewers noted problems with the outcome of the default calculation
3 used in the alternative approach. Two reviewers thought the lesions seen in the inhalation study
4 may represent portal-of-entry responses; one of these reviewers thought additional text should be
5 added to the document.
6 Response. Since the reviewers were in agreement with the extrapolation approach employed and
7 described in the main body of the document, Appendix G in the external peer review draft that
8 demonstrated the application of the Agency's default method for deriving an RfC for category 1
9 gases was removed. The alternative approach used default ratios of ventilation rate and surface
10 areas cited and often used in accordance with the Agency's RfC Methods (U.S. EPA. 1994b).
11 which are also supported by several sources including ICRP (2002), Guilmette et al. (1997). and
12 Liu et al. (2009).
13 5. The text corresponding to the dosimetric extrapolation approach applied for 1,4-dixoane has been
14 revised for clarity and transparency; however, no changes to the quantitative approach were made.
15 EPA agrees that 1,4-dioxane induces portal of entry effects. 1,4-Dioxane is miscible with water and
16 has a high blood:air partition coefficient. Unlike typical highly water soluble and reactive portal-of-
17 entry acting gases, 1,4-dioxane also induces lower respiratory tract and systemic effects and has been
18 measured in the blood after inhalation exposure. Thus, it is difficult to determine what contribution
19 circulating 1,4-doxane makes to the portal-of-entry effects observed. Therefore, for the purposes of
20 dosimetric extrapolation, 1,4-dioxane was treated as a systemic acting gas and a DAF of 1 was
21 applied. In addition, a robust CFD and PBPK modeling database supports the scientific rationale to
22 apply of DAF of 1 for both portal of entry and systemic effects irrespective of "gas categorization"
23 (U.S. EPA. 2012b). Please comment on the rationale for the selection of the UFs applied to the POD
24 for the derivation of the RfC. Are the UFs appropriate based on A Review of the Reference Dose and
25 Reference Concentration Processes ((U.S. EPA. 2002c): Section 4.4.5;
26 www.epa.gov/iris/backgrd.html) and clearly described? If changes to the selected UFs are proposed,
27 please identify and provide scientific support for the proposed changes.
28 Comment: Four reviewers agreed with the selection and justification of the UFs applied to the
29 POD for the derivation of the RfC. One of these reviewers, however, suggested that it be noted
30 that the reproductive toxicity and teratogenicity indices monitored in rats by Giavini et al.
31 (1985b) were unremarkable. Two reviewers agreed with the selection of the UFs but requested
32 clarification of the justification for the database uncertainty factor. One reviewer further
33 questioned the reliability of the UF of 10 to extrapolate to a NOAEL given the lack of an
34 exposure group below 50 ppm where one of the critical effects was noted with an incidence rate
35 of 80% (olfactory epithelium), and the lack of female rats exposure in the 2 year bioassay despite
36 evidence of increased responsiveness to 1,4-dioxane vapors following inhalation as compared to
37 the male rat in a 13 week bioassay. Additionally, one reviewer debated the application of the UF
38 of 10 for individual differences among human subjects given that dosimetric differences for
39 particles among human subjects is often 1.3 rather than 3.
A-24
DRAFT - DO NOT CITE OR QUOTE
-------
1 Response. In accordance with U.S. EPA (2002c), the database was characterized and applied to
2 the derivation of the RfC. The EPA has added clarification to Section 5.2.4 regarding the
3 strengths and limitations of the data which support an uncertainty factor of 3 for the database.
4 Giavini et al. (1985b) administered 1,4-dioxane by gavage in water to pregnant rats. The authors
5 found statistically significant changes in fetal body weight at the highest dose group and reduced
6 ossification of the sternebrae; however, the lack of a multigenerational reproductive study
7 warrants the use of a 3 for UFD. As outlined in detail in response to the inhalation assessment
8 charge question Bl, the available data do not support female rats as definitively more responsive
9 than male rats following 13 weeks of exposure to 1,4-dioxane vapors. A recent modeling study by
10 Valcke and Krishnan (2011) assessed the impact of exposure duration and concentration on the
11 human kinetic adjustment factor and estimated the neonate vs. adult 1,4-dioxane blood
12 concentration ratio to be 3.2.
A.3.3 Carcinogenicity of 1,4-dioxane and derivation of an inhalation unit risk
13 1. Under EPA's Guidelines for Carcinogen Risk Assessment ((U.S. EPA. 2005b): Section 2.5;
14 www. epa. gov/iris/backgrd.html). the draft IRIS assessment characterizes 1,4-dioxane as "likely to be
15 carcinogenic to humans" by all routes of exposure. Please comment on whether this characterization
16 of the human cancer potential of 1,4-dioxane is scientifically supported and clearly described.
17 Comment: Five out of six reviewers agreed with the characterization that 1,4-dioxane is "likely to
18 be carcinogenic to humans." However, one of these reviewers suggested a more transparent
19 application of the criteria to the inhalation cancer data to classify the compound as "likely" would
20 be beneficial. One reviewer disagreed with the cancer classification of "likely to be carcinogenic
21 to humans" and suggested that it should be classified as a "possible human carcinogen". This
22 reviewer provided several arguments as a basis for a different classification: 1) no evidence of
23 increased cancer incidence in humans exposed to 1,4-dioxane in the limited number of
24 epidemiology studies, 2) negative in vivo and in vitro genotoxicity experiments suggesting that
25 1,4-dioxane is, at most, a weak genotoxicant, 3) data demonstrating observed tumors in rodents
26 occur following high chronic exposures, 4) the parent compound is the proximate irritant,
27 cytotoxicant, and carcinogenic moiety, and 5) conclusions and classifications by other
28 organizations (i.e., German Commission for the Health Hazards of Chemical Compounds in the
29 Work Area, ACGIH, IARC and WHO).
30 Response: Five of the six reviewers agreed with the characterization of "likely to be
31 carcinogenic to humans" and no change was made to this conclusion in the final Toxicological
32 Review. With respect to the one reviewer who suggested applying the criteria more transparently
33 to the inhalation data alone; when considering the characterization of the carcinogenic potential
34 for a compound, the available data across all exposure routes is first considered. If, for example,
35 portal of entry effects are observed for one route of exposure and not the other, or there is
36 evidence that a chemical is not absorbed from a particular route of exposure, then separate cancer
A-25
DRAFT - DO NOT CITE OR QUOTE
-------
1 descriptors may be used to describe the cancer potential. In the case of 1,4-dioxane, the tumors
2 that were observed in animals were systemic and independent of the route of exposure.
3 The one reviewer that disagreed with the classification provided a suggested classification that
4 appears to be based on earlier 1986 U.S. EPA cancer classification terminology. As summarized
5 in Section 4.7.1, the available human studies with small cohorts and limited number of reported
6 cases are inconclusive. The Agency agrees with the reviewer that the majority of the genotoxicity
7 studies are negative, suggesting 1,4-dioxane is not genotoxic (Section 4.5.1), and that tumors
8 have been observed in rodents following chronic exposure (summarized in Section 4.7.2). A lack
9 of data to determine the toxic moiety (e.g., parent compound, intermediate, or terminal
10 metabolite), does not impact the Agency's cancer classification.
11 2. The draft assessment concludes that there is insufficient information to identify the mode(s) of
12 carcinogenic action for 1,4-dioxane. Please comment on whether this determination is appropriate and
13 clearly described. If it is judged that a mode of action can be established for 1,4-dioxane, please
14 identify the mode of action and its scientific support (i.e., studies that support the key events, and
15 specific data available to inform the shape of the exposure-response curve at low doses).
16 Comment: Five out of six reviewers agreed with EPA's conclusion that there is insufficient
17 scientific information to establish the mode(s) of carcinogenic action for 1,4-dioxane. However,
18 one of these reviewers suggested integrating the sequence of events for a possible mode of action
19 described in a public comment into the body of the Toxicological Review. Another one of these
20 five reviewers provided several examples of places in the toxicological review that could use
21 clarification of study limitations and consideration of pertinent data: impact of 1,4-dioxane
22 volatility on in vitro and skin/paint study results; mechanistic section needs more discussion and
23 analysis of a potential genotoxic mode of action; critical deficiencies in the database should be
24 noted in the discussion of cytotoxicity/cell proliferation mode of action; examine dose-response
25 relationships for effects seen in the 13-week studies and how they may predict tumor incidence;
26 the lack of mouse liver initiation-promotion studies should be noted; and data do not support
27 statements regarding metabolic saturation and subsequent toxicity. One of the six reviewers
28 disagreed with EPA's conclusion that there is insufficient information to identify a MOA for 1,4-
29 dioxane. This reviewer commented that data clearly support a cytotoxicity/inflammation/
30 regenerative hyperplasia MOA with a dose threshold, citing the Kociba et al. (1974b). Kano et al.
31 (2008). and Kasai et al. (2009: 2008) studies.
32 Response: The Agency agrees with five of the six reviewers that there is insufficient evidence to
33 establish a carcinogenic MOA for 1,4-dioxane. As seen in responses to the public comments
34 regarding the carcinogenicity of 1,4-dioxane (Section A.4.2.), the sequence of events proposed by
35 the public commenter are not supported by the available data. These key events for the
36 hypothesized MOA are visualized in Figure 4-1 of the Toxicological Review.
37 The available data do not clearly support a cytotoxic/inflammation/regenerative hyperplasia
38 MOA (Section 4.7.3). Specifically, the studies referenced by the reviewer (Kasai et al.. 2009;
39 Kano et al.. 2008; 2008; Kociba etal.. 1974b) do not examine cytotoxicity or regenerative cell
A-26
DRAFT - DO NOT CITE OR QUOTE
-------
1 proliferation in the nasal cavity. Further, the existing data examine a small number of exposures
2 and timepoints. Kasai et al. (2009) suggests either genotoxic or cytotoxic MOA for 1,4-dioxane,
3 but their data do not provide sufficient evidence to conclude one way or the other. Furthermore,
4 there is no evidence of cytotoxicity in the nasal cavity in the Kasai et al. (2009; 2008) studies.
5 Additionally, evidence of cytotoxicity in one tissue type, does not dictate that cytotoxicity will be
6 present in all tissues at the same dose. Thus, the database does not provide evidence for each
7 stage of a regenerative hyperplasia MOA.
8 A number of changes were made as a result of the specific comments made regarding clarity and
9 study limitations. Regarding the volatility of 1,4-dioxane and reliability of the negative in vitro
10 studies and skin paint studies, text was added to Section 4.5.1 noting the four negative in vitro
11 studies that reported using closed systems and to Section 4.2.3 regarding the reliability of the data
12 from unoccluded versus occluded skin paint initiation/promotion studies. Text was revised in
13 Section 4.5.1 to state clearly that half of the studies showed 1,4-dioxane was not genotoxic;
14 however, data are not sufficient to support a genotoxic MOA and no additional discussion
15 regarding this MOA was added to the document. Text was added to Section 4.7.3 noting
16 deficiencies in the database surrounding a cytotoxicity/cell proliferation MOA. As a result of the
17 peer review comment, the noncancer effects were reexamined in detail and how they may relate
18 to the cancer effects seen. An attempt was made to create new tables showing the noncancer and
19 cancer effects across the dose and time; however, these tables were found to introduce more
20 confusion. Therefore, only clarifying text was added (Sections 4.7.1, 4.7.3.1.2, and 4.7.3.3)
21 regarding the noncancer effects and their relation to the cancer effects and the temporal sequence
22 of events, as well as clarifying the. In response to another comment from the reviewer, a
23 statement was added to Section 4.7.3.1.1 to clearly state that no studies have been conducted to
24 specifically examine the mouse liver, thus precluding any determination on whether 1,4-dioxane
25 acts as a tumor promoter in the mouse liver. A thorough review of statements in the document
26 pertaining to metabolic saturation and its relation to toxicity was performed in response to the
27 reviewers comment. Several changes were made throughout the document (e.g., Section 3.5.1,
28 4.6.2.1, and 4.7.3.7.1) clarifying relationships observed (or not) between metabolic saturation and
29 toxicity. In general metabolic saturation was observed in single dose studies (Young et al.. 1978b:
30 1978a). We agree with the reviewer that a single dose study does not provide adequate
31 information to support metabolic saturation following repeated long-term exposures, and that
32 since 1,4-dioxane induces P450 enzymes it is likely to enhance metabolic elimination in long
33 term exposure scenarios. Additional kinetic information is needed to determine if metabolic
34 saturation is a precursor to a toxic effect. Kociba et al. (Kociba et al.. 1975b) that stated toxicity
35 was only observed after metabolism was saturated did not present data for repeated doses to
36 support this conclusion.
37 3. A two-year inhalation cancer bioassay in male rats (Kasai et al.. 2009) was selected as the basis for
38 the derivation of the inhalation unit risk (IUR). Please comment on whether the selection of this study
39 is scientifically supported and clearly described. If a different study is recommended as the basis for
40 the IUR, please indentify this study and provide scientific support for this choice.
A-27
DRAFT - DO NOT CITE OR QUOTE
-------
1 Comment: Five of the six reviewers agreed that the use of the two year inhalation cancer bioassay
2 in male rats Kasai et al. (2009) is the most appropriate study to use for the derivation of the IUR.
3 Five of the six reviewers also stated the selection was clearly described and justified or supported
4 within the toxicological review. The other reviewer neither disagreed or agreed with the selection
5 of the study; however, the reviewer noted that the Kasai et al. (2009) study is the only
6 comprehensive inhalation study available for this chemical, because the other study by Torkelson
7 et al. Q974a) used only one dose and did not perform histology on the nasal tissues.
8 Response. No dissenting opinions or comments warranting additional justification were provided
9 by the external review panel regarding selection of the principal study for derivation of the IUR.
10 Thus, no changes were made to the assessment related to the selection and justification of the
11 Kasai et al. (2009) study for derivation of the IUR.
12 4. The incidence of hepatocellular adenomas and carcinomas, nasal cavity squamous cell carcinoma,
13 renal cell carcinoma, peritoneal mesothelioma, mammary gland fibroadenoma, Zymbal gland
14 adenoma, and subcutis fibroma were selected to serve as the basis for the derivation of the IUR.
15 Please comment on whether this selection is scientifically supported and clearly described. If a
16 different health endpoint is recommended for deriving the IUR, please identify this endpoint and
17 provide scientific support for this choice.
18 Comment: Five of the six reviewers agreed with EPA's choice to combine these tumor types for
19 derivation of the IUR, noting the statistically significant tumor incidence rates and the dose
20 related increase in tumors. One of the five reviewers that agreed with the approach questioned if
21 data are available to fully justify the pooling of certain tumor types. One of these five reviewers
22 noted that the mice were more sensitive than rats to the hepatocarcinogenic effects of 1,4-dioxane
23 following drinking water exposure. Thus, since mice were not included in a 2-year inhalation
24 cancer bioassay, the IUR may be underestimated and this should be noted as a source of
25 uncertainty qualitatively and a quantitatively. This reviewer suggested a quantitative adjustment
26 to the IUR by multiplying the IUR by the ratio of hepatocellular neoplasms in male rats: female
27 mice from the oral study. The sixth reviewer disagreed with combining all of these tumor types,
28 arguing that Zymbal gland tumors are limited to male rats; and peritoneal mesothelioma, subcutis
29 fibroma, and mammary fibroadenoma are typical spontaneous tumors in F344 rats (Haseman et
30 al.. 1998: Hall. 1990).
31 Response: In agreement with five of the six reviewers, the Agency retained the combination of
32 the tumor types with statistically significant incidence rates different from control or a
33 statistically determined dose-related trend in the combined tumor analysis for the derivation of
34 the IUR. Data were not available to establish whether the tumor types were biologically
35 dependent, thus independence was assumed and is not expected to produce substantial error in the
36 risk estimates (!!! INVALID CITATION !!!). It is acknowledged that Zymbal gland tumors do
37 not occur in humans due to the lack of a Zymbal gland; however, site concordance is not always
38 assumed for animals and humans (U.S. EPA. 2005a) because events leading to Zymbal gland
39 tumors may occur at other sites in humans. Additional text was added to section 5.5.1.6 and
A-28
DRAFT - DO NOT CITE OR QUOTE
-------
1 6.2.3.8 to address the possible underestimation of the carcinogenic inhalation potential of 1,4-
2 dioxane since female mice were the most sensitive following oral administration and were not
3 included in the 2-year inhalation cancer bioassay. While the uncertainties were noted
4 qualitatively, a quantitative adjustment was not performed on the IUR as this is not a standard
5 approach conducted by the agency. The sixth reviewer raised objections to using peritoneal
6 mesothelioma, subcutis fibroma, and mammary fibroadenoma as the reviewer characterized them
7 as "very commonly observed, spontaneous tumors in control F344 rats." The study authors used
8 untreated, clean air exposed rats as an experimental control to account for any possible
9 spontaneous tumors that may arise. Furthermore, the Agency accounts for the background rate in
10 controls when using the multistage cancer model.
11 5. The IUR was derived based on multiple carcinogenic effects observed in rats exposed to 1,4-dioxane
12 via inhalation. A Bayesian approach was used to estimate a BMDLio associated with the occurrence
13 of these multiple tumors, and then a linear low-dose extrapolation from this POD was performed to
14 derive the IUR. Additionally, for comparative purposes only, a total tumor analysis was performed
15 with the draft BMDS (version 2.2Beta) MSCombo model that yielded similar results (See Appendix
16 H). Please comment on whether these approaches for deriving the IUR have been clearly described
17 and appropriately conducted?
18 Comment: Two reviewers commented that the approaches were clearly described and
19 appropriately conducted; however, the methods to quantitate cancer risk are outside of their areas
20 of expertise. Four of the reviewers commented that both methods, Bayesian and BMDS, are
21 clearly described and appear appropriately conducted since both methods yielded similar results.
22 However, one of these four reviewers noted that additional information to reproduce the Bayesian
23 analysis should be provided. Another of these four reviewers noted that IUR estimates may
24 actually be larger since survival was significantly reduced in the high exposure group and that the
25 cancer dose-response modeling did not use survival adjusted data. One reviewer commented that
26 the limitations and assumptions related to the risk of developing any combination of the tumor
27 types is not well documented in the toxicological review. Additionally, one reviewer noted that
28 the total tumor approach was not utilized in the derivation of the oral CSF and recommended a
29 total tumor analysis for male and female rats exposed to 1,4-dioxane in drinking water. One
30 reviewer did not support the Agency's default use of Haber's Law to make adjustments for the
31 exposure duration in the derivation of the IUR (or RfC). This reviewer suggested additional
32 examination of the 1,4-dioxane data to gain insights into a and (3, if possible to further describe
33 uncertainties associated with this duration adjustment.
34 Response. Overall, the reviewers were in support of the quantitative approaches to the
35 multitumor analysis for the derivation of the IUR. As a result of the public comments regarding
36 the documentation and reproducibility of the Bayesian WinBUGS approach, and the fact that the
37 BMDS MS_Combo model has completed peer review since the draft of this assessment was
38 released, the transparent, reproducible MS_Combo approach is now considered the primary
39 approach for derivation of the IUR and the Bayesian WinBUGS approach is a supporting analysis
40 with details in Appendix G (formerly Appendix H). Additional details on the WinBUGS analysis
A-29
DRAFT - DO NOT CITE OR QUOTE
-------
1 was added to the appendix and the model code was made available via HERO (EPA, 2013).
2 Using MS_Combo approach as the primary approach did not result in any quantitative changes to
3 the IUR.
4 As stated in response to general charge question 1, similar methods to analyze the total tumor risk
5 were not available at the time of the completion of the oral assessment. Additionally, the
6 multistage model was not the best fitting model for female mouse liver tumors and was not used
7 in derivation of the oral slope factor, whereas the inhalation unit risk derivation does utilize the
8 multistage model. However, in response to the reviewer's comment, the male and female rat data
9 were analyzed using the BMDS MS_Combo model. BMDLnEc values for male rat and female rat
10 combined tumors were determined to be 7.59 and 11.26 mg/kg-day, respectively. Using a BMR
11 of 0.1 oral CSFs of 0.013 and 0.0088 (mg/kg-day)"1 were calculated for the male and female rat
12 data, respectively. Thus the combined tumor analysis for the oral assessment does not impact the
13 selection of the gender/species or overall oral CSF for 1,4-dioxane.The Agency concurs with the
14 reviewer who states that the IUR estimates may actually be larger if survival adjusted data were
15 used and this was noted in Section 5.5.1.6. However, day of death data were not available in the
16 Kasai (2009) study, thus this analysis cannot be performed.
17 Data are not available to move away from the default value of 1 for a and (3 in the C x T duration
18 adjustment approach for inhalation exposure. Two, 13-week subchronic studies in laboratory
19 animals (Kasai et al.. 2008; Fairley et al.. 1934a) and two, 2-year chronic studies in rats (Kasai et
20 al.. 2009; Torkelson et al.. 1974a) were identified; however, these data did not report the severity
21 of the lesions for multiple timepoints.
A.4 Public Comments - Inhalation Update
22 The Toxicological Review of 1,4-Dioxane (with Inhalation Update) was released for a 60-day
23 public comment period in September 2011. A listening session was scheduled in October 2011;
24 however, no participants registered to speak, so the listening session was cancelled. EPA received
25 written public comments on the draft assessment from Toxicology Excellence for Risk Assessment
26 (TERA) and joint comments from the National Association of Manufacturers (NAM) the Aerospace
27 Industries Association (AIA) provided by ARCADIS. The major comments received have been
28 synthesized and paraphrased below. EPA's responses to the comments and information regarding
29 how the assessment has been revised, where applicable, are included.
A.4.1 Inhalation reference concentration (RfC) for 1,4-dioxane
30 Comment: The use of 3 for the database uncertainty factor (UFD) based on the lack of a
31 multigenerational reproductive study is not warranted. Statistically significant changes in fetal
32 weight and ossified sternebrae reported by Giavini et al. (1985b) are not lexicologically
A-30
DRAFT - DO NOT CITE OR QUOTE
-------
1 significant. No effects were seen on reproductive organs in the oral or inhalation subchronic and
2 chronic studies (Kano et al. 2009: Kasai et al. 2009: Kano et al. 2008: Kasai et al. 2008: NCI.
3 1978: Kociba et al.. 1974b: Torkelson et al.. 1974b). For these reasons the UFD should be
4 reconsidered in the derivation of the RfC.
5 Response: Giavini et al. (1985b) administered 1,4-dioxane by gavage in water to pregnant rats.
6 The authors found statistically significant changes in fetal body weight at the highest dose group
7 and reduced ossification of the sternebrae. The other studies were not designed to examine
8 reproductive or developmental outcomes, and thus cannot be used to infer the
9 reproductive/developmental toxicity of 1,4-dioxane. While Torkelson et al. (1974b) did examine
10 the testes and uterus for gross histopathological changes (e.g., tumor) and did not find increased
11 incidence of tumors, this does not indicate that 1,4-dioxane may not be a developmental toxicant.
12 The study of reproductive organs in subchronic and chronic studies is not a replacement for a
13 multigeneration reproductive/developmental study. A UFD of 3 was used for the oral assessment
14 and was retained for the inhalation assessment.
A.4.2 Carcinogenicity of 1,4-dioxane
15 Comment. Low dose linearity should not have been assumed to derive the proposed IUR since
16 sufficient data exist to support a cytotoxic-proliferative mode of action (MO A) based generally
17 on the following arguments: 1,4-dioxane is neither mutagenic nor an initiator, but it can act as a
18 promoter, "literature indicates that 1,4-dioxane is a weak genotoxic carcinogen", Kasai et al.
19 (2009) characterized the MOA as "cytotoxic-proliferative". Additionally, the Agency's statement
20 that there is insufficient evidence to support any hypothesized MOA is not supported by the
21 "open literature and the data summarized and interpreted in the draft TR". Histopathology results
22 for the nasal cavity/olfactory epithelium, liver, and kidney from Kasai et al. (2009) clearly
23 indicate that cytotoxicity precedes tumor development.
24 Response. The Kasai et al. (2009) study does not provide evidence of cytotoxicity in the nasal
25 cavity. Kasai et al. (2009) suggest either a genotoxic or cytotoxic MOA for 1,4-dioxane, but their
26 data do not provide sufficient evidence for one hypothesis over the other. There is no evidence of
27 cytotoxicity in the Kasai et al. (2009: 2008) study. For instance, inflammation by itself is not
28 direct evidence of cytotoxicity. For the liver and kidney, Kasai et al. (2009) provide direct
29 evidence of cytotoxicity including clinical pathology (liver) and histopathology (liver and kidney)
30 data. Additionally, evidence of cytotoxicity in one tissue type, does not dictate that cytotoxicity
31 will be present in all tissues at the same dose.
32 Due to a lack of information to inform the MOA, the Agency used the default linear extrapolation
33 approach per the EPA Guidelines for Carcinogen Risk Assessment (2005b). Specifically, the
34 Guidelines state that "nonlinear approaches generally should not be used in cases where the mode
35 of action has not been ascertained" and that linear extrapolation will be used as the default in
36 these cases.
A-31
DRAFT - DO NOT CITE OR QUOTE
-------
1 It is important to note that five of the six members on the independent expert peer review panel
2 for this draft assessment agreed with EPA's conclusions regarding the weight of evidence in
3 support of a linear approach to derive the IUR, and all reviewers, including the public
4 commenters, supported EPA's decision to use the Kasai et al. (2009) study as the basis for
5 determining the IUR.
6 Comment. 1,4-Dioxane dose not cause mutagenicity, initiation, or DNA repair. 1,4-Dioxane
7 dose cause promotion and DNA replication. Occurrence of respiratory tumors in rodents may be
8 caused by 1,4-dioxane exceeding the metabolic capacity of the tissue. 1,4-Dioxane does cause
9 liver tumors and liver toxicity precedes tumors in time in both sexes of rats and mice, and
10 precedes tumors in dose in both sexes of rats. Liver toxicity indicated by biochemical measures
11 does occur at similar tumorigenic doses in mice; however histopathological indication of liver
12 toxicity does not appear to precede tumors in either sex of mice. EPA needs to show the liver
13 hyperplasia noted in Kano et al. (2009) in Appendix E of the draft toxicological review. 1,4-
14 Dioxane does cause dose-dependent nasal toxicity as indicated in the histological analyses at all
15 time points in both sexes of rats and mice and this toxicity precedes tumors in time and dose. It is
16 hypothesized that 1,4-dioxane causes liver and nasal tumors in rats and mice through a
17 regenerative hyperplasia MOA, which demonstrates a threshold. The applicability of this MOA to
18 other tumor types is unknown, so a separate, default linear extrapolation may be appropriate for
19 those tumor types.
20 Response. The Agency's determination that the MOA has not been established is supported by
21 five of the six external peer reviewers. The samples associated with liver hyperplasia for rats and
22 mice in Yamazaki et al. (1994a) and JBRC (1998) were re-examined according to updated criteria
23 for liver lesions and were afterwards classified as either hepatocellular adenoma or altered
24 hepatocellular foci in Kano et al. (2009). therefore there are no liver hyperplasia incidence data
25 from Kano et al. (2009) to report in Appendix E as the commenter suggests.
26 Due to a lack of information to substantiate the MOA, the Agency used the default linear
27 extrapolation approach per the EPA Guidelines for Carcinogen Risk Assessment (2005b).
28 Specifically, the Guidelines state that "nonlinear approaches generally should not be used in cases
29 where the mode of action has not been ascertained" and that linear extrapolation will be used as
30 the default in these cases.
31 Comment. Peritoneal mesotheliomas found in male rats, but not female counterparts, is likely
32 due to the occurrence of tunica vaginalis mesotheliomas in male rats. Rats are much more
33 sensitive to developing mesotheliomas from the tunica vaginalis than humans.
34 Response. The etiology and origin of the peritoneal mesotheliomas reported in Kano et al. (2009)
35 and Kasai et al. (2009) are unknown. The commenter indicated a range of considerations
36 including human sensitivity and / or relevance for the peritoneal mesotheliomas observed in male
37 rats(Kano et al.. 2009; Kasai et al.. 2009). The EPA Guidelines for Carcinogen Risk Assessment
38 (2005b) state that all tumor types are to be analyzed in a dose-response assessment followed by a
39 synthesis that considers, among other things, human relevance of each tumor type. In the absence
A-32
DRAFT - DO NOT CITE OR QUOTE
-------
1 of scientific information to evaluate the human relevance of peritoneal mesotheliomas observed
2 in male rats exposed to 1,4-dioxane EPA is required to implement the approaches from the
3 guidance (U.S. EPA. 2005b). EPA concluded there continues to be uncertainty as to the etiology,
4 origin, and species sensitivity of the peritoneal mesotheliomas found in the rats, and the tumor is
5 relevant to humans and evaluated in the cancer assessment.
6 Comment. EPA should document a complete MOA evaluation for each relevant tumor type by
7 including a discussion on what is known about the key events in each tissue.
8 Response. MOA information available for tumors associated with exposure to 1,4-dioxane was
9 evaluated in the Toxicological Review (Section 4.7.3). The MOA by which 1,4-dioxane produces
10 liver, nasal, kidney, peritoneal (mesotheliomas), mammary gland, Zymbal gland, and subcutis
11 tumors is unknown, and the available data do not support any hypothesized mode of carcinogenic
12 action for 1,4-dioxane. Available data also do not identify whether 1,4-dioxane or one of its
13 metabolites is responsible for the observed effects. Thus, it is not possible to document a
14 complete MOA in any tissue. This conclusion is supported by five of the six external reviewers.
15 Comment. The parameters necessary to reproduce the total tumor analysis using the Bayesian
16 method (WinBUGS) are not provided; the analysis is poorly documented; and the rationale for
17 application of the analysis is incomplete.
18 Response. The BMDS (version 2.2Beta) MS_Combo approach for total tumor analysis that was
19 also included in support of the WinBUGS approach in the draft toxicological review, is now
20 highlighted as the main approach in the body of the document. The MS_Combo approach uses
21 the U.S. EPA's Benchmark Dose Software and is a transparent, reproducible approach that
22 provided similar to the output from the complex WinBUGS analysis. The WinBUGS analysis is
23 still included in this toxicological review as a supporting analysis in Appendix G. Additional
24 details on the WinBUGS analysis was included in Appendix G and the model code made
25 available via HERO (EPA, 2013). Using MS_Combo approach as the primary approach did not
26 result in any quantitative changes to the IUR.
27 Comment. The requirements for scientific data to support a MOA appear too stringent. EPA
28 should provide guidance on what would be considered sufficient scientific evidence to determine
29 a MOA.
30 Response. It is not feasible to describe the exact data that would be necessary to conclude that a
31 particular MOA was operating to induce the tumors observed following 1,4-dioxane exposure.
32 The data would fit the criteria described in the U.S. EPA's Guidelines for Carcinogen Risk
33 Assessment (U.S. EPA. 2005a).
34 Comment. The attribution of some tumor types to exposure to 1,4-dioxane is questionable based
35 on statistics, including subcutis fibromas and Zymbal Gland adenomas. There is also uncertainty
36 surrounding the origin of the tumors reported in the Kasai et al. (2009) study (e.g., may be the
37 result of metastatic deposition), and hence the assumption of biological independence among the
A-33
DRAFT - DO NOT CITE OR QUOTE
-------
1 tumor types included in the total tumor analysis is not supported. Thus, the pooling of tumor
2 types for derivation of the IUR in the draft TR leads to overestimation of the actual
3 carcinogenicity, and only tumor types with statistically significant differences in incidence rate
4 compared to control animals should be used. Additionally, the highest dose used in the Kasai et
5 al. (2009) study exceeds the maximum tolerated dose (MTD) and should be excluded from the
6 dose-response analysis to derive the IUR.
7 Response. The commenter suggested that Zymbal Gland adenomas should not be considered
8 related to 1,4-dioxane exposure because the incidence rate at the highest dose group was not
9 statistically different from control; however, the Peto test did find a statistically significant
10 increasing trend. Tumor types were included in the analysis if they showed a statistical difference
11 from control or a statistically significant trend was evident. Zymbal Gland adenomas were
12 included in the analysis because the Guidelines for Carcinogen Risk Assessment (U.S. EPA.
13 2005a) do not require site concordance and a statistically significant dose-response trend was
14 observed for these tumors. Similarly, subcutis fibromas were included in the total tumor analysis
15 because a statistically significant difference was seen in the mid dose group. The rationale for
16 inclusion of tumors in the multitumor analysis is described in Section 5.4.4.2. Additional
17 scientific information would be required to evaluate the hypothesis that the tumors "may be the
18 result of metastatic deposition."
19 The Kasai et al. (2008) study demonstrates that the high dose used in determining the IUR is
20 below the MTD for 1,4-dioxane. Kasai et al. (2008) state that the MTD is likely higher than the
7 \ / J O
21 111 ppm reported by Torkelson et al. The 3,200 ppm high dose for 13wks in Kasai et al. (2008) is
22 higher than the 1,250 ppm dose used in Kasai et al. (2009). and no overt toxicity was reported at
23 the 3,200 ppm.
A.4.3 PBPK modeling
24 Comment: PBPK models of sufficient quality are available and should have been used to reduce
25 uncertainty in both the oral and inhalation assessments. Technical errors were identified in the
26 PBPK analysis that should be addressed and the use of the models should be reevaluated for both
27 the oral and inhalation assessment.
28 Response. The model code errors noted in the public comments were addressed as noted below;
29 however, the changes did not significantly impact model predictions nor the overall decision on
30 model use in the assessment.
31 Comment. The permeation constant to describe the slowly perfused (diffusion-limited) tissue
32 compartment was improperly used in the PBPK model.
33 Response. If one assumes that the exiting venous concentration is at equilibrium with the tissue,
34 then the diffusion-limited tissue mass balance could be described as was shown in the model
35 code. It does slowly transport in/out of the tissue while having the property that the tissue moves
A-34
DRAFT - DO NOT CITE OR QUOTE
-------
1 toward equilibrium with the blood, so it is empirically correct, though it is acknowledged that this
2 was not the most common way to code this compartment. Therefore, to be up-to-date with current
3 modeling practices, the blood flow to the slowly perfused tissues (QS) was used instead of the
4 diffusion limited constant (SPDC) change was made to the model code; however, this had very
5 minimal quantitative impact on model output. Additionally, the fraction of fat and slowly
6 perfused tissue compartments was updated to be more similar to the values used in the values
7 used in the published models (see Table B-l).
8 Comment. The metabolism of 1,4-dioxane in misused a zero order rate constant in the equation.
9 Response. The metabolic constant was correctly used in the model code as a first order rate
10 constant; however, it was incorrectly described in the text and code comments as zero-order. The
11 description of the rate constant was corrected in the text and the model code to be clear it is a
12 first-order rate constant.
13 Comment. The model description for the urinary excretion of HEAA is not adjusted to the ratio
14 of the molecular weights, thus under predicting the concentration of HEAA in urine.
15 Response. The reviewer is correct that the molecular weight was not accounted for, and since the
16 model mass units are in milligrams, the urinary excretion was corrected to account for the mass
17 conversion to HEAA. The corrected model predicts the human urinary HEAA early time points
18 well and over predicts the latter time points (694 mg vs. 621 mg) - See Appendix B. Following
19 all updates to the model, metabolic parameters were re-optimized and the plots and predictions
20 updated in Appendix B. These changes improved the model fits, but the model predictions of
21 blood 1,4-dioxane were still 4- to 7-fold lower than the data.
22 Comment. Complete model code (including all .m and .csl files) should be included for the
23 public and reviewers to use. It should be clear what model code was used to generate each figure
24 in the appendix.
25 Response. New practice within NCEA for transparency is to make the model code accessible via
26 the Health and Environmental Research Online (HERO) database. The model code is now
27 available via the online database and has been removed from the appendix (EPA, 2013).
28 Comment. Although the Young et al. (1977b) paper does have value in the model development
29 process, there are issues with the study design and exposure estimation, so it should not be used
30 to dismiss the use of the PBPK model for the assessment.
31 Response. In the absence of evidence to the contrary, the Agency cannot discount the human
32 blood kinetic data published by Young et al. (1977a). As the commenter noted, the liquids likely
33 absorbed some 1,4-dioxane; however, if the volume of air they extract is much less than the
34 volume inhaled by a subject in an hour, then they won't contribute much to the overall
35 absorption. Thus, this reason presented by the commenter is not sufficient for the Agency to
36 discount the data for model validation.
A-35
DRAFT - DO NOT CITE OR QUOTE
-------
A.4.4 Other comments
1 Comment. There are other relevant data that are missing from this assessment. Reports that
2 should be referenced include: Takano et al. (2010) JHealth Sci 56(5): 557-565 and Department
3 of the Army (2010) Toxicology Report No., 87-XE-08WR-09, Studies on Metabolism of 1,4-
4 dioxane.
5 Response. These same references were mentioned by a member of the independent external peer
6 review panel - refer to the response to the inhalation assessment update general charge question
7 #2 above. Briefly, Takano et al. (2010) was evaluated and added to the assessment in Section
8 3.5.2.5. The Army study is a report that has not undergone formal peer-review and thus, is
9 generally not considered in the development of an IRIS assessment.
A-36
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX B. EVALUATION OF EXISTING
PHARMACOKINETIC MODELS 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.. 1978b: 1978a) and humans (Young et al. 1977a) 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 (Takano et al.. 2010; Fisher et al.. 1997; Leung and Paustenbach. 1990b:
8 Reitzetal. 1990b).
9 In developing toxicity values for 1,4-dioxane, the available PBPK models were evaluated for
10 their ability to predict observations made in experimental studies of rat and human exposures to
11 1,4-dioxane. The model of Reitz et al. (1990a) was identified for further consideration to assist in the
12 derivation of toxicity values. Issues related to the biological plausibility of parameter values in the Reitz
13 et al. (1990a) human model were identified. The model was able to predict the only available human
14 inhalation data set (Young et al.. 1977a) by increasing (i.e., doubling) parameter values for human
15 alveolar ventilation, cardiac output, and the blood:air partition coefficient above the measured values.
16 Furthermore, the measured value for the slowly perfused tissue:air partition coefficient (i.e., muscle) was
17 replaced with the measured liver value to improve the fit. Analysis of the Young et al. (1977a) human
18 data suggested that the apparent volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold
19 higher in rats than humans, presumably due to species differences in tissue partitioning or other process
20 not represented in the model. Subsequent exercising of the model demonstrated that selecting a human
21 slowly perfused tissue:air partition coefficient much lower than the measured rat value resulted in better
22 agreement between model predictions of 1,4-dioxane in blood and experimental observations. Based upon
23 these observations, several model parameters (e.g., metabolism/elimination parameters) were
24 re-calibrated using biologically plausible values for flow rates and tissue:air partition coefficients.
25 This appendix describes activities conducted in the evaluation of the empirical models (1978c. d;
26 Young et al.. 1977b) and re-calibration and exercising of the Reitz et al. (1990a) PBPK model using
27 parameter values identified by Leung and Paustenbach (1990b) and Sweeney et al. (2008b), as well as
28 optimized values, to determine the potential utility of the models for 1,4-dioxane for interspecies and
29 route-to-route extrapolation.
B-l
DRAFT - DO NOT CITE OR QUOTE
-------
B.2 Implementation of the Empirical Models in acsIX
1 The scope of this effort consisted of implementation of the Young et al. (1978c. d; 1977b)
2 empirical rat and human models using acsIX, version 3.0.2.1 (Aegis Technologies, Huntsville, AL).
3 Using the model descriptions and equations given in Young et al. (1978c. d; 1977b). model code was
4 developed for the empirical models and executed, simulating the reported experimental conditions. The
5 model output was then compared with the model output reported in Young et al. (1978c. d; 1977b). All
6 model files are available electronically via HERO (U.S. EPA, 2013).
B.2.1 Model Descriptions
7 The empirical model of Young et al. (1978b: 1978a) for 1,4-dioxane in rats is shown in
8 Figure B-l. This is a single-compartment model that describes the absorption and metabolism kinetics of
9 1,4-dioxane in blood and urine. Pulmonary absorption is described by a first-order rate constant (kiNH).
10 The metabolism of 1,4-dioxane and subsequent appearance of HEAA is described by Michaelis-Menten
11 kinetics governed by a maximum rate (Vmax, mg/hour and affinity constant (Km, mg). The elimination of
12 both 1,4-dioxane and HEAA were described with first-order elimination rate constants, ke and kme,
13 respectively (hour"1) by which 35% of 1,4-dioxane and 100% of HEAA appear in the urine, while 65% of
14 1,4-dioxane is exhaled. Blood concentration of 1,4-dioxane was determined by dividing the amount of
15 1,4-dioxane in blood by a volume of distribution (Vd) of 0.301 L, which was the average Vd determined
16 from the i.v. dose studies.
Inhalation (kINH)
i.v. admin
Diox,,^
dt
Km + Dioxbod!
- ke x Dioxbods
dt
Km+Dioxb
body
Exhaled
kexDioxbody S*
> Urine (35%)
kme xHEAA
•>• Urine
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1_978a).
Figure B-l Schematic representation of empirical model for 1,4-dioxane in rats.
17 Figure B-2 illustrates the Young et al. (1977b) human empirical model for 1,4-dioxane. Like the
18 rat model, the human model predicts blood 1,4-dioxane and urinary 1,4-dioxane and HEAA levels using a
19 single-compartment structure. However, the metabolism of 1,4-dioxane to HEAA in humans is modeled
20 as a first-order process governed by a rate constant, KM (hour"1). Urinary deposition of 1,4-dioxane and
21 HEAA is described using the first order rate constants, ke (dlox) and kme (HEAA), respectively. Pulmonary
22 absorption is described similar to the approach used in the rat empirical model. Blood concentrations of
23 1,4-dioxane and HEAA are calculated as instantaneous amount (mg) divided by volume of distribution
24 (Vd): Vd(diox) or Vd(HEAA) (104 and 480 mL/kg BW, respectively [calculated by Young et al. (1977b)]).
B-2
DRAFT - DO NOT CITE OR QUOTE
-------
Inhalation
Dioxane
K
M
HEAA
* x
e (diox)
(HEAA)
Urine
Cumulative
Dioxane and
HEAA
Source: Reprinted with permission of Taylor & Francis, Young et al. (1977a).
Figure B-2 Schematic representation of empirical model for 1,4-dioxane in humans.
B.2.2 Modifications to the Empirical Models
2 Several modifications were made to the empirical models. The need for the modifications arose in
3 some cases from incomplete reporting of the Young et al. (1978b: 1978a: 1977a) studies and in other
4 cases from the desire to add capabilities to the models to assist in the derivation of toxicity values.
5 For the rat model, no information was given by Young et al. (1978b: 1978a) regarding the
6 parameterization of pulmonary absorption (or exhalation) or i.v. administration of 1,4-dioxane. Therefore,
7 additional parameters were added to simulate these processes in the simplest form. To replicate
8 1,4-dioxane inhalation, a first-order rate constant, ki^n (hour"1), was introduced. kiNH was multiplied by
9 the inhalation concentration and the respiratory minute volume of 0.238 L/minute (Young etal.. 1978b:
10 1978a). The value for ki^n (0.43 hour"1) was estimated by optimization against the blood time course data
11 of Young et al. (1978b: 1978a). Intravenous (i.v.) administration was modeled as instantaneous
12 appearance of the full dose at the start of the simulation. Rat urinary HEAA data were reported by Young
13 et al. (1978b: 1978a) in units of concentration. To simulate urinary HEAA concentration, an estimate of
14 urine volume was required. Since observed urinary volumes were not reported by Young et al. (1978b:
15 1978a). a standard rat urine production rate of 0.00145 L/hour was used.
16 For humans, Young et al. (1977a) used a fixed 1,4-dioxane inhalation uptake rate of
17 76.1 mg/hour, which corresponded to observations during a 50 ppm exposure. In order to facilitate
18 user-specified inhalation concentrations, pulmonary absorption was modeled similar to the rat model
19 addition (e.g., using ki^n, 1.06 hour"1) but using a human minute volume of 7.5 L/minute. Urinary HEAA
20 data were reported by Young et al. (1977a) as a cumulative amount (mg) of HEAA. Cumulative amount
21 of HEAA in the urine is readily calculated from the rate of transfer of HEAA from plasma to urine, so no
22 modification was necessary to simulate this dose metric for humans.
B-3
DRAFT - DO NOT CITE OR QUOTE
-------
1 Neither empirical model of Young et al. (1978b: 1978a: 1977a) described oral uptake of
2 1,4-dioxane. Adequate data to estimate oral absorption parameters are not available for either rats or
3 humans; therefore, neither empirical model was modified to include oral uptake.
B.2.3 Results
4 The acslX implementation of the Young et al. (1978b: 1978a) rat empirical model is in good
5 agreement with the 1,4-dioxane blood levels from the i.v. experiments and the model output reported in
6 the published paper (Figure B-3). However, the acslX version predicts urinary HEAA following i.v. dose
7 to reach a maximum sooner than the measured and predicted levels reported in the paper (Figure B-4).
8 These discrepancies may be due, at least in part, to the reliance in the acslX implementation on a constant,
9 standard, urine volume rather than experimental measurements, which may have been different from the
10 assumed value and may have varied over time. Unreported model parameters (e.g., lag times for
11 appearance of excreted HEAA in bladder urine) may also contribute to the discrepancy.
10000
I
a
u
o
j>
m
1000
100
30 40
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 acslX implementation
(left) and published (right) empirical rat model simulations of i.v.
administration experiments.
B-4
DRAFT - DO NOT CITE OR QUOTE
-------
10000
1000
100
f°°0
UD°DO°Q_ _ B. ^£ —
J
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1_978a).
Figure B-4 Output of HEAA urine level data from acslXtreme implementation of
the empirical rat model (left) and published (right)data following i.v.
administration experiments. The lines in the figure from Young et al.
(1978c. d) are best fit lines, and do not represent empirical model
simulations.
1 The Young et al. (1978b: 1978a) report did not provide model predictions for the 50-ppm
2 inhalation experiment. However, the acslX implementation produces blood 1,4-dioxane predictions that
3 are similar to the reported observations (Figure B-5). As with the urine data from the i.v. experiment, the
4 amount of HEAA in urine predicted using the acslX implementation was approximately threefold lower
5 than the observations However, this prediction is the amount of HEAA excreted over time and does not
6 rely on an estimate of urine volume to calculate, thus the reason for the discrepancy is likely due
7 unreported model parameters (e.g., lag times for appearance of excreted HEAA in bladder urine) or to
8 more complex kinetics than described using this simple model structure.
B-5
DRAFT - DO NOT CITE OR QUOTE
-------
—.25
ttt
D Young et al. U978a, b)
observations
1"
J2
3
5 5
o
Time(hrs)
Figure B-5 acslX empirical rat model predictions of blood 1,4-dioxane
concentration and total amount of HEAA levels in the urine for a
6-hour, 50-ppm 1,4-dioxane inhalation exposure.
1 Further evaluation of the Young et al. (1978c. d) empirical model was conducted against
2 subchronic inhalation exposure data reported by Kasai et al. (2008). In the experimental study, male and
3 female F344 rats were exposed to 0, 100, 200, 400, 800, 1,600, 3,200, or 6,400ppm 1,4-dioxane in a
4 13-week inhalation study. With the exception of the 6,400ppm dose, the Kasai et al. (2008) doses were
5 within the range of the doses modeled by Young et al. (1978c. d); however, the model was unable to fit
6 the measured 1,4-dioxane plasma levels reported by Kasai et al. (2008) (Figure B-6). This is could be due
7 to a difference in metabolism of 1,4-dioxane following the single exposure (Young et al.. 1978c. d)
8 compared to the 13-week repeated exposure (Kasai et al.. 2008).
B-6
DRAFT - DO NOT CITE OR QUOTE
-------
3000
f 2500
i 2000
• Male Data
D Female Data
D Mode I prediction
400 800 1600
Inhalation Exposure (pprr't
3200
Figure B-6 acslX predictions of blood 1,4-dioxane levels using the Young et al.
(1978c, d) model compared with data from Kasai et al. (2008).
1 Inhalation data for a single exposure level (50 ppm) are available for humans. The acslX
2 predictions of the blood 1,4-dioxane observations are similar to the predictions reported in Young et al.
3 (1977a) (Figure B-). Limited blood HEAA data were reported (n = 2-3 animals), and the specimen
4 analysis was highly problematic (e.g., an analytical interference was sometimes present from which
5 HEAA could not be separated). For this reason, Young et al. (1977a) did not compare predictions of the
6 blood HEAA data to observations in their manuscript. Young et al. (1977a) only compared model
7 simulations to blood 1,4-dioxane in their report.
8 Data for cumulative urinary HEAA amounts are provided in Young et al. (1977b). and no
9 analytical problems associated with these data were reported. The acslX prediction of the HEAA kinetics
10 profile is similar to the observations (Figure B-8). Unlike urinary HEAA observations in the rat, human
11 observations were reported as cumulative amount produced, negating the need for urine volume data.
12 Therefore, discrepancies between model predictions and experimental observations were reduced.
B-7
DRAFT - DO NOT CITE OR QUOTE
-------
O
4
•14-Dio>'3ne predicted- acslX version
Young i!977) empirical model
D 1,4-Dioxane observed
- —— HEAA predicted- acslX uersionYoung
(1977) empirical model
• HEAA observed
\
0 2 4 6 8 10 12 T™.Houn
Time(hrs)
Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
Figure B-7 Output of 1,4-dioxane and HEAA blood concentrations from the acslX
implementation (left) and published (right) data of a 6-hour, 50-ppm
inhalation exposure.
D
C 500 -
<
E 100 -
g
u
O
^^acsIX version- Young et al. (1977)
empirical model
D Observed
0 5 10 15
Time (hrs)
Source: Reprinted with permission of Taylor & Francis, Young et al. (1977a).
B-8
DRAFT - DO NOT CITE OR QUOTE
-------
B.2.4 Figure B-8 Observations and acsIX predictions of the cumulative
amount of HEAA in human urine following a 6-hour, 50-ppm inhalation
exposure.Conclusions for Empirical Model Implementation
1 The empirical models described by Young et al. (1978b: 1978a: 1977a) for rats and humans were
2 implemented using acsIX. The models were modified to allow for user-defined inhalation exposures 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 to describe oral absorption as adequate data are not available for parameter
5 estimation. The acsIX predictions of 1,4-dioxane in the blood are similar to the published data and
6 simulations of 6-hour, 50-ppm inhalation exposures in rats (Figure B-5) and humans (Figure B-7) and 3 to
7 1,000 mg/kg i.v. doses in rats (Figure B-3). However, the acsIX version predicts lower urinary HEAA
8 amounts and concentrations in rats appearing earlier than either the Young et al. (1978b: 1978a) model
9 predictions or the experimental observations (Figures B-4 and B-5). The lower predicted urinary HEAA
10 concentrations in the acslXtreme implementation for rats are likely due to use of default values for urine
11 volume in the absence of measured volumes. The reason for the differences in time-to-peak levels or
12 amount of HEAA in urine is unknown, but may be the result of an unreported adjustment by Young et al.
13 (1978b: 1978a) in model parameter values or more complex kinetics than can be described with this
14 model structure. Additionally, the acsIX implementation of the Young et al. (1978c. d_) model failed to
15 provide adequate fit to blood data reported following subchronic inhalation of 1,4-dioxane in rats at the
16 two high doses (Kasai et al.. 2008).
17 For humans, Young et al. (1977a) did not report model predictions of urinary HEAA levels. The
18 urinary HEAA levels predicted by acsIX approximated the observations reasonably well (Figure B-8),
19 while the blood HEAA did not (Figure B-7). However, unlike the situation in rats, these urine data are not
20 dependent on urine volumes (observations were reported as cumulative HEAA amount rather than HEAA
21 concentration). Presently, there is no explanation for the lack of fit of the empirical model to the blood
22 HEAA data. Since no blood HEAA model fits were shown in Young et al. , it is unclear if the
23 discrepancy is in the original model or only in the acsIX implementation.
B.3 Initial Evaluation of the PBPK Models
24 The PBPK model of Reitz et al. (1990b) was selected for further evaluation of its potential
25 application in this assessment. The model was not sufficient as published, and thus was re-calibrated
26 using measured values for cardiac and alveolar flow rates and tissue:air partition coefficients (Sweeney et
27 al.. 2008b: Leung and Paustenbach. 1990b). The predictions of blood and urine levels of 1,4-dioxane and
28 HEAA, respectively, from the re-calibrated model were compared with the empirical model predictions of
29 the same dosimeters to determine whether the re-calibrated PBPK model could perform similarly to the
30 empirical model. As part of the PBPK model evaluation, EPA performed a sensitivity analysis to identify
31 the model parameters having the greatest influence on the primary dosimeter of interest, the blood level of
32 1,4-dioxane. Variability data for the experimental measurements of the tissue:air partition coefficients
33 were incorporated to determine a range of model outputs bounded by biologically plausible values for
B-9
DRAFT - DO NOT CITE OR QUOTE
-------
1 these parameters. Additionally, the models were tested using first-order metabolism (instead of Michaelis-
2 Menten saturable metabolism) to determine if better model predictions could be generated.
B.3.1 Initial Recalibration of the Reitz et al. PBPK Model
3 Concern regarding adjustments made to some of the parameter values in Reitz et al. (1990a)
4 prompted a re-calibration of the Reitz et al. (1990a) human PBPK model using more biologically
5 plausible values for all measured parameter values. Reitz et al. (1990a) doubled the measured
6 physiological flows and blood:air partition coefficient and substituted the slowly-perfused tissue:air
7 partition coefficient with the liverair value in order to attain an adequate fit to the observations. This
8 approach increases uncertainty in these parameter values, and in the utilization of the model for
9 extrapolation. Therefore, the model was re-calibrated using parameter values that are more biologically
10 plausible to determine whether an adequate fit of the model to the available data can be attained.
B.3.2 Flow Rates
11 The cardiac output of 30 L/hour/kg0 74 (Table B-l) reported by Reitz et al. (Reitz etal. \990a) is
12 approximately double the mean resting value of 14 L/hour/kg0 74 reported in the widely accepted
13 compendium of Brown et al. (1997). Resting cardiac output was reported to be 5.2 L/minute (or 14
14 L/hour/kg074), while strenuous exercise resulted in a flow of 9.9 L/minute (or 26 L/hour/kg074) (Brown et
15 al.. 1997). Brown et al. (1997) also cite the ICRP (1975) as having a mean respiratory minute volume of
16 7.5 L/minute, which results in an alveolar ventilation rate of 6.86 L/minute (assuming 8.5% lung dead
17 space, (Overton et al.. 2001)). or 17.7 L/minute/kg0 74. Again, this is roughly half the value of 30
18 L/hour/kg0 74 employed for this parameter by Reitz et al. (1990a). Young et al. (1977a) reported that the
19 human subjects exposed to 50 ppm for 6 hours were resting inside a walk-in exposure chamber. Thus, use
20 of cardiac output and alveolar ventilation rates of 30 L/hour/kg074 is not consistent with the experimental
21 conditions being simulated.
22 A minute volume of 7.5 L/minute (or 17 L/hour/kg074) was used in the acslX implementation of
23 the Young et al. (1977b) model for volunteers having a mean BW of 84 kg and fit the blood 1,4-dioxane
24 data reasonably well. Based on these findings, the cardiac output and alveolar ventilation rates of 17.0 and
25 17.7 L/hour/kg0 74 were biologically plausible for the experimental subjects. These rate estimates are based
26 on calculations made using empirical data and are consistent with standard human values and the
27 experimental conditions (i.e., subject exertion level) reported by Young et al. (1977b). Therefore, these
28 flow values were chosen for the model re-calibration.
29
B-10
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-1 Human PBPK model parameter values published in literature and values used by EPA
in this assessment for 1,4-dioxane
Parameter (Abbreviation)
Body weight (BW)
Cardiac output (QCC)a
Alveolar ventilation (QPC)a
Reitz et al.
(1990b)
70
30
30
Leung and
Paustenbach
(1990W
84.1
15
15
Sweeney et al.
(2008b)
70
13
13
EPAb
84.1
17.0
17.7
Fractional Blood Flows
Liver (QLC)
Fat (QFC)
Richly perfused (QRC)
Slowly perfused (QSC)
0.25
0.05
0.52
0.18
0.25
0.05
0.51
0.19
0.227
0.052
0.472P
0.249
0.25
0.05
0.52P
0.18
Fractional Tissue Volumes
Liver (VLC)
Fat (VFC)
Richly perfused (VRC)
Slowly perfused (VSC)
Blood (VBC)
Unperfused tissue (VUC)
0.031
0.231
0.037
0.561
0.05
-
0.04
0.20
0.05
0.62
—
-
0.033
0.214
0.1 66q
0.437
0.079
0.071
0.04
0.20
0.05q
0.57
0.05
0.09
Partition Coefficients (PCs)
Blood:air(PB)
Fatair(PFA)
Liver: air (PLA)
Rapidly perfused tissue:air (PRA)
Slowly perfused tissue:air(PSA)
3,650C
851
1,557
1,557
1.5571
1,825 ±94°
(77 = 14)
851 ± 118°
(77=8)
1,557± 114a
(n=4)
1,557g
997 ± 254°
(n=6)
1,666 ±287
(7J=36)
865e
1,862 ±739'
(77 = 14)
560 ± 175"
(n=7)
1,348±290T
(77=7)
1,825
851
1,557
1,557
260j'm
Metabolic Constants
Maximum rate for 1,4-dioxane
metabolism (Vmaxc; mg/hr-kg BW07)
Metabolic affinity constant (Km; mg/L)
HEAA urinary elimination rate constant
(kme, hour-1)
12.5"
3.00
0.56
13.3°
15
-
54, 75, or192k
29, 32, or 147'
0.35
5.8j
5.3J
0.30j
aL/hour/kg BW074
bValues utilized by EPA in this assessment. Body weight was mean weight reported by Young et al. (197_7_b).
°Doubled from experimental value (1825) to obtain better fit to human data (Reitz et al.. 1990bl
dLeung as Paustenbach (1_990b) did not state if the values were reported ± standard deviation or standard error.
eAverage of Reitz et al. (1990b) rat value and mouse value determined by Sweeney et al. (2008b).
'Assumed equal to the measurement for rat tissue determined by Sweeney et al. (2008b).
9Assumed equal to liverair partition coefficient.
hAssumed equal to mouse kidney determined by Sweeney et al. (2008b).
'Authors reported poor fits to the venous blood data for rats and humans when the experimentally determined muscle:air partition
coefficient was used (value not reported) and had improved fits of the data when the partition coefficient for liverair was used.
'Obtained by model optimization.
kUsed parallelogram scaling approach based on scaled in vitro data to give a range of values referred to by the authors as
"minimum, representative, and maximum."
'Scaled rat in vitro data according to in vitro human:rat ratios to give a similar range as Vmax, referred to by the authors as
"minimum, representative, and maximum."
"Value used in Figure B-11, estimated 4-fold lower value than Leung as Paustenbach (1_990b) because recalibrated model was
predictions were 4- to 7-fold lower than the data; however, this parameter value is not considered "biologically plausible."
"Reported in manuscript as 6.55 mg/hr-kg BW086
"Reported in manuscript as 6.55 mg/hr-kg BW086
"Calculated from QRC=1-(QFC+QSC+QLC)
Calculated from VRC=1-(VLC+VFC+VSC+VBC+VUC)
Converted to mg/hr-kg BW for consistency.
Converted to mg/hr-kg BW07 for consistency.
B-ll
DRAFT - DO NOT CITE OR QUOTE
-------
B.3.3 Partition Coefficients
1 Two data sources are available for the tissue:air equilibrium partition coefficients for 1,4-dioxane:
2 Leung and Paustenbach (1990a) and Sweeney et al. (2008a). Both investigators used vial equilibration
3 techniques for experimental determinations. The values reported in Leung and Paustenbach (1990b) were
4 also used, at least as starting points, by Reitz et al. (1990b). Leung and Paustenbach (1990b) reported
5 mean values and an indication of variance (it was not clear if the values were standard deviations or
6 standard errors) for human blood:air, rat blood:air, rat liverair, rat muscle:air (e.g., slowly perfused
7 tissue:air), and rat fatair (Table B-l). They assumed the rapidly perfused tissue:air partition coefficient
8 was equal to the value for the liver and that all human tissue partition coefficients were equivalent to the
9 rat, except where the separate determination was made for human blood:air partition coefficient.
10 Sweeney et al. (2008b) experimentally determined partition coefficients for blood:air (mouse, rat,
11 and human), liverair (mouse and rat), fatair (mouse), richly perfused tissue:air (mouse), and slowly
12 perfused tissue:air (mouse). Values for human tissue:air partition coefficients for the model were
13 estimated as averages of rat and mouse values (liverair, fatair, and slowly perfused tissue:air) or set
14 equal to the mouse value (richly perfused:air set equal to mouse kidney:air partition coefficient) (Sweeney
15 et al.. 2008b). For example, the human fatair partition coefficient, used an average (851) of the Reitz et
16 al. (1990b) rat value (851) and their experimentally determined mouse value (879) (Sweeney et al..
17 2QQ8b).
18 For the PBPK model implementation, tissue:blood partition coefficients for each compartment
19 were determined by dividing the tissue:air partition coefficients by the blood:air partition coefficient.
B.3.4 Calibration Method
20 The PBPK model was re-calibrated three times using the physiological values selected by EPA
21 (current assessment, Table B-l) and the (1) partition coefficients of Leung and Paustenbach (1990a). (2)
22 Sweeney et al. (2008a). and (3) biologically plausible values based on these two publications, separately.
23 For each calibration, the metabolic parameters VmaxC and Km, were simultaneously fit (using the parameter
24 estimation tool provided in the acslX software) to the output of 1,4-dioxane blood concentrations
25 generated by the acslX implementation of the Young et al. (1977a) empirical human model for a 6 hour,
26 50 ppm inhalation exposure. Subsequently, the HEAA urinary elimination rate constant, kme, was fitted to
27 the urine HEAA predictions from the empirical model. The empirical model predictions that were
28 validated against the experimental observations were used to provide a more robust data set for model
29 fitting, since the empirical model simulation provided 240 data points (one prediction every 0.1 hour)
30 compared with hourly experimental observations, and to avoid introducing error by calibrating the model
31 to data digitally captured from Young et al. (1977a).
B-12
DRAFT - DO NOT CITE OR QUOTE
-------
B.3.5 Results
1 Results of the model re-calibration are provided in Table B-2. The re-calibrated values for VmaxC
2 and kme associated with the Leung and Paustenbach (1990a) or Sweeney et al. (2008a) tissue:air partition
3 coefficients are very similar. Plots of predicted and experimentally observed blood 1,4-dioxane and
4 urinary HEAA levels are shown in Figure B-6 and B-10 for Leung and Paustenbach (1990b) and Sweeney
5 et al. (2008b) partition coefficients. Neither re-calibration resulted in an adequate fit to the blood
6 1,4-dioxane data from the empirical model output or the experimental observations. Re-calibration using
7 either the Leung and Paustenbach (1990b) or Sweeney et al. (2008b) partition coefficients resulted in
8 blood 1,4-dioxane predictions that were 4- to 7-fold lower than empirical model predictions or
9 observations.
10 The refitted values for kme resulted in HEAA levels in urine that were very similar to the
11 empirical model output (compare Figure B-7, Figure B-6, and Figure B-7), which was not surprising,
12 given the fitting of a single parameter to the data.
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
13
14
Source of Partition Coefficients
Maximum rate for 1,4-dioxane metabolism
(Vmaxc)
Metabolic affinity constant (Km)c
HEAA urinary elimination rate constant
(kme)d
Leung and Paustenbach
(1990a)
4.9
1.8
0.27
Sweeney et al. (2008a)
4.0
0.78
0.25
EPA
5.8
5.3
0.30
aCardiac output = 17.0 L/hour/kg BWU'4, alveolar ventilation = 17.7 L/hour/kg BWU'4
bmg/hour/kg BW07
°mg/L
"hour"1
B-13
DRAFT - DO NOT CITE OR QUOTE
-------
Hi
o
5
*
o
_0
ffi
PBPK predicted
Em p i r i c al p r e d i cte d
1.4-Dioxane observed
N
_£_
01 SuO
- M0
<
O
£ 300
<
Time (hrs)
PBPK predicted
~ ~™ Empirical predicted
D Observed
10 15
Time (hrs)
Figure B-6 Human predicted and observed blood 1,4-dioxane concentrations (left)
and urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-dioxane
exposure and re-calibration of the PBPK model with tissue:air
partition coefficient values from Leung and Paustenbach (1990b).
PBPK predicted
~ ~™ Enplrical predicted
D 1.4-Dioxane observe!
j
4
5
6
0246
Time (hrs
Figure B-7 Human predicted and observed blood 1,4-dioxane concentrations (left)
and urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-dioxane
exposure and re-calibration of the PBPK model with tissue:air partition
coefficient valuesfrom Sweeney et al. (2008b).
Model outputs of the blood 1,4-dioxane and urinary HEAA levels using the EPA suggested
(Table B-2) parameters are shown in Figure B-8. To obtain these improved fits, a very low value for the
slowly perfused tissue:air partition coefficient (22) was used. The value was 4- to 6-fold lower than the
measured values reported in Leung and Paustenbach (1990a) and Sweeney et al. (2008a), and 7-fold
lower than the value used by Reitz et al. (1990a). While the predicted maximum blood 1,4-dioxane levels
B-14
DRAFT - DO NOT CITE OR QUOTE
-------
1 are much closer to the observations (e.g., 2- to 3-fold lower than the observations), the value used for the
2 slowly perfused tissue partition coefficient is not supported by laboratory data.
PBPK predicted
~ Empirical predicted
D 1.4-Dio:-;ane observed
^— PBPK predicted
— Empirical predicted
D HEAA observed
Time(hrs)
10 15
Time(hrs)
Figure B-8 Human predicted and observed blood 1,4-dioxane concentrations (left)
and urinary HEAA levels (right) following a 6-hour, 50 ppm 1,4-dioxane
exposure using EPA estimated biologically plausible parameters (Table
B-l).
B.3.6 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.. 1977a) and measured mean tissue:air 1,4-dioxane
5 partition coefficients reported by Leung and Paustenbach (1990a) or Sweeney et al. (2008a). The resulting
6 predictions of 1,4-dioxane in blood following a 6-hour, 50-ppm inhalation exposure were 4- to 7-fold
7 lower than either the observations or the empirical model predictions, while the predictions of urinary
8 HEAA by the PBPK and empirical models were similar to each other (Figure B-6 and Figure B-7).
9 Output from the model using biologically plausible physiological parameter values (Table B-l),
10 Figure B-8 shows that application of a value for the slowly perfused tissue:air partition coefficient, which
11 is 6-fold lower than the measured value reported by Leung and Paustenbach (1990a). results in closer
12 agreement of the predictions to observations. Thus, model re-calibration using experiment-specific flow
13 rates and mean measured partition coefficients does not result in an adequate fit of the PBPK model to the
14 available data.
15 The Sweeney et al. (2008a) PBPK model consisted of compartments for fat, liver, slowly
16 perfused, and other well perfused tissues. Lung and stomach compartments were used to describe the
17 route of 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 rat
B-15
DRAFT - DO NOT CITE OR QUOTE
-------
1 PBPK model were derived by optimization data from an i.v. exposure of 1,000 mg/kg data (Young et al..
2 1978b: 1978a) for induced metabolism. For uninduced metabolism data generated by i.v. exposures to 3,
3 10, 30, and 100 mg/kg were used (Young et al.. 1978b: 1978a). Data generated from the 300 mg/kg i.v.
4 exposure were not used to estimate VmaxC and Km. The best fitting values for VmaxC to estimate the blood
5 data from the Young et al. (1978b: 1978a) study using the Sweeney et al. (2008a) model resulted in VmaxC
6 values of 12.7, 10.8, 7.4 mg/kg-hr0 7; suggesting a gradual dose dependent increase in metabolic rate with
7 dose. These estimates were for a range of doses between 3 and 1,000 mg/kg i.v. dose. Although the
8 Sweeney et al. (2008a) model utilized two values for VmaxC (induced and uninduced), the PBPK model
9 does not include dose-dependent function description of the change of Vmax for i.v. doses between 100
10 and 1,000 mg/kg. PBPK model outputs were compared with other data not used in fitting model
11 parameters by visual inspection. The model predictions gave adequate match to the 1,4-dioxane
12 exhalation data after a 1,000 mg/kg i.v. dose. 1,4-Dioxane exhalation was overpredicted by a factor of
13 about 3 for the 10 mg/kg i.v. dose. Similarly, the simulations of exhaled 1,4-dioxane after oral dosing
14 were adequate at 1,000 mg/kg, and 100 mg/kg (within 50%), but poor at 10 mg/kg (model overpredicted
15 by a factor of five). The fit of the model to the human data (Young et al.. 1977a) was also problematic
16 (Sweeney et al.. 2008a). Using physiological parameters of Brown et al. (1997) and measured partitioning
17 parameters (Sweeney et al.. 2008a: Leung and Paustenbach. 1990a) with no metabolism, measured blood
18 1,4-dioxane concentrations reported by Young et al. (1977a) could not be achieved using the reported
19 exposure concentrations. Inclusion of any metabolism further decreased predicted blood concentrations. If
20 estimated metabolism rates were used with the reported exposure concentration, urinary metabolite
21 (HEAA) excretion was underpredicted (Sweeney et al.. 2008a). Thus, the models were inadequate to use
22 for rat to human extrapolation.
B.3.7 Sensitivity Analysis
23 A sensitivity analysis of the Reitz et al. (1990a) model was performed, using the EPA values
24 listed in Table B-l, to determine which PBPK model parameters exert the greatest influence on the
25 outcome of dosimeters of interest—in this case, the concentration of 1,4-dioxane in blood. Knowledge of
26 model sensitivity is useful for guiding the choice of parameter values to minimize model uncertainty.
B.3.8 Method
27 A univariate sensitivity analysis was performed on all of the model parameters for two endpoints:
28 blood 1,4-dioxane concentrations after 1 and 4 hours of exposure. These time points were chosen to
29 assess sensitivity during periods of rapid uptake (1 hour) and as the model approached steady state
30 (4 hours) for blood 1,4-dioxane. Model parameters were perturbated 1% above and below nominal values
31 and sensitivity coefficients were calculated as follows:
B-16
DRAFT - DO NOT CITE OR QUOTE
-------
. f(x+&x)-f(x) -v
/ \x)~
Ax- /(.v)
1 where * 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.3.9 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-9. The three parameters with the highest sensitivity coefficients in
8 descending order are alveolar ventilation (QPC), the blood:air partition coefficient (PB), and the slowly
9 perfused tissue:air partition coefficient (PSA). Not surprisingly, these were the parameters that were
10 doubled or given surrogate values in the Reitz et al. (1990a) model in order to achieve an adequate fit to
11 the data. Because of the large influence of these parameters on the model, it is important to assign values
12 to these parameters in which high confidence is placed, in order to reduce model uncertainty.
Sensitivity Coefficients: CV - 1hr
0.01 0.10 1.00
QPC
PB
* PSA
'S
I QSC
ro
°- QCC
vmaxc
K.
I
I
|
I
I
|
Sensitivity Coefficients: CV - 4 hr
0.01 0.10
QPC
PB
* PSA
'S
ro VmaxC
ro
°- K*
PRA
QSC
1.00
I
I
]
I
Figure B-9 The highest seven sensitivity coefficients (and associated parameters)
for blood 1,4-dioxane concentrations (CV) at 1 (left) and 4 (right) hours
of a 50-ppm inhalation exposure.
B.4 PBPK Model Exercises Using Biologically Plausible Parameter
Boundaries
13 The PBPK model includes numerous physiological parameters whose values are typically taken
14 from experimental observations. In particular, values for the flow rates (cardiac output and alveolar
B-17
DRAFT - DO NOT CITE OR QUOTE
-------
1 ventilation) and tissue:air partition coefficients (i.e., mean and standard deviations) are available from
2 multiple sources as means and variances. The PBPK model was exercised by varying the partition
3 coefficients over the range of biological plausibility (parameter mean ± 2 standard deviations),
4 re-calibrating the metabolism and elimination parameters, and exploring the resulting range of blood
5 1,4-dioxane concentration time course predictions. Cardiac output and alveolar ventilation were not
6 varied because the experiment-specific values used did not include any measure of inter-individual
7 variation.
B.4.1 Observations Regarding the Volume of Distribution
8 Young et al. (1978b: 1978a) used experimental observations to estimate a Vd for 1,4-dioxane in
9 rats of 301 mL or 1,204 mL/kg BW. For humans, the Vd was estimated to be 104 mL/kg BW (Young et
10 al.. 1977a). It is possible that a very large volume of the slowly perfused tissues in the body of rats and
11 humans may be a significant contributor to the estimated 10-fold difference in distribution volumes for
12 the two species. This raises doubt regarding the appropriateness of using the measured rat slowly perfused
13 tissue:air partition coefficient as a surrogate values for humans in the PBPK model.
B.4.2 Defining Boundaries for Parameter Values
14 Given the possible 10-fold species differences in the apparent Vd for 1,4-dioxane in rats and
15 humans, boundary values for the partition coefficients were chosen to exercise the PBPK model across its
16 performance range to either minimize or maximize the simulated Vd. This was accomplished by defining
17 biologically plausible values for the partition coefficients as the mean ± 2 standard deviations of the
18 measured values. Thus, to minimize the simulated Vd for 1,4-dioxane, the selected blood:air partition
19 coefficient was chosen to be the mean + 2 standard deviations, while all of the other tissue:air partition
20 coefficients were chosen to be the mean - 2 standard deviations. This created conditions that would
21 sequester 1,4-dioxane in the blood, away from other tissues. To maximize the simulated 1,4-dioxane Vd,
22 the opposite selections were made: blood:air and other tissue:air partition coefficients were chosen as the
23 mean - 2 standard deviations and mean + 2 standard deviations, respectively. Subsequently, VmaxC, Km,
24 and kme were optimized to the empirical model output data as described in Section B.3.4. This procedure
25 was performed for both the Leung and Paustenbach (1990a) and Sweeney et al. (2008a) partition
26 coefficients (Table B-l). The two predicted time courses resulting from the re-calibrated model with
27 partition coefficients chosen to minimize or maximize the 1,4-dioxane Vd represent the range of model
28 performance as bounded by biologically plausible parameter values.
B.4.3 Results
29 The predicted time courses for a 6-hour, 50-ppm inhalation exposure for the re-calibrated human
30 PBPK model with mean (central tendency) and ± 2 standard deviations from the mean values for partition
B-18
DRAFT - DO NOT CITE OR QUOTE
-------
1 coefficients are shown in Figure B-10 for the Leung and Paustenbach (1990a) values and Figure B-l 1 for
2 the Sweeney et al. (2008a) values. The resulting fitted values for VmaxC, Km, and kme, are given in
3 Table B-3. By bounding the tissue:air partition coefficients with upper and lower limits on biologically
4 plausible values from Leung and Paustenbach (1990a) or Sweeney et al. (2008a), the model predictions
5 are still at least 2- to 4-fold lower than either the empirical model output or the experimental observations.
6 The range of possible urinary HEAA predictions approximate the prediction of the empirical model, but
7 this agreement is not surprising, as the cumulative rate of excretion depends only on the rate of
8 metabolism of 1,4-dioxane, and not on the apparent Vd for 1,4-dioxane. These data show that the PBPK
9 model cannot adequately reproduce the predictions of blood 1,4-dioxane concentrations of the Young et
10 al. (1977a) human empirical model or the experimental observations when constrained by biologically
11 plausible values for physiological flow rates and tissue:air partition coefficients.
J
o
a
*
o
_0
m
1 -I
D Young et al. (1977) Observed Data
— — Young etal. (1977) Empirical Model
— — PC- Upper
PC- Central
— — — PC- LO'A'er
\\
J 300 -
-------
n Younget al. i!977i Observed Data
— — Younget si. 11977} Empirical Mod
_ — PC-Upper
PC-Central
— — — PC-Lower
J! 600 -
5
c
< 500
ui
X
~0 400
O
J 300 -
-jj 200
1
D Youns at al. (1977! Observed Data
Young et al. (1977) Empirical Model
_ _ PC-Upper
PC- Central
— — — PC-Lower
Time (hrs)
Time (hrs)
Source: Reprinted with permission of Oxford Journals, Sweeney et al. (2008a); Used with permission of Taylor & Francis,
Young et al. (1977a).
Figure B-ll Comparisons of the range of PBPK model predictions from upper and
lower boundaries on partition coefficients from Sweeney et al (2008b)
with empirical model predictions and experimental observations for
blood 1,4-dioxane concentrations (left) and amount of HEAAin urine
(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
1
2
3
4
5
Source of partition
coefficients
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)b
Metabolic dissociation constant
(Km)c
HEAA urinary elimination rate
constant (kme)d
Leung and Paustenbach (1990a)
For maximal Vd
3.63
0.41
0.24
"Cardiac output = 17.0 L/hour/kg BWU'4' alveolar ventilation
bmg/hour/kg BW07
dhour"1
B.4.4 Alternative Model
Since the PBPK model
For minimal Vd
6.2
5.6
0.29
= 17.7 L/hour/kg BWU'4
Sweeney et
For maximal Vd
8.7
0.000038
0.18
Parameterization
does not predict the experimental observations of Young
al. (2008a)
For minimal Vd
5.3
3.8
0.28
et al. (1977a)
when parameterized by biologically plausible values, an exercise was performed to explore alternative
parameters and values capable of producing an adequate fit of the data. Since the metabolism of
1,4-dioxane appears to be linear in humans for a 50-ppm exposure (Young et al.. 1977a). the parameters
VmaxC and Km were replaced by a first-order, non-saturable metabolism rate constant, kLC. This rate
B-20
DRAFT - DO NOT CITE OR QUOTE
-------
1 constant was fitted to the experimental blood 1,4-dioxane data using partition coefficient values of
2 Sweeney et al. (2008a) to minimize the Vd (i.e., maximize the blood 1,4-dioxane levels). The resulting
3 model predictions are shown in Figure B-12. As before, the maximum blood 1,4-dioxane levels were
4 approximately sevenfold lower than the observed values.
I
0
4
D Yo u n g et al. 119 / /1 0 b ;e r/ed [> ata
— — YounEetal. (1977) Empirital Model
kLC (1.2)-fluid
Figure B-12 Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, kLC (1.2 hour"1), to the
experimental data.
5 A re-calibration was performed using only the data from the exposure phase of the experiment,
6 such that the elimination data did not influence the initial metabolism and tissue distribution. The model
7 predictions from this exercise are shown in Figure B-13. These predictions are more similar to the
8 observations made during the exposure phase of the experiment; however, this is achieved at greatly
9 reduced elimination rate and hence under predictions of urinary HEAA (compare Figure B-8 and
10 Figure B-13).
B-21
DRAFT - DO NOT CITE OR QUOTE
-------
e
o
o
.2
a
4
D YnungBt al. (1377J Bhlfi
_ _ Young et al. (1977i Empirical r.lodel
kLCiO.D-fltud
Time(hrs)
Figure B-13 Predictions of blood 1,4-dioxane concentration following calibration of
a first-order metabolism rate constant, kLC (0.1 hour"1), to only the
exposure phase of the experimental data.
1 Finally, the model was re-calibrated by simultaneously fitting kLC and the slowly perfused
1 tissue:air partition (PSA) coefficient to the experimental data with no bounds on possible values (except
3 that they be non-zero). The fitted slowly perfused tissue:air partition coefficient was a very low value of
4 10 (compared to experimentally determined values, see Table B-l). The resulting model predictions,
5 however, were closer to the observations (Figure B-l4). These exercises show that better fits to the
6 observed blood 1,4-dioxane kinetics are achieved only when parameter values are adjusted in a way that
7 corresponds to a substantial decrease in apparent Vd of 1,4-dioxane in the human, relative to the rat (e.g.,
8 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 and poor predictions of urinary elimination of HEAA.
13 These observations suggest that some other process not captured in the present PBPK model structure is
14 responsible for the species differences in 1,4-dioxane Vd and the inability to reproduce the human
15 experimental inhalation data with biologically plausible parameter values.
B-22
DRAFT - DO NOT CITE OR QUOTE
-------
1
o
a
4
o
J>
m
— — Younget al.
-------
1 14). Re-calibration of the model with upper bounds on the tissue:air partition coefficients results in
2 predictions that are still 2- to 4-fold lower than empirical model prediction or observations (Figure B-10
3 and Figure B-l 1). Exploration of the model space using an assumption of first-order metabolism (valid
4 for the 50-ppm inhalation exposure) showed that an adequate fit to the exposure and elimination data can
5 be achieved only when unrealistically low values are assumed for the slowly perfused tissue:air partition
6 coefficient (Figure B-14). Artificially low values for the other tissue:air partition coefficients are not
7 expected to improve the model fit, because blood 1,4-dioxane is less sensitive to these parameters than it
8 is to VmaxC and Km. This suggests that the model structure is insufficient to capture the apparent species
9 difference in the blood 1,4-dioxane Vd between rats and humans. Differences in the ability of rat and
10 human blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this is expected to be
11 evident in very different values for rat and human blood:air partition coefficients, which is not the case
12 (Table B-l). Additionally, the models do not account for induction in metabolism, which may be present
13 in animals exposed repeatedly to 1,4-dioxane. Therefore, some other modification(s) to the Reitz et al.
14 (1990b) model structure may be necessary. Sweeney et al. (2008a) PBPK model provided an overall
15 improvement on previous models; however, the Sweeney et al. (2008a) inhalation model predictions of
16 animal and human data were still problematic.
B.6 acsIX Model Code
17 The PBPK acsIX model code is made available electronically through EPA's Health and
18 Environmental Research Online (HERO) database. All model files may be downloaded in a zipped
19 workspace from HERO (U.S. EPA, 2013).
B-24
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-4, 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-4 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%)
530
27/33b
(82%)
Females (mg/kg-day)
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 (19781
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 (y? p > 0.1)
8 (Table C-5). Comparing across models, a better fit is indicated by a lower AIC value (U.S. EPA. 2012a).
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-5, Figure C-15) 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-5,) and could be used to derive a POD of 452.4 mg/kg-day for this endpoint. For
13 those models that exhibit adequate fit, models with the lower AIC values are preferred. Differences in
14 AIC values of less than 1 are generally not considered important. BMDS modeling results for all
15 dichotomous models are shown in Table C-5.
C-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table C-5 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
a p-Value from the x^
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
: goodness-of-fit test for the selected model.
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
Values < 0.1 indicate that the model exhibited
437.08
471.92
447.21
436.19
297.86
456.42
452.36
189.49
a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to > 1.
°Slope restricted to > 1.
dBetas restricted to >0.
Source: NCI (19781
C-2
DRAFT - DO NOT CITE OR QUOTE
-------
LogProbit 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
•
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-15BMD 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 Chi^2 = 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 Chi^2 = 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-6
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. BMDio and BMDLio 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. 2012a). 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-6. 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-7 and
12 Table D-8; Figure D-17)
13 • Hepatic adenomas and carcinomas in male F344 rats (Table D-9 and Table D-10;
14 Figure D-18 and Figure D-19)
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-l 1)
17 o Nasal cavity tumors in female F344 rats (Table D-12; Figure D-20)
18 o Nasal cavity tumors in male F344 rats (Table D-13; Figure D-21)
19 o Mammary gland adenomas in female F344 rats (Table D-14;
20 Figure D-22 and Figure D-23)
21 o Peritoneal mesotheliomas in male F344 rats (Table D-l5; Figure D-24
22 and Figure D-25)
23 • Hepatic adenomas and carcinomas in female BDF1 mice (Table D-l6,
24 Table D-17, and Table D-18; Figure D-26, Figure D-27, Figure D-28, and
25 Figure D-29)
26 • Hepatic adenomas and carcinomas in male BDF1 mice (Table D-l9 and
27 Table D-20; Figure D-30 and Figure D-31)
28 Data and BMD modeling results from the additional chronic bioassays (NCI. 1978; Kociba et al..
29 1974a) were evaluated for comparison with the data from Kano et al. (2009). These results are presented
30 as 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-21)
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.. 1974a) treated with
3 1,4-dioxane in the drinking water for 2 years (Table D-22)
4 o BMDS dose-response modeling results for incidence of hepatocellular
5 carcinoma in male and female Sherman rats (combined) (Kociba et al.
6 1974a) exposed to 1,4-dioxane in drinking water for 2 years (Table D-23;
7 Figure D-32 and Figure D-33)
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.. 1974a) exposed to 1,4-dioxane in the drinking water for 2 years
11 (Table D-24; Figure D-34)
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-25)
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-26; Figure D-35
18 and Figure D-36)
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-27;
22 Figure D-3 7 and Figure D-3 8)
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-28;
26 Figure D-3 9 and Figure D-40)
27 • Incidence of hepatocellular adenoma or carcinoma in male and female B6C3Fi
28 mice (NCI. 1978) exposed to 1,4-dioxane in drinking water (Table D-29)
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-30; Figure D-41)
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-31;
36 Figure D-42 and Figure D-43).
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. 2012a) 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-6.
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 BMDLnED values are summarized in Table D-6.
Table D-6 Recommended models for rodents exposed to 1,4-dioxane in drinking water (Kano et
al.. 2009)
Endpoint
Female F344
Hepatic
Tumors
Mammary
Gland
Tumors
Nasal
Cavity
Tumors
Model
selection
criterion
Rat
Lowest
Lowest
Lowest
AIC
AIC
AIC
Model Type
Multistage
(2 degree)
LogLogistic
Multistage
(3 degree)
AIC
91.5898
194.151
42.6063
p-value
0.4516
0.8874
0.9966
BMDa
mg/kg-day
79
161
381
.83
.01
.65
BMDLa BMDHEDa BMDLHEDa
mg/kg-day mg/kg-day mg/kg-day
58
81
282
.09
.91
.61
19.84
40.01
94.84
14.43
20.35
70.23
Male F344 Rat
Hepatic
Tumors
Peritoneal
Meso-thel
ioma
Nasal
Cavity
Tumors
Female BDF1
Hepatic
Tumors
Lowest
Lowest
Lowest
Mouse
Lowest
AIC
AIC
AIC
AIC
BMR 50%
Probit
Probit
Multistage
(3 degree)
LogLogistic
LogLogistic
147.787
138.869
24.747
176.214
176.214
0.9867
0.9148
0.9989
0.1421
0.1421
62
93
328
5
49.
.20
.06
.11
.54
88b
51
76
245
3
32.
.12
.32
.63
.66
93b
17.43
26.09
91.97
0.83
7.51°
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-7.
D-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-7 Data for hepatic adenomas and carcinomas in female 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
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 for the entire suite of models are presented in Table D-8.
Table D-8 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"
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)0
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
93.1067
91.7017
93.102
93.0762
114.094
91.5898
93.2682
91.8786
93.2255
114.094
4458.37
D value BMDl° BMDLio
p~ mg/kg-day mg/kg-day
0.3024
0.4459
0.3028
0.3074
0.0001
0.4516
0.2747
0.3839
0.2825
0.0001
NCa
89.46
93.02
88.34
87.57
25.58
79.83
92.81
85.46
92.67
25.58
NCa
62.09
71.60
65.52
66.19
19.92
58.09
59.31
67.84
59.89
19.92
NCa
x23
0.027
0.077
0.016
0.001
-1.827
-0.408
0.077
-0.116
0.088
-1.827
0
BMDlOHED
mg/kg-day
22.23
23.12
21.95
21.76
6.36
19.84
23.06
21.24
23.03
6.36
0
BMDLioHED
mg/kg-day
15.43
17.79
16.28
16.45
4.95
14.43
14.74
16.86
14.88
4.95
0
aMaximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
""Best-fitting model.
Value unable to be calculated (NC: not calculated) by BMDS.
DRAFT - DO NOT CITE OR QUOTE
D-5
-------
Multistage Cancer Model with 0.95 Confidence Level
I
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-17 Multistage BMD model (2 degree) for the combined incidence of hepatic
adenomas and carcinomas in female F344 rats.
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
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_hepato_adcar_Msc-BMR10-2poly.plt
Mon Oct 26 08:20:52 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.0281572
Beta(l) = 0
Beta(2) = 1.73306e-005
D-6
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Asymptotic Correlation Matrix of Parameter Estimates (*** The model parameter (s)
-Beta (1) have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Background Beta (2)
Background 1 -0.2
Beta(2) -0.2 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.0362773 * * *
Beta(l) 0 * * *
Beta(2) 1.65328e-005 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log (likelihood) # Param's Deviance Test d.f. P-value
Full model -42.9938 4
Fitted model -43.7949 2 1.60218 2 0.4488
Reduced model -120.43 1 154.873 3 <.0001
AIC: 91.5898
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0363 1.814 3.000 50 0.897
18.0000 0.0414 2.071 1.000 50 -0.760
83.0000 0.1400 7.001 6.000 50 -0.408
429.0000 0.9540 47.701 48.000 50 0.202
= 1.59 d.f. = 2 P-value = 0.4516
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level =0.95
BMD = 79.8299
BMDL = 58.085
BMDU = 94.0205
Taken together, (58.085 , 94.0205) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.00172161
D.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-9.
D-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-9
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).
4 Note that the incidence of rats with hepatic adenomas, carcinomas, and with either adenomas or
5 carcinomas are monotone non-decreasing functions of dose. These data therefore appear to be appropriate
6 for dose-response modeling using BMDS.
7 The results of the BMDS modeling for the entire suite of models tested using the data for hepatic
8 adenomas and carcinomas for male F344 rats are presented in Table D-10.
Table D-10 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"
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
149.884
147.813
149.886
149.913
152.836
149.814
149.772
147.787
149.856
152.836
4441.71
p-value
0.7257
0.9749
0.7235
0.6972
0.0978
0.8161
0.9171
0.9867
0.7576
0.0978
NCa
BMDio
mg/kg-day
62.41
68.74
62.10
61.70
23.82
61.68
63.62
62.20
62.63
23.82
NCa
BMDLio
mg/kg-day
30.79
55.39
34.61
37.49
18.34
28.26
27.49
51.12
30.11
18.34
NCa
x23
-0.03
0.097
-0.021
-0.003
-0.186
-0.063
-0.024
-0.05
-0.039
-0.186
0
BMDlOHED
mg/kg-day
17.49
19.27
17.41
17.29
6.68
17.29
17.83
17.43
17.56
6.68
0
BMDLio HED
mg/kg-day
8.63
15.53
9.70
10.51
5.14
7.92
7.71
14.33
8.44
5.14
0
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
""Best-fitting model.
Value unable to be calculated (NC: not calculated) by BMDS.
D-8
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
0.8
0.6
<
J 0.4
CO
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-18 Probit BMD model for the combined incidence of hepatic adenomas and
carcinomas in male F344 rats.
1
2
o
J
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-BMR10.plt
Mon Oct 26 08:32:08 2009
BMDS Model Run
The form of the probability function is:
P[response] = CumNorm(Intercept+Slope*Dose) ,
where CumNorm(.) is the cumulative normal distribution function
Dependent variable = Effect
Independent variable = Dose
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
background = 0 Specified
intercept = -1.51718
slope = 0.00831843
Asymptotic Correlation Matrix of Parameter Estimates
(*** 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-9
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 intercept slope
3 intercept 1 -0.69
4 slope -0.69 1
5
6
7 Parameter Estimates
8 95.0% Wald Confidence Interval
9 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
10 intercept 1.53138 0.160195 -1.84535 -1.2174
11 slope 0.00840347 0.000976752 0.00648907 0.0103179
12
13
14 Analysis of Deviance Table
15
16 Model Log(likelihood) # Param's Deviance Test d.f. P-value
17 Full model -71.8804 4
18 Fitted model -71.8937 2 0.0265818 2 0.9868
19 Reduced model -115.644 1 87.528 3 <.0001
20
21 AIC: 147.787
22
23
24 Goodness of Fit
25 Scaled
26 Dose Est._Prob. Expected Observed Size Residual
27
28 0.0000 0.0628 3.142 3.000 50 -0.083
29 11.0000 0.0751 3.754 4.000 50 0.132
30 55.0000 0.1425 7.125 7.000 50 -0.050
31 274.0000 0.7797 38.985 39.000 50 0.005
32
33 ChiA2 = 0.03 d.f. = 2 P-value = 0.9867
34
35 Benchmark Dose Computation
36
37 Specified effect =0.1
38 Risk Type = Extra risk
39 Confidence level =0.95
40 BMD = 62.1952
41 BMDL = 51.1158
42
D-10
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
T3
-------
1 Beta(2) = 0
2 Beta(3) = 5.14597e-008
3 Asymptotic Correlation Matrix of Parameter Estimates
4 (*** The model parameter(s) -Beta(2)have been estimated at a boundary point, or have
5 been specified by the user, and do not appear in the correlation matrix )
6
7 Background Beta(l) Beta(3)
8 Background 1 -0.67 0.58
9 Beta(l) -0.67 1 -0.95
10 Beta(3) 0.58 -0.95 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 0.0619918 * * *
18 Beta(l) 0.001449 * * *
19 Beta(2) 0 * * *
20 Beta(3) 5.11829e-008 * * *
21
22 * - Indicates that this value is not calculated.
23
24
25
26 Analysis of Deviance Table
27
28 Model Log(likelihood) # Param's Deviance Test d.f. P-value
29 Full model -71.8804 4
30 Fitted model -71.8858 3 0.0107754 1 0.9173
31 Reduced model -115.644 1 87.528 3 <.0001
32
33 AIC: 149.772
34
35
36 Goodness of Fit
37 Scaled
38 Dose Est._Prob. Expected Observed Size Residual
39
40 0.0000 0.0620 3.100 3.000 50 -0.058
41 11.0000 0.0769 3.844 4.000 50 0.083
42 55.0000 0.1412 7.059 7.000 50 -0.024
43 274.0000 0.7799 38.997 39.000 50 0.001
44
45 ChiA2 = 0.01 d.f. = 1 P-value = 0.9171
46
47
48 Benchmark Dose Computation
49
50 Specified effect =0.1
51 Risk Type = Extra risk
52 Confidence level =0.95
53 BMD = 63.6179
54 BMDL = 27.4913
55 BMDU = 123.443
56
57 Taken together, (27.4913, 123.443) is a 90% two-sided confidence interval for the BMD
58
59 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-l 1. Note that the incidence of rats with these endpoints are monotone
3 non-decreasing functions (except female peritoneal mesotheliomas). These data therefore appear to be
4 appropriate for dose-response modeling using BMDS.
Table D-ll 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-12
6 through Table D-l5 for tumors in the nasal cavity, mammary gland, and peritoneal cavity.
Table D-12 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
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)d
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
44.4964
44.4963
44.4963
44.4963
45.6604
43.0753
42.6063
44.4963
44.4963
45.6604
46.4963
p-value
1
1
1
1
0.6184
0.9607
0.9966
1
1
0.6184
0.9997
BMDio
mg/kg-day
403.82
421.54
413.69
400.06
375.81
366.07
381.65
414.11
414.86
375.81
413.96
BMDLio
mg/kg-day
269.03
351.74
268.85
260.38
213.84
274.63
282.61
333.31
273.73
213.84
372.57
x2b
0
0
0
0
0.595
0.109
0.021
0
0
0.595
1.64x10'B
BMDlOHED
mg/kg-day
100.35
104.75
102.80
99.42
93.39
90.97
94.84
102.91
103.09
93.39
102.87
BMDLio HED
mg/kg-day
66.85
87.41
66.81
64.71
53.14
68.24
70.23
82.83
68.02
53.14
92.58
"Nasal cavity tumors in female F344 rats include squamous cell carcinoma and esthesioneuro-epithelioma.
""Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
°Slope restricted > 1.
dBest-fitting model.
D-13
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with O.9t> Confidence Level
(D
"0
I
0.3
0.25
0.2
0.15
0.1
0.05
Multistage Cancer
Linear extrapolation
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-20 Multistage BMD model (3 degree) for nasal cavity tumors in female
F344 rats.
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
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-BMR10-3poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-BMR10-3poly.plt
Mon Oct 26 08:28:58 2009
BMDS Model Run
The form of the probability function is: P[response] = background +
(1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
Beta(3) = 1.91485e-009
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background -Beta(l) -Beta(2)
D-14
DRAFT - DO NOT CITE OR QUOTE
-------
1 have been estimated at a boundary point, or have been specified by the user,
2 and do not appear in the correlation matrix )
o
J
4 Beta(3)
5 Beta(3) 1
6
7 Parameter Estimates
8
9 95.0% Wald Confidence Interval
10 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
11 Background 0 * * *
12 Beta(l) 0 * * *
13 Beta(2) 0 * * *
14 Beta(3) 1.89531e-009 * * *
15
16 * - Indicates that this value is not calculated.
17
18
19 Analysis of Deviance Table
20
21 Model Log(likelihood) # Param's Deviance Test d.f. P-value
22 Full model -20.2482 4
23 Fitted model -20.3031 1 0.109908 3 0.9906
24 Reduced model -30.3429 1 20.1894 3 0.0001551
25
26 AIC: 42.6063
27
28
29 Goodness of Fit
30 Scaled
31 Dose Est._Prob. Expected Observed Size Residual
32
33 0.0000 0.0000 0.000 0.000 50 0.000
34 18.0000 0.0000 0.001 0.000 50 -0.024
35 83.0000 0.0011 0.054 0.000 50 -0.233
36 429.0000 0.1390 6.949 7.000 50 0.021
37
38 Chi^2 = 0.06 d.f. = 3 P-value = 0.9966
39
40
41 Benchmark Dose Computation
42
43 Specified effect = 0.1
44 Risk Type = Extra risk
45 Confidence level =0.95
46 BMD = 381.651
47 BMDL = 282.609
48 BMDU = 500.178
49
50 Taken together, (282.609, 500.178) is a 90% two-sided confidence interval for the BMD
51
52 Multistage Cancer Slope Factor = 0.000353846
D-15
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-13 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
BMDlOHED
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
aNasal cavity tumors in male F344 rats include squamous cell carcinoma, Sarcoma: NOS, rhabdomyosarcoma, and
esthesioneuro-epithelioma.
bMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
°Slope restricted > 1.
dBest-fitting model.
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-21 Multistage BMD model (3 degree) for nasal cavity tumors in male F344
rats.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
D-16
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-17
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
Table D-14 BMDS dose-response modeling results for the incidence of mammary gland
adenomas in female F344 rats (Kano et al.. 2009)
Model
Gamma
Logistic
LogLogistic"
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
NCa
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
3.49x10'3
BMDlOHED
mg/kg-day
43.90
57.24
40.01
67.28
43.90
43.90
43.90
55.43
43.90
43.90
23.37
BMDLio HED
mg/kg-day
24.63
39.69
20.35
43.41
24.63
24.63
24.63
37.67
24.63
24.63
3.48
Maximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
°Slope restricted > 1.
Value unable to be calculated (NC: not calculated) by BMDS.
D-18
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic 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
I
£2
0.5
0.4
0.3
0.2
0.1
400
dose
11:31 02/01 2010
Source: Use with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-22LogLogistic BMD model for mammary gland adenomas in female F344
rats.
Logistic Model. (Version: 2.12; Date: 05/16/2008)
Input Data File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.plt
Mon Feb 01 11:31:31 2010
BMDS Model Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.12
intercept = -7.06982
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
45(
D-19
DRAFT - DO NOT CITE OR QUOTE
-------
1 (*** The model parameter(s) -slope have been estimated at a boundary point, or have
2 been specified by the user, and do not appear in the correlation matrix )
o
J
4 background intercept
5 background 1 -0.53
6 intercept -0.53 1
7
8 Parameter Estimates
9
10 95.0% Wald Confidence Interval
11 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
12 background 0.130936 * * *
13 intercept -7.2787 * * *
14 slope 1 * * *
15
16 * - Indicates that this value is not calculated.
17
18
19
20 Analysis of Deviance Table
21
22 Model Log(likelihood) # Param's Deviance Test d.f. P-value
23 Full model -94.958 4
24 Fitted model -95.0757 2 0.235347 2 0.889
25 Reduced model -98.6785 1 7.4409 3 0.0591
26
27 AIC: 194.151
28
29
30 Goodness of Fit
31 Scaled
32 Dose Est._Prob. Expected Observed Size Residual
33 ~
34 0.0000 0.1309 6.547 6.000 50 -0.229
35 18.0000 0.1416 7.080 7.000 50 -0.032
36 83.0000 0.1780 8.901 10.000 50 0.406
37 429.0000 0.3294 16.472 16.000 50 -0.142
38
39 Chi^2 = 0.24 d.f. = 2 P-value = 0.8874
40
41
42 Benchmark Dose Computation
43 Specified effect = 0.1
44 Risk Type = Extra risk
45 Confidence level =0.95
46 BMD = 161.012
47 BMDL = 81.9107
D-20
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
I
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
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-23 Multistage BMD model (1 degree) for mammary gland adenomas in
female F344 rats.
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
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_maitim_ad_Msc-BMR10-lpoly. (d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_mairan_ad_Msc-BMR10-lpoly.plt
Mon Oct 26 08:27:02 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.136033
Beta(l) = 0.000570906
Asymptotic Correlation Matrix of Parameter Estimates
D-21
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 Variable Estimate Std. Err. Lower Conf. Limit Upper
9 Conf. Limit
10 Background .133161 * * *
11 Beta(l) 0.000596394 * * *
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 -94.958 4
21 Fitted model -95.111 2 0.305898 2 0.8582
22 Reduced model -98.6785 1 7.4409 3 0.0591
23
24 AIC: 194.222
25
26
27 Goodness of Fit
28 Scaled
29 Dose Est._Prob. Expected Observed Size Residual
30
31 0.0000 0.1332 6.658 6.000 50 -0.274
32 18.0000 0.1424 7.121 7.000 50 -0.049
33 83.0000 0.1750 8.751 10.000 50 0.465
34 429.0000 0.3288 16.442 16.000 50 -0.133
35
36 ChiA2 = 0.31 d.f. = 2 P-value = 0.8559
37
38
39 Benchmark Dose Computation
40
41 Specified effect =0.1
42 Risk Type = Extra risk
43 Confidence level = 0.95
44 BMD = 176.663
45 BMDL = 99.1337
46 BMDU = 501.523
47
48 Taken together, (99.1337, 501.523) is a 90% two-sided confidence interval for the BMD
49
50 Multistage Cancer Slope Factor = 0.00100874
D-22
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-15 BMDS dose-response modeling results for the incidence of peritoneal mesotheliomas
in male F344 rats (Kano et al.. 2009)
Model
Gamma
Logistic
LogLogistic
LogProbit0
AIC
140.701
139.016
140.699
140.69
p-value
0.9189
0.8484
0.9242
0.9852
BMDio
mg/kg-day
73.52
103.52
72.56
70.29
BMDLio
mg/kg-day
35.62
84.35
36.37
52.59
x2a
0.018
0.446
0.014
0.001
BMDlOHED
mg/kg-day
20.61
29.02
20.34
19.70
BMDLio HED
mg/kg-day
9.98
23.65
10.19
14.74
Multistage-Cancer
(1 degree)
140.826 0.3617
41.04
30.51
-1.066
11.50
8.55
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
140.747
140.747
138.869
140.709
140.826
2992
0.8135
0.8135
0.9148
0.8915
0.3617
NCa
77.73
77.73
93.06
74.77
41.04
NCa
35.43
35.43
76.32
35.59
30.51
NCa
0.067
0.067
0.315
0.027
-1.066
0
21.79
21.79
26.09
20.96
11.50
0
9.93
9.93
21.39
9.97
8.55
0
aMaximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
bSlope restricted > 1.
°Best-fitting model.
Value unable to be calculated (NC: not calculated) by BMDS.
Probit Model with 0.95 Confidence Level
1
2
o
6
4
o
2
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-24 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)
DRAFT - DO NOT CITE OR QUOTE
D-23
-------
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 Chi^2 = 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-24
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer 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
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-25 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-25
DRAFT - DO NOT CITE OR QUOTE
-------
1 Beta(l) = 0.000816174
2 Beta(2) = 7.47062e-006
o
J
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 ChiA2 = 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-16. 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-26
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-16 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-17. 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-l 1 and
13 D-l 2. A summary of the BMD results for BMRs of 10, 30, and 50% are shown in Table D-l 8. 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.
D-27
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-17 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
LogLogistic"
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
aMaximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
bBest-fitting model, lowest AIC value.
°Slope restricted > 1.
Value unable to be calculated (NC: not calculated) by BMDS.
Table D-18 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-28
DRAFT - DO NOT CITE OR QUOTE
-------
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
Log-Logistic Model with 0.95 Confidence Level
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-26 LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 10%.
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
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-29
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 ChiA2 = 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-30
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic 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
•
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-27 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-31
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 ChiA2 = 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-32
DRAFT - DO NOT CITE OR QUOTE
-------
•
I
C
o
13
ro
0.8
0.6
0.4
0.2
Log-Logistic Model with 0.95 Confidence Level
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-28 LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in female BDF1 mice with a BMR of 50%.
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
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-33
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 Chi^2 = 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-34
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
•
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-29 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*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
D-35
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 ChiA2 = 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-19. 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-36
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-19 Data for hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.. 2009)
6
7
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
The results of the BMDS modeling for the entire suite
carcinomas in male BDF1 mice are presented in Table D-20.
Dose (mg/kg-day)
0
9
15
23
27
50
of models
49
17
20
31
19
50
for hepatic
191
23
23
37
13
50
adenomas and
677
11
36
40
10
50
Table D-20 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
LogLogistic"
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
NCa
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
-1.25x10'3
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.
bBest-fitting model.
°Slope restricted > 1.
Value unable to be calculated (NC: not calculated) by BMDS.
Values much larger than 1 are undesirable.
D-37
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
"0
I
d
O
t3
cc
0.9
0.8
0.7
0.6
0.5
0.4
0.3
Log-Logistic
3MDL BMD
100 200 300 400
dose
500
600
700
07:30 10/262009
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
Figure D-30 LogLogistic BMD model for the combined incidence of hepatic
adenomas and carcinomas in male BDF1 mice.
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
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-38
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 ChiA2 = 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-39
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with O.9b 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
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-31 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-40
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 ChiA2 = 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 1974a) were evaluated for comparison with the Kano et al. (2009) study. These results are presented in
53 the 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-21.
D-41
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-21 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
Model
BMDio
Endpoint selection Model Type AIC p-value mg/kg-da
criterion y
RMnl RMn RMnl
BM°LJ° BM/^°HED BMDLionED
mg/kg-day mg/kg-day mg/kg-day
Kociba et al.. (1974a) Male and Female (combined) Sherman Rats
Hepatic Lowest AIC Probit 84.3126 0.606 1113.94
Tumors
920.62
290.78
0 LowestAIC LogLogistic 84.2821 0.7333 111.
Tumors
.46
72.41
28.75
240.31
Nasal .. ... .
Cavity Lowest AIC Multistage 26.4156 0.9999 1717.16 1306.29 448.24 340.99
Tumorsb _ <3 de9ree)
NCI. (1978) Female Osborne-Mendel Rats
18.68
Nasal
Cavity LowestAIC LogLogistic 84.2235 0.2486 155.32
Tumorsb
100.08
40.07
25.82
NCI. (1978) Male Osborne-Mendel Rats
Nasal
Cavity LowestAIC LogLogistic 92.7669 0.7809 56.26
Tumorsb
37.26
16.10
10.66
NCI. (1978) Female B6C3Fi Mice
Hepatic
model
' (^eS 85'3511
v ^ '
160'68
67'76
23'12
9'75
NCI. (1978) Male B6C3Fi Mice
Hepatic Lowest A|C Gamma 177.539 0.7571
Tumors
601.69
243.92
87.98
35.67
Incidence of hepatocellular carcinoma.
blncidence of nasal squamous cell carcinoma.
""Incidence of hepatocellular adenoma.
Incidence of hepatocellular adenoma or carcinoma.
D.7.1 Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma
(Kociba et al.. 1974a)
1 The incidence data for hepatocellular carcinoma and nasal squamous cell carcinoma are presented
2 in Table D-22. The predicted BMDioHED and BMDLioHED values are also presented in Table D-23 and
3 Table D-24 for hepatocellular carcinomas and nasal squamous cell carcinomas, respectively.
D-42
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-22 Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in male
and female Sherman rats (combined) (Kociba et al.. 1974a) treated with 1,4-dioxane in
the drinking water for 2 years
Animal Dose (mg/kg-day) Incidence of hepatocellular '"s'tfamous ceM3'
(average of male and female dose) carcinoma3 . a
carcinoma
0 1/106b 0/106C
14 0/110 0/110
121 1/106 0/106
1,307 10/66d 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. (1_974a).
Table D-23 BMDS dose-response modeling results for the incidence of hepatocellular carcinoma
in male and female Sherman rats (combined) (Kociba et al.. 1974a) exposed to
1,4-dioxane in the drinking water for 2 years
Model
Gamma
Logistic
LogLogistic
LogProbit0
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
86.2403
84.3292
86.2422
84.4246
85.1187
86.2868
86.2868
84.3126
86.2443
85.1187
1503.63
p-value
0.3105
0.6086
0.3103
0.5977
0.3838
0.3109
0.3109
0.606
0.3104
0.3838
NCd
BMDio
mg/kg-day
985.13
1148.65
985.62
1036.97
940.12
1041.72
1041.72
1113.94
998.33
940.12
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
x2a
-0.005
-0.004
-0.005
-0.011
0.279
-0.006
-0.006
-0.005
-0.005
0.279
0
BMDlOHED
mg/kg-day
257.15
299.84
257.28
270.68
245.40
271.92
271.92
290.78
260.60
245.40
0
BMDLio HED
mg/kg-day
164.05
256.06
159.53
198.46
152.33
164.07
164.08
240.31
164.43
152.33
0
aMaximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
bSlope restricted > 1.
""Best-fitting model.
Value unable to be calculated (NC: not calculated) by BMDS.
D-43
DRAFT - DO NOT CITE OR QUOTE
-------
Probit Model with 0.95 Confidence Level
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. (1_974a).
Figure D-32 Probit BMD model for the incidence of hepatocellular carcinoma in
male and female Sherman rats exposed to 1,4-dioxane in drinking
water.
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
Probit Model. (Version: 3.1; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.plt
Tue Oct 27 12:54:14 2009
BMDS Model Run
The form of the probability function is:
P[response] = CumNorm(Intercept+Slope*Dose) ,where CumNorm(.) is the cumulative normal
distribution function
Dependent variable = Effect
Independent variable = Dose
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
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-44
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 Chi^2 = 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-45
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
T3
•§
<
CZ
O
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. (1_974a).
Figure D-33 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*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 = 0.000925988
D-46
DRAFT - DO NOT CITE OR QUOTE
-------
1 Beta(l) = 0.000124518
2
o
J
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-47
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-24 BMDS dose-response modeling results for the incidence of nasal squamous cell
carcinoma in male and female Sherman rats (combined) (Kociba et al.. 1974a)
exposed to 1,4-dioxane in the drinking water for 2 years
Model
Gamma
Logistic
LogLogistic
LogProbit0
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
aMaximum absolute x2 residual deviation
bSlope restricted > 1 .
°Best-fitting model.
p-value
1
1
1
1
0.9163
0.9977
0.9999
1
1
0.9163
0.9997
between
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"
observed and predicted count. Values much
BMDlOHED
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
larger than 1 are
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
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
BMD
06:25 10/272009
0 200 400 600 800 1000 1200 1400 1600 1800
dose
Figure D-34 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-48
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
Multistage Cancer Model. (Version: 1.
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_
Gnuplot Plotting File:
L: \Priv\NCEA HPAG\14Dioxane\BMDS \rnsc
Tue Oct 27 07:25:02 2009
BMDS Model Run
The form of the probability function
P [response] = background +
7; Date: 05/16/2008)
kociba mf rat nasal car Msc-BMR10-3poly. (d)
kociba mf rat nasal car Msc-BMR10-3poly.plt
is :
(1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
The parameter betas are restricted to
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 4
Total number of records with missing
Total number of parameters in model =
Total number of specified parameters
Degree of polynomial = 3
Maximum number of iterations = 250
be positive
values = 0
4
= 0
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
Beta(3) = 2.08414e-011
: le-008
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background -Beta(l) -Beta(2)
have been estimated at a boundary point, or have been specified by the user,
and do not appear in the correlation matrix )
Beta(3)
Beta(3) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 0 * * *
Beta(2) 0 * * *
Beta(3) 2.08088e-011 * * *
* - Indicates that this value is not
Analysis of Deviance Table
calculated.
Model Log (likelihood) # Param's Deviance Test d.f. P-value
Full model -12.2039 4
Fitted model -12.2078 1 0.00783284 3
Reduced model -17.5756 1 10.7433 3 0
0.9998
.0132
D-49
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 AIC: 26.4156
o
J
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 Chi^2 = 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-25. 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-26 through Table D-28.
D-50
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-25 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)
240^
530
Nasal cavity squamous cell carcinoma
0/33C
12/26"
16/33"
Female rat Animal Dose (mg/kg-day)a
350
640
Nasal cavity squamous cell carcinoma
0/34C
10/30°
8/29°
Hepatocellular adenoma
0/31C
10/30"
11/29"
aTumor incidence values were adjusted for mortality (NCI. 19781
bGroup not included in statistical analysis by NCI (1978) because the dose group was started a year earlier without
appropriate controls.
°p & 0.001; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.
Source: NCI (1978).
Table D-26 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
LogLogistic"
LogProbit
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
"Maximum absolute x^
"Best-fitting model.
AIC
84.6972
92.477
84.2821
85.957
84.6972
84.6972
91.7318
84.6972
84.6972
p-value
0.5908
0.02
0.7333
0.3076
0.5908
0.5908
0.0251
0.5908
0.5908
: residual deviation between
BMDio
mg/kg-day
132.36
284.09
111.46
209.47
132.36
132.36
267.02
132.36
132.36
BMDLio
mg/kg-day
94.06
220.46
72.41
160.66
94.06
94.06
207.18
94.06
94.06
x2a
0
1.727
0
1.133
0
0
1.7
0
0
observed and predicted count. Values much
BMDlOHED
mg/kg-day
34.144
73.29
28.75
54.04
34.14
34.14
68.88
34.14
34.14
larger than 1 are
BMDLio HED
mg/kg-day
24.26
56.87
18.68
41.45
24.26
24.26
53.44
24.26
24.26
undesirable.
D-51
DRAFT - DO NOT CITE OR QUOTE
-------
O
t5
CO
0.5
0.4
0.3
0.2
0.1
Log-Logistic Model with O.9b Confidence Level
Log-Logistic
BMD
100
200
06:32 10/272009
Source: NCI (1978).
300
dose
400
500
600
Figure D-35 LogLogistic BMD model for the incidence of hepatocellular adenoma
in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
water.
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_nci_frat_hepato_ad_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_hepato_ad_Lnl-BMR10-Restrict.plt
Tue Oct 27 07:32:13 2009
BMDS Model Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0
intercept = -6.62889
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
D-52
DRAFT - DO NOT CITE OR QUOTE
-------
1 (*** The model parameter(s) -background -slope have been estimated at a boundary
2 point, or have been specified by the user, and do not appear in the correlation
3 matrix)
4
5 intercept
6 intercept 1
7
8 Parameter Estimates
9
10 95.0% Wald Confidence Interval
11 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
12 background 0 * * *
13 intercept -6.91086 * * *
14 slope 1 * * *
15
16 * - Indicates that this value is not calculated.
17
18
19 Analysis of Deviance Table
20
21 Model Log(likelihood) # Param's Deviance Test d.f. P-value
22 Full model -40.8343 3
23 Fitted model -41.141 1 0.613564 2 0.7358
24 Reduced model -50.4308 1 19.1932 2 <.0001
25
26 AIC: 84.2821
27
28
29 Goodness of Fit
30 Scaled
31 Dose Est._Prob. Expected Observed Size Residual
32
33 0.0000 0.0000 0.000 0.000 31 0.000
34 350.0000 0.2587 8.536 10.000 33 0.582
35 640.0000 0.3895 12.464 11.000 32 -0.531
36
37 ChiA2 = 0.62 d.f. = 2 P-value = 0.7333
38
39
40 Benchmark Dose Computation
41
42 Specified effect = 0.1
43 Risk Type = Extra risk
44 Confidence level =0.95
45 BMD = 111.457
46 BMDL = 72.4092
D-53
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer 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
28
29
30
31
32
I
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
_
BMDL
BMD
100
200
300
dose
400
500
600
06:32 10/27 2009
Source: NCI (19781
Figure D-36 Multistage BMD model (1 degree) for the incidence of hepatocellular
adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
drinking water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_hepato_ad_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_hepato_ad_Msc-BMR10-lpoly.plt
Tue Oct 27 07:32:16 2009
BMDS Model Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
D-54
DRAFT - DO NOT CITE OR QUOTE
-------
1 Background = 0.0385912
2 Beta(l) = 0.000670869
3 Asymptotic Correlation Matrix of Parameter Estimates
4
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)
9 Beta(l) 1
10
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) 0.00079602 * * *
19
20 * - Indicates that this value is not calculated.
21
22
23
24 Analysis of Deviance Table
25
26 Model Log(likelihood) # Param's Deviance Test d.f. P-value
27 Full model -40.8343 3
28 Fitted model -41.3486 1 1.02868 2 0.5979
29 Reduced model -50.4308 1 19.1932 2 <.0001
30
31 AIC: 84.6972
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 0.2432 8.024 10.000 33 0.802
40 640.0000 0.3992 12.774 11.000 32 -0.640
41
42 Chi^2 = 1.05 d.f. = 2 P-value = 0.5908
43
44
45 Benchmark Dose Computation
46
47 Specified effect = 0.1
48 Risk Type = Extra risk
49 Confidence level =0.95
50 BMD = 132.359
51 BMDL = 94.0591
52 BMDU = 194.33
53
54 Taken together, (94.0591, 194.33 ) is a 90% two-sided confidence interval for the BMD
55
56 Multistage Cancer Slope Factor = 0.00106316
D-55
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-27 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
LogLogistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
aMaximum absolute x^
bBest-fitting model.
°Slope restricted > 1 .
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
: residual deviation between
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
observed and predicted count.
x2a
1.466
2.148
0
1.871
1.466
1.466
2.136
1.466
1.466
BMDlOHED
mg/kg-day
45.47
90.68
40.07
65.71
45.47
45.47
84.73
45.47
45.47
Values much larger than 1
BMDLio HED
mg/kg-day
31.54
69.33
25.82
50.50
31.54
31.54
64.83
31.54
31.54
are undesirable.
Log-Logistic Model with 0.95 Confidence Level
1
2
o
J
4
0.5
0.4
0.3
0.2
0.1
06:30 10/27 2009
Source: NCI (19781
Figure D-37 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-56
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-57
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
o
J
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 6ahcer
Linear extrapolation
BMD
100
200
06:30 10/27 2009
Source: NCI (1978).
300
dose
400
500
600
Figure D-38 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*dose/xl)]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
D-58
DRAFT - DO NOT CITE OR QUOTE
-------
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
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
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 = 0.0569154
Beta(l) = 0.00042443
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Background have been estimated at a boundary point, or
have been specified by the user, and do not appear in the correlation matrix)
Beta(l)
Beta(l) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0 * * *
Beta(l) 0.000597685 * * *
* - Indicates that this value is not calculated.
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -39.7535 3
Fitted model -41.3998 1 3.29259 2 0.1928
Reduced model -47.9161 1 16.3252 2 0.0002851
AIC:
1.7996
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0.000 34 0.000
350.0000 0.1888 6.607 10.000 35 1.466
640.0000 0.3179 11.125 8.000 35 -1.134
Chi^2 = 3.44 d.f. = 2 P-value = 0.1795
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 176.281
BMDL = 122.274
BMDU = 271.474
Taken together, (122.274, 271.474) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000817837
D-59
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-28 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
Model
Gamma
Logistic
LogLogistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
aMaximum absolute x^
bBest-fitting model.
°Slope restricted > 1 .
AIC
93.6005
103.928
92.7669
95.0436
93.6005
93.6005
103.061
93.6005
93.6005
for 2 years
p-value
0.5063
0.0061
0.7809
0.2373
0.5063
0.5063
0.0078
0.5063
0.5063
: residual deviation between
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
observed and predicted count.
x2a
0
2.024
0
1.246
0
0
2.024
0
0
Values much
BMDlOHED
mg/kg-day
21.17
51.25
16.10
35.46
21.16
21.16
48.10
21.17
21.17
larger than 1
BMDLio HED
mg/kg-day
15.66
39.86
10.66
27.43
15.66
15.66
37.67
15.66
15.66
are undesirable.
Log-Logistic Model with 0.95 Confidence Level
1
2
o
6
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 (19781
Figure D-39 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-60
-------
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 ChiA2 = 0.49 d.f. = 2 P-value = 0.7809
D-61
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
o
J
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
CD
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-40 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-62
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 Chi^2 = 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-29. 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-41), while the gamma model
D-63
DRAFT - DO NOT CITE OR QUOTE
-------
1 provided the best fit to the male mouse data (Figure D-42). The results of the BMDS modeling for the
2 entire suite of models are presented in Table D-30 and Table D-31 for the female and male data,
3 respectively.
Table D-29 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
28/47c
Female mouse Animal Dose (mg/kg-day)a
0
0/50°
380
21/48C
860
35/37c
"Tumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
°p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.
Source: NCI (19781
Table D-30 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
Model
Gamma
Logistic
LogLogistic
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 x2 residual deviation between observed and predicted count.
"Slope restricted > 1.
°Best-fitting model.
Values much larger than 1 are undesirable.
D-64
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
1
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
I
0.8
0.6
0.4
0.2
Multistage Cancer
Linear extrapolation
BMDL
BMD
100
200
300
400
dose
500
600
700
800
900
06:36 10/27 2009
Source: NCI (19781
Figure D-41 Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in female B6C3Fi mice exposed to 1,4-dioxane in
drinking water.
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_fmouse_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_fmouse_hepato_adcar_Msc-BMR10-2poly.plt
Tue Oct 27 07:36:26 2009
BMDS Model Run
The form of the probability function is:
P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
The parameter betas are restricted to be positive
Dependent variable = Effect
Independent variable = Dose
Total number of observations = 3
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
DRAFT - DO NOT CITE OR QUOTE
D-65
-------
1 Default Initial Parameter Values
2 Background = 0
3 Beta(l) = 2.68591e-005
4 Beta(2) = 3.91383e-006
5
6
7 Asymptotic Correlation Matrix of Parameter Estimates
8 (*** The model parameter(s) -Background have been estimated at a boundary point, or
9 have been specified by the user, and do not appear in the correlation matrix)
10
11 Beta(l) Beta(2)
12 Beta(l) 1 -0.92
13 Beta(2) -0.92 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 * * *
21 Beta(l) 2.686e-005 * * *
22 Beta(2) 3.91382e-006 * * *
23
24 * - Indicates that this value is not calculated.
25
26
27 Analysis of Deviance Table
28
29 Model Log(likelihood) # Param's Deviance Test d.f. P-value
30 Full model -40.6756 3
31 Fitted model -40.6756 2 3.20014e-010 1 1
32 Reduced model -91.606 1 101.861 2 <.0001
33
34 AIC: 85.3511
35
36 Goodness of Fit
37 Scaled
38 Dose Est._Prob. Expected Observed Size Residual
39
40 0.0000 0.0000 0.000 0.000 50 0.000
41 380.0000 0.4375 21.000 21.000 48 0.000
42 860.0000 0.9459 35.000 35.000 37 0.000
43
44 ChiA2 = 0.00 d.f. = 1 P-value = 1.0000
45
46
47 Benchmark Dose Computation
48 Specified effect =0.1
49 Risk Type = Extra risk
50 Confidence level = 0.95
51 BMD = 160.678
52 BMDL = 67.7635
53 BMDU = 186.587
54
55 Taken together, (67.7635, 186.587) is a 90% two-sided confidence interval for the BMD
56
57 Multistage Cancer Slope Factor = 0.00147572
D-66
DRAFT - DO NOT CITE OR QUOTE
-------
Table D-31 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
LogProbit0
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
aMaximum absolute x residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
bBest-fitting model.
"Value unable to be calculated (NC: not calculated) by BMDS.
dSlope restricted > 1.
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-42 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
o
J
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)
D-67
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
f)
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
Gnuplot Plotting File:
L: \Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-BMR10-Restrict.plt
Tue Oct 27 07:34:35 2009
====================================================================
BMDS Model Run
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
The form of the probability function is:
P [response] = background+ (1-background) *CumGamma [slope*dose, power] ,
where CumGamma ( . ) is the cummulative Gamma distribution function
Dependent variable = Effect
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.17
Slope = 0.000671886
Power =1.3
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter (s) -Power have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
Background Slope
Background 1 -0.52
Slope -0.52 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.160326 0.0510618 0.060247 0.260405
Slope 0.0213093 0.000971596 0.019405 0.0232136
Power 18 NA
NA - Indicates that this parameter has hit a bound implied by some ineguality
constraint and thus has no standard error.
Analysis of Deviance Table
Model Log (likelihood) # Param's Deviance Test d.f. P-value
Full model -86.7213 3
Fitted model -86.7693 2 0.096042 1 0.7566
Reduced model -96.715 1 19.9875 2 <.0001
AIC: 177.539
Goodness of Fit
Scaled
Dose Est._Prob. Expected Observed Size Residual
------------------------------------------------------------------------
0.0000 0.1603 7.856 8.000 49 0.056
720.0000 0.3961 19.806 19.000 50 -0.233
830.0000 0.5817 27.339 28.000 47 0.196
ChiA2 = 0.10 d.f. = 1 P-value = 0.7571
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 601.692
D-68
DRAFT - DO NOT CITE OR QUOTE
-------
BMDL = 243.917
T3
•§
C=
o
CO
Multistage Cancer Model with 0.95 Confidence Level
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
BMDL
BMD
0
06:34 10/272009
Source: NCI (1978).
100
200
300
400
dose
500
600
700
800
Figure D-43 Multistage BMD model (2 degree) for the incidence of hepatocellular
adenoma or carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in
drinking water.
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
Multistage Cancer Model. (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-BMR10-2poly.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-69
DRAFT - DO NOT CITE OR QUOTE
-------
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
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
ChiA2 = 2.02 d.f. = 1 P-value = 0.1554
Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 354.409
BMDL = 126.241
BMDU = 447.476
Taken together, (126.241, 447.476) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000792138
D-70
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX E. COMPARISON OF SEVERAL DATA
REPORTS FOR THE JBRC 2-YEAR 1,4-DIOXANE
DRINKING WATER STUDY
1 As described in detail in Section 4.2.1.2.6 of this Toxicological Review of 1,4-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 1994a). 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 2009a) 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.. 1994a). Table E-32 and Table E-2 show the histological
16 nonneoplastic findings provided for male and female F344 rats, respectively. Table E-34 and Table E-4
17 show the histological nonneoplastic findings provided for male and female F344 rats, respectively.
18 Table E-34 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-32 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; liver9
Spongiosis hepatis; liver
Clear cell foci; liver9
Acidophilic cell foci; liver9
Basophilic cell foci; liver9
Mixed-cell foci; liver9
Nuclear enlargement;
kidney proximal tubule
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
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.
(1994af
JBRC (1998)"
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«ntr«i 8- 41- 209-
C°SV01 24 121 586
1°' (16) (81) (398)
0/50 0/50 0/50 26/50
0/40 0/45 0/35 12/22°
0/50 0/50 0/50 31/50
0/40 0/45 0/35 15/22°
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 20/22°
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 17/22°
0/50 0/50 0/50 46/50
0/40 0/45 0/35 20/22°
0/50 0/50 1/50 44/50
0/40 0/45 1/35 20/22°
0/50 0/50 0/50 48/50
0/40 0/45 0/35 21/22°
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 12/22°
12/50 20/50 25/50 40/50
12/40 20/45 21/35f 21/22°
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 14/35° 22/22e
0/50 0/50 0/50 50/50
0/40 0/45 0/35 22/22e
0 200 1,000 5,000
nr
n 11± 55± 274±
0 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 14/50e 13/50°
Not reported
Not reported
Not reported
Dose rates (mg/kg-day) were not provided in Yamazaki etal. (1_994a). Drinking water concentrations of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
JBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group + standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA. 2010b).
dJBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994a) and JBRC (19981 were re-examined according to updated
criteria for liver lesions and were afterwards classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009V
E-2
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-33 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in female F344 rats
Yamazaki et al. (1994a)a
Kano et al. (2009!
Drinking water concentration |
0 200 1,000 5,000 0
Calculated
Not reported
Nasal respiratory
epithelium; nuclear
enlargement
NasaT respiratory
epithelium;
squamous cell
metaplasia
Nasal respiratory
epithelium;
squamous cell
hyperplasia
Nasal gland;
proliferation
Nasal olfactory
epithelium; nuclear
enlargement
NasaTolfactory
epithelium;
respiratory
metaplasia
Nasal olfactory
epithelium; atrophy
Lamina propria;
hydropic change
Lamina propria;
slerosis
Nasal cavity;
adhesion
Nasal cavity;
inflammation
Liver; hyperplasia9
Liver; spongiosis
hepatis
Liver; cyst formation
Liver; clear cell foci9
Liver; acidophilic cell
foci9
Liver; basophilic cell
foci9
Liver; mixed-cell
foci9
Kidney proximal
tubule; nuclear
enlargement
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
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 24/38°
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
7/24°
35/50
18/24°
5/50
4/24f
11/50
8/24°
39/50
22/24°
42/50
24/24°
40/50
22/24°
46/50
23/24°
48/50
23/24°
46/50
24/24°
15/50
7/24°
47/50
24/24°
20/50
14/24°
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
22/24°
„ 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 etal. (1_994a). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups.
Statistical test results were not reported.
bJBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010bV
dJBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994al and JBRC (19981 were re-examined according to updated
criteria for liver lesions and were afterwards classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009V
Table E-34 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in male F344 rats
Yamazaki et al.
(1994af
0 200 1,000 5,000
JBRC (1998)!
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])"'
Not reported
Pnntrnl 8' 41' 209-
Co,ftrol 24 121 586
1°' (16) (81) (398)
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/bO
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
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
Liver
Hepatocellular adenoma'
Hepatocellular carcinoma
Hepatocellular adenoma
or carcinoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50 2/50
Not
0/50 0/50
Not
Not
Not
4/50 24/50
reported
0/50 14/50
reported
reported
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
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
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
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
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
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 etal. (1_994a). Drinking water concentrations of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
JBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010bl.
dps0.01 by Fisher's Exact test.
'Significantly increased by Peto test for trend p< 0.01.
The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994al and JBRC (19981 were re-examined according to updated
criteria for liver lesions and were afterwards classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (20091.
E-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-35 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in female F344 rats
Yamazaki et al. (1994af | JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
0 200 1,000 5,000
0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day])°'c
Not Reported
rnntrn, 12- 56- 307-
™SX" 29 149 720
1°' (21) (103) (514)
n 18± 83± 429±
0 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
7/50
Not reported
0/50
0/50 0/50
0/50
Not reported
0/50
0/50 0/50
0/50
Not reported
0/50
0/50 0/50
1/50
Not reported
0/50 0/50
Not
Not
Not
Not
Not
0/50 0/50
Not
0/50 7/50"'1
reported
reported
reported
reported
reported
0/50 1/50
reported
0/50
0/50 0/50
7/50"'1
Not reported
0/50
0/50 0/50
0/50
Not reported
0/50
0/50 0/50
0/50
Not reported
0/50
0/50 0/50
1/50
Not reported
Liver
Hepatocellular
adenoma9
Hepatocellular
carcinoma
Hepatocellular
adenoma or
carcinoma9
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
1/50 0/50 5/50 38/50
Not reported
0/50 0/50 0/50 10/50
Not reported
Not reported
Not reported
1/50 0/50 5/50 38/50e'f
Not reported
1/50 0/50 0/50 10/50e'f
Not reported
1/50 0/50 5/50 40/50e'f
Not reported
3/50 1/50 6/50 48/50°'
Not reported
0/50 0/50 0/50 10/5Cr
Not reported
3/50 1/50 6/50 48/50°'
Not reported
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
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
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
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 16/5°U'
Not reported
8/50 8/50 11/50 18/5°U'
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki etal. (1_994a). Drinking water concentrations of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
bJBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group + standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA. 2010b).
dps 0.05 by Fisher's Exact test.
ep < 0.01 by Fisher's Exact test.
'Significantly increased by Peto test for trend p< 0.01.
9The samples associated with liver hyperplasia for rats and mice in Yamazaki et al. (1994a) and JBRC (19981 were re-examined according to updated
criteria for liver lesions and were afterwards classified as either hepatocellular adenoma or altered hepatocellular foci in Kano et al. (2009V
E-5
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-36 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in male Crj:BDF1 mice
Yamazaki et al.
(1994a)a
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])
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
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
Not
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
reported
Control
0
0/50
0/31
0/50
0/31
0/50
0/31
1/50
1/31
0/50
0/31
0/50
0/31
0/50
0/31
0/50
0/31
1/50
1/31
2/50
2/31
0/50
0/31
40/50
28/31
37-
94
(66)
0/50
0/33
0/50
0/33
0/50
0/33
2/50
1/33
0/50
0/33
0/50
0/33
0/50
0/33
0/50
0/33
0/50
0/33
3/50
2/33
0/50
0/33
42/50
30/33
144-
358
(251)
0/50
0/25
9/50
7/25e
1/50
0/25
1/50
1/25
0/50
0/25
0/50
0/25
0/50
0/25
0/50
0/25
0/50
0/25
4/50
3/25
0/50
0/25
38/50
24/25f
451-
1086
(768)
31/50
19/266
49/50
26/26e
48/50
26/26e
25/50
15/266
42/50
24/26e
17/50
12/266
41/50
24/26e
43/50
26/26e
27/50
22/26e
16/50
8/26f
39/50
22/26e
34/50
21/26f
n 49± 191±
0 5 21
0/50 0/50 0/50
677±
74
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 etal. (1994a). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups.
Statistical test results were not reported.
bJBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group + standard deviation. The mean shown in this table was used in the 2010Toxicolgical Review
of 1,4-Dioxane (U.S. EPA, 2010bV
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-37 Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
JBRC drinking water study in female Crj:BDF1 mice
Yamazaki et al.
(1994a)a
JBRC (1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000 | 0 500 2,000 8,000 | 0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])°'c
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
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 ^Q* 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. (1994a). 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 (19981 report.
CJBRC (19981 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, 2009al.
dKano etal. (20091 reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA, 2010bV
ep< 0.01 by chi-squaretest.
E-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-38 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in male Crj:BDF1 mice
Yamazaki et al.
(1994a)a
JBRC(1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0 500 2,000 8,000
0
500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])°'c
Not reported
~ . , 37- 144- 451-
Control 94 358 1086
u (66) (251) (768)
0
49± 191± 677±
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 etal. (1_994a). Drinking water concentrations of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
bJBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group + standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA. 2010b).
dps 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-8
DRAFT - DO NOT CITE OR QUOTE
-------
Table E-39 Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
drinking water study in female Crj:BDF1 mice
Yamazaki et al.
(1994a)a
JBRC(1998)b
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000
0 500 2,000 8,000
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day])°'c
Not reported
~ . , 45- 192- 759-
Control 1og 454 1374
u (77) (323) (1066)
n 66+ 278 ± 964 ±
0 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 reported
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 1/50
Not reported
Liver
Hepatocellular adenomas
Hepatocellular carcinomas
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"'a
Not reported
4/50 34/50" 41/50" 46/50"'a
Not reported
5/50 31/50" 20/50" 3/50
Not reported
0/50 6/50' 30/50" 45/50"'a
Not reported
5/50 35/50" 41/50" 46/50"'a
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994aV Drinking water concentrations (ppm) of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
JBRC (19981 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, 2009al.
cKano etal. (20091 reported a mean intake dose for each group + standard deviation. The mean shown in this table was used in the 2010Toxicolgical
Review of 1,4-Dioxane (U.S. EPA. 2010b).
dps0.01 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-40, for centrilobular necrosis of the liver in male F344/DuCrj rats
3 exposed to 1,4-dioxane vapors for 2 years (Kasai et al. 2009). Doses associated with a BMR of a 10%
4 extra risk were calculated.
Table F-40 Incidence of centrilobular necrosis of the liver in male F344/DuCrj rats exposed to
1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
1/50
(2%)
50
3/50
(6%)
250
6/50
(12%)
1,250
12/50a
(24%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
5 As assessed by the %2 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 (%2 p > 0.1) (Table F-41).
7 Comparing across adequately fitting models, the BMDL estimates were not within threefold difference
8 of each other. Therefore, in accordance with EPA BMD technical guidance (U.S. EPA. 2012a). the
9 adequately fitting model that resulted in the lowest BMDL was selected as appropriate for deriving a POD
10 which was the Dichotomous-Hill model. BMDS modeling results for all dichotomous models are shown
11 in Table F-41 and the model plot (Figure F-44) and output for the selected Dichotomous-Hill model are
12 included immediately after the table.
F-l
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-41 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.
bPower restricted to > 1.
°Slope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit model based on lowest BMDL.
Source: Kasai et al. (2QQ9).
1
.1
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-44 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.
1 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-BMR10-Restrict.(d)
3 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
4 files/dhl_Centr_necrosis_liver_Dhl-BMR10-Restrict.pit
5 Wed Jan 12 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= 1
19
20 Total number of observations = 4
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 = 1
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 1 -0.25 -0.89
39 g -0.25 1 0.016
40 intercept -0.89 0.016 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 v 0.311077 0.156196 0.00493876 0.617216
48 g 0.0709966 0.0662298 -0.0588115 0.200805
49 intercept -6.06188 1.34538 -8.69878 -3.42498
50 slope 1 NA
51
52 NA - 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 1 0.7479
61 Reduced model -69.3031 1 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 50 -0.100
3 50.0000 0.0522 2.612 3.000 50 0.247
4 250.0000 0.1285 6.423 6.000 50 -0.179
5 1250.0000 0.2372 11.861 12.000 50 0.046
6
7 ChiA2 = 0.10 d.f. = 1 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 BMD = 219.51
15 BMDL = 59.5598
F.2 Squamous Cell Metaplasia
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-42, for squamous cell metaplasia of the respiratory epithelium in male
18 F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
19 a 10% extra risk were calculated.
Table F-42 Incidence of squamous cell metaplasia of the respiratory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
0
0/50
1,4-dioxane
50
0/50
vapor concentration (ppm)
250
7/50b
(14%)
1,250
44/50a
(88%)
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
Source: Kasai et al. (2009).
20 For incidence of squamous cell metaplasia in F344/DuCrj male rats, the logistic and probit
21 models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1; see Table F-43), and thus
22 should not be considered further for identification of a POD. All of the remaining models exhibited
23 adequate fit. The BMDL estimates for all appropriately fitting models were within threefold
24 difference of each other, indicating that BMDL selection should be made based on model fit (U.S.
25 EPA. 2012a). As assessed by the AIC, the Log-probit model provided the best fit to the squamous cell
26 metaplasia data for male rats (Table F-43, Figure F-45), and could be used to derive a POD for this
27 endpoint.
F-4
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-43 Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
data forsquamous 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
Dichotomous-Hillc
81.687
89.4148
81.5252
81.23
82.6875
87.9361
82.1236
92.9215
83.1888
0.8682
0.0464
0.9142
0.9894
0.6188
0.0779
0.7679
0.0198
0.9995
0.24
1.806
0.131
0.032
0.605
1.681
0.33
-1.76
0
218.38
370.443
218.218
217.79
231.294
337.732
218.435
87.682
240.867
150.329
288.535
158.293
159.619
141.025
268.424
145.383
68.8015
161.945
a p-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to a 1.
°Slope restricted to a 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-45 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-BMR10-Restrict.(d)
F-5
DRAFT - DO NOT CITE OR QUOTE
-------
1 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
2 files/lnp_sgu_cell_meta_re_Lnp-BMR10-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 >= 1
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 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 1 -0.99
35 slope -0.99 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 NA
42 intercept -8.86173 1.2226 -11.258 -6.46548
43 slope 1.40803 0.193057 1.02965 1.78642
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 -38.5944 4
52 Fitted model -38.615 2 0.041197 2 0.9796
53 Reduced model -113.552 1 149.916 3 <.0001
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 50 0.000
62 50.0000 0.0004 0.020 0.000 50 -0.141
63 250.0000 0.1384 6.922 7.000 50 0.032
64 1250.0000 0.8808 44.038 44.000 50 -0.017
65
66 ChiA2 = 0.02 d.f. = 2 P-value = 0.9894
67
F-6
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 BMD = 217.79
7 BMDL = 159.619
F.3 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-44, 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-44 Incidence of squamous cell hyperplasia of the respiratory epithelium in male
F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0 50 250 1,250
0/50 0/50 1/50 10/50a
(2%) (20%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2QQ9).
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., %2 p-value < 0.1; see
14 Table F-45), 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. 2012a). 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-45, Figure F-46 and subsequent textual model
19 output), and could be used to derive a POD for this endpoint.
F-7
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-45 Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
data forsquamous 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
Dichotomous-Hillc
81.687
89.4148
81.5252
81.23
82.6875
87.9361
82.1236
92.9215
83.1888
0.8682
0.0464
0.9142
0.9894
0.6188
0.0779
0.7679
0.0198
0.9995
0.24
1.806
0.131
0.032
0.605
1.681
0.33
-1.76
0
218.38
370.443
218.218
217.79
231.294
337.732
218.435
87.682
240.867
150.329
288.535
158.293
159.619
141.025
268.424
145.383
68.8015
161.945
a p-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to a 1.
°Slope restricted to a 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
LogProbit
BMDL
BMD
200
400
13:2501/132011
600
dose
800
1000
1200
Figure F-46 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-BMR10-Restrict.(d)
4 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
5 files/lnp_sgu_cell_hyper_re_Lnp-BMR10-Restrict.pit
DRAFT - DO NOT CITE OR QUOTE
F-8
-------
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 >= 1
13
14 Total number of observations = 4
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 = 0
24 intercept = -7.75604
25 slope = 1
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 1
34
35 Parameter Estimates
36
37 95.0% Wald Confidence Interval
38 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
39 background 0 NA
40 intercept -7.90911 0.186242 -8.27414 -7.54408
41 slope 1 NA
42
43 NA - 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 4
50 Fitted model -30.2589 1 0.673572 3 0.8794
51 Reduced model -42.5964 1 25.3487 3 <.0001
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 50 -0.040
61 250.0000 0.0085 0.424 1.000 50 0.889
62 1250.0000 0.2182 10.911 10.000 50 -0.312
63
64 Chi^2 = 0.89 d.f. = 3 P-value = 0.8282
65
66
67 Benchmark Dose Computation
F-9
DRAFT - DO NOT CITE OR QUOTE
-------
1 Specified effect = 0.1
2 Risk Type = Extra risk
3 Confidence level = 0.95
4 BMD = 755.635
5 BMDL = 560.86
F.4 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-46, 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-46 Incidence of respiratory metaplasia of the olfactory epithelium in male F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
11/50
(22%)
50 250
34/50 49/50 a
(68%) (98%)
1,250
48/50a
(96%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
10 As assessed by the %2 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 (%2 p > 0.1)
12 (Table F-47). However, given that first non-control dose had a response level substantially above the
13 desired BMR (i.e. 10%), the use of BMD methods included substantial model uncertainty. The model
14 uncertainty associated with this dataset is related to low-dose extrapolation and consistent with BMD
15 Technical Guidance Document (U.S. EPA. 2012a) all available dichotomous models in the Benchmark
16 Dose Software (version 2.1.2) were fit to the incidence data shown in Table F-46 with the highest dose
17 group omitted. As assessed by the %2 goodness-of-fit test, the logistic, log-logistic, log-probit, and probit
18 models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0. l;See Table F-48), 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. 2012a). The AIC values for gamma, multistage, quantal-linear,
22 and Weibull models in Table F-48 are equivalent and the lowest and, in this case, essentially represent the
23 same model. Therefore, consistent with the Benchmark Dose Technical Guidance (U.S. EPA. 2012a).
24 any of them with equal AIC values (gamma, multistage, quantal-linear, or Weibull) could be used to
25 identify a POD for this endpoint. The model plot for the gamma model (Figure F-47) and output are
26 included immediately after the table.
F-10
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-47 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
Dichotomous-Hillc
179.68
191.339
152.72
161.267
179.68
198.785
179.68
179.68
150.466
0
0
0.0285
0
0
0
0
0
NA
-2.07
1.788
0.039
-0.39
-2.07
1.479
-2.07
-2.07
0
17.4082
34.2946
4.05465
14.3669
17.4082
61.4378
17.4082
17.4082
38.8552
12.3829
24.5917
1.90233
10.3023
12.3829
45.9091
12.3829
12.3829
31.4727
ap-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to a 1.
°Slope restricted to a 1.
dBetas restricted to >0.
Source: Kasai et al. (2009).
Table F-48 Goodness-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
0
-0.106
-1.511
-0.106
-0.106
6.46848
12.5197
14.2075
12.2114
6.46847
15.2883
6.46847
6.46847
4.73742
9.34421
3.77044
7.80131
4.73742
11.6855
4.73742
4.73742
a p-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to a 1.
°Slope restricted to a 1.
dBetas restricted to >0.
eBold indicates best-fit models based on lowest AIC.
Source: Kasai et al. (2009).
F-ll
DRAFT - DO NOT CITE OR QUOTE
-------
Gamma Multi-Hit Model with 0.95 Confidence Level
I
0.8
0.6
0.4
0.2
Gamma Multi-Hit
E MDL BMD
50
16:2401/132011
100 150
dose
200
250
Figure F-47 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-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/gam_resp_meta_no high dose_Gam-BMR10-Restrict.pit
Thu Jan 13 16:24:15 2011
BMDS_Model_Run
The form of the probability function is:
P[response]= background+(1-background)*CumGamma[siope*dose,power] ,
where CumGamma(.) is the cummulative Gamma distribution function
Dependent variable = Effect
Independent variable = Dose
Power parameter is restricted as power >=1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.230769
Slope = 0.022439
Power = 1.3
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Power have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)
Background Slope
Background 1 -0.33
Slope -0.33 1
F-12
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Parameter Estimates
o
J
4 95.0% Wald Confidence Interval
5 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
6 Background 0.226249 0.0588535 0.110898 0.3416
7 Slope 0.0162883 0.00320976 0.00999729 0.0225793
8 Power 1 NA
9
10 NA - 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 1 0.5961
18 Reduced model -99.1059 1 73.0301 2 <.0001
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 50 0.338
28 250.0000 0.9868 49.341 49.000 50 -0.422
29
30 Chi^2 = 0.30 d.f. = 1 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 BMD = 6.46848
37 BMDL = 4.73742
F.5 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-49, for atrophy of the olfactory epithelium in
40 male F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (Kasai et al. 2009).
41 Doses associated with a BMR of a 10% extra risk were calculated.
F-13
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-49 Incidence of atrophy of the olfactory epithelium in male F344/DuCrj rats exposed to
1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50 250
40/50 a 47/50 a
(80%) (94%)
1,250
48/50a
(96%)
ap < 0.01 by Fisher's exact test.
Source: Kasai et al. (2009).
1 As assessed by the %2 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., %2 p-value < 0.1;
3 see Table F-50), 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, indicating
5 that BMDL selection should be made based on model fit (U.S. EPA. 2012a). As assessed by the AIC, the
6 Log-logistic model provided the best fit to the atrophy data for male rats (Table F-50, Figure F-48), and
7 could be used to derive a POD for this endpoint. However, given that first non-control dose had a
8 response level substantially above the desired BMR (i.e. 10%), the use of BMD methods included
9 substantial model uncertainty.
Table F-50 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-logistic0'6
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
Dichotomous-Hillc
159.444
190.692
93.9074
117.337
159.444
200.626
159.444
159.444
95.5314
0
0
0.3023
0
0
0
0
0
1
0
4.342
0
0
0
3.943
0
0
0
9.93187
33.9373
1.67195
9.42745
9.9319
61.9146
9.9319
9.9319
2.93951
8.14152
25.4454
1.01633
7.20318
8.14152
47.107
8.14152
8.14152
0.544697
a p-Value from the x^ goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to a 1.
°Slope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit model based on lowest AIC.
Source: Kasai et al. (2009).
F-14
DRAFT - DO NOT CITE OR QUOTE
-------
Log-Logistic Model with 0.95 Confidence Level
0.8
0.6
0.4
0.2
Log-Logistic
BMDLBMD
0
200
400
600
dose
800
1000
1200
09:53 01/14 2011
Figure F-48 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
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
Logistic Model. (Version: 2.13; Date: 10/28/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl_atrophy_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl_atrophy_Lnl-BMR10-Restrict.plt
Fri Jan 14 09:53:22 2011
BMDS_Model_Run
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]
Dependent variable = Effect
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0
intercept = -3.48908
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -background -slope have been estimated at a boundary
point, or have been specified by the user, and do not appear in the correlation
matrix)
F-15
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
9S
£j
26
27
28
29
30
31
32
33
34
35
36
37
38
intercept
intercept
Parameter
1
Estimates
95.0% Wald Confidence Interval
Variable
background
intercept
slope 1 *
Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
o * * *
-2.71122 * * *
* *
* - Indicates that this value is not calculated.
Analysis
Model Log
of Deviance Table
(likelihood) # Param's Deviance Test d.f. P-value
Full model -44.7657 4
Fitted model -45.9537 1 2.37596 3 0.4981
Reduced model -126.116 1 162.701 3 <.0001
AIC: 93.9074
Goodness
Scaled
Dose Est.
0.0000 0.
50.0000 0
250.0000
1250.0000
ChiA2 = 3
Benchmark
Specified
Risk Type
Confidence
of Fit
Prob. Expected Observed Size Residual
0000 0.000 0.000 50 0.000
.7687 38.433 40.000 50 0.525
0.9432 47.161 47.000 50 -0.099
0.9881 49.405 48.000 50 -1.833
.65 d.f. = 3 P-value = 0.3023
Dose Computation
effect =0.1
= Extra risk
level =0.95
BMD = 1.67195
BMDL = 1.
01633
F.6 Hydropic Change
39 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
40 incidence data shown in Table F-51, for hydropic change of the lamina propria in the nasal cavity of male
41 F344/DuCrj rats exposed to 1,4-dioxane vapors for 2 years (Kasai et al. 2009). Doses associated with a
42 BMR of a 10% extra risk were calculated.
F-16
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-51 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
0
0/50
1,4-dioxane
50
2/50
(4%)
vapor concentration (ppm)
250
36/50 a
(72%)
1,250
49/50a
(98%)
p < 0.01 by Fisher's exact test.
Source: Kasai et al., (2QQ9).
1 For incidence of hydropic change of the lamina propria in F344/DuCrj male rats, the gamma,
2 logistic, multistage, probit, Weibull, and quantal-linear models all exhibited a statistically significant lack
3 of fit (i.e., x2/"-value < 0.1; see Table F-53), and thus should not be considered further for identification
4 of a POD. The BMDL estimates for all appropriately fitting models were within threefold difference
5 of each other, indicating that BMDL selection should be made based on model fit (U.S. EPA. 2012a).
6 As assessed by the AIC, the Log-logistic model provided the best fit to the hydropic change of the lamina
7 propria data for male rats (Table F-52, Figure F-49 and subsequent text output), and could be used to
8 derive a POD of for this endpoint.
F-17
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-52 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-logistic0'6
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
Dichotomous-Hillc
98.3441
117.957
90.5388
91.5881
99.3482
136.585
100.225
99.3482
91.8937
0.0002
0
0.6819
0.3458
0.0256
0
0.0033
0.0256
1
-1.321
-1.143
-0.333
-0.538
-2.411
-2.099
-1.899
-2.411
0
51.979
89.2909
68.5266
63.0852
28.7899
92.6118
39.1371
28.7899
73.1032
28.7632
70.6131
46.7808
44.5657
22.6831
74.3784
23.9762
22.6831
49.2687
ap-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to a 1.
°Slope restricted to a 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-49 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-BMR10-Restrict.(d)
4 Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
5 files/lnl_hydrpic_Lnl-BMR10-Restrict.plt
DRAFT - DO NOT CITE OR QUOTE
F-18
-------
1 Fri Jan 14 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-slope*Log(dose))]
7
8 Dependent variable = Effect
9 Independent variable = Dose
10 Slope parameter is restricted as slope >= 1
11
12 Total number of observations = 4
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 = -11.5745
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 1 -0.99
31 slope -0.99 1
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 *
38 slope 2.3501 * * *
39
40 * - 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 4
46 Fitted model -43.2694 2 0.645129 2 0.7243
47 Reduced model -136.935 1 187.976 3 <.0001
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 50 -0.333
57 250.0000 0.6994 34.969 36.000 50 0.318
58 1250.0000 0.9903 49.515 49.000 50 -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 BMD = 68.5266
67 BMDL = 46.7808
F-19
DRAFT - DO NOT CITE OR QUOTE
* *
* *
-------
F.7 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-53, 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-53 Incidence of sclerosis of the lamina propria in the nasal cavity of F344/DuCrj rats
exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0 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. (2QQ9).
5 As assessed by the %2 goodness-of-fit test, all models with the exception of the dichotomous-hill
6 model, exhibited a statistically significant lack of fit (i.e., %2 /"-value < 0. l;See Table F-54), 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 %2 goodness-of-fit
9 test (Table F-54, Figure F-50 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-20
DRAFT - DO NOT CITE OR QUOTE
-------
Table F-54 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-HillCl
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* goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to > 1.
°Slope restricted to > 1.
dBetas restricted to >0.
eModel output warned that the BMDL estimate was "imprecise at best".
Source: Kasai et al. (2QQ9).
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
Dichotomous Hill Model. (Version: 1.2; Date: 12/11/2009)
Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl_sclerosis_Dhl-BMR10-Restrict.(d)
Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/dhl_sclerosis_Dhl-BMR10-Restrict.plt
Fri Jan 14 10:53:28 2011
BMDS_Model_Run
The form of the probability function is:
P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
where: 0 <= g < 1, 0= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
v = -9999
g = -9999
intercept = -11.4511
slope = 1.86444
Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -g have been estimated at a boundary point, or have been
specified by the user, and do not appear in the correlation matrix)
F-21
DRAFT - DO NOT CITE OR QUOTE
-------
1 v intercept slope
2 v 1 0.00074 -0.00078
3 intercept 0.00074 1 -1
4 slope -0.00078 -1 1
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 0 NA
12 intercept -62.1804 4133.38 -8163.46 8039.1
13 slope 11.2979 748.603 -1455.94 1478.53
14
15 NA - 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 1 0.9991
24 Reduced model -123.82 1 129.007 3 <.0001
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 50 -0.001
34 250.0000 0.4400 22.000 22.000 50 0.000
35 1250.0000 0.8000 40.000 40.000 50 -0.000
36
37 ChiA2 = 0.00 d.f. = 1 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 BMD = 206.74
44
45 Warning: BMDL computation is at best imprecise for these data
46 BMDL = 167.46
F-22
DRAFT - DO NOT CITE OR QUOTE
-------
I
0.8
0.6
0.4
0.2
Dichotomous-Hill Model with 0.95 Confidence Level
Dichotomous-Hill
BMDL BMD
200
400
10:5301/142011
600
dose
800
1000
1200
Figure F-50 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-23
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX G. 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 BMCLio values from the best fitting model, determined by adequate global- fit (%p > 0.1)
7 and AIC values, are reported for each endpoint (U.S. EPA. 2012a). Given the multiplicity of tumor sites,
8 basing the IUR on one tumor site will underestimate the carcinogenic potential of 1,4-dioxane.
9 Multitumor BMD analysis was conducted using BMDS (version 2.2beta) MS_Combo program; model
10 output is shown in Section G.3. Additionally, a Bayesian analysis was performed using WinBUGS
11 (Spiegelhalter et al.. 2003). freeware developed by the MRC Biostatistical Unit, Cambridge, United
12 Kingdom (available at http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml) and reported in
13 detail in Section G.3. The results of both analyses were comparable and resulted in equivalent lURs.
14 A summary of the BMDS model predictions for the Kasai et al. (2009) study are shown in
15 Table G-l.
G.1 General Issues and Approaches to BMDS and Multitumor
Modeling
G.1.1 Combining Data tumor types
16 The incidence of adenomas and the incidence of carcinomas within a dose group at a site or tissue
17 in rodents are sometimes combined. This practice is based upon the hypothesis that adenomas may
18 develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a: McConnell et al..
19 1986). In the same manner and was done for the oral cancer assessment (Appendix D), the incidence of
20 hepatic adenomas and carcinomas was summed without double-counting them so as to calculate the
21 combined incidence of either a hepatic carcinoma or a hepatic adenoma in rodents.
22 The remaining of the tumor types were assumed to occur independently.
G.1.2 Summary
23 The BMDS models recommended to calculate rodent BMCio and BMCLio values for individual
24 tumor types and combined tumor analysis are summarized in Table G-l. The first order multistage models
25 for most tumor types were selected because they resulted in the lowest AIC values; however, for renal cell
G-l
DRAFT - DO NOT CITE OR QUOTE
-------
1 carcinoma and Zymbal gland adenoma, the lowest AIC model was not the first order model. In BMDS,
2 the third order model resulted in the lowest AIC (first (1°)-, second (2°)-, and third (3°)-degree models
3 were evaluated); however, using the MCMC approach in WinBUGS, the third order (3°) multistage
4 model did not converge while the second order(2°) model did converge. Thus, for renal cell carcinoma
5 and Zymbal gland adenoma, the second order (2°) multistage model was used in both the MCMC
6 (WinBugs) approach and the BMDS (Version 2.2 beta) MS_Combo approach for direct comparison of
7 results. These results are shown below in Table G-l.
Table G-55 Summary of BMC10 and BMCL10 model results for individual tumor types and
combined tumor analysis for male rats exposed to 1,4-dioxane vapors (Kasai et al..
2009)
Endpoint
Nasal squamous cell carcinoma
Hepatocellular
adenoma/carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Zymbal gland adenoma
Subcutis fibroma3
BMDS Version 2.2beta MS Combo
WinBUGS multitumor analysis"
Multistage
Model
Degree
First (1°)
First (1°)
Third (3°)
First (1°)
First (1°)
Third (3°)
First (1°)
AIC
49.03
127.9
29.99
155.4
86.29
29.99
89.2
p-value
0.9607
0.6928
0.9984
0.8509
0.7904
0.9984
0.5245
X2 Residual
of Interest
0.176
-0.763
0.017
-0.204
-0.149
0.017
0.537
BMCIO(ppm)
1107.04
252.80
1355.16
82.21
1635.46
1355.16
141.762
40.4
39.2
BMCL10
(ppm)
629.95
182.26
16.15
64.38
703.03
16.15
81.9117
30.3
31.4
aHigh-dose dropped. See Section G.2.6 for details.
bln MCMC approach, the simulations for the four-parameter third order(3°) multistage model did not converge for renal cell carcinomas
and Zymbal gland adenomas. Second order (2°) multistage model was used instead.
G.2 BMDS Model Output for Multistage Cancer Models for Individual
Tumor Types
8 For tumor incidence data reported in the Kasai et al. (2009) 2-year inhalation bioassay, multistage
9 cancer models of first (1°)-, second (2°)-, and third (3°)degrees were implemented BMDS (Version
10 2.2Beta). Incidence data used for BMD analysis are shown in Table G-2. Tumor incidence for mammary
11 gland adenoma was excluded from this analysis since only 1 tumor of this type was found across all
12 doses.
G-2
DRAFT - DO NOT CITE OR QUOTE
-------
Table G-56 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)
ec
Nasal squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Zymbal gland adenoma
Subcutis fibroma
0 (clean air)
0/50
1/50
0/50
1/50
0/50
2/50
1/50
0/50
1/50
50
0/50
2/50
0/50
2/50
0/50
4/50
2/50
0/50
4/50
250
1/50
3/50
1/50
4/50
0/50
14/50a
3/50
0/50
9/50a
1,250
6/50D'c
21/50a'c
2/50
22/50a'c
4/50c
41/50a'c
5/50°
4/50c
5/50
ap < 0.01 by Fisher's exact test.
bp < 0.05 by Fisher's exact test.
°p < 0.01 by Peto's test for dose-related trend.
p < 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)
G.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 G-3. The first (l°)-degree polynomial was the best fitting model based on
5 AIC. The plot (Figure G-l) and model output for the first (l°)-degree model are shown below.
G-3
DRAFT - DO NOT CITE OR QUOTE
-------
Table G-57 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
yf 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
aBest-fitting model based on AIC.
I
C
o
0.25
0.2
0.15
0.1
0.05
10:26 11/172010
Multistage Cancer Model with 0.95 Confidence Level
Multistage Cancer
Linear extrapolation
BMDL
BMD
200
400
600
dose
800
1000
1200
Figure G-51 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.pit
Wed Nov 17 10:57:55 2010
BMDSModelRun
G-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 = 4
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 = 1
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) 1
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 * - 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 4
42 Fitted model -23.5154 1 0.534383 3 0.9113
43 Reduced model -30.3429 1 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 0 50 0.000
55 50.0000 0.0047 0.237 0 50 -0.488
56 250.0000 0.0235 1.176 1 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
G-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 BMD = 1107.04
7 BMDL = 629.948
8 BMDU = 2215.11
9
10 Taken together, (629.948, 2215.11) is a 90% two-sided confidence interval for the BMD
G.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 G. 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 G-4. The lst-degree polynomial was the best fitting model based on AIC.
16 The plot (Figure G-2) and model output for the lst-degree model are shown below.
G-6
DRAFT - DO NOT CITE OR QUOTE
-------
Table G-58 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
aBest-fitting model based on AIC.
Multistage Cancer Model with 0.95 Confidence Level
I
C
o
0.6
0.5
0.4
0.3
0.2
0.1
Multistage Cancer
Linear extrapolation
1200
10:24 11/172010
Figure G-52Multistage 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.pit
6 Wed Nov 17 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) ]
G-7
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 The parameter betas are restricted to be positive
o
J
4 Dependent variable = EFFECT
5 Independent variable = DOSE
6
7 Total number of observations = 4
8 Total number of records with missing values = 0
9 Total number of parameters in model = 2
10 Total number of specified parameters = 0
11 Degree of polynomial = 1
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(l)
24 Background 1 -0.53
25 Beta(l) -0.53 1
26
27 Parameter Estimates
28
29 95.0% Wald Confidence Interval
30 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
31 Background 0.0170678 * * *
32 Beta(l) 0.000416776 * * *
33
34 * - 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 1 42.5066 3 <.0001
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 1 50 0.160
52 50.0000 0.0373 1.867 2 50 0.099
53 250.0000 0.1143 5.716 4 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 BMD = 252.799
64 BMDL = 182.256
65 BMDU = 371.457
66
67 Taken together, (182.256, 371.457) is a 90% two-sided confidence interval for the BMD
G-
DRAFT - DO NOT CITE OR QUOTE
-------
G.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 G-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 G-3 and Figure G-4)
8 and model outputs for both the second (2°)- and third-degree (3°) models are shown below.
G-9
DRAFT - DO NOT CITE OR QUOTE
-------
Table G-59 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
aBest-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
BMID
200
400
600 800
dose
1000
1200
1400
10:1702/102011
Figure G-53Multistage model (Second-degree (2°)) for male rat renal cell
carcinomas and Zymbal gland adenomas.
1
2
3
4
5
6
7
8
9
10
11
Multistage Cancer Model. (Version: 1.9; Date: 05/26/2010)
Input Data File: C:/Documents and
Settings/emclanah/Desktop/BMD_14D_Cancer/Data/msc_Kasai2009_renal_Msc2-BMR10.(d)
Gnuplot Plotting File: C:/Documents and
Settings/emclanah/Desktop/BMD_14D_Cancer/Data/msc_Kasai2009_renal_Msc2-BMR10.pit
Thu Feb 10 10:17:39 2011
BMDS_Model_Run
The form of the probability function is:
G-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 = 4
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 = 0
20 Beta(l) = 0
21 Beta(2) = 5.40386e-008
22
23 Asymptotic Correlation Matrix of Parameter Estimates
24 (*** Thg model parameter(s) -Background -Beta(l) 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) 1
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) 0 * * *
36 Beta(2) 5.11454e-008 * * *
37
38 * - 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 1 0.339554 3 0.9524
45 Reduced model -19.6078 1 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 50 -0.080
55 250.0000 0.0032 0.160 0.000 50 -0.400
56 1250.0000 0.0768 3.840 4.000 50 0.085
57
58 ChiA2 = 0.17 d.f. = 3 P-value = 0.9817
59
G-ll
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1,435.28
BMDL = 999.44
BMDU = 3,666.87
Taken together, (999.44 , 3,666.87) is a 90% two-sided confidence interval for the BMD
Multistage Cancer Slope Factor = 0.000100056
Multistage Cancer Model with 0.95 Confidence Level
13
14
15
16
17
18
19
20
21
22
23
24
25
26
0.2
0.15
T3
-------
1
2 Dependent variable = EFFECT
3 Independent variable = DOSE
4
5 Total number of observations = 4
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) = 0
18 Beta(2) = 0
19 Beta(3) = 4.2804e-011
20
21
22 Asymptotic Correlation Matrix of Parameter Estimates
23 (*** The model parameter(s) -Background -Beta(l) -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) 1
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) 0 * * *
36 Beta(2) 0 * * *
37 Beta(3) 4.23353e-011 * * *
38
39 * - 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 1 0.0669578 3 0.9955
46 Reduced model -19.6078 1 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 50 -0.016
59 250.0000 0.0007 0.033 0 50 -0.182
60 1250.0000 0.0794 3.968 4 50 0.017
61
62 ChiA2 = 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
G-13
DRAFT - DO NOT CITE OR QUOTE
-------
1 Confidence level = 0.95
2 BMD = 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 BMD
G.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 G-6. The lst-degree polynomial was the best fitting model based on AIC. The plot (Figure G-5) and
11 model output for the lst-degree model are shown below.
Table G-60 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
AIC
p-value
X^ Residual of
Interest
BMCio
(ppm)
BMCLio
(ppm)
(1°) First3 155.433 0.8509 -0.204 82.21 64.38
(2°) Second 157.168 0.8053 -0.204 96.23 65.15
(3°) Third 157.168 0.8053 0 96.23 65.15
a Best-fitting model based on AIC.
G-14
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
T3
-------
1 Background = 0.0172414
2 Beta(l) = 0.00135351
o
J
4 Asymptotic Correlation Matrix of Parameter Estimates
5
6 Background Beta(l)
7 Background 1 -0.45
8 Beta(l) -0.45 1
9
10 Parameter Estimates
11 95.0% Wald Confidence Interval
12 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
13 Background 0.033631 * * *
14 Beta(l) 0.00128167 * * *
15
16 * - Indicates that this value is not calculated.
17
18 Analysis of Deviance Table
19
20 Model Log(likelihood) # Param's Deviance Test d.f. P-value
21 Full model -75.553 4
22 Fitted model -75.7165 2 0.326905 2 0.8492
23 Reduced model -123.008 1 94.9105 3 <.0001
24
25 AIC: 155.433
26
27 Log-likelihood Constant 68.666413125908832
28
29 Goodness of Fit
30 Scaled
31 Dose Est._Prob. Expected Observed Size Residual
32
33 0.0000 0.0336 1.682 2 50 0.250
34 50.0000 0.0936 4.681 4 50 -0.331
35 250.0000 0.2986 14.928 14 50 -0.287
36 1,250.0000 0.8053 40.265 41 50 0.263
37
38 Chi^2 = 0.32 d.f. = 2 P-value = 0.8509
39
40 Benchmark Dose Computation
41 Specified effect =0.1
42 Risk Type = Extra risk
43 Confidence level = 0.95
44 BMD = 82.2057
45 BMDL = 64.3808
46 BMDU = 107.497
47
48 Taken together, (64.3808, 107.497) is a 90% two-sided confidence interval for the BMD
G.2.5 Mammary Gland Fibroadenoma
49 The incidence data for mammary gland fibroadenomas were monotonic non-decreasing functions
50 of dose; therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
51 BMDS modeling for the multistage cancer model for first (1°)-, second (2°), and third (3°)-degree
52 polynomials are shown in Table G-7. Since quadratic and cubic terms of the multistage models evaluated
53 resulted in the estimates on the boundary, i.e. equal to 0, the first (l°)-degree polynomial was selected
54 based on model parsimony. The plot (Figure G-6) and model output for the first (l°)-degree model are
55 shown below.
G-16
DRAFT - DO NOT CITE OR QUOTE
-------
Table G-61 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 1s-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 G-56Multistage 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.pit
6 Wed Nov 17 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) ]
G-17
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 The parameter betas are restricted to be positive
o
J
4 Dependent variable = EFFECT
5 Independent variable = DOSE
6
7 Total number of observations = 4
8 Total number of records with missing values = 0
9 Total number of parameters in model = 2
10 Total number of specified parameters = 0
11 Degree of polynomial = 1
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(l)
24 Background 1 -0.61
25 Beta(l) -0.61 1
26
27 Parameter Estimates
28
29 95.0% Wald Confidence Interval
30 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
31 Background 0.0315836 * * *
32 Beta(l) 6.44224e-005 * * *
33
34 * - 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 1 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 1 50 -0.468
52 50.0000 0.0347 1.735 2 50 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 BMD = 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 BMD
G-18
DRAFT - DO NOT CITE OR QUOTE
-------
G.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 G-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 G-7) and model output for the first (l°)-degree model are shown below.
Table G-62 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 1s-degree model based on parsimony.
G-19
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Cancer Model with 0.95 Confidence Level
T3
-------
1 Default Initial Parameter Values
2 Background = 0.0327631
3 Beta(l) = 0.000673665
4
5
6 Asymptotic Correlation Matrix of Parameter Estimates
7
8 Background Beta(l)
9 Background 1 -0.68
10 Beta(l) -0.68 1
11
12 Parameter Estimates
13
14 95.0% Wald Confidence Interval
15 Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
16 Background 0.0262054 * * *
17 Beta(l) 0.00074322 * * *
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 -42.4101 3
25 Fitted model -42.6047 2 0.389155 1 0.5327
26 Reduced model -46.5274 1 8.23466 2 0.01629
27
28 AIC: 89.2094
29
30 Log-likelihood Constant 37.900888781466982
31
32 Goodness of Fit
33 Scaled
34 Dose Est._Prob. Expected Observed Size Residual
35
36 0.0000 0.0262 1.310 1 50 -0.275
37 50.0000 0.0617 3.086 4 50 0.537
38 250.0000 0.1913 9.566 9 50 -0.204
39 Chi^2 = 0.41 d.f. = 1 P-value = 0.5245
40
41
42 Benchmark Dose Computation
43 Specified effect = 0.1
44 Risk Type = Extra risk
45 Confidence level =0.95
46 BMD = 141.762
47 BMDL = 81.9117
48 BMDU = 364.364
49
50 Taken together, (81.9117, 364.364) is a 90% two-sided confidence interval for the BMD
G.3 Multitumor Analysis Using BMDS MS_Combo
51 The combined tumor analysis was also performed with beta version of the MS_Combo model in
52 BMDS (Version 2.2beta). The model resulted in similar results to the Bayesian method and model output
53 is shown below for the combined calculation.
54
55 **** start of combined BMD and BMDL Calculations.****
56 Combined Log-Likelihood -277.79874987953076
57 Combined Log-likelihood Constant 246.62591390071873
58
G-21
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 BMD = 40.4937
7 BMDL = 32.331
G.4 Multitumor analysis using Bayesian Methods
8 Given the multiplicity of tumor sites, basing the IUR on one tumor site will likely underestimate
9 the carcinogenic potential of 1,4-dioxane. Simply pooling the counts of animals with one or more tumors
10 (i.e., counts of tumor bearing animals) would tend to underestimate the overall risk when tumors are
11 independent across sites and ignores potential differences in the dose-response relationships across the
12 sites (NRC. 1994; Bogen. 1990). NRC (1994) also noted that the assumption of independence across
13 tumor types is not likely to produce substantial error in the risk estimates unless tumors are known to be
14 biologically dependent.
15 Kopylev et al. (2009) describe a Markov Chain Monte Caro (MCMC) computational approach to
16 calculating the dose associated with a specified composite risk under assumption of independence of
17 tumors. The current Guidelines for Carcinogen Risk Assessment recommend calculation of an upper
18 bound to account for uncertainty in the estimate (U.S. EPA. 2005a). For uncertainty characterization,
19 MCMC methods have the advantage of providing information about the full distribution of risk and/or
20 benchmark dose, which can be used in generating a confidence bound. This MCMC approach building on
21 the re-sampling approach recommended by Bogen (1990). and also provides a distribution of the
22 combined potency across sites.
23 For individual tumor data modeled using the multistage model:
24 P(d q) = 1 - exp[-(q0 + q,d + q2cf + ... + q^)], q£0
25 the model for the combined tumor risk is still multistage, with a functional form that has the sum of
26 stage-specific multistage coefficients as the corresponding multistage coefficient;
27 Pc(d q) = 1
28 The resulting equation for fixed extra risk (BMR) is polynomial in dose (when logarithms of both
29 sides are taken) and can be straightforwardly solved for a combined BMC. Computation of the confidence
30 bound on combined risk BMC can be accomplished via likelihood methods (BMDS-MS_Combo),
31 re-sampling (bootstrap) or Bayesian methods.
32 The MCMC computations were conducted using WinBUGS (Spiegelhalter et al.. 2003)(freeware
33 developed by the MRC Biostatistical Unit, Cambridge, United Kingdom, available at
34 http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml). The model code was checked and
35 compiled within, and the data read into, WinBUGS. Three chains were used for the analysis. Initial values
G-22
DRAFT - DO NOT CITE OR QUOTE
-------
1 for each variable were generated using a Uniform(0,1) distribution and read into WinBUGS. The
2 WinBUGS code calculates the BMC directly (U.S. EPA, 2013).
3 In a Bayesian analysis, the choice of an appropriate prior probability is important. In the
4 examples developed by Kopylev et al. (2009). a diffuse (i.e., high variance or low tolerance) Gaussian
5 prior restricted to be nonnegative was used; such diffuse priors performed reasonably well.
6 The mean and the 5th percentile of the posterior distribution of combined BMC provide estimates
7 of the mean BMC and the lower bound on the BMC (BMCL), respectively, for the combined tumor risk.
8 The values calculated using this method were: mean BMCio 39.2 ppm, and BMCLio 31.4 ppm.
G-23
DRAFT - DO NOT CITE OR QUOTE
------- |