DRAFT - DO NOT CITE OR QUOTE
                                                       EPA/635/R-11/003A
  — ^_ ^_ _                                                www. ep a. gov/iris
oEPA
           TOXICOLOGICAL REVIEW


                                 OF


                        1,4-DIOXANE

               (WITH INHALATION UPDATE)

                          (CAS No. 123-91-1)

               In Support of Summary Information on the
               Integrated Risk Information System (IRIS)

                           September 2011
                              NOTICE


This document is an External Review draft. This information is distributed solely for the purpose of
pre-dissemination peer review under applicable information quality guidelines. It has not been formally
disseminated by EPA. It does not represent and should not be construed to represent any Agency
determination or policy. It is being circulated for review of its technical accuracy and science policy
implications.
                      U.S. Environmental Protection Agency
                             Washington, DC

-------
DISCLAIME  R
          This document is a preliminary draft for review purposes only. This
          information is distributed solely for the purpose of pre-dissemination peer
          review under applicable information quality guidelines. It has not been
          formally disseminated by EPA. It does not represent and should not be
          construed to represent any Agency determination or policy. Mention of trade
          names or commercial products does not constitute endorsement or
          recommendation for use.
                     DRAFT - DO NOT CITE OR QUOTE

-------
CONTENTS:   TOX 1C 0 L  0 G  1C A L   REVIEW   OF

1,4-DIOXANE  (CAS  NO.  123-91-1)


 LIST OF ABBREVIATIONS AND ACRONYMS	XIV

 FOREWORD	XVI

 AUTHORS, CONTRIBUTORS, AND REVIEWERS	XVII
      1.1  INTRODUCTION	1

 2    CHEMICAL AND PHYSICAL INFORMATION                                                 3
              Figure 2-1    1,4-Dioxane chemical structure.	3
              Table 2-1    Physical properties and chemical identity of 1,4-dioxane	3
      TOXICOKINETICS
3.1
3.2
3.3


3.4
3.5

3,
3,




3,
Absorption
Distribution
Metabolism
Figure 3-1 Suggested metabolic pathways of 1 ,4-dioxane in the rat.
Figure 3-2 Plasma 1,4-dioxane levels in rats following i.v. doses of
3-5,600 mg/kg
Elimination
Physiologically Based Pharmacokinetic Models
Figure 3-3 General PBPK model structure.
.5.1 Available Pharmacokinetic Data
.5.2 Published PBPK Models for 1 ,4-Dioxane
3.5.2.1 Leung and Paustenbach
3.5.2.2 Reitzetal.
3.5.2.3 Fisher et al.
3.5.2.4 Sweeney et al.
.5.3 Implementation of Published PBPK Models for 1 ,4-Dioxane
5
6
7
8
9
10
11
12
12
14
14
15
16
16
17
      3.6   Rat Nasal Exposure via Drinking Water	20

      HAZAR D IDE NTIFICATION	21

      4.1   Studies in Humans - Epidemiology, Case Reports, Clinical Controls 	21
        4.1.1 Thiessetal.	22
        4.1.2 Buffleretal.	23
      4.2   Subchronic and Chronic Studies and Cancer Bioassays in Animals - Oral and
           Inhalation	24
        4.2.1 Oral Toxicity	25
           4.2.1.1  Subchronic Oral Toxicity	25
              Table 4-1    Incidence of histopathological lesions in F344/DuCrj rats exposed to
                          1,4-dioxane in drinking water for 13 weeks	28
              Table 4-2    Incidence of histopathological lesions in Crj:BDF1 mice exposed to
                          1,4-dioxane in drinking water for 13 weeks	29
           4.2.1.2  Chronic Oral Toxicity and Carcinogenicity	30
              Table 4-3    Number of incipient liver tumors and hepatomas in male
                          Sprague-Dawley rats exposed to 1,4-dioxane in drinking water for
                          13 months	32
              Table 4-4    Incidence of liver and nasal tumors in male and female Sherman rats
                          (combined) treated with 1,4-dioxane in the drinking water for 2 years	34
              Table 4-5    Incidence of nonneoplastic lesions in Osborne-Mendel rats exposed
                          to 1,4-dioxane in drinking water	36
              Table 4-6    Incidence of nasal cavity sguamous cell carcinoma and liver
                          hepatocellular adenoma in Osborne-Mendel rats exposed to
                          1,4-dioxane in drinking water	37
              Table 4-7    Incidence of hepatocellular adenoma or carcinoma in B6C3F-:  mice
                          exposed to 1,4-dioxane in drinking water	38
                                                                                  in
                       DRAFT - DO NOT CITE OR QUOTE

-------
         Table 4-8    Incidence of histopathological lesions in male F344/DuCrj rats
                      exposed to 1,4-dioxane in drinking water for 2 years	42
         Table 4-9    Incidence of histopathological lesions in female F344/DuCrj rats
                      exposed to 1,4-dioxane in drinking water for 2 years	42
         Table 4-10   Incidence of nasal cavity, peritoneum, and mammary gland tumors in
                      F344/DuCrj rats exposed to 1,4-dioxane in drinking water for 2 years	44
         Table 4-11   Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane in
                      drinking water for 2 years	44
         Table 4-12   Incidence of histopathological lesions in male Crj:BDF1 mice exposed
                      to 1,4-dioxane in drinking water for 2 years	46
         Table 4-13   Incidence of histopathological lesions in female Crj:BDF1 mice
                      exposed to 1,4-dioxane in drinking water for 2 years	46
         Table 4-14   Incidence of tumors in Crj:BDF1 mice exposed to 1,4-dioxane in
                      drinking water for 2 years	47
   4.2.2  Inhalation Toxicity	48
      4.2.2.1   Subchronic Inhalation Toxicity	48
         Table 4-15   Terminal body weights and relative organ weights of F344/DuCrj rats
                      exposed to 1,4-dioxane vapor by whole-body inhalation for 13 weeks	50
         Table 4-16   Hematology and clinical chemistry of F344/DuCrj rats exposed to
                      1,4-dioxane vapor by whole-body inhalation for 13 weeks	51
         Table 4-17   Incidence data of histopathological lesions in F344/DuCrj rats
                      exposed to 1,4-dioxane vapor by whole-body inhalation for 13 weeks	51
      4.2.2.2   Chronic Inhalation Toxicity and Carcinogenicity	52
         Table 4-18   Terminal body and relative organ weights of F344/DuCrj male rats
                      exposed to 1,4-dioxane vapor by whole-body inhalation for 2 years	56
         Table 4-19   Hematology and clinical chemistry of F344/DuCrj male rats exposed
                      to 1,4-dioxane vapor by whole-body inhalation for 2 years	56
         Table 4-20   Incidence of pre-and nonneoplastic lesions in male F344/DuCrj rats
                      exposed to 1,4-dioxane vapor by whole-body inhalation for 2 years	57
         Table 4-21   Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
                      vapor by whole-body inhalation for 2 years	58
   4.2.3  Initiation/Promotion Studies	58
      4.2.3.1   Bulletal.	58
      4.2.3.2   Kingetal.	58
      4.2.3.3   Lundberg et al.	59
4.3  Reproductive/Developmental Studies—Oral and Inhalation	60
   4.3.1  Giavini et al.	60
4.4  Other Duration or Endpoint Specific Studies	61
   4.4.1  Acute and Short-term Toxicity	61
      4.4.1.1   Oral Toxicity	61
      4.4.1.2   Inhalation Toxicity	61
         Table 4-22   Acute and short-term toxicity studies of 1,4-dioxane	62
   4.4.2  Neurotoxicity	63
      4.4.2.1   Frantiketal.	64
      4.4.2.2   Goldberg et al.	64
      4.4.2.3   Kanada et al. 	65
      4.4.2.4   Knoefel	65
4.5  Mechanistic Data and Other Studies in Support of the Mode of Action	65
   4.5.1  Genotoxicity	65
         Table 4-23   Genotoxicity studies of 1,4-dioxane; in vitro	69
         Table 4-24   Genotoxicity studies of 1,4-dioxane; mammalian in vivo	71
   4.5.2  Mechanistic Studies 	72
      4.5.2.1   Free Radical Generation 	72
      4.5.2.2   Induction of Metabolism	73
      4.5.2.3   Mechanisms of Tumor Induction 	73
4.6  Synthesis of Major Noncancer Effects	75
   4.6.1  Oral 75
         Table 4-25   Oral toxicity studies (noncancer effects) for 1,4-dioxane	77
   4.6.2  Inhalation	79
         Table 4-26   Inhalation toxicity studies (noncancer  effects) for 1,4-dioxane	81
      4.6.2.1   Mode of Action Information	81
4.7  Evaluation of Carcinogenicity	83
   4.7.1  Summary of Overall Weight of Evidence	83
   4.7.2  Synthesis of Human, Animal, and Other Supporting Evidence	84
   4.7.3  Mode of Action Information	86
      4.7.3.1   Identification of Key Events for Carcinogenicity	87

                                                                                   iv
                   DRAFT - DO NOT CITE OR QUOTE

-------
         Figure 4-1   A schematic representation of the possible key events in the delivery
                     of 1,4-dioxane to the liver and the hypothesized MOA(s) for liver
                     carcinogenicity_
      4.7.3.2   Strength, Consistency, Specificity of Association	89
      4.7.3.3   Dose-Response Relationship	90
         Table 4-27  Temporal sequence and dose-response relationship for possible key
                     events and liver tumors in rats and mice                                  91
Table 4-28 Temporal sequence and dose-response relationship for possible key
events and nasal tumors in rats and mice
4.7.3.4 Temporal Relationship
4.7.3.5 Bioloqical Plausibility and Coherence
4.7.3.6 Other Possible Modes of Action
4.7.3.7 Conclusions About the Hypothesized Mode of Action
4.7.3.8 Relevance of the Mode of Action to Humans
Susceptible Populations and Life Stages
93
94
95
96
97
97
98
4.8

DOS E-RESPONSEASSESSMENTS	99

5.1  Oral Reference Dose (RfD)	99
   5.1.1  Choice of Principal Studies and Critical Effect with Rationale and Justification	99
   5.1.2  Methods of Analysis—Including Models (PBPK, BMD, etc.)	100
         Table 5-1    Incidence of cortical tubule degeneration in Osborne-Mendel rats
                     exposed to 1,4-dioxane in drinking water for 2 years	102
         Table 5-2   BMD and BMDL values derived from BMD modeling of cortical tubule
                     degeneration in male and female Osborne-Mendel rats exposed to
                     1,4-dioxane in drinking water for 2 years	102
         Table 5-3   Incidence of liver hyperplasia in F344/DuCrj rats exposed to
                     1,4-dioxane in drinking water for 2 years	102
         Table 5-4   BMD and BMDL values derived from BMD modeling of liver
                     hyperplasia in male and female F344/DuCrj  rats exposed to
                     1,4-dioxane in drinking water for 2 years	103
   5.1.3  RfD Derivation - Including Application of Uncertainty Factors (UFs)	103
   5.1.4  RfD Comparison Information	104
         Figure 5-1   Potential points of departure (POD) for liver toxicity endpoints with
                     corresponding applied uncertainty factors and derived RfDs following
                     oral exposure to 1,4-dioxane.	105
         Figure 5-2   Potential points of departure (POD) for kidney toxicity endpoints with
                     corresponding applied uncertainty factors and derived RfDs following
                     oral exposure to 1,4-dioxane.	106
         Figure 5-3   Potential points of departure (POD) for nasal inflammation with
                     corresponding applied uncertainty factors and derived sample RfDs
                     following oral exposure to 1,4-dioxane.	107
         Figure 5-4   Potential points of departure (POD) for organ specific toxicity
                     endpoints with corresponding applied uncertainty factors and derived
                     sample RfDs following oral exposure to 1,4-dioxane. 	108
   5.1.5  Previous RfD Assessment	108
5.2  Inhalation  Reference Concentration (RfC)	108
   5.2.1  Choice of Principal Study and Candidate Critical Effect(s) with Rationale and
         Justification	108
         Table 5-5   Incidences  of nonneoplastic lesions resulting from chronic exposure
                     (ppm) to 1,4-dioxane considered for identification of a critical effect.	111
   5.2.2  Methods of Analysis	111
   5.2.3  Exposure Duration and Dosimetric Adjustments	111
         Table 5-6   Duration adjusted POD estimates for BMDLs (from best fitting BMDS
                     models) or  NOAELs/LOAELs from chronic exposure to 1,4-dioxane	112
   5.2.4  RfC Derivation- Including Application of Uncertainty Factors (UFs)	115
   5.2.5  RfC Comparison Information	115
         Figure 5-5   Potential points of departure (POD) for candidate endpoints with
                     corresponding applied uncertainty factors and derived sample RfCs
                     following inhalation exposure to 1,4-dioxane. 	116
   5.2.6  Previous RfC Assessment	116
5.3  Uncertainties in the Oral  Reference Dose and Inhalation Reference Concentration	116
5.4  Cancer Assessment	118
   5.4.1  Choice of Study/Data - with Rationale and Justification	118
      5.4.1.1   Oral Study/Data	118
                  DRAFT - DO NOT CITE OR QUOTE

-------
              Table 5-7    Incidence of liver, nasal cavity, peritoneal, and mammary gland
                          tumors in rats and mice exposed to 1,4-dioxane in drinking water for
                          2 years (based on survival to 12 months)	120
           5.4.1.2   Inhalation Study/Data	120
              Table 5-8    Incidence of liver, nasal cavity, kidney, peritoneal, and mammary
                          gland, Zymbal gland, and subcutis tumors in rats exposed to
                          1,4-dioxane vapors for 2 years.	122
        5.4.2  Dose-Response Data	122
           5.4.2.1   Oral Data	122
              Table 5-9    Incidence of hepatocellular adenoma or carcinoma in rats and mice
                          exposed to 1,4-dioxane in drinking water for2 years	123
           5.4.2.2   Inhalation Data	123
              Table 5-10   Incidence of tumors in F344 male rats exposed to 1,4-dioxane for 104
                          weeks (6 hours/day, 5 days/week)	124
        5.4.3  Dose Adjustments and Extrapolation Method(s)	124
           5.4.3.1   Oral	124
              Table 5-11   Calculated HEDs for the tumor incidence data  used for
                          dose-response modeling	125
           5.4.3.2   Inhalation	126
        5.4.4  Oral  Slope Factor and Inhalation Unit Risk	128
           5.4.4.1   Oral Slope Factor	128
              Table 5-12   BMD HED and BMDI_HED values from models fit to tumor incidence data
                          for rats and mice exposed to 1,4-dioxane in drinking  water for 2 years
                          and corresponding oral CSFs	129
           5.4.4.2   Inhalation Unit Risk	130
              Table 5-13   Dose-response modeling summary results for male rat tumors
                          associated with inhalation exposure to 1,4-dioxane for 2 years	131
        5.4.5  Previous Cancer Assessment	132
     5.5  Uncertainties in Cancer Risk Values 	133
        5.5.1  Sources of Uncertainty	133
           5.5.1.1   Choice of Low-Dose Extrapolation Approach	133
           5.5.1.2   Dose Metric	134
           5.5.1.3   Cross-Species Scaling	134
           5.5.1.4   Statistical Uncertainty at the POD	135
           5.5.1.5   Bioassay Selection	135
           5.5.1.6   Choice of Species/Gender	135
           5.5.1.7   Relevance to Humans	136
           5.5.1.8   Human Population Variability	136
              Table 5-14   Summary of uncertainty in the 1,4-dioxane cancer risk estimation	137

6    MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD AND DOSE
     RESPONSE 	1 38

     6.1  Human Hazard Potential 	138
     6.2  DOSE RESPONSE 	139
        6.2.1  Noncancer/Oral	139
        6.2.2  Noncancer/lnhalation  	140
        6.2.3  Cancer	140
           6.2.3.1   Oral	140
           6.2.3.2   Inhalation	140
           6.2.3.3   Choice of Low-Dose Extrapolation Approach	141
           6.2.3.4   Dose Metric	142
           6.2.3.5   Cross-Species Scaling	142
           6.2.3.6   Statistical Uncertainty at the POD	142
           6.2.3.7   Bioassay Selection	143
           6.2.3.8   Choice of Species/Gender	143
           6.2.3.9   Relevance to Humans	143
           6.2.3.10  Human Population Variability	144

REFERENCES	145

APPENDIX A.       SUMMARY OF EXTERNAL PEER RE VIEW AND PUBLIC COMMENTS
     AND DIS P OS ITION	A-1

     A.1  External Peer Review Panel Comments	A-1
        A. 1.1  General Charge Questions	A-1

                                                                                     vi
                       DRAFT - DO NOT CITE OR QUOTE

-------
        A. 1.2 Oral reference dose (RfD) for 1,4-dioxane 	A-5
        A.1.3 Carcinogenicity of 1,4-dioxane	A-10
     A.2  Public Comments	A-15
        A.2.1 Oral reference dose (RfD) for 1,4-dioxane 	A-15
        A.2.2 Carcinogenicity of 1,4-dioxane	A-15
        A.2.3 PBPK Modeling	A-18
        A.2.4 Other Comments	A-18

APPENDIX B.      EVALUATION OF EXISTING PBPK MODELS FOR  1,4-DIOXANE	B-1

     B.1  Background 	B-1
     B.2  Scope	B-1
     B.3  Implementation of the Empirical Models in aclsXtreme	B-2
        B.3.1 Model Descriptions	B-2
              Figure B-1   Schematic representation of empirical model for 1,4-dioxane in rats.	B-3
              Figure B-2   Schematic representation of empirical model for 1,4-dioxane in
                          humans.	B-3
        B.3.2 Modifications to the Empirical Models	B-3
        B.3.3 Results 	B-4
              Figure B-3   Output of 1,4-dioxane blood level data from the acslXtreme
                          implementation (left) and published (right) empirical rat model
                          simulations of i.v. administration experiments.	B-5
              Figure B-4   Output of HEAA urine level data from  acslXtreme implementation
                          (left) and published (right) empirical rat model simulations of i.v.
                          administration experiments.	B-5
              Figure B-5   acslXtreme predictions  of blood 1,4-dioxane and urine HEAA levels
                          from the empirical rat model simulations of a 6-hour, 50-ppm
                          inhalation exposure.	B-6
              Figure B-6   Output of 1,4-dioxane blood level data from the acslXtreme
                          implementation (left) and published (right) empirical human model
                          simulations of a 6-hour, 50-ppm inhalation exposure.	B-7
              Figure B-7   Observations and acslXtreme  predictions of cumulative HEAA in
                          human urine following a 6-hour, 50-ppm inhalation exposure.	B-7
              Figure B-8   EPA-modified Young et al. empirical model prediction (line) of plasma
                          1,4-dioxane levels in rats following exposure to 1,4-dioxane for 13
                          weeks compared to data from  Kasai et al. (2008).	B-8
        B.3.4 Conclusions for Empirical Model Implementation	B-8
     B.4  Initial Recalibration of the PBPK Model	B-9
        B.4.1 Sources of Values for Flow Rates	B-9
              Table B-1    Human PBPK model parameter values for 1,4-dioxane	B-10
        B.4.2 Sources of Values for Partition Coefficients	B-10
        B.4.3 Calibration Method 	B-10
        B.4.4 Results 	B-11
              Table B-2   PBPK metabolic and elimination parameter values resulting from
                          re-calibration of the human model using alternative values for
                          physiological flow rates3 and tissue:air partition coefficients	B-11
              Figure B-9   Predicted and observed blood 1,4-dioxane concentrations (left) and
                          urinary HEAA levels (right) following re-calibration  of the human
                          PBPK model with tissue:air partition coefficient values.	B-12
              Figure B-10 Predicted and observed blood 1,4-dioxane concentrations (left) and
                          urinary HEAA levels (right) following re-calibration  of the human
                          PBPK model with tissue:air partition coefficient values.	B-12
              Figure B-11 Predicted and observed blood 1,4-dioxane concentrations (left) and
                          urinary HEAA levels (right)	B-13
              using EPA estimated biologically plausible parameters (Table B-1).	B-13
        B.4.5 Conclusions for PBPK Model Implementation	B-13
        B.4.6 Sensitivity Analysis	B-14
        B.4.7 Method 	B-14
        B.4.8 Results 	B-15
              Figure B-12 The highest seven sensitivity coefficients (and associated
                          parameters) for blood 1,4-dioxane concentrations (CV) at 1 (left) and
                          4 (right) hours of a 50-ppm inhalation exposure.	B-15
     B.5  PBPK Model Exercises Using Biologically Plausible Paramter Boundaries	B-15
        B.5.1 Observations Regarding the Volume  of Distribution	B-16
        B.5.2 Defining Boundaries for Parameter Values	B-16
        B.5.3 Results 	B-16

                                                                                       vii
                        DRAFT - DO NOT CITE OR QUOTE

-------
              Figure B-13 Comparisons of the range of PBPK model predictions from upper and
                          lower boundaries on partition coefficients with empirical model
                          predictions and experimental observations for blood 1,4-dioxane
                          concentrations (left) and urinary HEAA levels (right) from a 6-hour,
                          50-ppm inhalation exposure.
              Figure B-14 Comparisons of the range of PBPK model predictions from upper and
                          lower boundaries on partition coefficients with empirical model
                          predictions and experimental observations for blood 1,4-dioxane
                          concentrations (left) and urinary HEAA levels (right) from a 6-hour,
                          50-ppm inhalation exposure.	
              Table B-3
        B.5.4
              Figure B-17 Predictions of blood 1,4-dioxane concentration following simultaneous
                          calibration of a zero-order metabolism rate constant, ki_c, and slowly
                          perfused tissue:air partition coefficient to the experimental  data.	
     B.6  Conclusions	
     B.7  aclsXtreme Code for the Young et al. Empricial Model for 1,4-Dioxane in Rats	
     B.8  aclsXtreme Code for the Young et al. Empricial Model for 1,4-Dioxane in Humans_
     B.9  aclsXtreme Code for the Reitz et al. PBPK Model For 1,4-Dioxane	
                                                                                 B-17
                                                                                 B-18
            PBPK metabolic and elimination parameter values resulting from
            recalibration of the human model using biologically plausible values
            for physiological flow rates3 and selected upper and lower boundary
            values fortissue:air partition coefficients	B-18
Alternative Model Parameterization	B-18
Figure B-15   Predictions of blood 1,4-dioxane concentration following calibration of
            a zero-order metabolism rate constant, ki_c, to the experimental data. 	B-19
Figure B-16   Predictions of blood 1,4-dioxane concentration following calibration of
            a zero-order metabolism rate constant, ki_c, to only the exposure
            phase of the experimental data._                                        B-20
                                                                                 B-21
                                                                                 B-21
                                                                                "B-22
                                                                                 B-24
                                                                                "B-25
APPENDIX C.
     DETAILS OF BMD ANALYSIS  FOR ORAL RFD FOR 1,4-DIOXANE	C-1
     C.1  Cortical Tubule Degeneration_
              Table C-1    Incidence of cortical tubule degeneration in Osborne-Mendel rats
                           exposed to 1,4-dioxane in drinking water for 2 years	
              Table C-2    Goodness-of-fit statistics and BMDio and BMDL-io values from models
                           fit to incidence data for cortical tubule degeneration in male and
                           female Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in
                           drinking water	
              Figure C-1    BMD Log-probit model of cortical tubule degeneration incidence data
                           for male  rats exposed to 1,4-dioxane in drinking water for 2 years to	
              Figure C-2    BMD Weibull model of cortical tubule degeneration incidence data for
                           female rats exposed to 1,4-dioxane in drinking water for 2 years to
     C.2
              Table C-4
              Figure C-3
              Figure C-4
              Table C-4.
              Figure C-5
              Table C-4.
              Figure C-6


              Table C-4.
              Figure C-7
            Benchmark dose modeling results based on the incidence of liver
            hyperplasias in male and female F344 rats exposed to 1,4-dioxane in
            drinking water for 2 years_
                                                                                 C-1
                                                                                  C-1
                                                                                  C-2
                                                                                 C-3
Liver hyperplasi;
Table C-3
support the results in Table C-2
a
Incidence of
1,4-dioxane
liver hyperplasia in
in drinking water3


F344/DuCrj


rats exposed


to
C-5
C-6
C-7
                                                                                                C-8
            BMD gamma model of liver hyperplasia incidence data for F344 male
            rats exposed to 1,4-dioxane in drinking water for 2 years to support
            results Table C-4.	C-9
            BMD multistage (2 degree) model of liver hyperplasia incidence data
            for F344 male rats exposed to 1,4-dioxane in drinking water for 2
            years to support results	C-11
            C-12
            BMD Weibull model of liver hyperplasia incidence data for F344 male
            rats exposed to 1,4-dioxane in drinking water for 2 years to support
            the results in	C-13
            C-14
            BMD quantal-linear model of liver hyperplasia incidence data for F344
            male rats exposed to 1,4-dioxane in drinking water for 2 years to
            support the results in	C-15
            C-16
            BMD Multistage model (third (3°)of liver hyperplasia incidence data
            for F344 female rats exposed to 1,4-dioxane in drinking water for 2
            years to support the results in	C-17
                                                                                      vm
                        DRAFT - DO NOT CITE OR QUOTE

-------
             Table C-4.   C-18
APPENDIX D.
        DETAILS OF BMD ANALYSIS FOR ORAL CSF FOR 1,4-DIOXANE	D-1
     D.1   General Issues and Approaches to BMDS Modeling _
        D.1.1 Combining Data on Adenomas and Carcinomas_
        D.1.2 Model Selection Criteria	
        D.1.3 Summary	
             Table D-1
               Recommended models for rodents exposed to 1,4-dioxane in drinking
               water (Kano et al., 2009)
     D.2  Female F344 Rats: Hepatic Carcinomas and Adenomas_
             Table D-2
             Table D-3
               Data for hepatic adenomas and carcinomas in female F344 rats
               (Kano et al., 2009)	
               BMDS dose-response modeling results for the combined incidence of
               hepatic adenomas and carcinomas in female F344 rats (Kano et al.,
               2009).
     D.3
    Figure D-1   Multistage BMD model (2 degree) for the combined incidence of
               hepatic adenomas and carcinomas in female F344 rats.	
Male F344 Rats: Hepatic Carcinomas and Adenomas_
             Table D-5


             Figure D-2

             Figure D-3
     D.4
               Multistage BMD model (3 degree) for the combined incidence of
               hepatic adenomas and carcinomas in male F344 rats.	
F344 Rats: Tumors at Other Sites	
    Table D-6
             Table D-7

             Figure D-4

             Table D-8

             Figure D-5

             Table D-9

             Figure D-6

             Figure D-7

             Table D-10

             Figure D-8
             Figure D-9
Data for significant tumors at other sites in male and female F344 rats
(Kano et al., 2009)	
BMDS dose-response modeling results for the incidence of nasal
cavity tumors in female F344 rats3 (Kano et al., 2009)
               Multistage BMD model (3 degree) for nasal cavity tumors in female
               F344 rats.	
               BMDS dose-response modeling results for the incidence of nasal
               cavity tumors in male F344 rats3 (Kano et al., 2009) _
               Multistage BMD model (3 degree) for nasal cavity tumors in male
               F344 rats.	
               BMDS dose-response modeling results for the incidence of mammary
               gland adenomas in female F344 rats (Kano et al., 2009)
               LogLogistic BMD model for mammary gland adenomas in female
               F344 rats.	
               Multistage BMD model (1 degree) for mammary gland adenomas in
               female F344 rats.	
               BMDS dose-response modeling results for the incidence of peritoneal
               mesotheliomas in male F344 rats (Kano et al., 2009)	
               Probit BMD model for peritoneal mesotheliomas in male F344 rats.
               Multistage BMD (2 degree) model for peritoneal mesotheliomas in
               male F344 rats.
     D.5
Female BDF1 Mice: Hepatic Carcinomas and Adenomas	
    Table D-11  Data for hepatic adenomas and carcinomas in female BDF1 mice
               (Kano etal., 2009).
             Table D-12
             Table D-13
             Figure D-10
     D.6
                LogLogistic BMD model for the combined incidence of hepatic
               adenomas and carcinomas in female BDF1 mice with a BMR of 10%.
    Figure D-11   LogLogistic BMD model for the combined incidence of hepatic
               adenomas and carcinomas in female BDF1 mice with a BMR of 30%.
    Figure D-12   LogLogistic BMD model for the combined incidence of hepatic
               adenomas and carcinomas in female BDF1 mice with a BMR of 50%.
    Figure D-13   Multistage BMD model (1  degree) for the combined incidence of
               hepatic adenomas and carcinomas in female BDF1 mice. 	
Male BDF1 Mice: Hepatic Carcinomas and Adenomas	
                                                                 D-2
                                                                 D-2
                                                                "D-3
                                                                -D-4

                                                                 D-4
                                                                -D-4

                                                                 D-5
                                                                _D-6

                                                                 D-6
                                                                "D-8
               BMDS dose-response modeling results for the combined incidence of
               adenomas and carcinomas in livers of male F344 rats (Kano et al.,
               2009)	D-9
               Probit BMD model for the combined incidence of hepatic adenomas
               and carcinomas in male F344 rats.                                    D-9
 D-11
.D-13

.D-13

.D-14

.D-14

.D-17

.D-17

.D-20

.D-20

.D-22

 D-24
~_D-24

 D-26
~D-27

 D-28
               BMDS dose-response modeling results for the combined incidence of
               hepatic adenomas and carcinomas in female BDF1 mice (Kano et al.,
               2009)	
               BMDS LogLogistic dose-response modeling results using BMRs of
               10, 30, and 50% for the combined incidence of hepatic adenomas
               and carcinomas in female BDF1 mice (Kano et al., 2009).
                                                                                         D-29
                                                               .D-29

                                                               .D-30

                                                               .D-32

                                                               .D-34

                                                                D-36
                                                                D-37
                                                                                  IX
                      DRAFT - DO NOT CITE OR QUOTE

-------
      Table D-14

      TableD-15


      Figure D-14

      Figure D-15
                    Data for hepatic adenomas and carcinomas in male BDF1 mice
                    (Kanoetal., 2009).
                     Multistage BMD model (1 degree) for the combined incidence of
                    hepatic adenomas and carcinomas in male BDF1 mice.	
D.7  BMD Modeling Results from Additional Chronic Bioassays	
         Table D-16  Summary of BMDS dose-response modeling estimates associated
                    with liver and nasal tumor incidence data resulting from chronic oral
                    exposure to 1,4-dioxane in rats and mice
D.7.1
         Hepatocellular Carcinoma and Nasal Squamous Cell Carcinoma (Kociba et al.,
         1974)	
      TableD-17
      TableD-18
                    Incidence of hepatocellular carcinoma and nasal squamous cell
                    carcinoma in male and female Sherman rats (combined) (Kociba et
                    al., 1974) treated with 1,4-dioxane in the drinking water for 2 years _
                    BMDS dose-response modeling results for the incidence of
                    hepatocellular carcinoma in male and female Sherman rats
                    (combined) (Kociba et al., 1974) exposed to 1,4-dioxane in the
                    drinking water for 2 years_
      TableD-19
                    BMDS dose-response modeling results for the incidence of nasal
                    squamous cell carcinoma in male and female Sherman rats
                    (combined) (Kociba et al., 1974) exposed to 1,4-dioxane in the
                    drinking water for 2 years_
      Figure D-18  Multistage BMD model (3 degree) for the incidence of nasal
                 squamous cell carcinoma in male and female Sherman rats exposed
                 to 1,4-dioxane in drinking water.  	
D.7.2  Nasal Cavity Squamous Cell Carcinoma and Liver Hepatocellular Adenoma in
      Osborne-Mendel Rats (NCI, 1978)	
      Table D-20  Incidence of nasal cavity squamous cell carcinoma and hepatocellular
      Table D-21
                 adenoma in Osborne-Mendel rats (NCI, 1978) exposed to
                 1,4-dioxane in the drinking water	
                    BMDS dose-response modeling results for the incidence of
                    hepatocellular adenoma in female Osborne-Mendel rats (NCI, 1978)
                    exposed to 1,4-dioxane in the drinking water for 2 years_
      Table D-22
      Figure D-21   LogLogistic BMD model for the incidence of nasal cavity squamous
                 cell carcinoma in female Osborne-Mendel rats exposed to
                 1,4-dioxane in drinking water.
      Figure D-22  Multistage BMD model (1 degree) for the incidence of nasal cavity
                 squamous cell carcinoma in female Osborne-Mendel rats exposed to
                 1,4-dioxane in drinking water. 	
      Table D-23
                    BMDS dose-response modeling results for the incidence of nasal
                    cavity squamous cell carcinoma in male Osborne-Mendel rats (NCI,
                    1978) exposed to 1,4-dioxane in the drinking water for 2 years_
      Figure D-23  LogLogistic BMD model for the incidence of nasal cavity squamous
                 cell carcinoma in male Osborne-Mendel rats	
      exposed to 1,4-dioxane in drinking water.	
 D-38
                    BMDS dose-response modeling results for the combined incidence of
                    hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.,
                    2009)	
                     LogLogistic BMD model for the combined incidence of hepatic
                    adenomas and carcinomas in male BDF1  mice.
.D-38

.D-39

 D-41
 D-42
                                                                                   D-43
                                                                                   D-43
                                                                                   D-44
                                                                                   D-44
      Figure D-16  Probit BMD model for the incidence of hepatocellular carcinoma in
                 male and female Sherman rats exposed to 1,4-dioxane in drinking
                 water.	D-45
      Figure D-17  Multistage BMD model (1 degree) for the incidence of hepatocellular
                 carcinoma in male and female Sherman rats exposed to 1,4-dioxane
                 in drinking water.                                                   D-47
                                                                                   D-49
                                                                                      D-49
                                                                                      D-51
                                                                                      D-52
                                                                                   D-53
      Figure D-19  LogLogistic BMD model for the incidence of hepatocellular adenoma
                 in female Osborne-Mendel rats exposed to 1,4-dioxane in drinking
                 water.	D-53
      Figure D-20  Multistage BMD model (1 degree) for the incidence of hepatocellular
                 adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
                 drinking water.	D-55
                    BMDS dose-response modeling results for the incidence of nasal
                    cavity squamous cell carcinoma in female Osborne-Mendel rats (NCI,
                    1978) exposed to 1,4-dioxane in the drinking water for 2 years_            D-57
                                                                                      D-57
                                                                                      D-59
.D-61

 D-61
"D-61
               DRAFT - DO NOT CITE OR QUOTE

-------
              Figure D-24  Multistage BMD model (1 degree) for the incidence of nasal cavity
                          squamous cell carcinoma in male Osborne-Mendel ratsexposed to
                          1,4-dioxane in drinking water. 	
        D.7.3 Hepatocellular Adenoma or Carcinoma in B6C3F-I Mice (NCI, 1978)	
              Table D-24  Incidence of hepatocellular adenoma or carcinoma in male and
                          female B6C3Fi mice (NCI, 1978) exposed to 1,4-dioxane in drinking
                          water
                                                                                  D-63
                                                                                  "D-64
                                                                                  D-66
              Table D-25
                BMDS dose-response modeling results for the combined incidence of
                hepatocellular adenoma or carcinoma in female B6C3F-: mice (NCI,
                1978) exposed to 1,4-dioxane in the drinking water for 2 years	D-66
              Figure D-25  Multistage BMD model (2 degree) for the incidence of hepatocellular
                          adenoma or carcinoma in female B6C3F-: mice exposed to
                          1,4-dioxane in drinking water.
                                                                                  D-67
              Table D-26
                BMDS dose-response modeling results for the combined incidence of
                hepatocellular adenoma or carcinoma in male B6C3F-: mice (NCI,
                1978) exposed to 1,4-dioxane in drinking water_
                                                                                            D-69
              Figure D-26  Gamma BMD model for the incidence of hepatocellular adenoma or
                          carcinoma in male B6C3F-: mice exposed to 1,4-dioxane in drinking
                          water.	D-69
              Figure D-27  Multistage BMD model (2 degree) for the incidence of hepatocellular
                          adenoma or carcinoma in  male B6C3F-: mice exposed to 1,4-dioxane
                          in drinking water.	D-71
APPENDIX E.       COMPARISON OF SEVERAL DATA REPORTS FOR THE JBRC 2-YEAR
     1,4-DIOXANE DRINKING WATER STUDY	E-1
              Table E-1

              Table E-2

              Table E-3

              Table E-4

              Table E-5

              Table E-6

              Table E-7

              Table E-8
                Nonneoplastic lesions: Comparison of histological findings reported
                for the 2-year JBRC drinking water study in male F344 rats
 E-2
                Nonneoplastic lesions: Comparison of histological findings reported
                for the 2-year JBRC drinking water study in female F344 rats_
 E-3
                Neoplastic lesions: Comparison of histological findings reported for
                the 2-year JBRC drinking water study in male F344 rats_
 E-4
                Neoplastic lesions: Comparison of histological findings reported for
                the 2-year JBRC drinking water study in female F344 rats	
 E-5
                Nonneoplastic lesions: Comparison of histological findings reported
                for the 2-year JBRC drinking water study in male Crj:BDF1 mice	E-6
                Nonneoplastic lesions: Comparison of histological findings reported
                for the 2-year JBRC drinking water study in female Crj:BDF1 mice	E-7
                Neoplastic lesions: Comparison of histological findings reported for
                the 2-year JBRC drinking water study in male Crj:BDF1 mice_               E-8
                Neoplastic lesions: Comparison of histological findings reported for
                the 2-year JBRC drinking water study in female Crj:BDF1 mice	
                                                                                              E-9
APPENDIX F.
         DETAILS OF BMD ANALYSIS FOR INHALATION RFC FOR 1,4-DIOXANE	F-1
     F.1  Centrilobular Necrosis of the Liver
                                                                                    F-1
              Table F-1
              Table F-2
                Incidence of centrilobular necrosis of the liver in F344/DuCrj rats
                exposed to 1,4-dioxane via inhalation for 2 years	
                                                                                              F-1
     F.2
                Goodness-of-fit statistics and BMD-io and BMDL-io values from models
                fit to incidence data for centrilobular necrosis of the liver in male
                F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai et al., 2009)	
    Figure F-1   BMD Dichotomous Hill model of centrilobular necrosis incidence data
                for male rats exposed to 1,4-dioxane vapors for 2 years to support the
                results in Table F-2.	
Spongiosis Hepatis	
                                                                                              F-2
              Table F-3
              Table F-4
              Figure F-2
              Figure F-3
                Incidence of spongiosis hepatis of the liver in F344/DuCrj rats
                exposed to 1,4-dioxane via inhalation for 2 years	
 F-2
.F-4

 F-4
                Goodness-of-fit statistics and BMD-io and BMDL-io values from models
                fit to incidence data for spongiosis hepatis of the liver in male
                F344/DuCrj rats (NCI,  1978) exposed to 1,4-dioxane vapors	
                                                                                              F-5
                BMD Dichotomous-Hill model of spongiosis hepatis incidence data for
                male rats exposed to 1,4-dioxane vapors for 2 years to support the
                results in Table F-4.
                                                                                              F-5
                BMD Log-Logistic model of spongiosis hepatis incidence data for
                male rats exposed to 1,4-dioxane vapors for 2 years to support the
                results in Table F-4.
                                                                                              F-7
                                                                                     XI
                       DRAFT - DO NOT CITE OR QUOTE

-------
F.3  Squamous Cell Metaplasia_
                                                                                      F-8
         Table F-5
         Table F-6
                Incidence of squamous cell metaplasia of the respiratory epithelium in
                F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years	F-9
                Goodness-of-fit statistics and BMD-io and BMDL-io values from models
                fit to incidence data for squamous cell metaplasia of the respiratory
                epithelium in male F344/DuCrj rats exposed to 1,4-dioxane vapors
                (Kasaietal.,2009)_                                                   F-10
F.4
    Figure F-4   BMD Log-probit model of squamous cell metaplasia of the respiratory
                epithelium incidence data for male rats exposed to 1,4-dioxane
                vapors for 2 years to support the results  in Table F-6.	
Squamous Cell Hyperplasia_
         Table F-7
         Table F-8
F.5
    Figure F-5   BMD Log-probit model of squamous cell hyperplasia of the respiratory
                epithelium incidence data for male rats exposed to 1,4-dioxane
                vapors for 2 years to support the results  in Table F-8.	
Respiratory Metaplasia	
         Table F-9
         Table F-10
         Table F-11
         Figure F-6
F.6  Atrophy_
         Table F-12
         TableF-13
         Figure F-7
F.7
                BMD Gamma model of respiratory metaplasia of olfactory epithelium
                incidence data for male rats exposed to 1,4-dioxane vapors for 2
                years	

                Incidence of respiratory metaplasia of the olfactory epithelium in
                F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years	
                Goodness-of-fit statistics and BMD-io and BMDL-io values from models
                fit to incidence data for atrophy of olfactory epithelium in male
                F344/DuCrj rats (Kasai et al., 2009) exposed to 1,4-dioxane vapors	
                BMD Log-Logistic model of atrophy of olfactory epithelium incidence
                data for male rats exposed to 1,4-dioxane vapors for 2 years to
                support the results in Table F-13.  	
Hydropic Change
    Table F-14  incidence of hydropic change of the lamina propria in the nasal cavity
                of F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years
         TableF-15
         Figure F-8
F.8  Sclerosis
                Goodness-of-fit statistics and BMD-io and BMDL-io values from models
                fit to incidence data for hydropic change of the lamina propria in the
                nasal cavity of male F344/DuCrj rats exposed to 1,4-dioxane vapors
                (Kasai et al., 2009)	
                BMD Log-logistic model of hydropic change of lamina propria (nasal
                cavity) incidence data for male rats exposed to 1,4-dioxane vapors for
                2 years to support the results in Table F-16.	
                                                                                          F-10
                Incidence of squamous cell hyperplasia of the respiratory epithelium
                in F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years	F-12
                Goodness-of-fit statistics and BMDio and BMDLio values from models
                fit to incidence data for squamous cell hyperplasia of the respiratory
                epithelium in male F344/DuCrj rats exposed to 1,4-dioxane vapors
                (Kasai et al., 2009)	F-13
                                                                                          F-13
                Incidence of respiratory metaplasia of the olfactory epithelium in
                F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years	F-15
                Goodness-of-fit statistics and BMDio and BMDLio values from models
                fit to incidence data for respiratory metaplasia of olfactory epithelium
                in male F344/DuCrj rats (Kasai et al., 2009) exposed to 1,4-dioxane
                vapors                                                               F-16
                Goodness-of-fit statistics and BMDio and BMDL-io values from models
                fit to incidence data for respiratory metaplasia of olfactory epithelium
                with high dose group dropped in male F344/DuCrj rats (Kasai et al.,
                2009) exposed to 1,4-dioxane vapors	F-16
 F-17
.F-18

 F-19
                                                                                          F-20
 F-20
.F-22

 F-22
                                                                                          F-23
 F-23
.F-25

 F-25
         Table F-16  Incidence of sclerosis of the lamina propria in the nasal cavity of
                     F344/DuCrj rats exposed to 1,4-dioxane via inhalation for 2 years	
         Table F-17  Goodness-of-fit statistics and BMDio and BMDL-io values from models
                     fit to incidence data for sclerosis of the lamina propria in the nasal
                     cavity of male F344/DuCrj rats exposed to 1,4-dioxane vapors (Kasai
                     et al., 2009)	F-26
         Figure F-9   BMD Log-logistic model of sclerosis of lamina propria (nasal cavity)
                     incidence data for male rats exposed to 1,4-dioxane vapors for 2
                     years to support the results in Table F-18.	F-28
                                                                                 Xll
                  DRAFT - DO NOT CITE OR QUOTE

-------
APPENDIXG.       RFC DERIVATION: ALTERNATIVE APPROACH IN THE APPLICATION OF
     THE  DOSIMETRIC ADJUSTMENT FACTOR  	G-1

     G.1  Application of DAF for Category 1 Gases	G-2
     G.2  Application of Uncertainty Factors	G-2

APPENDIX H.       DETAILS OF  BMD ANALYSIS FOR INHALATION UNIT RISK FOR
     1,4-DIOXANE  H-1

     H.1  General Issues and Approaches to BMDS and Multitumor Modeling	H-1
        H.1.1  Combining Data tumor types	H-1
        H.1.2  Summary	H-1
              Table H-1   Summary of BMC-io and BMCL-io model results for individual tumor
                         types  and combined tumor analysis for male rats exposed to
                         1,4-dioxane vapors (Kasai etal., 2009)	H-2
     H.2  BMDS Model Output for Multistage Cancer Models for Inidividual Tumor Types	H-2
              Table H-2   Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane
                         vapor by whole-body inhalation for 2 years	H-3
        H.2.1  Nasal Squamous Cell Carcinoma	H-3
              Table H-3   BMDS Multistage cancer dose-response modeling results for the
                         incidence of nasal squamous cell carcinomas in male rats exposed to
                         1,4-dioxane vapors for2-years (Kasai etal., 2009)	H-4
              Figure H-1   Multistage model  (First (1°)-degree) for male rat nasal squamous cell
                         carcinomas.	H-4
        H.2.2  Hepatocellular Adenoma and Carcinoma	H-6
              Table H-4   BMDS Multistage cancer dose-response modeling results for the
                         incidence of either hepatocellular adenoma or carcinoma in male rats
                         exposed to 1,4-dioxane vapors for2-years (Kasai et al., 2009)	H-7
              Figure H-2  Multistage model  (First-degree (1°)) for male rat hepatocellular
                         adenomas and carcinomas.	H-7
        H.2.3  Renal Cell Carcinoma and Zymbal Gland Adenoma	H-9
              Table H-5   BMDS Multistage cancer dose-response modeling results for the
                         incidence of renal cell carcinomas and Zymbal gland adenomas in
                         male rats exposed to 1,4-dioxane vapors for2-years (Kasai et al.,
                         2009)	H-10
              Figure H-3  Multistage model  (Second-degree (2°)) for male rat renal cell
                         carcinomas and Zymbal gland  adenomas.	H-10
              Figure H-4  Multistage model  (Third-degree (3°)) for male rat renal cell
                         carcinomas.	H-12
        H.2.4  Peritoneal Mesothelioma	H-14
              Table H-6   BMDS Multistage cancer dose-response modeling results for the
                         incidence of peritoneal mesothelioma in male rats exposed to
                         1,4-dioxane vapors for 2-years (Kasai et al., 2009)	H-15
              Figure H-5  Multistage model  (First-degree (1°)) for male rat peritoneal
                         mesotheliomas.	H-15
        H.2.5  Mammary Gland  Fibroadenoma	H-17
              Table H-7   BMDS Multistage cancer dose-response modeling results for the
                         incidence of mammary gland fibroadenoma in male rats exposed to
                         1,4-dioxane vapors for 2-years (Kasai etal., 2009)	H-18
              Figure H-6  Multistage model  (First-degree (1°)) for male rat mammary gland
                         fibroadenoma.	H-18
        H.2.6  Subcutis Fibroma 	H-20
              Table H-8   BMDS Multistage cancer dose-response modeling results for the
                         incidence of subcutis fibromas in male rats exposed to 1,4-dioxane
                         vapors for 2-years (Kasai et al., 2009)	H-21
              Figure H-7  Multistage model  (First-degree (1°)) for male rat subcutis fibroma
                         (high dose dropped).	H-21
        H.2.7  Multitumor analysis using Bayesian Methods	H-23
     H.3  Multitumor Analysis Using BMDS MSCOMBO (BETA)  	H-24
                                                                                  xm
                       DRAFT - DO NOT CITE OR QUOTE

-------
LIST  OF  ABBREVIATIONS  AND  ACRONYMS
AIC            Akaike's Information Criterion
ALP            alkaline phosphatase
ALT            alanine aminotransferase
AST            aspartate aminotransferase
ATSDR         Agency for Toxic Substances and Disease Registry
BMC	benchmark concentration
BMCL	benchmark concentration, lower 95% confidence limit
BMCLio	benchmark concentration, lower 95% confidence limit at 10% extra risk
BMD           benchmark dose
BMD-io          benchmark dose at 10% extra risk
BMDso          benchmark dose at 30% extra risk
BMDso          benchmark dose at 50% extra risk
BMDL          benchmark dose, lower 95% confidence limit
BMDL-io        benchmark dose, lower 95% confidence limit at 10% extra risk
BMDLso        benchmark dose, lower 95% confidence limit at 30% extra risk
BMDLso        benchmark dose, lower 95% confidence limit at 50% extra risk
BMDS          Benchmark Dose Software
BMR           benchmark response
BrdU           5-bromo-2'-deoxyuridine
BUN           blood urea nitrogen
BW(s)          body weight(s)
CASE          computer automated structure evaluator
CASRN        Chemical Abstracts Service Registry Number
CHO           Chinese hamster ovary (cells)
Cl             confidence interval(s)
CNS           central nervous system
CPK           creatinine phosphokinase
CREST        antikinetochore
CSF           cancer slope factor
CV            concentration in venous blood
CYP450        cytochrome P450
DEN           diethylnitrosamine
FISH           fluorescence in situ hybridization
G-6-Pase       glucose-6-phosphatase
GC            gas chromatography
GGT           Y-g|utamyl transpeptidase
GST-P          qlutathione S-transferase. placental form
HEAA          (3-hydroxyethoxy acetic acid
HED(s)         human equivalent dose(s)
HPLC          high-performance liquid chromatography
HSDB          Hazardous Substances Data Bank
Hz             Hertz
IARC           International  Agency for Research on Cancer
i.p.             intraperitoneal
i.v.             intravenous
IRIS           Integrated Risk Information System
JBRC          Japan Bioassay Research Center
ke             1 st order elimination rate of 1,4-dioxane
kiNH            1 st order 1,4-dioxane inhalation rate constant
ki_c            1st order, non-saturable metabolism rate constant for 1,4-dioxane in the liver
Km             Michaelis constant for metabolism of 1,4-dioxane in the liver
kme            1st order elimination rate of HEAA (1,4-dioxane metabolite)
koc            soil organic carbon-water portioning coefficient
LAP           leucine aminopeptidase
LDso           median lethal dose
LDH           lactate dehydrogenase
LOAEL         lowest-observed-adverse-effect-level
MCH	mean corpuscular hemoglobin
MCV
MOA
MS
mean corpuscular volume
mode of action
mass spectrometry, multi-stage
                                                                                       xiv
                            DRAFT - DO NOT CITE OR QUOTE

-------
MTD            maximum tolerated dose
MVK            Moolgavkar-Venzon-Knudsen (model)
NCE            normochromatic erythrocyte
NCI             National Cancer Institute
ND             no data, not detected
NE             not estimated
NOAEL         no-observed-adverse-effect-level
NRC            National Research Council
NTP            National Toxicology Program
OCT            ornithine carbamyl transferase
ODC            ornithine decarboxylase
OECD          Organization for Economic Co-operation and Development
PB             blood:air partition coefficient
PBPK           physiologically based pharmacokinetic
PC             partition coefficient
PCS            polychlorinated biphenyl
PCE            polychromatic erythrocyte
PFA            fat:air partition coefficient
PLA            liverair partition coefficient
POD            point of departure
ppm            parts per million
PRA            rapidly perfused tissue:air partition coefficient
PSA            slowly perfused tissue:air partition coefficient
QCC            normalized cardiac output
QPC            normalized alveolar ventilation rate
RBC            red blood cell
RfC             inhalation reference concentration
RfD             oral reference dose
SCE            sister chromatid exchange
SDH            sorbitol dehydrogenase
SMR            standardized mortality ratio
SRC            Syracuse Research Corporation
TPA            12-O-tetradecanoylphorbol-13-acetate
TWA            time-weighted average
UF             uncertainty factor
UNEP           United Nations  Environment Programme
U.S.            United States of America
U.S. EPA        U.S.  Environmental Protection Agency
V               volts
VAS            visual analogue scale
Vd              volume of distribution
Vmax            maximal rate of metabolism
Vmaxc            normalized maximal rate of metabolism of 1,4-dioxane in liver
VOC(s)         volatile organic compound(s)
WBC            white blood cell
X2              Chi-squared
                                                                                                    xv
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      FOR  EWOR  D

 1           The purpose of this Toxicological Review is to provide scientific support and rationale for the
 2    hazard and dose-response assessment in IRIS pertaining to chronic exposure to 1,4-dioxane. It is not
 3    intended to be a comprehensive treatise on the chemical or toxicological nature of 1,4-dioxane.

 4           The intent of Section 6, Major Conclusions in the Characterization of Hazard and Dose
 5    Response, is to present the major conclusions reached in the derivation of the reference dose, reference
 6    concentration and cancer assessment, where applicable, and to characterize the overall confidence in the
 7    quantitative and qualitative aspects of hazard and dose response by addressing the quality of data and
 8    related uncertainties. The discussion is intended to convey the limitations of the  assessment and to aid
 9    and guide the risk assessor in the ensuing steps of the risk assessment process.

10           For other general information about this assessment or other questions relating to IRIS, the reader
11    is referred to EPA's IRIS Hotline at (202) 566-1676 (phone), (202) 566-1749 (fax), or
12    hotline.iris@epa.gov (email address).

13    NOTE: New studies (Kasai et al., 2009;  Kasai et al., 2008) regarding the toxicity of 1.4-dioxane through
14    the inhalation route of exposure are available that were not included in the 1.4-dioxane assessment that
15    was posted on the IRIS database in 2010 (U.S. EPA. 2010).

16           These studies have been incorporated into the previously posted assessment (U.S. EPA. 2010) for
17    review. Sections including new information can be identified by the red underlined text in the document.
18    The entire document is provided for completeness.
                                     DRAFT-DO NOT CITE OR QUOTE                                      xvi

-------
AUTHORS,  CONTRIBUTORS,  AND  REVIEWERS
CHEMICAL  MANAGERS/AUTHORS

                  Patricia Gillespie. Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Eva D. McLanahan, Ph.D.
                    Lieutenant Commander, U.S. Public Health Service
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  ReederSams II, Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC
CO-AUTHORS  AND  CONTRIBUTORS

                  J. Allen Davis, MSPH
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Hisham EI-Masri, Ph.D.
                    National Health and Environmental Effects Research Laboratory
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  JeffS. Gift, Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Karen Hogan
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Washington, DC

                  Leonid Kopvlev. Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Washington, DC

                  William Lefew. Ph.D.
                    National Health and Environmental Effects Research Laboratory
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Fernando Llados
                    Environmental  Science Center
                    Syracuse Research Corporation
                    Syracuse, NY

                  Michael Lumpkin, Ph.D.
                    Environmental  Science Center
                    Syracuse Research Corporation
                    Syracuse, NY
                                DRAFT-DO NOT CITE OR QUOTE                                      xvii

-------
                  Allan Marcus, Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Marc Odin, Ph.D.
                    Environmental Science Center
                    Syracuse Research Corporation
                    Syracuse, NY

                  Susan Rieth
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Washington, DC

                  Andrew Rooney, Ph.D.*
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC
               'Currently at National Toxicology Program , National Institute of Environmental Health Sciences; Research
                  Triangle Park, NC

                  Paul Schlosser, Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  John Stanek. Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Julie Stickney, Ph.D.
                    Environmental Science Center
                    Syracuse Research Corporation
                    Syracuse, NY

                  John Vandenberg, Ph.D.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC

                  Debra Walsh, M.S.
                    National Center for Environmental Assessment
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC
TECHNICAL  SUPPORT

                   Ellen Lorang, M.S.
                    National Center for Environmental Assessment
                    Research Triangle Park, NC
                   J. Sawyer Lucy, B.A.
                    Student Services Contractor
                    National Center for Environmental Assessment
                    Research Triangle Park, NC
                   Deborah Wales
                    National Center for Environmental Assessment
                    Research Triangle Park, NC
                                 DRAFT-DO NOT CITE OR QUOTE                                       xvii
                                                                                                           i

-------
         R EVIE WE RS



1           This document has been provided for review to EPA scientists, interagency reviewers from other

2    federal agencies and White House offices. The comments and responses in Appendix A were in regards to

3    the oral assessment previously reviewed. A summary of external peer review and public comments and

4    disposition following review of the inhalation assessment for 1.4-dioxane will be included when they

5    become available.



     INTERNAL   EPA  REVIEWERS  (ORAL  AS S E S S ME NT)

                      Anthony DeAngelo, Ph.D.
                        National Health and Environmental Effects Research Laboratory
                        Office of Research and Development

                      Nagu Keshava, Ph.D.
                        National Center for Environmental Assessment
                        Office of Research and Development

                      Jason Lambert, Ph.D.
                        National Center for Environmental Assessment
                        Office of Research and Development

                      Connie Meacham, M.S.
                        National Center for Environmental Assessment
                        Research Triangle Park, NC

                      Douglas Wolf, Ph.D.
                        National Health and Environmental Effects Research Laboratory
                        Office of Research and Development
                                    DRAFT-DO NOT CITE OR QUOTE                                     xix

-------
     EXTERNAL   PEER   REVIEWERS
(ORAL  AS  S ES S ME NT)
                  George V. Alexeeff, Ph.D., DABT
                    Office of Environmental Health Hazard Assessment (OEHHA)
                    California  EPA

                  Bruce C. Allen, M.S.
                    Bruce Allen Consulting

                  James V. Bruckner, Ph.D.
                    Department of Pharmaceutical and Biomedical Sciences
                    College of Pharmacy
                    The University of Georgia

                  Harvey J. Clewell III, Ph.D., DABT
                    Center for Human Health Assessment
                    The Hamner Institutes for Health Sciences

                  Lena Ernstgard, Ph.D.
                    Institute of Environmental Medicine
                    Karolinska Institutet

                  Frederick J. Kaskel, M.D., Ph.D.
                    Children's Hospital at Montefiore
                    Albert Einstein College of Medicine of Yeshiva University

                  Kannan Krishnan, Ph.D., DABT
                    Inter-University Toxicology Research Center (CIRTOX)
                    Universite de Montreal

                  Raghubir P. Sharma, DVM, Ph.D.
                    Department of Physiology and Pharmacology
                    College of Veterinary Medicine (retired)
                    The University of Georgia
                                 DRAFT-DO NOT CITE OR QUOTE                                       xx

-------
      1.1   INTRODUCTION

 1           This document presents background information and justification for the Integrated Risk
 2    Information System (IRIS) Summary of the hazard and dose-response assessment of 1,4-Dioxane.
 3    IRIS Summaries may include oral reference dose (RfD) and inhalation reference concentration (RfC)
 4    values for chronic and other exposure durations, and a carcinogenicity assessment.

 5           The RfD and RfC, if derived, provide quantitative information for use in risk assessments for
 6    health effects known or assumed to be produced through a nonlinear (presumed threshold) mode of
 7    action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with uncertainty spanning
 8    perhaps an order of magnitude) of a daily exposure to the human population (including sensitive
 9    subgroups) that is likely to be without an appreciable risk of deleterious effects during a lifetime.  The
10    inhalation RfC (expressed in units of mg/m3) is analogous to the oral RfD, but provides a continuous
11    inhalation exposure estimate.  The inhalation RfC considers toxic effects for both the respiratory system
12    (portal-of-entry) and for effects peripheral to the respiratory system (extrarespiratory or systemic effects).
13    Reference values are generally derived for chronic exposures (up to a lifetime), but may also be derived
14    for acute (< 24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
15    lifetime) exposure durations, all of which are derived based on an assumption of continuous exposure
16    throughout the duration specified. Unless specified otherwise, the RfD and RfC are derived for chronic
17    exposure duration.

18           The carcinogenicity assessment provides information on the carcinogenic hazard potential of the
19    substance in question and quantitative estimates of risk from oral and inhalation exposure may be derived.
20    The information includes a weight-of-evidence judgment of the likelihood that the agent is a human
21    carcinogen and the conditions under which the carcinogenic effects may be expressed. Quantitative risk
22    estimates may be derived from the application of a low-dose extrapolation procedure.  If derived, the oral
23    slope factor is a plausible upper bound on the estimate of risk per mg/kg-day of oral exposure. Similarly,
24    an inhalation unit risk is a plausible upper bound on the estimate of risk per ug/m3 air breathed.

25           Development of these hazard identification and dose-response assessments for 1,4-dioxane has
26    followed the general  guidelines for  risk assessment as set forth by the National Research Council (NRC,
27    1983). U.S. Environmental Protection Agency (U.S. EPA) Guidelines and Risk Assessment Forum
28    technical panel reports that may have been used in the  development of this assessment include the
29    following Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S. EPA. 1986b).
30    Guidelines for Mutagenicity Risk Assessment (U.S. EPA. 1986a). Recommendations for and
31    Documentation of Biological Values for Use in Risk Assessment (U.S. EPA. 1988), Guidelines for
32    Developmental Toxicity Risk Assessment (U.S. EPA. 1991). Interim Policy for Particle Size and Limit
33    Concentration Issues in Inhalation Toxicity (U.S. EPA. 1994a). Methods for Derivation of Inhalation
34    Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA. 1994b). Use of the
35    Benchmark Dose Approach in Health Risk Assessment (U.S. EPA. 1995). Guidelines for Reproductive
36    Toxicity Risk Assessment (U.S. EPA. 1996). Guidelines for Neurotoxicity Risk Assessment (U.S. EPA.
                                                                                                   1
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    1998). Science Policy Council Handbook: Risk Characterization (U.S. EPA. 2000a). Benchmark Dose
 1    Technical Guidance Document (External Review Draft) (U.S. EPA. 2000c). Supplementary Guidance for
 3    Conducting Health Risk Assessment of Chemical Mixtures (U.S. EPA. 2000b). A Review of the Reference
 4    Dose and Reference Concentration Processes (U.S. EPA. 2002b). Guidelines for Carcinogen Risk
 5    Assessment (U.S. EPA, 2005b), Supplemental Guidance for Assessing Susceptibility from Early-Life
 6    Exposure to Carcinogens (U.S. EPA. 2005a). Science Policy Council Handbook: Peer Review (U.S. EPA.
 7    2006b). and A Framework for Assessing Health Risks of Environmental Exposures to Children (U.S.
 8    EPA. 2006a).

 9           In 2010. an updated health assessment for oral exposures to 1.4-dioxane was released (U.S. EPA.
10    2010). During the development of the 2010 health assessment, new studies (Kasai et al.. 2009; Kasai et
11    al.. 2008) regarding the toxicity of 1.4-dioxane through the inhalation route of exposure became available
12    that were not included in the 1.4-dioxane assessment that was posted on the IRIS database in 2010.  These
13    new inhalation studies have been incorporated into the previously posted assessment for this review.
14    Sections including new information can be identified in this draft assessment by underlined red text.
15    Tables containing new information can be identified by red text, but for improved legibility the new
16    information presented in the tables has not been underlined.  The entire document is provided for
17    completeness.

18           The literature search strategy employed for 1,4-dioxane was based on the chemical name,
19    Chemical Abstracts Service Registry Number (CASRN), and multiple common synonyms. Any pertinent
20    scientific information submitted by the public to the IRIS Submission Desk was also considered in the
21    development of this document. Primary, peer-reviewed-literature was reviewed through September 2009
22    for the oral assessment and through July 2011 for the inhalation assessment and was included where that
23    literature was determined to be critical to the assessment. The relevant literature included publications on
24    1,4-dioxane which were identified through Toxicology Literature Online  (TOXLINE), PubMed, the Toxic
25    Substance Control Act Test Submission Database (TSCATS), the Registry of Toxic Effects of Chemical
26    Substances (RTECS), the Chemical Carcinogenesis Research Information System (CCRIS), the
27    Developmental and Reproductive Toxicology/Environmental Teratology Information Center
28    (DART/ETIC), the Environmental Mutagens Information Center (EMIC) and Environmental Mutagen
29    Information Center Backfile (EMICBACK) databases, the Hazardous Substances Data Bank (HSDB), the
30    Genetic Toxicology Data Bank (GENE-TOX), Chemical abstracts, and Current Contents.  Other peer-
31    reviewed information, including health assessments developed by other organizations, review articles, and
32    independent analyses of the health effects data were retrieved and may be included in the assessment
33    where appropriate.
34
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     2   CHEMICAL  AND  PHYSICAL  INFORMATION


1           1,4-Dioxane, a volatile organic compound (VOC), is a colorless liquid with a pleasant odor
2    (Hawley and Lewis. 2001; Lewis. 2000). Synonyms include diethylene ether, 1,4-diethylene dioxide,
3    diethylene oxide, dioxyethylene ether, and dioxane (Hawley and Lewis. 2001). The chemical structure of
4    1,4-dioxane is shown in Figure 2-1. Selected chemical and physical properties of this substance are in
5    Table 2-1:
                                                   O
                                                   O

            Figure 2-1  1,4-Dioxane chemical structure.

     Table 2-1    Physical properties and chemical identity of 1,4-dioxane


     CASRN:                            123-91-1 (CRC Handbook (Lide, 2000))
     Molecular weight:                    88.10 (Merck Index (2001))
     Chemical formula:                    C4H8O2 (Merck Index (2001))
     Boiling point:                        101.1°C (Merck Index (2001))
     Melting point:                        11.8°C (CRC Handbook (Lide. 2000))
     Vapor pressure:                     40 mmHg at 25°C (Lewis, 2000)
     Density:                            1.0337 q/mL at 20°C (CRC Handbook (Lide, 2000))
     Vapor density:                       3.03 (air= 1) (Lewis. 2000)
     Water solubility:                     Miscible with water (Hawlev and Lewis. 2001)
     Other solubilities:                    Miscible with ethanol, ether, acetone (CRC Handbook (Lide. 2000))
     Log Kow:                            -0.27 (Hansch et al., 1995)
     Henry's Law constant:                4.80 x 10'6 atm-m3/molecule at25°C (Parket al.. 1987)
     OH reaction rate constant:             1.09 x 10"11 cm3/molecule sec at 25°C (Atkinson. 1989)
     Koc:                               17 (estimated using log Kow) (ACS Handbook (Lvman et al.. 1990))
     Bioconcentration factor:               0.4 (estimated using log Kow) (Meylan et al., 1999)
                                       1 ppm = 3.6 mg/m ; 1 mg/m  = 0.278 ppm
     Conversion factors (in air):
                     v     '             (25 C and 1 atm) (HSDB, 2007)
6           1,4-Dioxane is produced commercially through the dehydration and ring closure of diethylene
7    glycol (Surprenant. 2002). Concentrated sulfuric acid is used as a catalyst (Surprenant. 2002). This is a
8    continuous distillation process with operating temperatures and pressures of 130-200°C and  188-
9    825 mmHg, respectively (Surprenant 2002). During the years 1986 and 1990, the U.S. production of


                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    1,4-dioxane reported by manufacturers was within the range of 10-50 million pounds (U.S. EPA. 2002b).
 2    The production volume reported during the years  1994, 1998, and 2002 was within the range of 1-
 3    10 million pounds (U.S. EPA. 2002b).

 4           Historically, 1,4-dioxane has been used as a stabilizer for the solvent 1,1,1-trichloro-ethane
 5    (Surprenant 2002). However, this use is no longer expected to be important due to the 1990 Amendments
 6    to the Clean Air Act and the Montreal Protocol, which mandate the eventual phase-out of
 7    1,1,1 -trichloroethane production in the U.S. (ATSDR. 2007; U.N. Environment Programme. 2000;
 8    "Amendments to the Clean Air Act.  Sec.  604. Phase-out of production and consumption of class I
 9    substances." 1990). 1,4-Dioxane is a contaminant of some ingredients used in the manufacture of personal
10    care products and cosmetics. 1,4-Dioxane is also used as a solvent for cellulosics, organic products,
11    lacquers, paints, varnishes, paint and varnish removers, resins, oils, waxes, dyes, cements, fumigants,
12    emulsions, and polishing compositions (Hawley and Lewis, 2001; Merck Index. 2001; IARC. 1999).
13    1,4-Dioxane has been used as a solvent in the formulation of inks, coatings, and adhesives and in the
14    extraction of animal and vegetable oil (Surprenant 2002). Reaction products of 1,4-dioxane are  used in
15    the manufacture of insecticides, herbicides, plasticizers, and monomers (Surprenant. 2002).

16           When 1,4-dioxane enters the air,  it will exist as a vapor, as indicated by its vapor pressure
17    (HSDB. 2007). It is expected to be degraded in the atmosphere through photooxidation with hydroxyl
18    radicals (HSDB. 2007; Surprenant 2002). The estimated half-life for this reaction is 6.7 hours (HSDB.
19    2007). It may also be broken down by reaction with  nitrate radicals, although this removal process is not
20    expected to compete with hydroxyl radical photooxidation (Grosjean. 1990). 1,4-Dioxane is not expected
21    to undergo direct photolysis (Wolfe and Jeffers. 2000).  1,4-Dioxane is primarily photooxidized to
22    2-oxodioxane and through reactions  with nitrogen oxides (NOX) results in the formation of ethylene
23    glycol diformate (Platz et al.. 1997).  1,4-Dioxane  is  expected to be highly mobile in soil based on its
24    estimated Koc and is expected to leach to lower soil horizons and groundwater (ATSDR, 2007; ACS
25    Handbook (LymanetaL 1990). This substance may volatilize from dry soil surfaces based on its vapor
26    pressure (HSDB. 2007). The estimated bioconcentration factor value indicates that 1,4-dioxane will not
27    bioconcentrate in aquatic or marine organisms (Tvleylan et al.. 1999; Franke etal.. 1994). 1,4-Dioxane is
28    not expected to undergo hydrolysis or to biodegrade readily in the environment (ATSDR, 2007; HSDB,
29    2007). Therefore, volatilization is expected to be the dominant removal process for moist soil and surface
30    water. Based on a Henry's  Law constant of 4.8x 10"6  atm-m3/mole, the half-life for volatilization of
31    1,4-dioxane from a model  river is 5 days and that  from a model lake is 56 days (HSDB. 2007; Lyman et
32    al.. 1990; Park etal.. 1987). 1,4-Dioxane may be more persistent in groundwater where volatilization is
33    hindered.

34           Recent environmental monitoring data for 1,4-dioxane are lacking. Existing data indicate that
35    1,4-dioxane may leach from hazardous waste sites into drinking water sources located nearby (Yasuhara
36    etal.. 2003; Yasuhara et al.. 1997; Lesage et al.. 1990). 1,4-Dioxane has been detected in contaminated
37    surface and groundwater samples collected near hazardous waste sites and industrial facilities (Derosa et
38    al.. 1996).
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      3   TOXICOKINETICS

 1           Data for the toxicokinetics of 1,4-dioxane in humans are very limited. However, absorption,
 2    distribution, metabolism, and elimination of 1,4-dioxane are well described in rats exposed via the oral,
 3    inhalation, or intravenous (i.v.) routes. 1,4-Dioxane is extensively absorbed and metabolized in humans
 4    and rats. The metabolite most often measured and reported is (3-hydroxyethoxy acetic acid (HEAA),
 5    which is predominantly excreted in the urine; however, other metabolites have also been identified.
 6    Saturation of 1,4-dioxane metabolism has been observed in rats and would be expected in humans;
 7    however, human exposure levels associated with nonlinear toxicokinetics are not known.
 8           Important data elements that have contributed to our current understanding of the toxicokinetics
 9    of 1,4-dioxane are summarized in the following sections.
      3.1  Absorption

10           Absorption of 1,4-dioxane following inhalation exposure has been qualitatively demonstrated in
11    workers and volunteers. Workers exposed to a time-weighted average (TWA) of 1.6 parts per
12    million (ppm) of 1,4-dioxane in air for 7.5 hours showed a HEAA/1,4-dioxane ratio of 118:1 in urine
13    (Young et al. 1976). The authors assumed lung absorption to be 100% and calculated an average
14    absorbed dose of 0.37 mg/kg, although no exhaled breath measurements were taken. In a study with four
15    healthy male volunteers, Young et al. (1977) reported 6-hour inhalation exposures of adult volunteers to
16    50 ppm of 1,4-dioxane in a chamber, followed by blood and urine analysis for 1,4-dioxane and HEAA.
17    The study protocol was approved by a seven-member Human Research Review Committee of the Dow
18    Chemical Company, and written informed consent of study participants was obtained. At a concentration
19    of 50 ppm, uptake of 1,4-dioxane into plasma was rapid and approached steady-state conditions by
20    6 hours. The authors reported a calculated absorbed dose of 5.4 mg/kg. However, the exposure chamber
21    atmosphere was kept at a constant concentration of 50 ppm and exhaled breath was not analyzed.
22    Accordingly, gas uptake  could not be measured. As a result, the absorbed fraction of inhaled 1,4-dioxane
23    could not be accurately determined in humans. Rats inhaling 50 ppm for 6 hours exhibited 1,4-dioxane
24    and HEAA in urine with an HEAA to 1,4-dioxane ratio of over 3,100:1 (Young etal., 1978a; 1978b).
25    Plasma concentrations at the end of the 6-hour exposure period averaged 7.3  ug/mL. The authors
26    calculated an absorbed 1,4-dioxane dose of 71.9 mg/kg; however, the lack of exhaled breath data and
27    dynamic exposure chamber precluded the accurate determination of the absorbed fraction of inhaled
28    1,4-dioxane.

29           No human data are available to evaluate the oral absorption of 1,4-dioxane. Gastrointestinal
30    absorption was  nearly complete in  male Sprague Dawley rats orally dosed with 10-1,000 mg/kg of
31    [14C]-l,4-dioxane given as a single dose or as 17 consecutive daily doses (Young et al., 1978a; 1978b).
32    Cumulative recovery of radiolabel  in the feces was 
-------
 1           No human data are available to evaluate the dermal absorption of 1,4-dioxane; however,
 2    Bronaugh (1982) reported an in vitro study in which 1,4-dioxane penetrated excised human skin 10 times
 3    more under occluded conditions (3.2% of applied dose) than unoccluded conditions (0.3% of applied
 4    dose). [14C]-l,4-Dioxane was dissolved in lotion, applied to the excised skin in occluded and unoccluded
 5    diffusion cells, and absorption of the dose was recorded 205 minutes after application. Bronaugh (1982)
 6    also reported observing rapid evaporation, which further decreased the small amount available for skin
 7    absorption.

 8           Dermal absorption data in animals are also limited. Dermal absorption in animals was reported to
 9    be low following exposure of forearm skin of monkeys (Marzulli et al.. 1981). In this study, Rhesus
10    monkeys were exposed to [14C]-l,4-dioxane in methanol or skin lotion vehicle for 24 hours (skin was
11    uncovered/unoccluded). Only 2-3% of the original radiolabel was cumulatively recovered in urine over a
12    5-day period.
      3.2   Distribution

13           No data are available for the distribution of 1,4-dioxane in human tissues. No data are available
14    for the distribution of 1,4-dioxane in animals following oral or inhalation exposures.

15           Mikheev et al. (1990) studied the distribution of [14C]-l,4-dioxane in the blood, liver, kidney,
16    brain, and testes of rats (strain not reported) for up to 6 hours following intraperitoneal (i.p.) injection of
17    approximately one-tenth the median lethal dose (LD50) (actual dose not reported). While actual tissue
18    concentrations were not reported, tissue:blood ratios were given for each tissue at six time points ranging
19    from 5 minutes to 6 hours. The time to reach maximum accumulation of radiolabel was shorter for liver
20    and kidney than for blood or the other tissues, which the authors suggested was indicative of selective
21    membrane transport. Tissue:blood ratios were less than one for all tissues except testes, which had a ratio
22    greater than one at the 6-hour time point. The significance of these findings is questionable since the
23    contribution of residual blood in the tissues was unknown (though saline perfusion may serve to clear
24    tissues of highly water-soluble 1,4-dioxane), the tissue concentrations of radiolabel were not reported, and
25    data were collected from so few time points.

26           Woo et al. (1977a) administered i.p. doses of [3H]-1,4-dioxane (5 mCi/kg body weight [BW]) to
27    male Sprague Dawley rats with and without pretreatment using mixed-function oxidase inducers
28    (phenobarbital, 3-methylcholanthrene, or polychlorinated biphenyls [PCBs]). Liver, kidney, spleen, lung,
29    colon, and skeletal muscle tissues were collected from 1,2,6, and 12 hours after dosing. Distribution was
30    generally uniform across tissues, with blood concentrations higher than tissues at all times except for
31    1 hour post dosing, when kidney levels were approximately 20% higher than blood. Since tissues were
32    not perfused prior to analysis, the contribution of residual blood to radiolabel measurements is unknown,
33    though loss of 1,4-dioxane from tissues would be unknown had saline perfusion been performed.
34    Covalent binding reached peak percentages at 6 hours after dosing in liver (18.5%), spleen (22.6%), and
35    colon (19.5%). At 16 hours after dosing, peak covalent binding percentages were observed in whole blood
36    (3.1%), kidney (9.5%), lung (11.2%), and skeletal muscle  (11.2%). Within hepatocytes, radiolabel

                                                                                                     6
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    distribution at 6 hours after dosing was greatest in the cytosolic fraction (43.8%) followed by the
 2    microsomal (27.9%), mitochondrial (16.6%), and nuclear (11.7%) fractions. While little covalent binding
 3    of radiolabel was measured in the hepatic cytosol (4.6%), greater binding was observed at 16 hours after
 4    dosing in the nuclear (64.8%), mitochondrial (45.7%), and microsomal (33.4%) fractions. Pretreatment
 5    with inducers of mixed-function oxidase activity did not significantly change the extent of covalent
 6    binding in subcellular fractions.
      3.3   Metabolism

 7           The major product of 1,4-dioxane metabolism appears to be HEAA, although there is one report
 8    that identified l,4-dioxane-2-one as a major metabolite (Woo et al.. 1977a). However, the presence of this
 9    compound in the sample was believed to result from the acidic conditions (pH of 4.0-4.5) of the
10    analytical procedures. The reversible conversion of HEAA and p-l,4-dioxane-2-one is pH-dependent
11    (Braun and Young. 1977). Braun and Young (1977) identified HEAA (85%) as the major metabolite,
12    with most of the remaining dose excreted as unchanged 1,4-dioxane in the urine of Sprague Dawley rats
13    dosed with 1,000 mg/kg of uniformly labeled l,4-[14C]dioxane. In fact, toxicokinetic studies of
14    1,4-dioxane in humans and rats (Young et al. (1978b: 1978a:  1977)) employed an analytical technique
15    that converted HEAA to the more volatile l,4-dioxane-2-one prior to gas chromatography (GC); however,
16    it is still unclear as to whether HEAA or l,4-dioxane-2-one is the major metabolite of 1,4-dioxane.

17           A proposed metabolic scheme for 1,4-dioxane metabolism (Woo et al.. 1977a) in
18    Sprague Dawley rats is shown in Figure 3-1. Oxidation of 1,4-dioxane to diethylene glycol (pathway a),
19    l,4-dioxane-2-ol (pathway c), or directly to l,4-dioxane-2-one (pathway b) could result in the production
20    of HEAA.  1,4-Dioxane oxidation appears to be cytochrome P450 (CYP450)-mediated, as CYP450
21    induction with phenobarbital or Aroclor 1254 (a commercial PCB mixture) and suppression with
22    2,4-dichloro-6-phenylphenoxy ethylamine or cobaltous chloride were effective in significantly increasing
23    and decreasing, respectively, the appearance of HEAA in the urine of male Sprague Dawley rats
24    following 3 g/kg i.p. dose (Woo et al.. 1978. 1977c). 1,4-Dioxane itself induced CYP450-mediated
25    metabolism of several barbiturates in Hindustan mice given i.p. injections of 25 and 50 mg/kg
26    1,4-dioxane (Mungikar and Pawar. 1978). Of the three possible pathways proposed in this scheme,
27    oxidation to diethylene glycol and  HEAA appears to be the most likely, because diethylene glycol was
28    found as a minor metabolite in Sprague Dawley rat urine following a single 1,000 mg/kg gavage dose of
29    1,4-dioxane (Braun and Young,  1977). Additionally, i.p. injection of 100-400 mg/kg diethylene glycol in
30    Sprague Dawley rats resulted in urinary elimination of HEAA (Woo etal.. 1977b).
                                     DRAFT - DO NOT CITE OR QUOTE

-------
                                        C>     OH         ^OH   0
                       °-         (a)      HOH2C     CH2OH     HOH2C     COOH
                       O
                       [I]      ""---....,
                                     (b)"
                                                                      + hi,O
                                                                              -hLO
                                                                           ~0
                                                                           [IV]
             Source: Adapted with permission of Elsevier Ltd., Woo et al. (1977a: 1977c).
             Figure 3-1   Suggested metabolic pathways of 1,4-dioxane in the rat.
             Legend: I = 1,4-dioxane; II = diethylene glycol; III = P-hydroxyethoxy acetic acid (HEAA);
             IV = l,4-dioxane-2-one; V = l,4-dioxane-2-ol; VI = P-hydroxyethoxy acetaldehyde. Note:
             Metabolite [V] is a likely intermediate in pathway b as well as pathway c. The proposed
             pathways are based on the metabolites identified; the enzymes responsible for each reaction
             have not been determined. The proposed pathways do not account for metabolite
             degradation to the labeled carbon dioxide (CO2) identified in expired air after labeled
             1,4-dioxane exposure.

 1           Metabolism of 1,4-dioxane in humans is extensive. In a survey of 1,4-dioxane plant workers
 2   exposed to a TWA of 1.6 ppm of 1,4-dioxane for 7.5 hours, Young et al. (1976) found HEAA and
 3   1,4-dioxane in the worker's urine at a ratio of 118:1. Similarly, in adult male volunteers exposed to
 4   50 ppm for 6 hours (Young et al.. 1977). over 99% of inhaled 1,4-dioxane (assuming negligible exhaled
 5   excretion) appeared in the urine as HEAA. The linear elimination of 1,4-dioxane in both plasma and urine
 6   indicated that 1,4-dioxane metabolism was a nonsaturated, first-order process at this exposure level.

 7           Like humans, rats extensively metabolize inhaled  1,4-dioxane, as HEAA content in urine was
 8   over 3,000-fold higher than that of 1,4-dioxane following exposure to 50 ppm for 6 hours (Young et al..
 9   1978a: 1978b). 1,4-Dioxane metabolism in rats was a saturable process, as exhibited by oral and i.v.
10   exposures to various doses of [14C]-l,4-dioxane (Young et al.. 1978a; 1978b). Plasma data from
11   Sprague Dawley rats given single i.v. doses of 3,  10, 30, 100, 300, or 1,000 mg [14C]-l,4-dioxane/kg
12   demonstrated a dose-related shift from linear, first-order to nonlinear, saturable metabolism of
13   1,4-dioxane between plasma 1,4-dioxane levels of 30 and  100 ug/mL (Figure 3-2). Similarly, in rats
14   given, via gavage in distilled water, 10,  100, or 1,000 mg [14C]-l,4-dioxane/kg singly or 10 or 1,000 mg
15   [14C]-l,4-dioxane/kg in 17 daily doses, the percent urinary excretion of the radiolabel decreased
16   significantly with dose while radiolabel in expired air increased. Specifically, with single
17   [14C]-l,4-dioxane/kg doses, urinary radiolabel decreased from 99 to 76% and expired 1,4-dioxane
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    increased from <1 to 25% as dose increased from 10 to 1,000 mg/kg. Likewise, with multiple daily doses
 2    10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary radiolabel decreased from 99 to 82% and expired
 3    1,4-dioxane increased from 1 to 9% as dose increased. The differences between single and multiple doses
 4    in urinary and expired radiolabel support the notion that 1,4-dioxane may induce its own metabolism.

 5           Induction of 1.4-dioxane metabolism was evaluated in a 13 week inhalation study by Kasai et al.
 6    (2008). In this study, male and female F344 rats were exposed daily to concentrations of 0 (control). 100.
 7    200. 400. 1.600. and 3.200 ppm. Plasma levels of 1.4-dioxane linearly increased with increasing
 8    inhalation concentration, suggesting that metabolic saturation was not achieved during the course of the
 9    experiments for plasma levels up to 730 and 1.054 ug/mL in male and female rats, respectively, at the
10    highest exposure concentration (3.200 ppm). In contrast. Young et al. (1978b) single dose experiments of
11    inhalation exposure to 50 ppm in male rats showed possible saturation of metabolism at plasma levels of
12    100 u,g/mL. Therefore, lack of the metabolic saturation of 1.4-dioxane found in the Kasai et al. (2008)
13    study is likely attributed to enhanced metabolism by the induction of P450 enzymes, including CYP2E1.
14    by 13 weeks of repeated inhalation exposure to 1.4-dioxane at concentrations up to 3.200 ppm (Kasai et
15    al.. 2008).
                         10,000
                             O   5   10   >5
                                                25  3O   35   4O  45  5O  55   60   6S  7C
             Source: Reprinted with permission of Taylor and Francis, Young et al. (1978b).
             Figure 3-2  Plasma 1,4-dioxane levels in rats following i.v. doses of 3-5,600 mg/kg

             [y-axis is plasma concentration of 1,4-dioxane (jig/mL) and x-axis is time (hr)]
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           1,4-Dioxane has been shown to induce several isoforms of CYP450 in various tissues following
 2    acute oral administration by gavage or drinking water (Nannelli et al.. 2005). Male Sprague Dawley rats
 3    were exposed to either 2,000 mg/kg 1,4-dioxane via gavage for 2 consecutive days or by ingestion of a
 4    1.5% 1,4-dioxane drinking water solution for 10 days. Both exposures resulted in significantly increased
 5    CYP2B1/2, CYP2C11, and CYP2E1 activities in hepatic microsomes. The gavage exposure alone
 6    resulted in increased CYP3A activity. The increase in 2C11 activity was unexpected, as that isoform has
 7    been observed to be under hormonal control and was typically suppressed in the presence of 2B1/2 and
 8    2E1 induction. In the male rat, hepatic 2C11 induction is associated with masculine pulsatile plasma
 9    profiles of growth hormone (compared to the constant plasma levels in the female), resulting in
10    masculinization of hepatocyte function ("Waxman et al..  1991). The authors postulated that 1,4-dioxane
11    may alter plasma growth hormone levels, resulting in the observed 2C11 induction. However, growth
12    hormone induction of 2C11 is primarily dependent on the duration between growth hormone pulses and
13    secondarily on growth hormone plasma levels (Agrawal and Shapiro. 2000; Waxman etal., 1991). Thus,
14    the induction of 2C11 by  1,4-dioxane may be mediated by changes in the time interval between growth
15    hormone pulses rather than changes in growth hormone levels. This may be accomplished by  1,4-dioxane
16    temporarily influencing the presence of growth hormone cell surface binding sites (Agrawal and Shapiro.
17    2000). However, no studies are available to confirm the influence of 1,4-dioxane on either growth
18    hormone levels or changes in growth hormone pulse  interval.

19           In nasal and renal mucosal cell microsomes,  CYP2E1 activity, but not CYP2B1/2 activity, was
20    increased. Pulmonary mucosal CYP450 activity levels were not significantly altered. Observed increases
21    in 2E1 mRNA in rats exposed by gavage and i.p. injection suggest that 2E1 induction in kidney and nasal
22    mucosa is controlled by a transcriptional activation of 2E1 genes. The lack of increased mRNA in
23    hepatocytes suggests that induction is regulated via a post-transcriptional mechanism. Differences in 2E1
24    induction mechanisms in liver, kidney, and nasal mucosa suggest that induction is controlled in a
25    tissue-specific manner.
      3.4  Elimination

26           In workers exposed to a TWA of 1.6 ppm for 7.5 hours, 99% of 1,4-dioxane eliminated in urine
27    was in the form of HEAA (Young et al., 1976). The elimination half-life was 59 minutes in adult male
28    volunteers exposed to 50 ppm 1,4-dioxane for 6 hours, with 90% of urinary 1,4-dioxane and 47% of
29    urinary HEAA excreted within 6 hours of onset of exposure (Young et al., 1977). There are no data for
30    1,4-dioxane elimination in humans from oral exposures.

31           Elimination of 1,4-dioxane in rats (Young et al.. 1978a: 1978b). was primarily via urine. As
32    comparably assessed in humans, the elimination half-life in rats exposed to 50 ppm 1,4-dioxane for
33    6 hours was calculated to be 1.01 hours.  In Sprague Dawley rats given single daily doses of 10, 100, or
34    1,000 mg [14C]-l,4-dioxane/kg or multiple doses of 10 or 1,000 mg [14C]-l,4-dioxane/kg, urinary
35    radiolabel ranged from 99% down to 76% of total radiolabel. Fecal elimination was less than 2% for all
36    doses. The effect of saturable metabolism on expired 1,4-dioxane was apparent, as expired 1,4-dioxane in

                                                                                                 10
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    singly dosed rats increased with dose from 0.4 to 25% while expired 14CO2 changed little (between 2 and
 2    3%) across doses. The same relationship was seen in Sprague Dawley rats dosed i.v. with 10 or 1,000 mg
 3    [14C]-l,4-dioxane/kg. Higher levels of 14CO2 relative to 1,4-dioxane were measured in expired air of the
 4    10 mg/kg group, while higher levels of expired  1,4-dioxane relative to 14CO2 were measured in the
 5    1,000 mg/kg group.
      3.5  Physiologically Based  Pharmacokinetic  Models

 6           Physiologically based pharmacokinetic models (PBPK) models have been developed for
 7    1.4-dioxane in rats (Sweeney et al.. 2008; Leung and Paustenbach.  1990; Reitz et al.. 1990). mice (Reitz
 8    etal.. 1990). humans (Sweeney et al. 2008; Leung and Paustenbach. 1990; Reitz etal.. 1990). and
 9    lactating women (Fisher etal.. 1997). Each of the models simulates the body as a series of compartments
10    representing tissues or tissue groups that receive  blood from the central vascular compartment
11    (Figure 3-3). Modeling was conducted under the premise that transfers of 1,4-dioxane between blood and
12    tissues occur sufficiently fast to be effectively blood flow-limited, which is consistent with the available
13    data (Ramsey and Andersen. 1984).  Blood time course and metabolite production data in rats and humans
14    suggest that absorption and metabolism are accomplished through common mechanisms in both species
15    (Young et al. (1978b; 1978a;  1977)). allowing identical model structures to be used for both species (and
16    by extension, for mice as well). In all three models, physiologically relevant, species-specific parameter
17    values for tissue volume, blood flow, and metabolism and elimination are used. The models  and
18    supporting data are reviewed below, from the perspective of assessing their utility for predicting internal
19    dosimetry and for cross-species extrapolation of  exposure-response relationships for critical  neoplastic
20    and nonneoplastic endpoints (also see Appendix  B).
                                                                                                 11
                                    DRAFT - DO NOT CITE OR QUOTE

-------
                                     IV
                                  infusion
                                     J
            inhalation
            4      t
                                              Lung blood



-Q
O
CD
Z3
O
£=
CD
>



L J

f Rapidly 1
[^ perfused J

[Slowly 1
perfused J

4 F^t L


^ Liver L
                                                                        "D
                                                                        O
                                                                        m
                                                                        cu
                                                                        t;
    Gl
absorption
                                                    metabolism
            Figure 3-3  General PBPK model structure.

            Consisting of blood-flow limited tissue compartments connected via arterial and venous
            blood flows. Note: Orally administered chemicals are absorbed directly into the liver while
            inhaled and intravenously infused chemicals enter directly into the arterial and venous
            blood pools, respectively.
     3.5.1  Available Pharmacokinetic Data

 1          Animal and human data sets available for model calibration derive from Young et al. (1978b:
 2   1978a: 1977). Mikheev et al. (1990). and Woo et al. (1977a; 1977b). Young et al. Q978b; 1978a) studied
 3   the disposition of radiolabeled [14C]-l,4-dioxane in adult male Sprague Dawley rats following i.v.,
 4   inhalation, and single and multiple oral gavage exposures. Plasma concentration-time profiles were
 5   reported for i.v. doses of 3, 10, 30, 100, and 1,000 mg/kg. In addition, exhaled 14CO2 and urinary
 6   1,4-dioxane and HEAA profiles  were reported following i.v. doses of 10 and  1,000 mg/kg. The plasma
 7   1,4-dioxane concentration-time course, cumulative urinary 1,4-dioxane and cumulative urinary HEAA
 8   concentrations were reported following a 6-hour inhalation exposure to 50 ppm. Following oral gavage
 9   doses of 10-1,000 mg/kg, percentages of total orally administered radiolabel were measured in urine,
10   feces, expired air, and the whole body.

11          Oral absorption of 1,4-dioxane was extensive, as only approximately 1% of the administered dose
12   appeared in the feces within 72 hours of dosing (Young et al.. 1978a: 1978b). Although it may be
                                                                                               12
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    concluded that the rate of oral absorption was high enough to ensure nearly complete absorption by
 2    72 hours, a more quantitative estimate of the rate of oral absorption is not possible due to the absence of
 3    plasma time course data by oral exposure.

 4           Saturable metabolism of 1,4-dioxane was observed in rats exposed by either the i.v. or oral routes
 5    (Young et al.. 1978a: 1978b). Elimination of 1,4-dioxane from plasma appeared to be linear following i.v.
 6    doses of 3-30 mg/kg, but was nonlinear following doses of 100-1,000 mg/kg. Accordingly, 10 mg/kg i.v.
 7    doses resulted in higher concentrations of 14CO2 (from metabolized 1,4-dioxane) in expired air relative to
 8    unchanged 1,4-dioxane, while 1,000 mg/kg i.v. doses resulted in higher concentrations of expired
 9    1,4-dioxane relative to 14CO2. Thus, at higher i.v. doses, a higher proportion of unmetabolized
10    1,4-dioxane is available for exhalation. Taken together, the i.v. plasma and expired air data from Young et
11    al. (1978b: 1978a) corroborate previous studies describing the saturable nature of 1,4-dioxane metabolism
12    in rats (1977a; Woo et al.. 1977b) and are useful for optimizing metabolic parameters (Vmax and Km) in a
13    PBPK model.

14           Similarly, increasing single or multiple oral doses of 10-1,000 mg/kg resulted in increasing
15    percentage of 1,4-dioxane in exhaled air and decreasing percentage of radiolabel (either as  1,4-dioxane or
16    a metabolite) in the urine, with significant differences in both metrics being observed between doses of 10
17    and 100 mg/kg (Young et al., 1978a;  1978b). These data identify the region (10-100 mg/kg) in which oral
18    exposures will result in nonlinear metabolism of 1,4-dioxane and can be used to test whether metabolic
19    parameter value estimates derived from i.v. dosing data are adequate for modeling oral exposures.

20           Post-exposure plasma data from a single 6-hour, 50 ppm inhalation exposure in rats were reported
21    (Young et al., 1978a; 1978b). The observed linear elimination of 1,4-dioxane after inhalation exposure
22    suggests that, via this route, metabolism is in the linear region at this exposure level.

23           The only human data adequate for use in PBPK model development (Young etal.  1977) come
24    from adult male volunteers exposed to 50 ppm 1,4-dioxane for 6 hours. Plasma 1,4-dioxane and HEAA
25    concentrations were measured both during and after the exposure period, and urine concentrations were
26    measured following exposure. Plasma levels of 1,4-dioxane approached steady-state at 6 hours. HEAA
27    data were insufficient to describe the  appearance or elimination of HEAA in plasma.  Data on elimination
28    of 1,4-dioxane and HEAA in the urine up to 24 hours from the beginning of exposure were reported. At
29    6 hours from onset of exposure, approximately 90% and 47% of the cumulative (0-24 hours) urinary
30    1,4-dioxane and HEAA, respectively, were measured in the urine. The ratio of HEAA to 1,4-dioxane in
31    urine 24 hours after onset of exposure was 192:1 (similar to the ratio of 118:1 observed by  Young et al.
32    (1976) in workers  exposed to 1.6 ppm for 7.5 hours), indicating  extensive metabolism of 1,4-dioxane As
33    with Sprague Dawley rats, the elimination of 1,4-dioxane from plasma was linear across all observations
34    (6 hours following end of exposure),  suggesting that human metabolism of 1,4-dioxane is linear for a
35    50 ppm inhalation exposure to steady-state. Thus, estimation of human Vmax and Km from these data will
36    introduce uncertainty into internal dosimetry performed in the nonlinear region of metabolism.

37           Further data were reported for the tissue distribution of 1,4-dioxane in rats. Mikheev et al. (1990)
38    administered i.p. doses of [14C]-l,4-dioxane to white rats (strain not reported) and reported time-to-peak
39    blood, liver, kidney, and testes concentrations. They also reported ratios of tissue to blood concentrations
                                                                                                   13
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    at various time points after dosing. Woo et al. (1977a; 1977b) administered i.p. doses of [14C]-l,4-dioxane
 2    to Sprague Dawley rats and measured radioactivity levels in urine. However, since i.p. dosing is not
 3    relevant to human exposures, these data are of limited use for PBPK model development.
      3.5.2  Published PBPK Models for 1,4-Dioxane
      3.5.2.1   Leung  and  Paustenbach
 4           Leung and Paustenbach (1990) developed a PBPK model for 1,4-dioxane and its primary
 5    metabolite, HEAA, in rats and humans. The model, based on the structure of a PBPK model for styrene
 6    (Ramsey and Andersen. 1984). consists of a central blood compartment and four tissue compartments:
 7    liver, fat, slowly perfused tissues (mainly muscle and skin), and richly perfused tissues (brain, kidney, and
 8    viscera other than the liver). Tissue volumes were calculated as percentages of total BW, and blood flow
 9    rates to each compartment were calculated as percentages of cardiac output. Equivalent cardiac output
10    and alveolar ventilation  rates were allometrically scaled to a power (0.74) of BW for each species. The
11    concentration  of 1,4-dioxane in alveolar blood was assumed to be in equilibrium with alveolar air at a
12    ratio equal to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane
13    between blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
14    between blood and tissue, governed by tissue:blood  equilibrium partition coefficients. The latter were
15    derived from the quotient of blood:air and tissue:air partition coefficients, which were measured in vitro
16    (Leung and Paustenbach. 1990) for blood, liver, fat, and skeletal muscle (slowly perfused tissue).
17    Blood:air partition coefficients were measured for both humans and rats. Rat tissue:air partition
18    coefficients were used as surrogate values for humans, with the exception of slowly perfused tissue:blood,
19    which was estimated by optimization to the plasma time-course data. Portals of entry included i.v.
20    infusion (over a period of 36 seconds) into the venous blood, inhalation by diffusion from the alveolar air
21    into the lung blood at the rate of alveolar ventilation, and oral administration via zero-order absorption
22    from the gastrointestinal tract to the liver.  Elimination of 1,4-dioxane was accomplished through
23    pulmonary exhalation and saturable hepatic metabolism. Urinary excretion of HEAA was assumed to be
24    instantaneous  with the generation of HEAA from the hepatic metabolism of 1,4-dioxane.

25           The parameter values for hepatic metabolism of 1,4-dioxane, Vmax and Km, were optimized and
26    validated against plasma and/or urine time course data for 1,4-dioxane and HEAA in rats following i.v.
27    and inhalation exposures and humans following inhalation exposure (Young et al. (1978b: 1978a: 1977)):
28    the exact data (i.e., i.v., inhalation, or both) used for the optimization and calibration were not reported.
29    Although the liver and fat were represented by tissue-specific compartments, no tissue-specific
30    concentration  data were available for model development, raising uncertainty as the model's ability to
31    adequately predict exposure to these tissues. The human inhalation exposure of 50 ppm for 6 hours
32    (Young et al..  1977) was reported to be in the linear range for metabolism; thus, uncertainty exists in the
33    ability of the allometrically-scaled value for the human metabolic Vmax to accurately describe 1,4-dioxane
34    metabolism from exposures resulting in metabolic saturation. Nevertheless, these values resulted in the

                                                                                                   14
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    model producing good fits to the data. For rats, the values for Vmax had to be adjusted upwards by a factor
 2    of 1.8 to reasonably simulate exposures greater than 300 mg/kg. The model authors attributed this to
 3    metabolic enzyme induction by high doses of 1,4-dioxane.
      3.5.2.2   Reitzetal.

 4           Reitz et al. (1990) developed a model for 1,4-dioxane and HEAA in the mouse, rat, and human.
 5    This model, also based on the styrene model of Ramsey and Andersen (1984). included a central blood
 6    compartment and compartments for liver, fat, and rapidly and slowly perfused tissues. Tissue volumes
 7    and blood flow rates were defined as percentages of total BW and cardiac output, respectively.
 8    Physiological parameter values were similar to those used by Andersen et al. (1987). except that flow
 9    rates for cardiac output and alveolar ventilation were doubled in order to produce a better fit of the model
10    to human blood level data (Young et al.. 1977). Portals of entry included i.v. injection into the venous
11    blood, inhalation, oral bolus dosing, and oral dosing via drinking water. Oral absorption of 1,4-dioxane
12    was simulated, in all three species, as a first-order transfer to liver (halftime approximately 8 minutes).

13           Alveolar blood levels of 1,4-dioxane were assumed to be in equilibrium with alveolar air at a
14    ratio equal to the experimentally measured blood:air partition coefficient. Transfers of 1,4-dioxane
15    between blood and tissues were assumed to be blood flow-limited and to achieve rapid equilibrium
16    between blood and tissue, governed by tissue:blood equilibrium partition coefficients. These coefficients
17    were derived by dividing experimentally measured (Leung and Paustenbach. 1990) in vitro blood:air and
18    tissue:air partition coefficients for blood, liver, fat. Blood:air partition coefficients were measured for both
19    humans and rats. The mouse blood:air partition coefficient was different from rat or human values; the
20    source of the partition coefficient for blood in mice  was not reported. Rat tissue:air partition coefficients
21    were used as surrogate values for humans. Rat tissue partition coefficient values were the same  values as
22    used in the Leung  and Paustenbach (1990) model (with the exception of slowly perfused tissues) and were
23    used in the models for all three species. The liver value was used for the rapidly perfused tissues, as well
24    as slowly perfused tissues. Although slowly perfused tissue:air partition coefficients for rats were
25    measured, the authors suggested that  1,4-dioxane in the muscle and air may not have reached equilibrium
26    in the highly gelatinous tissue homogenate (Reitzetal.. 1990). Substitution of the liver value provided
27    much closer agreement to the plasma data than when the muscle value was used. Further, doubling of the
28    measured human blood:air partition coefficient improved the fit of the model to the human blood level
29    data compared to the fit resulting from the measured value (Reitzetal..  1990). The Reitz et al. (1990)
30    model simulated three routes of 1,4-dioxane elimination: pulmonary exhalation, hepatic metabolism to
31    HEAA, and urinary excretion of HEAA. The elimination of HEAA was modeled as a first-order transfer
32    of 1,4-dioxane metabolite to urine.

33           Values for the metabolic rate constants, Vmax and Km, were optimized to achieve agreement with
34    various observations. Reitz et al. (1990)  optimized values for human Vmax and Km against the
35    experimental human 1,4-dioxane inhalation data (Young et al.. 1977). As noted previously, because the
36    human exposures were below the level needed to exhibit nonlinear kinetics, uncertainty exists in the

                                                                                                    15
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    ability of the optimized value of Vmax to simulate human 1,4-dioxane metabolism above the concentration
 2    that would result in saturation of metabolism. Rat metabolic rate constants were obtained by optimization
 3    to simulated data from a two compartment empirical pharmacokinetic model, which was fitted to i.v.
 4    exposure data (Young etal.. 1978a: 1978b). As with the Leung and

 5           The Leung and Paustenbach model (1990) and the Reitz et al. (1990) model included
 6    compartments for the liver and fat, although no tissue-specific concentration data were available to
 7    validate dosimetry for these organs. The derivations of human and rat HEAA elimination rate constants
 8    were not reported. Since no pharmacokinetics data for 1,4-dioxane in mice were available, mouse
 9    metabolic rate constants were allometrically scaled from rat and human values.
      3.5.2.3  Fisher etal.

10           A PBPK model was developed by Fisher et al. (1997) to simulate a variety of volatile organic
11    compounds (VOCs, including 1,4-dioxane) in lactating humans. This model was similar in structure to
12    those of Leung and Paustenbach (1990) and Reitz et al. (1990) with the addition of elimination of
13    1,4-dioxane to breast milk. Experimental measurements were made for blood:air and milk:air partition
14    coefficients. Other partition coefficient values were taken from Reitz et al. (1990). The model was not
15    optimized, nor was performance tested against experimental exposure data. Thus, the ability of the model
16    to simulate 1,4-dioxane exposure data is unknown.
      3.5.2.4  Sweeney et al.

17           The Sweeney et al. (2008) model consisted of fat liver, slowly perfused, and other well perfused
18    tissue compartments. Lung and stomach compartments were used to describe the route of exposure, and
19    an overall volume of distribution compartment was used for calculation of urinary excretion levels of
20    1.4-dioxane and HEAA. Blood, saline, and tissue to air partition coefficient values for 1.4-dioxane were
21    experimentally determined for rats and mice. Average values of the rat and mouse partition coefficients
22    were used for humans. Metabolic constants (VmaxC and Km) for the rat were derived by. optimization of
23    data from an i.v. exposure of 1.000 mg/kg (Young et al.. 1978b) for inducible metabolism. For uninduced
24    VmaxC estimation, data generated by_ i.v. exposures to 3^ 10. 30. and 100 mg/kg were used (Young et al..
25    1978b). Sweeney et al. (2008) determined best fit values for VmaxC by fitting to blood data in Young et
26    al. (1978b). The best fit VmaxC values were  7.5. 10.8. and 12.7 mg/hr-kg°75 for i.v. doses of 3 to 100.
27    300. and 1.000 mg/kg. suggesting a gradual dose dependent increase in metabolic rate over i.v. doses
28    ranging from 3 to 1.000 mg/kg. Although the Sweeney et al. (2008) model utilized two values for VmaxC
29    (induced and uninduced). the PBPK model does not include a dose-dependent function description of the
30    change of Vmax for i.v. doses between metabolic induced and uninduced exposures. Mouse VmaxC and
31    absorption constants were derived by optimizing fits to the blood 1.4-dioxane concentrations in mice
32    administered nominal doses of 200 and 2.000 mg/kg 1.4-dioxane via gavage in a water vehicle (Young et
33    al.. 1978b). The in vitro Vmax values  for rats and mice were scaled to estimate in vivo rates. The scaled

                                                                                                  16
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    and optimized rat VmaxC values were similar. The discrepancy between the scaled and optimized mouse
 2    values was larger, which was attributed to possible induction in mice at the lowest dose tested (200
 3    mg/kg). The ratio of optimized/scaled values for the rat was used to adjust the scaled human VmaxC and
 4    Km values to projected in vivo values.

 5           The Sweeney et al. (2008) model outputs were compared, by visual inspection, with data not used
 6    in fitting model parameters. The model predictions gave adequate match to the 1.4-dioxane exhalation
 7    data in rats after a 1.000 mg/kg i.v. dose.  1.4-Dioxane exhalation was overpredicted by a factor of about 3
 8    after a 10 mg/kg i.v. dose. Similarly, the simulations of exhaled 1.4-dioxane after oral dosing were
 9    adequate at 1.000 mg/kg and 100 mg/kg (within 50%). but poor at 10 mg/kg (model over predicted by a
10    factor of 5). The model did not adequately fit the human data (Young et al., 1977). Using physiological
11    parameters of Brown et al. (1997) and measured partitioning parameters (Sweeney et al.. 2008; Leung and
12    Paustenbach. 1990) with no metabolism, measured blood 1.4-dioxane concentrations reported by Young
13    et al. (1977) could not be achieved unless the estimated exposure concentration was increased by 2-fold.
14    As expected, inclusion of any metabolism resulted in a decrease in predicted blood concentrations. If
15    estimated metabolism rates were used with the reported exposure concentration, urinary metabolite
16    excretion was also underpredicted (Sweeney et al.. 2008).
      3.5.3  Implementation of Published  PBPK Models for 1,4-Dioxane

17           As previously described, several pharmacokinetic models have been developed to predict the
18    absorption, distribution, metabolism, and elimination of 1,4-dioxane in rats and humans. Single
19    compartment, empirical models for rats (Young et al.. 1978a: 1978b) and humans (Young et al.. 1977)
20    were developed to predict blood levels of 1,4-dioxane and urine levels of the primary metabolite, HEAA.
21    PBPK models that describe the kinetics of 1,4-dioxane using biologically realistic flow rates, tissue
22    volumes, enzyme affinities, metabolic processes, and elimination behaviors were also developed
23    (Sweeney et al.. 2008: Fisher etal.. 1997: Leung and Paustenbach. 1990: Reitzetal.. 1990).

24           In developing updated toxicity values for 1,4-dioxane the available PBPK models were evaluated
25    for their ability to predict observations made in experimental studies of rat and human exposures to
26    1,4-dioxane (Appendix B). The Reitz et al. (1990) and Leung and Paustenbach (1990) PBPK models were
27    both developed from a PBPK model  of styrene (Ramsey and Andersen. 1984). with the exception of
28    minor differences in the use of partition coefficients and biological parameters. The model code for Leung
29    and Paustenbach (1990) was unavailable in contrast to Reitz et al. (1990). The model of Reitz et al.
30    (1990) was identified for further consideration to assist in the derivation of toxicity values, and the
31    Sweeney et al. (2008) PBPK model was also evaluated.

32           The biological plausibility of parameter values in the Reitz et al. (1990) human model were
33    examined. The model published by Reitz et al. (1990) was able to predict the only available human
34    inhalation data (50 ppm 1,4-dioxane  for 6 hours; Young et al., (1977)) by increasing (i.e., approximately
35    doubling) the parameter values for human alveolar ventilation (30 L/hour/kg°74), cardiac output (30
36    L/hour/kg°74), and the blood:air partition coefficient (3,650) above the measured values of

                                                                                                  17
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    13 L/minute/kg074 (Brown et al.. 1997). 14 L/hour/kg°74 (Brown etal.. 1997). and 1,825 (Leung and
 2    Paustenbach. 1990). respectively. Furthermore, Reitz et al. (1990) replaced the measured value for the
 3    slowly perfused tissue:air partition coefficient (i.e., muscle—value not reported in manuscript) with the
 4    measured liver value (1,557) to improve the fit. Analysis of the Young et al. (1977) human data suggested
 5    that the apparent volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold higher in rats
 6    than humans, presumably due to species differences in tissue partitioning or other process not represented
 7    in the model. Based upon these observations, several model parameters (e.g., metabolism/elimination
 8    parameters) were re-calibrated using biologically plausible values for flow rates and tissue:air partition
 9    coefficients.

10           Appendix B describes all activities that were conducted in the evaluation of the empirical models
11    and the re-calibration and evaluation of the Reitz et al. (1990) PBPK model to determine the adequacy
12    and preference for the potential use of the models.

13           The evaluation consisted of implementation of the Young et al. (1978b; 1978a; 1977) empirical
14    rat and human models using the acslXtreme simulation software, re-calibration of the Reitz et al. (1990)
15    human PBPK model, and evaluation of the model parameters published by Sweeney et al. (2008). Using
16    the model descriptions and equations given in Young et al. (1978b: 1978a: 1977). model code was
17    developed for the empirical models and executed, simulating the reported experimental conditions. The
18    model output was then compared with the model output reported in Young et al. (1978b; 1978a; 1977).

19           The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for cardiac and
20    alveolar flow rates and tissue:air partition coefficients. The predictions of blood and urine levels of
21    1,4-dioxane and HEAA, respectively, from the re-calibrated model were compared with the empirical
22    model predictions of the same dosimeters to  determine whether the re-calibrated PBPK model could
23    perform similarly to the empirical model. As part of the PBPK model evaluation, EPA performed a
24    sensitivity analysis to identify the model parameters having the greatest influence on the primary
25    dosimeter of interest, the blood level of 1,4-dioxane. Variability data for the experimental measurements
26    of the tissue: air partition coefficients were incorporated to determine a range of model outputs bounded
27    by biologically plausible values for these parameters. Model parameters from Sweeney et al. (2008) were
28    also tested to evaluate the ability of the PBPK model to predict human data following exposure to
29    1,4-dioxane.

30           The rat and human empirical models of Young et al. (1978b; 1978a; 1977) were successfully
31    implemented in acslXtreme and perform identically to the models reported in the published papers
32    (Figure B-3 through Figure B-7), with the exception of the lower predicted HEAA concentrations and
33    early appearance of the peak HEAA levels in rat urine. The early appearance of peak HEAA levels cannot
34    presently be explained, but may result from manipulations of kme or other parameters by Young et al.
35    (1978b: 1978a) that were not reported.  The lower predictions of HEAA levels are likely due to reliance on
36    a standard urine volume production rate in the absence of measured (but unreported) urine volumes.
37    While the human urinary HEAA predictions  were lower than observations, this is due to parameter fitting
38    of Young et al. (1977). No model output was published in Young et al. (1977) for comparison. The
39    empirical models were modified to allow for user-defined inhalation exposure levels. However, no

                                                                                                  18
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    modifications were made to model oral exposures as adequate data to parameterize such modifications do
 2    not exist for rats or humans. Further evaluations of the Young et aL (1977) modified model were
 3    conducted against data from the Kasai et al. (2008) subchronic inhalation study. The results of this
 4    evaluation are shown in Appendix B (Figure B-8). It shows that the Young et al. (1977) inhalation
 5    empirical model failed to provide an adequate simulation of the 13 week inhalation exposure blood data
 6    of Kasai et al. (2008). Since the Young et al. (1977) model consistently overpredicted the Kasai et al.
 7    (2008) data, the lack of model fit is most likely due to the lack of inclusion of other metabolic processes
 8    or parameters.

 9           Several procedures were applied to the  Reitz et al. (1990) human PBPK model to determine if an
10    adequate fit of the model to the empirical model output or experimental observations could be attained
11    using biologically plausible values for the model parameters. The re-calibrated model predictions for
12    blood 1,4-dioxane levels do not come within 10-fold of the experimental values using measured tissue:air
13    partition coefficients from Leung and Paustenbach (1990) or Sweeney et al. (2008) (Figure B-9 and
14    Figure B-10). The utilization of a slowly perfused tissue:air partition coefficient 10-fold lower than
15    measured values produces exposure-phase predictions that are much  closer to observations, but does not
16    replicate the elimination kinetics (Figure B-l 1). Recalibration of the  model with upper bounds on the
17    tissue:air partition coefficients results in predictions that are still six- to sevenfold lower than empirical
18    model prediction or observations (Figure B-13 and Figure B-14). Exploration of the model space using  an
19    assumption of zero-order metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate
20    fit to the exposure and elimination data can be achieved only when unrealistically low values are assumed
21    for the slowly perfused tissue:air partition coefficient (Figure B-17).  Artificially low values for the other
22    tissue:air partition coefficients are not expected to improve the model fit,  as these parameters are shown
23    in the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km. In the absence
24    of actual measurements for the human slowly perfused tissue:air partition coefficient, high uncertainty
25    exists for this model parameter value. Differences in the ability of rat and human blood to  bind
26    1,4-dioxane may contribute to the difference in Vd. However, this is expected to be evident in very
27    different values for rat and human blood:air partition coefficients, which is not the case (Table B-l).
28    Therefore, some other, as yet unknown, modification to model structure may be necessary.

29           Similarly, Sweeney et al. (2008) also evaluated the available PBPK models (Leung and
30    Paustenbach. 1990; Reitz etal., 1990) for 1,4-dioxane. To address uncertainties and deficiencies in these
31    models, the investigators conducted studies to fill data gaps and reduce uncertainties pertaining to the
32    pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. The following studies were
33    performed:

34       •   Partition coefficients, including measurements for mouse blood and tissues (liver, kidney, fat, and
35           muscle) and confirmatory measurements for human blood and rat blood and muscle.
36       •   Blood time course measurements in mice conducted for gavage administration of nominal single
37           doses (20, 200, or 2,000 mg/kg) of 1,4-dioxane administered in water.
38       •   Metabolic rate constants for rat, mouse, and human liver based on incubations of 1,4-dioxane
39           with rat, mouse, and human hepatocytes and measurement of HEAA.
                                                                                                   19
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           The studies conducted by Sweeney et al. (2008) resulted in partition coefficients that were
 2    consistent with previously measured values and those used in the Leung and Paustenbach (1990) model.
 3    Of noteworthy significance, the laboratory results of Sweeney et al. (2008) did not confirm the human
 4    blood:air partition coefficient Reitz et al. (1990) reported. Furthermore, Sweeney et al. (2008) estimated
 5    metabolic rate constants (VmaxC and Km) within the range used in the previous models (Leung and
 6    Paustenbach. 1990; Reitz etal.. 1990). Overall, the Sweeney et al. (2008) model utilized more rodent in
 7    vivo and in vitro data in model parameterization and refinement; however, the model was still unable to
 8    adequately predict the human blood data from Young et al. (1977).

 9           Updated PBPK models were developed based on these new data and data from previous kinetic
10    studies in rats, workers, and human volunteers reported by Young et al. (1978b;  1978a; 1977; 1976). The
11    optimized rate of metabolism for the mouse was significantly higher than the value previously estimated.
12    The optimized rat kinetic parameters were similar to those in the 1990 models. Of the two available
13    human studies (Young (1977; 1976). model predictions were consistent with one study, but did not fit the
14    second as well.
      3.6   Rat Nasal Exposure via  Drinking Water

15           Sweeney et al. (2008) conducted a rat nasal exposure study to explore the potential for direct
16    contact of nasal tissues with 1,4-dioxane-containing drinking water under bioassay conditions. Two
17    groups of male Sprague Dawley rats (5/group) received drinking water in 45-mL drinking water bottles
18    containing a fluorescent dye mixture (Cell Tracker Red/FluoSpheres). The drinking water for one of these
19    two groups also contained 0.5% 1,4-dioxane, a concentration within the range used in chronic toxicity
20    studies. A third group of five rats received tap water alone (controls). Water was provided to the rats
21    overnight. The next morning, the water bottles were weighed to estimate the amounts of water consumed.
22    Rats were sacrificed and heads were split along the midline for evaluation by fluorescence microscopy.
23    One additional rat was dosed twice by gavage with 2 mL of drinking water containing fluorescent dye
24    (the second dose was 30 minutes after the first dose; total of 4 mL administered) and sacrificed 5 hours
25    later to evaluate the potential for systemic delivery of fluorescent dye to the nasal tissues.

26           The presence of the fluorescent dye mixture had no measurable impact on water consumption;
27    however, 0.5% 1,4-dioxane reduced water consumption by an average of 62% of controls following a
28    single, overnight exposure. Fluorescent dye was detected in the  oral cavity and nasal airways of each
29    animal exposed to the Cell Tracker Red/FluoSpheres mixture in their drinking water, including numerous
30    areas of the anterior third of the nose along the nasal vestibule, maxillary turbinates, and dorsal
31    nasoturbinates. Fluorescent dye was occasionally detected in the ethmoid turbinate region and
32    nasopharynx.  1,4-Dioxane had no effect on the detection of the  dye. Little or no fluorescence at the
33    wavelength associated with the dye mixture was detected in control animals or in the single animal that
34    received the dye mixture by oral gavage. The investigators concluded that the findings indicate rat nasal
35    tissues are exposed by direct contact with drinking water under  bioassay conditions.
                                                                                                  20
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      4   HAZAR D  IDE  NOTIFICATION
      4.1  Studies in  Humans - Epidemiology, Case Reports, Clinical
           Controls

 1           Case reports of acute occupational poisoning with 1,4-dioxane indicated that exposure to high
 2    concentrations resulted in liver, kidney, and central nervous system (CNS) toxicity (Johnstone.  1959;
 3    Barber. 1934). Barber (1934) described four fatal cases of hemorrhagic nephritis and centrilobular
 4    necrosis of the liver attributed to acute inhalation exposure to high (unspecified) concentrations of
 5    1,4-dioxane. Death occurred within 5-8 days of the onset of illness. Autopsy findings suggested that the
 6    kidney toxicity may have been responsible for lethality, while the liver effects may have been compatible
 7    with recovery. Jaundice was not observed in subjects and fatty change was not apparent in the liver.
 8    Johnstone (1959) presented the fatal case of one worker exposed to high concentrations of 1,4-dioxane
 9    through both inhalation and dermal exposure for a 1 week exposure duration. Measured air concentrations
10    in the work environment of this subject were 208-650 ppm, with a mean value of 470 ppm. Clinical signs
11    that were observed following hospital admission included severe epigastric pain, renal failure, headache,
12    elevation in blood pressure, agitation and restlessness, and coma. Autopsy findings revealed significant
13    changes in the liver, kidney, and brain. These included centrilobular necrosis of the liver and hemorrhagic
14    necrosis of the kidney cortex. Perivascular widening was observed in the brain with small foci of
15    demyelination in several regions (e.g., cortex, basal nuclei). It was suggested that these neurological
16    changes may have been secondary to anoxia and cerebral edema.

17           Several studies examined the effects of acute inhalation exposure in volunteers. In a study
18    performed at the Pittsburgh Experimental Station of the U.S. Bureau of Mines, eye irritation and a
19    burning sensation in the nose and throat were reported in five men exposed to  5,500 ppm of 1,4-dioxane
20    vapor for  1 minute (Yant et al.. 1930). Slight vertigo was also reported by three of these men. Exposure to
21    1,600 ppm of 1,4-dioxane vapor for 10 minutes resulted in similar symptoms with a reduced intensity of
22    effect. In a study conducted by the Government Experimental Establishment at Proton, England (Fairley
23    etal.. 1934). four men were exposed to 1,000 ppm of 1,4-dioxane for 5 minutes. Odor was detected
24    immediately and one volunteer noted a constriction in the throat. Exposure of six volunteers to 2,000 ppm
25    for 3 minutes resulted in no symptoms of discomfort. Wirth and Klimmer (1936). of the Institute of
26    Pharmacology, University of Wurzburg, reported slight mucous membrane irritation in the nose and
27    throat of several human subjects exposed to concentrations greater than 280 ppm for several minutes.
28    Exposure to approximately 1,400 ppm for several minutes caused a prickling sensation in the nose and a
29    dry and scratchy throat. Silverman et al. (1946) exposed 12 male and 12 female subjects to varying air
30    concentrations of 1,4-dioxane for 15 minutes. A 200 ppm concentration was reported to be tolerable,
31    while a concentration of 300 ppm caused irritation to the eyes, nose, and throat. The study conducted by
32    Silverman et al. (1946) was conducted by the Department of Industrial Hygiene, Harvard School of
33    Public Health, and was sponsored and supported by a grant from the Shell Development Company. These
34    volunteer  studies published in the 1930s and 1940s (Silverman et al.. 1946; Wirth and Klimmer. 1936;
35    Fairley et  al.. 1934; Yantetal.. 1930) did not provide information on the human subjects research ethics
                                                                                                 21
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    procedures undertaken in these studies; however, there is no evidence that the conduct of the research was
 2    fundamentally unethical or significantly deficient relative to the ethical standards prevailing at the time
 3    the research was conducted.

 4           Young et al. (1977) exposed four healthy adult male volunteers to a 50-ppm concentration of
 5    1,4-dioxane for 6 hours. The investigators reported that the protocol of this study was approved by a
 6    seven-member Human Research Review Committee of the Dow Chemical Company and was followed
 7    rigorously. Perception of the odor of 1,4-dioxane appeared to diminish over time, with two of the four
 8    subjects reporting inability to detect the odor at the end of the exposure period. Eye irritation was the only
 9    clinical sign reported in this study. The pharmacokinetics and metabolism of 1,4-dioxane in humans were
10    also evaluated in this study (see Section 3.3). Clinical findings were not reported in four workers exposed
11    in the workplace to a TWA concentration of 1.6 ppm for 7.5 hours (Young et al.. 1976).

12           Ernstgard et al. (2006) examined the acute effects of 1,4-dioxane vapor in male and female
13    volunteers. The study protocol was approved by the Regional Ethics Review Board in Stockholm, and
14    performed following informed consent and according to the Helsinki declaration. In a screening study by
15    these investigators, no self-reported symptoms (based on a visual analogue scale (VAS) that included
16    ratings for discomfort in eyes, nose, and throat, breathing difficulty, headache, fatigue, nausea, dizziness,
17    or feeling of intoxication) were observed at concentrations up to 20 ppm; this concentration was selected
18    as a tentative no-observed-adverse-effect-level (NOAEL) in the main study. In the main study, six male
19    and six female healthy volunteers were exposed to 0 or 20 ppm 1,4-dioxane, at rest, for 2 hours. This
20    exposure did not significantly affect symptom VAS ratings, blink frequency, pulmonary function or nasal
21    swelling (measured before and at 0 and 3 hours after exposure), or inflammatory markers in the plasma
22    (C-reactive protein and interleukin-6) of the volunteers. Only ratings for "solvent smell" were
23    significantly increased during exposure.

24           Only two well documented epidemiology studies were  available for occupational workers
25    exposed to 1,4-dioxane (Buffler et al.. 1978; Thiess et al., 1976). These studies did not provide evidence
26    of effects in humans; however, the cohort size and number of reported cases were small.
      4.1.1  Thiess etal.

27           A cross-sectional survey was conducted by Thiess et al. (1976) in German workers exposed to
28    1,4-dioxane. The study evaluated health effects in 74 workers, including 24 who were still actively
29    employed in 1,4-dioxane production at the time of the investigation, 23 previously exposed workers who
30    were still employed by the manufacturer, and 27 retired or deceased workers. The actively employed
31    workers were between 32 and 62 years of age and had been employed in  1,4-dioxane production for 5-
32    41 years.  Former workers (age range not given) had been exposed to 1,4-dioxane for 3-38 years and
33    retirees (age range not given) had been exposed for 12-41 years. Air concentrations in the plant at the
34    time of the study were 0.06-0.69 ppm. A simulation of previous exposure conditions (prior to 1969)
35    resulted in air measurements between 0.06 and 7.2 ppm.
                                                                                                   22
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Active and previously employed workers underwent a thorough clinical examination and X-ray,
 2    and hematological and serum biochemistry parameters were evaluated. The examination did not indicate
 3    pathological findings for any of the workers and no indication of malignant disease was noted.
 4    Hematology results were generally normal. Serum transaminase levels were elevated in 16 of the
 5    47 workers studied; however, this finding was consistent with chronic consumption of more than
 6    80 grams of alcohol per day, as reported for these workers. No liver enlargement or jaundice was found.
 7    Renal function tests and urinalysis were normal in exposed workers. Medical records of the 27 retired
 8    workers (15 living at the time of the study) were reviewed. No symptoms of liver or kidney disease were
 9    reported and no cancer was detected. Medical reasons for retirement did not appear related to 1,4-dioxane
10    exposure (e.g., emphysema, arthritis).

11           Chromosome  analysis was performed on six actively  employed workers and six control persons
12    (not characterized). Lymphocyte cultures were  prepared and chromosomal aberrations were evaluated. No
13    differences were noted in the percent of cells with gaps or other chromosome aberrations. Mortality
14    statistics were calculated for 74 workers of different ages and varying exposure periods. The proportional
15    contribution of each of the exposed workers to  the total time of observation was calculated as the sum of
16    man-years per 10-year age group. Each person  contributed one man-year per calendar year to the specific
17    age group in which he was included at the  time. The expected number of deaths for this population was
18    calculated from the age-specific mortality  statistics for the German Federal Republic for the years 1970-
19    1973. From the total of 1,840.5 person-years, 14.5 deaths were expected; however, only 12 deaths were
20    observed in exposed workers between 1964 and 1974. Two cases of cancer were reported, including one
21    case of lamellar epithelial carcinoma and one case of myelofibrosis leukemia. These cancers were not
22    considered to be the cause of death in these cases and other severe illnesses were present. Standardized
23    mortality ratios (SMRs) for cancer did not significantly differ from the control population (SMR for
24    overall  population = 0.83; SMR for 65-75-year-old men = 1.61; confidence intervals (CIs) were not
25    provided).
      4.1.2  Buffleretal.

26           Buffler et al. (1978) conducted a mortality study on workers exposed to 1,4-dioxane at a chemical
27    manufacturing facility in Texas. 1,4-Dioxane exposure was known to occur in a manufacturing area and
28    in a processing unit located 5 miles from the manufacturing plant. Employees who worked between April
29    1, 1954, and June 30, 1975, were separated into two cohorts based on at least 1 month of exposure in
30    eitherthe manufacturing plant (100 workers) or the processing area (65 workers). Company records and
31    follow-up techniques were used to compile information on name, date of birth, gender, ethnicity, job
32    assignment and duration, and employment status at the time of the study. Date and cause of death were
33    obtained from copies of death certificates and autopsy reports (if available). Exposure levels for each job
34    category were estimated using the 1974 Threshold Limit Value for 1,4-dioxane (i.e., 50 ppm) and
35    information from area and personal monitoring. Exposure levels were classified as low (<25 ppm),
36    intermediate (50-75 ppm), and high (>75 ppm). Monitoring was not conducted prior to 1968 in the
37    manufacturing areas or prior to 1974 in the processing area; however, the study authors assumed that

                                                                                                  23
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    exposures would be comparable, considering that little change had been made to the physical plant or the
 2    manufacturing process during that time. Exposure to 1,4-dioxane was estimated to be below 25 ppm for
 3    all individuals in both cohorts. Manufacturing area workers were exposed to several other additional
 4    chemicals and processing area workers were exposed to vinyl chloride.

 5           Seven deaths were identified in the manufacturing cohort and five deaths were noted for the
 6    processing cohort. The average exposure duration was not greater for those workers who died, as
 7    compared to those still living at the time of the study. Cancer was the underlying cause of death for two
 8    cases from the manufacturing area (carcinoma of the stomach, alveolar cell carcinoma) and one case from
 9    the processing area (malignant mediastinal tumor). The workers from the manufacturing area were
10    exposed for 28 or 38 months and both  had a positive smoking history (>1 pack/day). Smoking history was
11    not available for processing area workers. The single case of cancer in this area occurred in a 21-year-old
12    worker exposed to 1,4-dioxane for 1 year. The mortality data for both industrial cohorts were compared to
13    age-race-sex specific death rates for Texas (1960-1969). Person-years  of observation contributed by
14    workers were determined over five age ranges with each worker contributing one person-year for each
15    year of observation in a specific age group. The expected number of deaths was determined by applying
16    the Texas 1960-1969 death rate statistics to the number of person years calculated for each cohort. The
17    observed and expected number of deaths for overall mortality (i.e., all causes) was comparable for both
18    the manufacturing area (7 observed versus 4.9 expected) and the processing area (5 observed versus
19    4.9 expected). No significant excess in cancer-related deaths was identified for both areas of the facility
20    combined (3 observed versus 1.7 expected). A separate analysis was performed to evaluate mortality in
21    manufacturing area workers exposed to 1,4-dioxane for more than 2 years. Six deaths occurred in this
22    group as compared to 4.1 expected deaths. The use of a conditional Poisson distribution indicated no
23    apparent excess in mortality or death due to malignant neoplasms in this study. It is important to note that
24    the cohorts evaluated were limited in size. In addition, the mean exposure duration was less than 5 years
25    (<2 years for 43% of workers) and the latency period for evaluation was less than 10 years for 59% of
26    workers. The study authors recommended a follow-up investigation to allow for a longer latency period;
27    however, no follow-up study of these workers has been published.
      4.2  Subchronic and Chronic Studies and Cancer Bioassays  in
           Animals - Oral  and Inhalation

28           The majority of the subchronic and chronic studies conducted for 1.4-dioxane were drinking
29    water studies. To date, there are only two subchronic inhalation studies (Kasai et al., 2008; Fairley et al.
30    1934) and two chronic inhalation studies (Kasai et al.. 2009; Torkelson et al.. 1974). The effects
31    following oral and inhalation exposures are described in detail below.
                                                                                                 24
                                    DRAFT - DO NOT CITE OR QUOTE

-------
      4.2.1  OralToxicity
      4.2.1.1   S ubchronic Oral Toxicity

 1           Six rats and six mice (unspecified strains) were given drinking water containing 1.25%
 2    1,4-dioxane for up to 67 days (Fairley et al.. 1934). Using reference BWs and drinking water ingestion
 3    rates for rats and mice (U.S. EPA. 1988). it can be estimated that these rats and mice received doses of
 4    approximately 1,900 and 3,300 mg/kg-day, respectively. Gross pathology and histopathology were
 5    evaluated in all animals. Five of the six rats in the study died or were killed in extremis prior to day 34 of
 6    the study. Mortality was lower in mice, with five of six mice surviving up to 60 days. Kidney enlargement
 7    was noted in 5/6 rats and 2/5 mice. Renal cortical degeneration was observed in all rats and 3/6 mice.
 8    Large areas of necrosis were observed in the cortex, while cell degeneration in the medulla was slight or
 9    absent. Tubular casts were observed and vascular congestion and hemorrhage were present throughout the
10    kidney. Hepatocellular degeneration with vascular congestion was also noted in five rats and three mice.
11    For this assessment, EPA identified the tested doses of 1,900 mg/kg-day in rats and 3,300 mg/kg-day in
12    mice as the lowest-observed-adverse-effect-levels (LOAELs) for liver and kidney degeneration in this
13    study.

14    4.2.1.1.1   S toner et a I. 1,4-Dioxane was evaluated by Stoner et al. (1986) for its ability to
15    induce lung adenoma formation in A/J mice.  Six- to 8-week-old male and female A/J mice (16/sex/group)
16    were given 1,4-dioxane by gavage or i.p. injection, 3 times/week for 8 weeks. Total cumulative dose
17    levels were given as 24,000 mg/kg (oral), and 4,800, 12,000, or 24,000 mg/kg (i.p.). Average daily dose
18    estimates were calculated to be 430 mg/kg-day (oral), and 86, 210, or 430 mg/kg-day (i.p.) by assuming
19    an exposure duration of 56  days. The authors indicated that i.p. doses  represent the maximum tolerated
20    dose (MTD), 0.5 times the MTD, and 0.2 times the MTD. Mice were killed 24 weeks after initiation of
21    the bioassay, and lungs, liver, kidney, spleen, intestines, stomach, thymus, salivary, and endocrine glands
22    were examined for gross lesions. Histopathology examination was performed if gross lesions were
23    detected. 1,4-Dioxane did not induce lung tumors in male or female A/J mice in this study.

14    4.2.1.1.2  Stottetal.  In the Stott et al. (1981) study, male Sprague Dawley rats
15    (4-6/group) were given average doses of 0, 10, or 1,000 mg/kg-day 1,4-dioxane (>99% pure) in their
16    drinking water, 7 days/week for 11 weeks. It should be noted that the methods description in this report
17    stated that the high  dose was 100 mg/kg-day, while the abstract, results, and discussion sections indicated
18    that the high dose was 1,000 mg/kg-day. Rats were implanted with a [6"3H]thymidine loaded osmotic
19    pump 7 days prior to sacrifice. Animals were sacrificed by cervical dislocation and livers were removed,
20    weighed, and prepared for histopathology evaluation. [3F£]-Thymidine incorporation was measured by
21    liquid scintillation spectroscopy.

14           An increase in the liver to BW ratio was observed in rats from the high dose group (assumed to
15    be 1,000 mg/kg-day). Histopathological alterations, characterized as minimal centrilobular swelling, were
16    also seen in rats from this dose group (incidence values were not reported). Hepatic DNA synthesis,
17    measured by [3H]-thymidine incorporation, was increased 1.5-fold in high-dose rats. No changes relative
                                                                                                   25
                                     DRAFT - DO NOT CITE  OR QUOTE

-------
 1    to control were observed for rats exposed to 10 mg/kg-day. EPA found a NOAEL value of 10 mg/kg-day
 2    and a LOAEL value of 1,000 mg/kg-day for this study based on histopathological changes in the liver.

 3           Stott et al. (1981) also performed several acute experiments designed to evaluate potential
 4    mechanisms for the carcinogenicity of 1,4-dioxane. These experiments are discussed separately in Section
 5    4.5.2 (Mechanistic Studies).

 6    4.2.1.1.3   Kanoetal. In the Kano et al. (2008) study, groups of 6-week-old F344/DuCrj
 7    rats (10/sex/group) and Crj:BDFl mice (10/sex/group) were administered 1,4-dioxane (>99% pure) in the
 8    drinking water for 13 weeks. The animals were observed daily for clinical signs of toxicity. Food
 9    consumption and BWs were measured once per week and water consumption was measured twice
10    weekly. Food and water were available ad libitum. The concentrations of 1,4-dioxane in the water for rats
11    and mice were 0, 640, 1,600, 4,000, 10,000, or 25,000 ppm. The investigators used data from water
12    consumption and BW changes to calculate a daily intake of 1,4-dioxane by the  male and female animals.
13    Thus, male rats received doses of approximately 0, 52, 126, 274, 657, and 1,554 mg 1,4-dioxane/kg-day
14    and female rats received 0,  83, 185, 427, 756, and 1,614 mg/kg-day. Male mice received 0, 86, 231, 585,
15    882, or 1,570 mg/kg-day and female mice received 0, 170, 387, 898, 1,620, or 2,669 mg/kg-day.

 6           No information was provided as to when the blood and urine samples were collected.
 7    Hematology analysis included red blood cell (RBC) count, hemoglobin, hematocrit, mean corpuscular
 8    volume (MCV), platelet count, white blood cell (WBC) count, and differential WBCs. Serum
 9    biochemistry included total protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only),
10    alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), leucine
11    aminopeptidase (LAP), alkaline phosphatase (ALP), creatinine phosphokinase (CPK) (rat only), urea
12    nitrogen, creatinine (rat only), sodium, potassium, chloride, calcium (rat only),  and inorganic phosphorous
13    (rat only). Urinalysis parameters were pH, protein, glucose, ketone body, bilirubin (rat only), occult
14    blood, and urobilinogen. Organ weights (brain, lung, liver, spleen, heart, adrenal, testis, ovary, and
15    thymus) were measured, and gross necropsy and histopathologic examination of tissues and organs were
16    performed on all animals (skin, nasal cavity, trachea, lungs, bone marrow, lymph nodes, thymus, spleen,
17    heart, tongue, salivary glands, esophagus, stomach, small and large intestine, liver, pancreas, kidney,
18    urinary bladder, pituitary thyroid adrenal, testes, epididymis, seminal vesicle, prostate, ovary, uterus,
19    vagina, mammary gland, brain, spinal cord, sciatic nerve, eye, Harderian gland, muscle, bone, and
20    parathyroid). Dunnett's test and %2 test were used to assess the statistical significance of changes in
21    continuous and discrete variables, respectively.

22           Clinical signs of toxicity in rats were not discussed in  the study report.  One female rat in the high
23    dose group (1,614 mg/kg-day) group died, but cause and time  of death were not specified. Final BWs
24    were reduced at the two highest dose levels in  females (12 and 21%) and males (7 and 21%), respectively.
25    Food consumption was reduced 13% in females at 1,614 mg/kg-day and 8% in 1,554 mg/kg-day males. A
26    dose-related decrease in water consumption was observed in male rats starting at 52 mg/kg-day (15%)
27    and in females starting at 185 mg/kg-day (12%). Increases in RBCs, hemoglobin, hematocrit, and
28    neutrophils, and a decrease  in lymphocytes were observed in males at 1,554 mg/kg-day. In females, MCV
29    was decreased at doses > 756 mg/kg and platelets were decreased at 1,614 mg/kg-day. With the exception

                                                                                                   26
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1   of the 30% increase in neutrophils in high-dose male rats, hematological changes were within 2-15% of
 2   control values. Total serum protein and albumin were significantly decreased in males at doses >
 3   274 mg/kg-day and in females at doses > 427 mg/kg-day. Additional changes in high-dose male and
 4   female rats included decreases in glucose, total cholesterol, triglycerides, and sodium (and calcium in
 5   females), and increases in ALT (males only), AST, ALP, and LAP. Serum biochemistry parameters in
 6   treated rats did not differ more than twofold from control values. Urine pH was decreased in males at >
 7   274 mg/kg-day and in females at > 756 mg/kg-day.

 8           Kidney weights were increased in females at >185 mg/kg-day with a maximum increase of 15%
 9   and 44% at 1,614 mg/kg-day for absolute and relative kidney weight, respectively. No organ weight
10   changes  were noted in male rats. Histopathology findings in rats that were  related to exposure included
11   nuclear enlargement of the respiratory epithelium, nuclear enlargement of the olfactory epithelium,
12   nuclear enlargement of the tracheal epithelium, hepatocyte swelling of the  centrilobular area of the liver,
13   vacuolar changes in the liver, granular changes in the liver, single cell necrosis in the liver, nuclear
14   enlargement of the proximal tubule of the kidneys, hydropic changes in the proximal tubule of the
15   kidneys, and vacuolar changes in the brain. The incidence data for histopathological lesions in rats are
16   presented in Table 4-1. The effects that occurred at the lowest doses were nuclear enlargement of the
17   respiratory epithelium in the nasal cavity and hepatocyte swelling in the central area of the liver in male
18   rats. Based on these histopathological findings the study authors identified the LOAEL as 126 mg/kg-day
19   and the NOAEL as 52 mg/kg-day.
                                                                                                   27
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     Table 4-1   Incidence of histopathological lesions in F344/DuCrj rats exposed to 1,4-dioxane in
                 drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; renal proximal tubule
Hydropic change; renal proximal tubule
Vacuolic change; brain
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
52
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
126
9/10"
0/10
0/10
9/10"
0/10
0/10
0/10
0/10
0/10
0/10
Female dose

Nuclear enlargement; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Nuclear enlargement; tracheal epithelium
Hepatocyte swelling
Vacuolic change; liver
Granular change; liver
Single cell necrosis; liver
Nuclear enlargement; proximal tubule
Hydropic change; proximal tubule
Vacuolic change; brain
0
0/10
0/10
0/10
0/10
0/10
2/10
2/10
0/10
0/10
0/10
83
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
185
5/1 Oc
0/10
0/10
0/10
0/10
1/10
1/10
0/10
0/10
0/10
274
10/10"
10/10"
10/10"
10/10"
0/10
5/1 Oc
5/1 Oc
1/10
0/10
0/10
657
9/10"
9/10"
10/10"
10/10"
10/10"
2/10
2/10
5/1 Oc
0/10
0/10
1,554
10/10"
10/10"
10/10"
10/10"
10/10"
10/10"
10/10"
9/10"
7/10"
10/10"
(mg/kg-day)a
427
10/10"
9/10"
9/10"
0/10
0/10
5/1 Oc
5/10
0/10
0/10
0/10
756
10/10"
10/10"
10/10"
9/10"
0/10
5/1 Oc
5/10
8/10"
0/10
0/10
1,614
8/9"
8/9"
9/9"
9/9"
9/9"
8/9"
8/9"
9/9"
5/9c
9/9"
aData are presented for sacrificed animals.
bp < 0.01 by x2 test.
°p < 0.05.
Source: Kano et al. (2008)
 1           Clinical signs of toxicity in mice were not discussed in the study report One male mouse in the
 2   high-dose group (1,570 mg/kg-day) died, but no information was provided regarding cause or time of
 3   death. Final BWs were decreased 29% in male mice at 1,570 mg/kg-day, but changed less than 10%
 4   relative to controls in the other male dose groups and in female mice. Food consumption was not
 5   significantly reduced in any exposure group. Water consumption was reduced 14-18% in male mice
 6   exposed to 86, 231, or 585 mg/kg-day. Water consumption was further decreased by 48 and 70% in male
 7   mice exposed to 882 and 1,570 mg/kg-day, respectively. Water consumption was also decreased 31 and
 8   57% in female mice treated with 1,620 and 2,669 mg/kg-day, respectively. An increase in MCV was
 9   observed in the two highest dose groups in both male (882 and 1,570 mg/kg-day) and female mice (1,620
10   and 2,669 mg/kg-day). Increases in RBCs, hemoglobin, and hematocrit were also observed in high dose
11   males (1,570 mg/kg-day). Hematological changes were within 2-15% of control values. Serum
12   biochemistry changes in exposed mice included decreased total protein (at 1,570 mg/kg-day in males,
13   >1,620 mg/kg-day in females), decreased glucose (at 1,570 mg/kg-day in males, >1,620 mg/kg-day in
14   females), decreased albumin (at 1,570 mg/kg-day in males, 2,669 mg/ kg-day in females), decreased total
15   cholesterol (> 585 mg/kg-day in males, >1,620 mg/kg-day in females), increased serum ALT (at
16   1,570 mg/kg-day in males, > 620 mg/kg-day in females), increased AST (at 1,570 mg/kg-day in males,
17   2,669 mg/kg-day in females), increased ALP (> 585 mg/kg-day in males, 2,669 mg/kg-day in females),
18   and increased LDH (in females only at doses > 1,620 mg/kg-day). With the exception of a threefold
                                                                                                28
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1   increase in ALT in male and female mice, serum biochemistry parameters in treated rats did not differ
 2   more than twofold from control values. Urinary pH was decreased in males at > 882 mg/kg-day and in
 3   females at > 1,620 mg/kg-day.

 4           Absolute and relative lung weights were increased in males at 1,570 mg/kg-day and in females at
 5   1,620 and 2,669 mg/kg-day. Absolute kidney weights were also increased in females at 1,620 and
 6   2,669 mg/kg-day and relative kidney weight was elevated at 2,669 mg/kg-day. Histopathology findings in
 7   mice that were related to exposure included nuclear enlargement of the respiratory epithelium, nuclear
 8   enlargement of the olfactory epithelium, eosinophilic change in the olfactory epithelium, vacuolic change
 9   in the olfactory nerve, nuclear enlargement of the tracheal epithelium,  accumulation of foamy cells in the
10   lung and bronchi, nuclear enlargement and degeneration of the bronchial epithelium, hepatocyte swelling
11   of the centrilobular area of the liver, and single cell necrosis in the liver. The incidence data for
12   histopathological lesions in mice are presented in Table 4-2. Based on the changes in the bronchial
13   epithelium in female mice, the authors identified the dose level of 387 mg/kg-day as the LOAEL for
14   mice; the NOAEL was 170 mg/kg-day (Kano et al., 2008).

15
Table 4-2 Incidence of histopathological lesions in
drinking water for 13 weeks
Effect
Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Single cell necrosis; liver

Nuclear enlargement; nasal respiratory epithelium
Eosinophilic change; nasal respiratory epithelium
Nuclear enlargement; nasal olfactory epithelium
Eosinophilic change; nasal olfactory epithelium
Vacuolic change; olfactory nerve
Nuclear enlargement; tracheal epithelium
Accumulation of foamy cells; lung/bronchi
Nuclear enlargement; bronchial epithelium
Degeneration; bronchial epithelium
Hepatocyte swelling
Crj:BDF1 mice exposed to
1,4-dioxane in
Male dose (mg/kg-day)a
0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10

0
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
86
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10

170
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
1/10
231
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
Female dose
387
0/10
1/10
0/10
0/10
0/10
2/10
0/10
10/10C
0/10
1/10
585
2/10
0/10
9/1 Oc
0/10
0/10
7/1 Oc
0/10
9/1 Oc
0/10
10/10C
5/1 Ob
882
5/1 Ob
0/10
10/10C
0/10
0/10
9/1 Oc
0/10
9/1 Oc
0/10
10/10C
10/10C
1,570
0/9
5/9b
9/9c
6/9c
9/9c
9/9c
6/9c
9/9c
8/9c
9/9c
9/9c
(mg/kg-day)a
898
3/10
1/10
6/1 Ob
1/1 Oc
0/10
9/1 Oc
0/10
10/10C
0/10
10/10C
1,620
3/10
5/1 Ob
10/10C
6/1 Ob
2/10
10/10C
10/10C
10/10C
7/1 Oc
10/10C
2,669
7/1 Oc
9/1 Oc
10/10C
6/1 Ob
8/1 Oc
10/10C
10/10C
10/10C
10/10C
9/1 Ob
                                     DRAFT - DO NOT CITE OR QUOTE
                                                                                                  29

-------
      Single cell necrosis; liver
0/10
0/10
0/10
7/1 Oc
10/10C
9/1 Oc
      aData are presented for sacrificed animals.
      bp < 0.01 by x2 test.
      °p < 0.05.
      Source: Kano et al (2008).
 1   4.2.1.1.4   Yamamoto eta I.  Studies (Yamamoto et al., 1998a; Yamamoto et al., 1998b) in
 2   rasH2 transgenic mice carrying the human prototype c-Ha-ras gene have been investigated as a bioassay
 3   model for rapid carcinogenicity testing. As part of validation studies of this model, 1,4-dioxane was one
 4   of many chemicals that were evaluated. RasH2 transgenic mice were Fl offspring of transgenic male
 5   C57BLr6J and normal female BALB/cByJ mice. CB6F] mice were used as a nontransgenic control.
 6   Seven- to nine-week-old mice (10-15/group) were exposed to 0, 0.5, or 1% 1,4-dioxane in drinking water
 7   for 26 weeks. An increase in lung adenomas was observed in treated transgenic mice, as compared to
 8   treated nontransgenic mice. The tumor incidence in transgenic animals, however, was not greater than
 9   that observed in vehicle-treated transgenic mouse controls. Further study details were not provided.
      4.2.1.2  Chronic Oral Toxicity and Carcinogenicity
 1    4.2.1.2.1   A rg us et a I. Twenty-six adult male Wistar rats (Argus et al., 1965) weighing
 2    between 150 and 200 g were exposed to 1,4-dioxane (purity not reported) in the drinking water at a
 3    concentration of 1% for 64.5 weeks. A group of nine untreated rats served as control. Food and water
 4    were available ad libitum. The drinking water intake for treated animals was reported to be 30 mL/day,
 5    resulting in a dose/rat of 300 mg/day. Using a reference BW of 0.462 kg for chronic exposure to male
 6    Wistar rats (U.S. EPA. 1988). it can be estimated that these rats received daily doses of approximately
 7    640 mg/kg-day.  All animals that died or were killed during the study underwent a complete necropsy. A
 8    list of specific tissues examined microscopically was not provided; however, it is apparent that the liver,
 9    kidneys, lungs, lymphatic tissue, and spleen were examined. No statistical analysis of the results was
10    conducted.

 1           Six of the 26 treated rats developed hepatocellular carcinomas, and these rats had been treated for
 2    an average of 452 days (range, 448-455 days). No liver tumors were observed in control rats. In two rats
 3    that died after 21.5 weeks of treatment, histological  changes appeared to involve the entire liver. Groups
 4    of cells were found that had enlarged hyperchromic  nuclei. Rats that died or were killed at longer
 5    intervals showed similar changes, in addition to large cells with reduced cytoplasmic basophilia. Animals
 6    killed  after 60 weeks of treatment showed small neoplastic nodules or multifocal hepatocellular
 7    carcinomas. No  cirrhosis was observed in this study. Many rats had extensive changes in the kidneys
 8    often resembling glomerulonephritis, however, incidence data was not reported for these findings. This
 9    effect  progressed from increased cellularity to thickening of the glomerular capsule followed by
10    obliteration of the glomeruli. One treated rat had an  early transitional cell carcinoma in the kidney's
11    pelvis; this rat also had a large tumor in the liver. The lungs from many treated and control rats (incidence
12    not reported) showed severe bronchitis with epithelial hyperplasia and marked peribronchial infiltration,
13    as well as multiple abscesses. One rat treated with 1,4-dioxane developed leukemia with infiltration of all
                                                                                                   30
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    organs, particularly the liver and spleen, with large, round, isolated neoplastic cells. In the liver, the
 2    distribution of cells in the sinusoids was suggestive of myeloid leukemia. The dose of 640 mg/kg-day
 3    tested in this study was a free-standing LOAEL, identified by EPA, for glomerulonephritis in the kidney
 4    and histological changes in the liver (hepatocytes with enlarged hyperchromic nuclei, large cells with
 5    reduced cytoplasmic basophilia).

 6    4.2.1.2.2   Argus et al.; Hoch-Ligeti et al.  Five groups (28-3 2/dose group) of male
 7    Sprague Dawley rats (2-3 months of age) weighing 110-230 g at the beginning of the experiment were
 8    administered 1,4-dioxane (purity not reported) in the drinking water for up to 13 months at concentrations
 9    of 0, 0.75, 1.0, 1.4, or 1.8% (Argus etal., 1973; Hoch-Ligeti et al.. 1970). The drinking water intake was
10    determined for each group over a 3-day measurement period conducted at the beginning of the study and
11    twice during the study (weeks were not specified).  The rats were killed with ether at 16 months or earlier
12    if nasal tumors were clearly observable. Complete  autopsies were apparently performed on all animals,
13    but only data from the nasal cavity and liver were presented and discussed. The nasal cavity was studied
14    histologically only from rats in which gross tumors in these locations were present; therefore, early
15    tumors may have been missed and pre-neoplastic changes were not studied. No statistical analysis of the
16    results was conducted. Assuming a BW of 0.523 kg for an adult male Sprague Dawley rat (U.S. EPA.
17    1988) and a drinking water intake of 30 mL/day as reported by the study authors, dose estimates were 0,
18    430, 574, 803, and 1,032 mg/kg-day. The progression of liver tumorigenesis was evaluated by an
19    additional group of 10 male rats administered 1% 1,4-dioxane in the drinking water (574 mg/kg-day), 5 of
20    which were sacrificed after 8 months of treatment and 5 were sacrificed after 13 months of treatment.
21    Liver tissue from these rats and control  rats was processed for electron microscopy examination.

 6           Nasal cavity tumors were observed upon gross examination in six rats (1/30 in the 0.75% group,
 7    1/30 in the 1.0% group, 2/30 in the 1.4% group, and 2/30 in the 1.8% group). Gross observation showed
 8    the tumors visible either at the tip of the nose, bulging out of the nasal cavity, or on the back of the nose
 9    covered by intact or later ulcerated skin. As the tumors obstructed the nasal passages, the rats had
10    difficulty breathing and lost weight rapidly. No neurological signs or compression of the brain were
11    observed. In all cases, the tumors were squamous cell carcinomas with marked keratinization and
12    formation of keratin pearls. Bony structure was extensively destroyed in some animals with  tumors, but
13    there was no invasion into the brain. In addition to  the squamous carcinoma, two adenocarcinomatous
14    areas were present.  One control rat had  a small, firm, well-circumscribed tumor on the back  of the nose,
15    which proved to be subcutaneous fibroma. The latency period for tumor onset was 329-487  days.
16    Evaluation of the latent periods and doses received did not suggest an inverse relationship between these
17    two parameters.

18           Argus et al. (1973) studied the progression of liver tumorigenesis by electron microscopy of liver
19    tissues obtained following interim sacrifice at 8 and 13 months of exposure (5 rats/group,
20    574 mg/kg-day). The first change observed in the liver was an increase in the size of the nucleus of the
21    hepatocytes, mostly in the periportal area. Precancerous changes were characterized by disorganization of
22    the rough endoplasmic reticulum, an increase in smooth endoplasmic reticulum, and a decrease in
23    glycogen and increase in lipid droplets in hepatocytes. These changes increased in severity in the
24    hepatocellular carcinomas in rats exposed to 1,4-dioxane for 13 months.
                                                                                                   31
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1           Three types of liver nodules were observed in exposed rats at 13-16 months. The first consisted
 2    of groups of cells with reduced cytoplasmic basophilia and a slightly nodular appearance as viewed by
 3    light microscopy. The second type of circumscribed nodule was described consisting of large cells,
 4    apparently filled and distended with fat. The third type of nodule was described as finger-like strands, 2-
 5    3 cells thick, of smaller hepatocytes with large hyperchromic nuclei and dense cytoplasm. This third type
 6    of nodule was designated as an incipient hepatoma, since it showed all the histological characteristics of a
 7    fully developed hepatoma. All three types of nodules were generally present in the same liver. Cirrhosis
 8    of the liver was not observed. The numbers of incipient liver tumors and hepatomas in rats from this study
 9    (treated for 13 months and observed at 13-16 months) are presented in Table 4-3.
      Table 4-3    Number of incipient liver tumors and hepatomas in male Sprague-Dawley rats
                  exposed to 1,4-dioxane in drinking water for 13 months
     	Dose (mg/kg-day)	Incipient tumors	Hepatomas	Total	
     	430	4	0	4
     	574	9	0	9
     	803	13	3	16
     	1,032	11	12	23
      aPrecise incidences cannot be calculated since the number of rats per group was reported as 28-32; incidence in control rats was
      not reported; no statistical analysis of the results was conducted in the study.
      Source: Argus et al. (1973).

10           Treatment with all dose levels of 1,4-dioxane induced marked kidney alterations, but quantitative
11    incidence data were not provided. Qualitatively, the changes indicated glomerulonephritis and
12    pyelonephritis, with characteristic epithelial proliferation of Bowman's capsule, periglomerular fibrosis,
13    and distension of tubules. No kidney tumors were found. No tumors were found in the lungs. One rat at
14    the  1.4% treatment level showed early peripheral adenomatous change of the alveolar epithelium and
15    another rat in the same group showed papillary hyperplasia of the bronchial epithelium. The lowest dose
16    tested (430 mg/kg-day) was considered a LOAEL by EPA for hepatic and renal effects in this study.

17    4.2.1.2.3   Hoch-Ligeti and Argus. Hoch-Ligeti and Argus (1970) provided a brief
18    account of the results of exposure of guinea pigs to 1,4-dioxane. A group of 22 male guinea pigs (neither
19    strain nor age provided) was administered 1,4-dioxane (purity not provided) in the drinking water for at
20    least 23 months and possibly up to 28 months. The authors stated that the concentration of 1,4-dioxane
21    was regulated so that normal growth of the guinea pigs was maintained, and varied 0.5-2% (no further
22    information provided). The investigators further stated that the amount of 1,4-dioxane received by the
23    guinea pigs over a 23-month period was 588-635 g. Using a reference BW of 0.89 kg for male guinea
24    pigs in a chronic study (U.S. EPA. 1988) and assuming an exposure period of 700 days (23 months), the
25    guinea pigs received doses between 944 and 1,019 mg 1,4-dioxane/kg-day. A group often untreated
26    guinea pigs served as controls. All animals were sacrificed within 28 months, but the scope of the
27    postmortem examination was not provided.

17           Nine treated guinea pigs showed peri- or intrabronchial epithelial hyperplasia and nodular
18    mononuclear infiltration in the lungs. Also, two guinea pigs had carcinoma of the gallbladder, three had

                                                                                                   32
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    early hepatomas, and one had an adenoma of the kidney. Among the controls, four guinea pigs had
 2    peripheral mononuclear cell accumulation in the lungs, and only one had hyperplasia of the bronchial
 3    epithelium. One control had formation of bone in the bronchus. No further information was presented in
 4    the brief narrative of this study. Given the limited reporting of the results, a NOAEL or LOAEL value
 5    was not provided for this study.

 6    4.2.1.2.4   Kociba et al. Groups of 6-8-week-old Sherman rats (60/sex/dose level) were
 7    administered 1,4-dioxane (purity not reported) in the drinking water at levels of 0 (controls), 0.01, 0.1, or
 8    1.0% for up to 716 days (Kociba et al.. 1974). The drinking water was prepared twice weekly during the
 9    first year of the study and weekly during the second year of the study. Water samples were collected
10    periodically and analyzed for 1,4-dioxane content by routine gas liquid chromatography. Food and water
11    were available ad libitum. Rats were observed daily for clinical signs of toxicity, and BWs were measured
12    twice weekly during the first month, weekly during months 2-7, and biweekly thereafter. Water
13    consumption was recorded at three different time periods during the study: days 1-113, 114-198, and
14    446-460. Blood samples were collected from a minimum of five male and five female control and
15    high-dose rats during the 4th, 6th,  12th, and 18th months of the study and at termination. Each sample
16    was analyzed for packed cell volume, total erythrocyte count, hemoglobin, and total and differential WBC
17    counts. Additional endpoints evaluated included organ weights (brain, liver, kidney, testes, spleen, and
18    heart) and gross and microscopic examination of major tissues and organs (brain, bone and bone marrow,
19    ovaries, pituitary, uterus, mesenteric lymph nodes, heart, liver, pancreas, spleen, stomach, prostate, colon,
20    trachea, duodenum, kidneys, esophagus, jejunum, testes, lungs, spinal cord, adrenals, thyroid,
21    parathyroid, nasal turbinates, and urinary bladder). The number of rats with tumors, hepatic tumors,
22    hepatocellular carcinomas, and nasal carcinomas were analyzed for statistical significance with Fisher's
23    Exact test (one-tailed), comparing each treatment group against the respective control group. Survival
24    rates were compared using %2 Contingency Tables and Fisher's Exact test. Student's test was used to
25    compare hematological parameters, body and organ weights, and water consumption of each treatment
26    group with the respective control group.

 6           Male and female rats in the high-dose group (1% in drinking water) consumed slightly less water
 7    than controls. BW gain was depressed in the high-dose groups relative to the other groups almost from
 8    the beginning of the study (food consumption data were not provided). Based on water consumption and
 9    BW data for specific exposure groups, Kociba et al. (1974) calculated mean daily doses of 9.6, 94, and
10    1,015 mg/kg-day for male rats and 19, 148, and 1,599 mg/kg-day for female rats during days  114-198 for
11    the 0.01, 0.1, and 1.0% concentration levels, respectively. Treatment with 1,4-dioxane significantly
12    increased mortality among high-dose males and females beginning at about 2-4 months of treatment.
13    These rats showed degenerative changes in both the liver and kidneys. From the 5th month on, mortality
14    rates of control and treated groups were not different. There were no treatment-related alterations in
15    hematological parameters. At termination, the only alteration in organ weights noted by the authors was a
16    significant increase in absolute and relative liver weights in male and female high-dose rats (data not
17    shown). Fiistopathological lesions were restricted to the liver and kidney from the mid- and high-dose
18    groups  and consisted of variable degrees of renal tubular epithelial and hepatocellular degeneration and
19    necrosis (no quantitative incidence data were provided). Rats from these groups also showed evidence of

                                                                                                   33
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    hepatic regeneration, as indicated by hepatocellular hyperplastic nodule formation and evidence of renal
 2    tubular epithelial regenerative activity (observed after 2 years of exposure). These changes were not seen
 3    in controls or in low-dose rats. The authors determined a LOAEL of 94 mg/kg-day based on the liver and
 4    kidney effects in male rats. The corresponding NOAEL value was 9.6 mg/kg-day.

 5           Histopathological examination of all the rats in the study revealed a total of 132 tumors in
 6    114 rats. Treatment with 1%  1,4-dioxane in the drinking water resulted in a significant increase in the
 7    incidence of hepatic tumors (hepatocellular carcinomas in six males and four females). In addition, nasal
 8    carcinomas (squamous cell carcinoma of the  nasal turbinates) occurred in one high-dose male and two
 9    high-dose females. Since 128 out of 132 tumors occurred in rats from the  12th to the 24th month, Kociba
10    et al. (1974) assumed that the effective number of rats was the number surviving at 12 months, which was
11    also when the first hepatic tumor was noticed. The incidences of liver and nasal tumors from Kociba et al.
12    (1974) are presented in Table 4-4. Tumors in other organs were not elevated when compared to control
13    incidence and did not appear to be related to  1,4-dioxane administration.
      Table 4-4   Incidence of liver and nasal tumors in male and female Sherman rats (combined)
                  treated with 1,4-dioxane in the drinking water for 2 years
Dose in mg/kg-day
(average of male and
female dose)
0
14
121
1307
Effective
number of
animals3
106
110
106
66
Number of
tumor-bearing
animals
31
34
28
21
Number of animals
Hepatic tumors
(all types)
2
0
1
12°
Hepatocellular
carcinomas
1
0
1
10C
Nasal
carcinomas
0
0
0
3a
      "Rats surviving until 12 months on study.
      bp = 0.00022 by one-tailed Fisher's Exact test.
      °p = 0.00033 by one-tailed Fisher's Exact test.
      dp = 0.05491 by one-tailed Fisher's Exact test.
      Source: Reprinted with permission of Elsevier, Ltd., Kociba et al. (1974).
14           The high-dose level was the only dose that increased the formation of liver tumors over control
15    (males 1,015 mg/kg-day; females 1,599 mg/kg-day) and also caused significant liver and kidney toxicity
16    in these animals. The mid-dose group (males 94 mg/kg-day; females  148 mg/kg-day) experienced hepatic
17    and renal degeneration and necrosis, as well as regenerative proliferation in hepatocytes and renal tubule
18    epithelial cells. No increase in tumor formation was seen in the mid-dose group. No toxicity or tumor
19    formation was observed in either sex in the low-dose (males 9.6 mg/kg-day; females 19 mg/kg-day) group
20    of rats.
                                                                                                    34
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1   4.2.1.2.5   National Cancer Institute (NCI). Groups of Osborne-Mendel rats
 2   (35/sex/dose) and B6C3F] mice (50/sex/dose) were administered 1,4-dioxane (> 99.95% pure) in the
 3   drinking water for 110 or 90 weeks, respectively, at levels of 0 (matched controls), 0.5, or 1% (NCI.
 4   1978). Solutions of 1,4-dioxane were prepared with tap water. The report indicated that at 105 weeks
 5   from the earliest starting date, a new necropsy protocol was instituted. This affected the male controls and
 6   high-dose rats, which were started a year later than the original groups of rats and mice. Food and water
 7   were available ad libitum. Endpoints monitored in this bioassay included clinical signs (twice daily), BWs
 8   (once every 2 weeks for the first 12 weeks and every month during the rest of the study), food and water
 9   consumption (once per month in 20% of the animals in each group during the second year of the study),
10   and gross and microscopic appearance of all major organs and tissues (mammary gland, trachea, lungs
11   and bronchi, heart, bone marrow, liver, bile duct, spleen, thymus, lymph nodes, salivary gland, pancreas,
12   kidney, esophagus, thyroid, parathyroid, adrenal, gonads, brain, spinal cord, sciatic nerve, skeletal muscle,
13   stomach, duodenum, colon, urinary bladder, nasal septum, and skin). Based on the measurements of water
14   consumption and BWs, the investigators calculated average daily intakes of 1,4-dioxane of 0, 240, and
15   530 mg/kg-day in male rats, 0, 350, and 640 mg/kg-day in female rats, 0, 720, and 830 mg/kg-day in male
16   mice, and 0, 380, and 860 mg/kg-day in female mice. According to the report, the doses of 1,4-dioxane in
17   high-dose male mice were only slightly higher than those of the low-dose group due to decreased fluid
18   consumption in high-dose male mice.

 1           During the second year of the study, the  BWs of high-dose rats were lower than controls, those of
 2   low-dose males were higher than controls, and those of low-dose females were comparable to controls.
 3   The fluctuations in the growth curves were attributed to mortality by the investigators; quantitative
 4   analysis of BW changes was not done. Mortality was significantly increased in treated rats, beginning at
 5   approximately 1 year of study. Analysis of Kaplan-Meier curves (plots of the statistical estimates of the
 6   survival probability function)  revealed significant positive dose-related trends (p < 0.001, Tarone test). In
 7   male rats, 33/35 (94%) in the control group, 26/35 (74%) in the mid-dose group, and 33/35 (94%) in the
 8   high-dose group were alive on week 52  of the study. The corresponding numbers for females were 35/35
 9   (100%), 30/35 (86%), and 29/35 (83%). Nonneoplastic lesions associated with treatment with 1,4-dioxane
10   were seen in the kidneys (males and females), liver (females only), and stomach (males only). Kidney
11   lesions consisted of vacuolar degeneration and/or focal tubular epithelial regeneration in the proximal
12   cortical tubules and occasional hyaline casts. Elevated incidence of hepatocytomegaly also occurred in
13   treated female rats. Gastric ulcers occurred in treated males, but none were seen in controls. The
14   incidence of pneumonia was increased above controls in high-dose female rats. The incidence of
15   nonneoplastic lesions in rats following drinking water exposure to 1,4-dioxane is presented in Table 4-5.
16   EPA identified the LOAEL in rats from this study as 240 mg/kg-day for increased incidence of gastric
17   ulcer and cortical tubular degeneration in the kidney in males; a NOAEL was not established.
                                                                                                   35
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      Table 4-5    Incidence of nonneoplastic lesions in Osborne-Mendel rats exposed to 1,4-dioxane in
                  drinking water
Males (mg/kg-day)

Cortical tubule degeneration
Hepatocytomegaly
Gastric ulcer
Pneumonia
0
0/3 1a
5/31
(16%)
0/30a
8/30
(27%)
240
20/3 1b
(65%)
3/32
(9%)
5/28b
(18%)
15/31
(48%)
530
27/33b
(82%)
11/33
(33%)
5/30b
(17%)
14/33
(42%)
Females (mg/kg-day)
0
0/3 1a
7/3 1a
(23%)
0/31
6/30a
(20%)
350
0/34
11/33
(33%)
1/33
(3%)
5/34
(15%)
640
10/32b
(31%)
17/32b
(53%)
1/30
(3%)
25/32b
(78%)
      Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this review.
      "Incidence significantly elevated compared to control by Fisher's Exact test (p < 0.05) performed for this review.
      Source: NCI (1978).

 1           Neoplasms associated with 1,4-dioxane treatment were limited to the nasal cavity (squamous cell
 2    carcinomas, adenocarcinomas, and one rhabdomyoma) in both sexes, liver (hepatocellular adenomas) in
 3    females, and testis/epididymis (mesotheliomas) in males. The first tumors were seen at week 52 in males
 4    and week 66 in females. The incidence of squamous cell carcinomas in the nasal turbinates in male and
 5    female rats is presented in Table 4-6. Squamous cell carcinomas were first seen on week 66 of the study.
 6    Morphologically, these tumors varied from minimal foci of locally invasive squamous cell proliferation to
 7    advanced growths consisting of extensive columns of epithelial cells projecting either into free spaces of
 8    the nasal cavity and/or infiltrating into the submucosa. Adenocarcinomas of the nasal cavity were
 9    observed in 3 of 34 high-dose male rats,  1 of 35 low-dose female rats, and 1 of 35 high-dose female rats.
10    The single rhabdomyoma (benign skeletal muscle tumor) was observed in the nasal cavity of a male rat
11    from the low-dose group. A subsequent re-examination of the nasal tissue sections by Goldsworthy et al.
12    (1991) concluded that the location of the tumors in the nasal apparatus was consistent with the possibility
13    that the nasal tumors resulted from inhalation of water droplets by the rats (see Section 4.5.2 for more
14    discussion of Goldsworthy et al. (1991)).
                                                                                                    36
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      Table 4-6   Incidence of nasal cavity squamous cell carcinoma and liver hepatocellular adenoma
                  in Osborne-Mendel rats exposed to 1,4-dioxane in drinking water
Males (mg/kg-day)a

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

Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
0
0/34 (0%)°
0/31 (0%)T
350
10/35(29%)e
10/33(30%)e
640
8/35 (23%)c
11/32(34%)e
      aTumor incidence values were not adjusted for mortality.
      bGroup not included in statistical analysis by NCI because the dose group was started a year earlier without
         appropriate controls.
      °p < 0.003 by Fisher's Exact test pair-wise comparison with controls.
      dp = 0.008 by Cochran-Armitage test.
      ep £ 0.001 by Fisher's Exact test pair-wise comparison with controls.
      fp = 0.001 by Cochran-Armitage test.
      Source: NCI (1978).

 1           The incidence of hepatocellular adenomas in male and female rats is presented in Table 4-6.
 2    Hepatocellular adenomas were first observed in high-dose females in week 70 of the study. These tumors
 3    consisted of proliferating hepatic cells oriented as concentric cords. Hepatic cell size was variable;
 4    mitoses and necrosis were rare. Mesothelioma of the vaginal tunics of the testis/epididymis was seen in
 5    male rats (2/33, 4/33, and 5/34 in controls, low-,  and high-dose animals, respectively). The difference
 6    between the treated groups and controls was not statistically significant. These tumors were characterized
 7    as rounded and papillary projections  of mesothelial cells, each supported by a core of fibrous tissue. Other
 8    reported neoplasms were considered  spontaneous lesions not related to treatment with 1,4-dioxane.

 9           In mice, mean BWs of high-dose female mice were lower than controls during the second year of
10    the study, while those of low-dose females were higher than controls. In males, mean BWs of high-dose
11    animals were higher than controls during the second year of the study. According to the investigators,
12    these fluctuations could have been due to mortality; no quantitative analysis of BWs was done. No other
13    clinical signs were reported. Mortality was significantly increased in female mice (p < 0.001, Tarone test),
14    beginning at approximately 80 weeks on study. The numbers of female mice that survived to 91 weeks
15    were 45/50 (90%) in the control group, 39/50 (78%) in the low-dose group, and 28/50 (56%) in the
16    high-dose group. In males, at least 90% of the mice in each group were still alive at week 91.
17    Nonneoplastic lesions that increased  significantly due to treatment with 1,4-dioxane were pneumonia in
18    males and females  and rhinitis in females. The incidences of pneumonia were 1/49 (2%), 9/50 (18%), and
19    17/47 (36%) in control, low-dose, and high-dose males, respectively; the corresponding incidences in
20    females were 2/50 (4%), 33/47 (70%), and 32/36 (89%). The incidences of rhinitis in female mice were
21    0/50, 7/48 (14%), and 8/39 (21%) in  control, low-dose, and high-dose groups, respectively. Pair-wise
22    comparisons of low-dose and high-dose incidences with  controls  for incidences of pneumonia and rhinitis
23    in females using Fisher's Exact test (done for this review) yielded p-values < 0.001 in all cases.
24    Incidences of other lesions were considered to be similar to those seen in aging mice. The authors stated
25    that hepatocytomegaly was commonly found in dosed mice, but the incidences were not significantly
26    different from controls and showed no dose-response trend. EPA  concluded the LOAEL for 1,4-dioxane

                                                                                                    37
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    in mice was 380 mg/kg-day based on the increased incidence of pneumonia and rhinitis in female mice; a
 2    NOAEL was not established in this study.
 3           As shown in Table 4-7, treatment with 1,4-dioxane significantly increased the incidence of
 4    hepatocellular carcinomas or adenomas in male and female mice in a dose-related manner. Tumors were
 5    first observed on week 81 in high-dose females and in week 58 in high-dose males. Tumors were
 6    characterized by parenchymal cells of irregular size and arrangement, and were often hypertrophic with
 7    hyperchromatic nuclei. Mitoses were seldom seen. Neoplasms were locally invasive within the liver, but
 8    metastasis to the lungs was rarely observed.
      Table 4-7    Incidence of hepatocellular adenoma or carcinoma in B6C3F1 mice exposed to
                  1,4-dioxane in drinking water
Males (mg/kg-day)a

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

Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
0
0/50 (0%)c
0/50 (0%)D
380
12/48(25%)c
21/48 (44%)c
860
29/37 (78%)c
35/37 (95%)c
      aTumor incidence values were not adjusted for mortality.
      bp < 0.001, positive dose-related trend (Cochran-Armitage test).
      °p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
      dp = 0.014.
      Source: NCI (1978).

 9           In addition to liver tumors, a variety of other benign and malignant neoplasms occurred.
10    However, the report (NCI. 1978) indicated that each type had been encountered previously as a
11    spontaneous lesion in the B6C3FJ mouse. The report further stated that the incidences of these neoplasms
12    were unrelated by type, site, group, or sex of the animal, and hence, not attributable to exposure to
13    1,4-dioxane. There were a few nasal adenocarcinomas (1/48 in low-dose females and 1/49 in high-dose
14    males) that arose from proliferating respiratory epithelium lining of the nasal turbinates. These growths
15    extended into the nasal cavity, but there was minimal local tissue infiltration. Nasal mucosal polyps were
16    rarely observed. The polyps were derived from mucus-secreting epithelium and were otherwise
17    unremarkable. There was a significant negative trend for alveolar/bronchiolar adenomas or carcinomas of
18    the lung in male mice, such that the incidence in the matched controls was higher than in the dosed
19    groups. The report (NCI. 1978) indicated that the probable reason for this occurrence was that the dosed
20    animals did not live as long as the controls, thus diminishing the possibility of the development of tumors
21    in the dosed groups.
                                                                                                   38
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    4.2.1.2.6   Kano et al.; Japan Bioassay Research Center; Yamazaki et al. The
 2    Japan Bioassay Research Center (JBRC) conducted a 2-year drinking water study determining the effects
 3    of 1,4-dioxane on both sexes of rats and mice. The study results have been reported several times: once as
 4    conference proceedings (Yamazaki et al.. 1994). once as a laboratory report (JBRC. 1998). and most
 5    recently as a peer-reviewed manuscript (Kano et al.. 2009). Dr. Yamazaki also provided some detailed
 6    information (Yamazaki. 2006). Variations in the data between these three reports were noted and
 7    included:  (1) the level of detail on dose information reported; (2) categories for incidence data reported
 8    (e.g., all animals or sacrificed animals); and (3) analysis of non- and neoplastic lesions.

 1           The 1,4-dioxane dose information provided in the reports varied. Specifically, Yamazaki et al.
 2    (1994) only included drinking water concentrations for each dose group. In contrast, JBRC (1998)
 3    included drinking water concentrations (ppm), in addition using body weights and water consumption
 4    measurements to calculate daily chemical intake (mg/kg-day). JBRC (1998) reported daily chemical
 5    intake for each dose group as a range. Thus, for the External Peer Review draft of this  Toxicological
 6    Review ofl,4-Dioxane (U.S. EPA. 2009b). the midpoint of the range was used. Kano et al. (2009) also
 7    reported a calculation of daily chemical intake based on body weight and water consumption
 8    measurements; however, for each dose group they reported a mean and standard deviation estimate.
 9    Therefore, because the mean more accurately represents the delivered dose than the midpoint of a range,
10    the Kano et al. (2009) calculated mean chemical intake (mg/kg-day) is used for quantitative analysis of
11    this data.

12           The categories for which incidence rates were described also varied among the reports. Yamazaki
13    et al. (1994) and Kano et al. (2009) reported histopathological results for all animals, including dead and
14    moribund animals; however, the detailed JBRC laboratory findings (1998) included separate incidence
15    reports for dead and moribund animals, sacrificed animals, and all animals.

16           Finally, the criteria used to evaluate some of the data were updated when JBRC published the
17    most recent manuscript by Kano et al. (2009). The manuscript by Kano et al. (2009) stated that the lesions
18    diagnosed in the earlier reports (JBRC. 1998; Yamazaki et al.. 1994) were re-examined and recategorized
19    as appropriate according to current pathological diagnostic criteria (see references in Kano et al. (2009)).

20           Groups of F344/DuCrj rats (50/sex/dose level) were exposed to 1,4-dioxane (>99% pure) in the
21    drinking water at levels of 0, 200, 1,000, or 5,000 ppm for 2 years. Groups of Crj:BDFl mice
22    (50/sex/dose level) were similarly exposed in the drinking water to 0, 500, 2,000, or 8,000 ppm of
23    1,4-dioxane. The high doses were selected based on results from the Kano et al. (2008) 13-week drinking
24    water study so as not to exceed the maximum tolerated dose (MTD) in that study. Both rats and mice
25    were 6 weeks old at the beginning of the study. Food and water were available ad libitum. The animals
26    were observed daily for clinical signs of toxicity; and BWs were measured once per week for 14 weeks
27    and once every 2 weeks until the end of the study. Food consumption was measured once a week for
28    14 weeks and once every 4 weeks for the remainder of the study. The investigators used data from water
29    consumption and BW to calculate an estimate of the daily intake of 1,4-dioxane (mg/kg-day) by male and
30    female rats and mice. Kano et al. (2009) reported a calculated mean ± standard deviation for the daily
31    doses of 1,4-dioxane for the duration of the study. Male rats received doses of approximately 0, 11±1,

                                                                                                  39
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    55±3, or 274±18 mg/kg-day and female rats received 0, 18±3, 83±14, or 429±69 mg/kg-day. Male mice
 2    received doses of 0, 49±5, 191±21, or 677±74 mg/kg-day and female mice received 0, 66±10, 278±40, or
 3    964±88 mg/kg-day. For the remainder of this document, including the dose-response analysis, the mean
 4    calculated intake values are used to identify dose groups. The Kano et al. (2009) study was conducted in
 5    accordance with the Organization for Economic Co-operation and Development (OECD) Principles for
 6    Good Laboratory Practice (GLP).

 7           No information was provided as to when urine samples were collected. Blood samples were
 8    collected only at the end of the 2-year study (Yamazaki. 2006).  Hematology analysis included RBCs,
 9    hemoglobin, hematocrit, MCV, platelets, WBCs and differential WBCs. Serum biochemistry included
10    total protein, albumin, bilirubin, glucose, cholesterol, triglyceride (rat only), phospholipid, ALT, AST,
11    LDH, LAP, ALP, y-glutamyl transpeptidase (GGT), CPK, urea  nitrogen, creatinine (rat only), sodium,
12    potassium, chloride, calcium, and inorganic phosphorous. Urinalysis parameters were pH, protein,
13    glucose, ketone body, bilirubin (rat only), occult blood, and urobilinogen. Organ weights (brain, lung,
14    liver, spleen, heart, adrenal, testis, ovary, and thymus) were measured, and gross necropsy and
15    histopathologic examination of tissues and organs were performed on all animals (skin, nasal cavity,
16    trachea, lungs, bone marrow, lymph nodes, thymus, spleen, heart, tongue, salivary glands, esophagus,
17    stomach, small and large  intestine, liver, pancreas, kidney,  urinary bladder, pituitary, thyroid, adrenal,
18    testes, epididymis, seminal vesicle, prostate, ovary, uterus,  vagina, mammary gland, brain, spinal cord,
19    sciatic nerve, eye, Harderian gland, muscle, bone, and parathyroid). Dunnett's test and %2 test were used to
20    assess the statistical significance of changes in continuous and discrete variables, respectively.

21           For rats, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
22    study. Both male and female rats in the high dose groups (274 and 429 mg/kg-day, respectively) exhibited
23    slower growth rates and terminal body weights that were significantly different (p < 0.05) compared to
24    controls. A statistically significant reduction in terminal BWs was observed in high-dose male rats (5%, p
25    < 0.01) and in high-dose female rats  (18%,  p < 0.01) (Kanoetal.. 2009). Food consumption was not
26    significantly affected by treatment in male or female rats; however, water consumption in female rats
27    administered 18 mg/kg-day was significantly greater (p < 0.05)  .

28           All  control and exposed rats  lived at least 12 months following study initiation (Yamazaki. 2006);
29    however, survival at the end of the 2-year study in the high dose group of male and female rats (274 and
30    429 mg/kg-day, respectively) was approximately 50%, which was significantly different compared to
31    controls. The investigators attributed these early deaths to the increased incidence in nasal tumors and
32    peritoneal mesotheliomas in male rats and nasal and hepatic tumors in female rats. (Yamazaki. 2006).

33           Several hematological  changes were noted in the JBRC report (1998): Decreases in RBC (male
34    rats only), hemoglobin, hematocrit, and MCV; and increases in  platelets in high-dose groups were
35    observed (JBRC.  1998). These changes (except for MCV) also occurred in mid-dose males. With the
36    exception of a 23% decrease in hemoglobin in high-dose male rats and a 27% increase in platelets in
37    high-dose female rats, hematological changes were within  15% of control values. Significant changes in
38    serum chemistry parameters occurred only in high-dose rats (males:  increased phospholipids, AST, ALT,
39    LDH, ALP, GGT, CPK, potassium, and inorganic phosphorus and decreased total protein, albumin, and

                                                                                                  40
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    glucose; females: increased total bilirubin, cholesterol, phospholipids, AST, ALT, LDH, GOT, ALP,
 2    CPK, and potassium, and decreased blood glucose) (JBRC. 1998). Increases in serum enzyme activities
 3    ranged from <2- to 17-fold above control values, with the largest increases seen for ALT, AST, and GGT.
 4    Urine pH was significantly decreased at 274 mg/kg-day in male rats (not tested at other dose levels) and
 5    at 83 and 429 mg/kg-day in female rats (JBRC. 1998). Also, blood in the urine was seen in female rats at
 6    83 and 429 mg/kg-day (JBRC. 1998). In male rats, relative liver weights were increased at 55 and
 7    274 mg/kg-day (Kano et al., 2009). In female rats, relative liver weight was increased at 429 mg/kg-day
 8    (Kano etal.. 2009).

 9           Microscopic examination of the tissues showed nonneoplastic alterations in the nasal cavity, liver,
10    and kidneys mainly in high-dose rats and, in a few cases, in mid-dose rats (Table 4-8 and Table 4-9).
11    Alterations in high-dose (274 mg/kg-day) male rats consisted of nuclear enlargement and metaplasia of
12    the olfactory and respiratory epithelia, atrophy of the olfactory epithelium, hydropic changes and sclerosis
13    of the lamina propria, adhesion, and inflammation. In female rats, nuclear enlargement of the olfactory
14    epithelium occurred at doses > 83 mg/kg-day, and nuclear enlargement and metaplasia of the respiratory
15    epithelium, squamous cell hyperplasia, respiratory metaplasia of the olfactory epithelium, hydropic
16    changes and sclerosis of the lamina propria, adhesion, inflammation, and proliferation of the nasal gland
17    occurred at 429 mg/kg-day. Alterations were  seen in the liver at > 55 mg/kg-day in male rats (spongiosis
18    hepatis, hyperplasia, and clear and mixed cell foci) and at 429 mg/kg-day in female rats (hyperplasia,
19    spongiosis hepatis, cyst formation, and mixed cell foci). Nuclear enlargement of the renal proximal tubule
20    occurred in males at 274 mg/kg-day and in females at > 83 mg/kg-day (JBRC. 1998).
                                                                                                   41
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-8    Incidence of histopathological lesions in male F344/DuCrj rats exposed to 1,4-dioxane
             in drinking water for 2 years
Dose (mg/kg-day)a'D

Nuclear enlargement; nasal respiratory epithelium0
Squamous cell metaplasia; nasal respiratory epithelium0
Squamous cell hyperplasia; nasal respiratory epithelium0
Nuclear enlargement; nasal olfactory epithelium0
Respiratory metaplasia; nasal olfactory epithelium0
Atrophy; nasal olfactory epithelium0
Hydropic change; lamina propria0
Sclerosis; lamina propria0
Adhesion; nasal cavity0
Inflammation; nasal cavity0
Hyperplasia; liver0
Spongiosis hepatis; liver0
Clear cell foci; liver0
Acidophilic cell foci; liver0
Basophilic cell foci; liver0
Mixed-cell foci; liver0
Nuclear enlargement; kidney proximal tubule0
0
0/50
0/50
0/50
0/50
12/50
0/50
0/50
0/50
0/50
0/50
3/50
12/50
3/50
12/50
7/50
2/50
0/50
11
0/50
0/50
0/50
0/50
11/50
0/50
0/50
0/50
0/50
0/50
2/50
20/50
3/50
8/50
11/50
8/50
0/50
55
0/50
0/50
0/50
5/50'
20/50
0/50
0/50
1/50
0/50
0/50
10/50
25/50'
9/50
7/50
8/50
14/508
0/50
274
26/508
31/508
2/50
38/508
43/50
36/50
46/50
44/50
48/50
13/50
24/50
40/50
8/50
5/50
16/50'
13/508
50/50
"Data presented for all animals, including animals that became moribund or died before the end of the study.
"Dose levels from Kano et al. (2009).
°Data from Kano et al. (2009).
dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
ep < 0.01 by Y2 test.
fp < 0.05 by x test.

Sources: Kano et al. (2009) and JBRC (1998).
Table 4-9    Incidence of histopathological lesions in female F344/DuCrj rats exposed to
             1,4-dioxane in drinking water for 2 years
Dose (mg/kg-day)a'D

Nuclear enlargement; nasal respiratory epithelium0
Squamous cell metaplasia; nasal respiratory epithelium0
Squamous cell hyperplasia; nasal cavity0
Nuclear enlargement; nasal olfactory epithelium °
Respiratory metaplasia; nasal olfactory epithelium0
Atrophy; nasal olfactory epithelium0
Hydropic change; lamina propria0
Sclerosis; lamina propria0
Adhesion; nasal cavity0
Inflammation; nasal cavity0
Proliferation; nasal gland0
Hyperplasia; liver0
Spongiosis hepatis; liver0
Cyst formation; liver0
Acidophilic cell foci; liver0
Basophilic cell foci; liver0
Clear cell foci; liver0
Mixed-cell foci; liver0
Nuclear enlargement; kidney proximal tubule0
0
0/50
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
0/50
0/50
3/50
0/50
0/50
1/50
23/50
1/50
1/50
0/50
18
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
2/50
0/50
1/50
1/50
27/50
1/50
1/50
0/50
83
0/50
0/50
0/50
28/508
2/50
1/50
0/50
0/50
0/50
1/50
0/50
11 /508
1/50
1/50
1/50
31/50
5/50
3/50
6/50
429
13/508
35/50 8
5/50
39/50
42/50
40/50
46/50
48/50
46/50
15/50
11/50
47/50
20/50
8/50
1/50
8/508
4/50
11/50'
39/50
                                 DRAFT - DO NOT CITE OR QUOTE
                                                                                                  42

-------
      aData presented for all animals, including animals that became moribund or died before the end of the study.
      "Dose levels from Kano et al. (2009).
      °Data from Kano et al. (2009).
      dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
      ep < 0.01 by Y2 test.
      fp < 0.05 by x test.
      Sources: Kano et al. (2009) and JBRC (1998).

 1           NOAEL and LOAEL values for rats in this study were identified by EPA as 55 and
 2    274 mg/kg-day, respectively, based on toxicity observed in nasal tissue of male rats (i.e., atrophy of
 3    olfactory epithelium, adhesion, and inflammation). Metaplasia and hyperplasia of the nasal epithelium
 4    were also observed in high-dose male and female rats. These effects are likely to be associated with the
 5    formation of nasal cavity tumors in these dose groups. Nuclear enlargement was observed in the nasal
 6    olfactory epithelium and the kidney proximal tubule at a dose of 83 mg/kg-day in female rats; however, it
 7    is unclear whether these  alterations represent adverse toxicological effects.  Hematological effects noted in
 8    male rats given 55 and 274 mg/kg-day (decreased RBCs, hemoglobin, hematocrit, increased platelets)
 9    were within 20% of control values. In female rats decreases in hematological effects were observed in the
10    high dose group (429 mg/kg-day). A reference range database for hematological effects in laboratory
11    animals (Wolford et al..  1986) indicates that a 20% change in these parameters may fall within a normal
12    range (10th-90th percentile values) and may not represent a treatment-related effect of concern. Liver
13    lesions were also seen at a dose of 55 mg/kg-day in male rats; these changes are likely to be associated
14    with liver tumorigenesis. Clear and mixed-cell foci are commonly considered preneoplastic changes and
15    would not be considered evidence of noncancer toxicity. The nature of spongiosis hepatis as a
16    preneoplastic change is less well understood (Bannasch. 2003; Karbe and Kerlin, 2002; Stroebel et al.,
17    1995). Spongiosis hepatis is a cyst-like lesion that arises from the perisinusoidal (Ito) cells (PSC) of the
18    liver. It is commonly seen in aging rats, but has been shown to increase in incidence following exposure
19    to hepatocarcinogens. Spongiosis hepatis can be seen in combination with preneoplastic foci in the liver
20    or with hepatocellular adenoma or carcinoma and has been considered a preneoplastic lesion (Bannasch.
21    2003; Stroebel etal.. 1995). This change can also be associated with hepatocellular hypertrophy and liver
22    toxicity and has been regarded as a secondary effect of some liver carcinogens (Karbe and Kerlin. 2002).
23    In the case of the JBRC (1998) study, spongiosis hepatis was associated with other preneoplastic changes
24    in the liver (clear and mixed-cell foci). No other lesions indicative of liver toxicity were seen in this
25    study; therefore, spongiosis hepatis was not considered indicative of noncancer effects. Serum chemistry
26    changes (increases in total protein, albumin, and glucose; decreases in AST, ALT, LDH, and ALP,
27    potassium, and inorganic phosphorous) were observed in both male and female rats (JBRC. 1998) in the
28    high dose groups, 274 and 429 mg/kg-day, respectively. These serum chemistry changes seen in terminal
29    blood samples from high-dose male and female rats are likely related to tumor formation in these dose
30    groups.

31           Significantly increased incidences of liver tumors (adenomas and carcinomas) and tumors of the
32    nasal cavity occurred in high-dose male and female rats (Table 4-10 and Table 4-11) treated with
33    1,4-dioxane for 2 years (Kano et al.. 2009). The first liver tumor was seen at 85 weeks in high-dose male
34    rats and 73 weeks in high-dose female rats (vs. 101-104 weeks in lower dose groups and controls)
35    (Yamazaki. 2006). In addition, a significant increase (p < 0.01, Fisher's Exact test) in mesotheliomas of

                                                                                                     43
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    the peritoneum was seen in high-dose males (28/50 versus 2/50 in controls). Mesotheliomas were the
 2    single largest cause of death among high-dose male rats, accounting for 12 of 28 pretermination deaths
 3    (Yamazaki. 2006). Also, in males, there were increasing trends in mammary gland fibroadenoma and
 4    fibroma of the subcutis, both statistically significant (p < 0.01) by the Peto test of dose-response trend.
 5    Females showed a significant increasing trend in mammary gland adenomas (p < 0.01 by Peto's test). The
 6    tumor incidence values presented in Table 4-10 and Table 4-11 were not adjusted for survival.
      Table 4-10  Incidence of nasal cavity, peritoneum, and mammary gland tumors in F344/DuCrj rats
                  exposed to 1,4-dioxane in drinking water for 2 years
Males
Dose (mg/kg-day)
0
11
55
274
0
Females
18
83
429
Nasal cavity
Squamous cell carcinoma
Sarcoma
Rhabdomyosarcoma
Esthesioneuroepithelioma
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
3/50a
2/50
1/50
1/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
0/50
7/50a'D
0/50
0/50
1/50
Peritoneum
Mesothelioma
2/50
2/50
5/50
28/50a'c
1/50
0/50
0/50
0/50
Mammary gland
Fibroadenoma
Adenoma
Either adenoma or fibroadenoma
1/50
0/50
1/50
1/50
1/50
2/50
0/50
2/50
2/50
4/50a
2/50
6/50a
3/50
6/50
8/50
2/50
7/50
8/50
1/50
10/50
11/50
3/50
16/50a'c
18/50a'c
      aStatistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
      bSignificantly different from control by Fisher's exact test (p < 0.01).
      cSignificantly different from control by Fisher's exact test (p < 0.05).
      Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
      Table 4-11  Incidence of liver tumors in F344/DuCrj rats exposed to 1,4-dioxane in drinking water
                  for 2 years
Males
Dose (mg/kg-day)
Hepatocellular adenoma
Hepatocellular carcinoma
Either adenoma or carcinoma
0
3/50
0/50
3/50
11
4/50
0/50
4/50
55
7/50
0/50
7/50
274
32/50a'D
14/50a'°
39/50a'c
0
3/50
0/50
3/50
Females
18
1/50
0/50
1/50
83
6/50
0/50
6/50
429
48/50a'°
10/50a'°
48/50a'D
      "Significantly different from control by Fisher's exact test (p < 0.01).
      bStatistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
      Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).

 7           For mice, growth and mortality rates were reported in Kano et al. (2009) for the duration of the
 8    study. Similar to rats, the growth rates of male and female mice were slower than controls and terminal
 9    body weights were lower for the mid (p < 0.01 for males administered 191 mg/kg-day and p < 0.05 for
10    females administered 278 mg/kg-day) and high doses (p < 0.05 for males and females administered 677
11    and 964 mg/kg-day, respectively). There were no differences in survival rates between control and treated
12    male mice; however, survival rates were significantly decreased compared to controls for female mice in
13    the mid (278 mg/kg-day, approximately 40% survival) and high (964 mg/kg-day, approximately 20%
14    survival)  dose groups. The study authors attributed these early female mouse deaths to the significant
                                                                                                      44
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    incidence of hepatic tumors, and Kano et al. (2009) reported tumor incidence for all animals in the study
 2    (N=50), including animals that became moribund or died before the end of the study. Additional data on
 3    survival rates of mice were provided in a personal communication from Dr. Yamazaki (2006). who
 4    reported that the survival of mice was low in all male groups (31/50, 33/50, 25/50 and 26/50 in control,
 5    low-, mid-, and high-dose groups, respectively) and particularly low in high-dose females (29/50, 29/50,
 6    17/50, and 5/50 in control, low-, mid-, and high-dose groups, respectively). These deaths occurred
 7    primarily during the second year of the study. Survival at 12 months in male mice was 50/50, 48/50,
 8    50/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively. Female mouse survival at
 9    12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-, mid-, and high-dose groups, respectively
10    (Yamazaki.  2006). Furthermore, these deaths were primarily tumor related. Liver tumors were listed as
11    the cause of death for 31 of the 45 pretermination deaths in high-dose female Crj:BDFl mice (Yamazaki.
12    2006). For mice, growth and mortality rates were reported in Kano et al. (2009)  for the duration of the
13    study. Similar to rats, the growth rates of male and female mice were slower than controls and terminal
14    body weights were lower for the mid (p < 0.01 for males administered 191 mg/kg-day and p < 0.05 for
15    females administered 278 mg/kg-day) and high doses (p < 0.05 for males and females administered 677
16    and 964 mg/kg-day, respectively).

17           Food consumption was not significantly affected, but water consumption was reduced 26% in
18    high-dose male mice and 28% in high-dose female mice. Final BWs were reduced 43% in high-dose male
19    mice and 15 and 45% in mid- and high-dose female mice, respectively. Male mice showed increases in
20    RBC counts, hemoglobin, and hematocrit, whereas in female mice, there was a decrease in platelets in
21    mid- and high-dose rats. With the exception of a 60% decrease in platelets in high-dose female mice,
22    hematological changes were within 15% of control values. Serum AST, ALT, LDH, and ALP activities
23    were significantly increased in mid- and high-dose male mice, whereas LAP and CPK were increased
24    only in high-dose male mice. AST, ALT, LDH, and ALP activities were increased in mid- and high-dose
25    female mice, but CPK activity was increased only in high-dose female mice. Increases in serum enzyme
26    activities ranged from less than two- to sevenfold above control values. Glucose and triglycerides were
27    decreased in high-dose males and in mid- and high-dose females. High-dose female mice also showed
28    decreases in serum phospholipid and albumin concentrations (not reported in males). Blood calcium was
29    lower in high-dose females and was not reported in males. Urinary pH was decreased in high-dose males,
30    whereas urinary protein, glucose, and occult blood were increased in mid- and high-dose female mice.
31    Relative and absolute lung weights were increased in high-dose males and in mid- and high-dose females
32    (JBRC. 1998). Microscopic examination of the tissues for nonneoplastic lesions showed significant
33    alterations in the epithelium of the respiratory tract, mainly in high-dose animals, although some changes
34    occurred in mid-dose mice (

35           Table  4-12 and Table 4-13). Commonly seen alterations included nuclear enlargement, atrophy,
36    and inflammation of the epithelium. Other notable changes observed included nuclear enlargement of the
37    proximal tubule of the kidney and angiectasis in the liver in high-dose male mice.
                                                                                                  45
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     Table 4-12   Incidence of histopathological lesions in male Crj:BDF1 mice exposed to 1,4-dioxane
                  in drinking water for 2 years
                                                                        Dose (mg/kg-day)
                                                                          49
                                                                                 191
                         677
     Nuclear enlargement; nasal respiratory epithelium0
                                                        0/50
0/50
0/50
31/508
     Nuclear enlargement; nasal olfactory epithelium
                                                        0/50
0/50
9/50e
49/508
     Atrophy; nasal olfactory epithelium
                                                        0/50
0/50
1/50
48/50
     Inflammation; nasal cavity
                                                        1/50
2/50
1/50
25/50
     Atrophy; tracheal epithelium
                                                        0/50
0/50
0/50
42/50
     Nuclear enlargement; tracheal epithelium
                                                        0/50
0/50
0/50
 17/50
     Nuclear enlargement; bronchial epithelium
                                                        0/50
0/50
0/50
41/50
     Atrophy; lung/bronchial epithelium
                                                        0/50
0/50
0/50
43/50
     Accumulation of foamy cells; lung
                                                        1/50
0/50
0/50
27/50
     Angiectasis; liver
                                                       2/50
3/50
4/50
 16/50
     Nuclear enlargement; kidney proximal tubule
                                                        0/50
0/50
0/50
     "Data from JBRC (1j
     ep < 0.01 by x2 test.

     Sources: Kano et al. (2009) and JBRC (1998).
39/50
aData presented for all animals, including animals that became moribund or died before the end of the study.
"Dose levels from Kano et al. (2009).
°Data from Kano et al. (2009).
                   . JBRC did not report statistical significance for the "All animals" comparison.
     Table 4-13   Incidence of histopathological lesions in female Crj:BDF1 mice exposed to
                  1,4-dioxane in drinking water for 2 years
Dose (mg/kg-day)a'D

Nuclear enlargement; nasal respiratory epithelium0
Nuclear enlargement; nasal olfactory epithelium0
Atrophy; nasal olfactory epithelium0
Inflammation; nasal cavity0
Atrophy; tracheal epithelium0
Nuclear enlargement; bronchial epithelium0
Atrophy; lung/bronchial epithelium0
Accumulation of foamy cells; lung0
0
0/50
0/50
0/50
2/50
0/50
0/50
0/50
0/50
66
0/50
0/50
0/50
0/50
0/50
1/50
0/50
1/50
278
0/50
41/508
1/50
7/50
2/50
22/50
7/50
4/50
964
41/508
33/508
42/50
42/50
49/50
48/50
50/50
45/50
     aData presented for all animals, including animals that became moribund or died before the end of the study.
     "Dose levels from Kano et al. (2009).
     °Data from Kano et al. (2009).
     dData from JBRC (1998). JBRC did not report statistical significance for the "All animals" comparison.
     ep < 0.01 by x2 test.

     Sources: Kano et al. (2009) and JBRC (1998).
1            NOAEL and LOAEL values for mice in this study were identified by EPA as 66 and

2    278 mg/kg-day, respectively, based on nasal inflammation observed in female mice. Nuclear enlargement

3    of the nasal olfactory epithelium and bronchial epithelium was also observed at a dose of 278 mg/kg-day

4    in female mice; however, it is unclear whether these alterations represent adverse toxicological effects.

5    The serum chemistry changes seen in terminal blood samples from male and female mice (mid- and

6    high-dose groups) are likely related to tumor formation in these animals. Liver angiectasis, an abnormal
                                                                                                        46
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    dilatation and/or lengthening of a blood or lymphatic vessel, was seen in male mice given 1,4-dioxane at a
 2    dose of 677 mg/kg-day.

 3           Treatment with 1,4-dioxane resulted in an increase in the formation of liver tumors (adenomas
 4    and carcinomas) in male and female mice. The incidence of hepatocellular adenoma was statistically
 5    increased in male mice in the mid-dose group only. The incidence of male mice with hepatocellular
 6    carcinoma or either tumor type (adenoma or carcinoma) was increased in the low, mid, and high-dose
 7    groups. The appearance of the first liver tumor occurred in male mice at 64, 74, 63, and 59 weeks in the
 8    control, low- mid-, and high-dose groups, respectively (Yamazaki, 2006). In female mice, increased
 9    incidence was observed for hepatocellular carcinoma in all treatment groups, while an increase in
10    hepatocellular adenoma incidence was only seen in the 66 and 278 mg/kg-day dose groups (Table 4-14).
11    The appearance of the first liver tumor in female mice occurred at 95, 79, 71, and 56 weeks in the control,
12    low-, mid-, and high-dose groups, respectively (Yamazaki, 2006). The tumor incidence data presented for
13    male and female mice in Table 4-14 are based on reanalyzed sample data presented in Kano et al. (2009)
14    that included lesions in animals that became moribund or died prior to the completion of the 2-year study.

15           Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and carcinomas in
16    control male and female BDF1 mice from ten 2-year bioassays at the JBRC. For female mice, out of 499
17    control mice, the incidence rates  were 4.4% for hepatocellular adenomas and 2.0% for hepatocellular
18    carcinomas. Kano et al. (2009) reported a 10% incidence rate for hepatocellular adenomas and a 0%
19    incidence rate for hepatocellular  carcinomas in control female BDF1. The background incidence rates for
20    male BDF1 mice were 15% and 22.8% for hepatocellular adenomas and carcinomas, respectively, out of
21    500 control mice in ten 2-year bioassays (Katagiri et al.. 1998). Background rates for B6C3Fi mice
22    evaluated by the National Toxicology Program are similar (10.3% and 21.3% for hepatocellular
23    adenomas and carcinomas in male mice, respectively; 4.0% and 4.1% for hepatocellular adenomas and
24    carcinomas in female mice, respectively) to the BDF1 mice background rates observed by JBRC
25    (Haseman et al.. 1984). Thus, the BDF1 mouse is not particularly sensitive compared to the commonly
26    used B6C3FJ strain and indicates that the results obtained by JBRC are reasonable.

      Table 4-14 Incidence of tumors in Crj:BDF1 mice exposed to 1,4-dioxane in drinking water for
                 2  years
Males
Dose (mg/kg-day)
0
49
191
677
0
Females
66
278
964
Nasal Cavity
Adenocarcinoma
Esthesioneuroepithelioma
0/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
0/50
0/50
0/50
0/50
0/50
0/50
1/50
0/50
Liver
Hepatocellular adenoma
Hepatocellular carcinoma
Either hepatocellular
adenoma or carcinoma
9/50
15/50
23/50
17/50
20/50
31/50
23/50a
23/50
37/50c
11/50
36/50a'c
40/50a'b
5/50
0/50
5/50
31/50a
6/50c
35/50a
20/50a
30/50a
41/50a
3/50
45/50a'c
46/50a'b
      "Significantly different from control by Fisher's exact test (p < 0.01).
      bStatistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
      °Significantly different from control by Fisher's exact test (p < 0.05).
      Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).

27           A weight of evidence evaluation of the carcinogenicity studies presented in Section 4.2.1.2is
28    located in Section 4.7 and Table 4-19.

                                                                                                   47
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      4.2.2  Inhalation Toxicity
      4.2.2.1  Subchronic Inhalation Toxicity

 1    4.2.2.1.1   Fairleyetal. Rabbits, guinea pigs, rats, and mice (3-6/species/group) were
 2    exposed to 1,000, 2,000, 5,000, or 10,000 ppm of 1,4-dioxane vapor two-times a day for 1.5 hours
 3    (3 hours/day) for 5 days/week and 1.5 hours on the 6th day (16.5 hours/week) (Fairleyetal.. 1934).
 4    Animals were exposed until death occurred or were sacrificed at varying time periods. At the 10,000 ppm
 5    concentration, only one animal (rat) survived a 7-day exposure. The rest of the animals (six guinea pigs,
 6    three mice, and two rats) died within the first five exposures. Severe liver and kidney damage and acute
 7    vascular congestion of the lungs were observed in these animals. Kidney damage was described as patchy
 8    degeneration of cortical tubules with vascular congestion and hemorrhage. Liver lesions varied from
 9    cloudy hepatocyte swelling to large areas of necrosis. At 5,000 ppm, mortality was observed in two mice
10    and one guinea pig following 15-34 exposures. The remaining animals were sacrificed following
11    49.5 hours (3 weeks) of exposure (three rabbits) or 94.5 hours (5 weeks) of exposure (three guinea pigs).
12    Liver and kidney damage in both dead and surviving animals was similar to that described for the
13    10,000 ppm concentration. Animals (four rabbits, four guinea pigs, six rats, and five mice) were exposed
14    to 2,000 ppm for 45-102 total exposure hours (approximately 2-6 weeks). Kidney and liver damage was
15    still apparent in animals exposed to this concentration. Animals exposed to 1,000 ppm were sacrificed at
16    intervals with the total exposure duration ranging between 78 and 202.5 hours (approximately 4-
17    12 weeks). Cortical kidney degeneration and hepatocyte degeneration and liver necrosis were observed in
18    these animals (two rabbits, three guinea pigs, three rats, and four mice).  The low concentration of
19    1,000 ppm was identified by EPA as a LOAEL for liver and kidney degeneration in  rats, mice, rabbits,
20    and guinea pigs in this study.
                                                                                                  48
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    4.2.2.1.2   Kasai et a I. Male and female 6-week-old F344/DuCrj rats (10/sex/group) were
 2    exposed to nominal concentrations of 0 (clean air). 100. 200. 400. 800. 1.600. 3.200. or 6.400 ppm (0.
 3    360. 720. 1.400. 2.900. 5.800. 1.2000. and 23.000 mg/m3. respectively) of vaporized 1.4-dioxane (>99%
 4    pure) for 6 hours/day. 5 days/week, for 13 weeks in whole body inhalation chambers (Kasai et al.. 2008).
 5    Each inhalation chamber housed 20 individual cages for 10 males and 10 females. During exposure, the
 6    concentration of 1.4-dioxane vapor was determined every 15 minutes by gas chromatography. In addition.
 7    during exposure, animals received food and water ad libitum and the following data were collected: 1)
 8    clinical signs and mortality (daily): 2) BW and food intake (weekly): 3) urinary parameters using Ames
 9    reagent strips (measured during week 13  of the exposure); and 4) 1.4-dioxane content in plasma from
10    three rats of both sexes (measured on the third day of exposure during weeks 12 and 13 at 1 hour
11    postmortem). At the end of the 13-week exposure period or at the time of an animal's death during
12    exposure, all organs were collected, weighed, and evaluated for macroscopic lesions. Histopathological
13    evaluations of organs and tissues were conducted in accordance with the OECD test guidelines, including
14    all tissues of the respiratory tract. Liver sections from male and female rats exposed to 800.  1.600 and
15    3.200 ppm of 1.4-dioxane were  also analyzed for foci (in the absence of tumor formation) by
16    immunohistochemical expression of glutathione S-transferase placental form (GST-P). Hematological and
17    clinical chemistry parameters were measured using blood collected from the abdominal aorta of rats
18    following an overnight fasting at the end of the 13-week exposure period. The measured hematological
19    and clinical chemistry parameters included: red blood cell count, hemoglobin, hematocrit. MCV. AST.
20    ALT, glucose, andtriglyceride.  Statistically significant differences (p-value of 0.05) between 1.4-dioxane
21    and clean air exposed groups were determined by study authors using Dunnett's test or y2test.

 1           All rats exposed to 6.400 ppm of 1.4-dioxane died by the end of the first week of exposure: the
 2    determined cause of death was renal failure and diagnosed as necrosis of the renal tubules. At
 3    concentrations lower than 6.400 ppm. mortality was not observed and all exposed rats were absent of
 4    clinical signs. Exposure-related  effects on final BWs. organ weights, and hematological and clinical
 5    chemistry parameters were reported as compared to controls and these changes are outlined in Table 4-15
 6    and Table 4-16. Briefly, terminal BWs were  significantly decreased in both sexes at 200 ppm; and
 7    additionally in females at 800 and 1.600 ppm. Statistically significant increases in several organ weights
 8    were observed, including lung (> 1.600 ppm. males; >200 ppm. females); liver (>800 ppm. both sexes).
 9    and kidneys (3.200 ppm. males: >800 ppm. females). Statistically significant changes in hematological
10    parameters and clinical chemistry were observed in both sexes at 3.200 ppm including increased levels of
11    hemoglobin ALT. RBC. AST .and MCV. In  females only, at 3.200 ppm. increased levels of hematocrit
12    was noted; and in males at this exposure concentration decreased levels of glucose and triglyceride were
13    observed, in addition to slightly decreased urinary protein. However, the urinary protein data were not
14    shown in this study. At 200 ppm. an increased AST level in females was noted. Blood plasma levels of
15    1.4-dioxane were also evaluated and in both  sexes, a linear increase  in 1.4-dioxane levels was detected at
16    exposure concentrations of 400 ppm and above. The highest blood levels of 1.4-dioxane were detected in
17    females.

18           Exposure and/or sex-related histopathology findings also reported by the  study authors included
19    nuclear enlargement of the nasal respiratory, nasal olfactory, tracheal. and bronchial epithelium; vacuolic

                                                                                                  49
                                     DRAFT - DO NOT CITE OR QUOTE

-------
1    change in the olfactory and bronchial epithelium; atrophy of the nasal epithelium; hydropic change in the
2    proximal tubules of the kidney: and single-cell necrosis and centrilobular swelling in the liver. Table 4-17
3    presents a summary of these histopathological lesions, including incidence and severity data. Further
4    microscopic evaluation of liver tissue revealed GST-P positive liver foci in both sexes at 3.200 ppm (3/10
5    males. 2/10 females) and in females at 1.600 ppm (4/10).

6           The study authors determined nuclear enlargement in the respiratory epithelium as the most
7    sensitive lesion and a LOAEL value of 100 ppm was identified by the study authors based on the
8    incidence data of this lesion in both male and female rats.
     Table 4-15  Terminal body weights and relative organ weights of F344/DuCrj rats exposed to
                 1,4-dioxane vapor by whole-body inhalation for 13 weeks
Males

Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
Females

Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)



Males3



1,4-dioxane vapor concentration (ppm)
lean air)
323 ± 14
0.310±0.011
2.610 ±0.069
0.589± 0.016

100
323 ± 14
0.312 ±
0.007
2.697 ±
0.092
0.596 ±
0.021

200
304 ± 11C
0.325 ±
0.008C
2.613 ±
0.084
0.612 ±
0.013

400
311 ± 19
0.320 ±
0.009
2.666 ±
0.080
0.601 ±
0.020
Females3
800
317± 12
0.321 ±
0.011
2.726 ±
0.082C
0.610 ±
0.015

1,600
312 ± 14
0.333 ±
0.009b
2.737 ±
0.077b
0.606 ±
0.021

3,200
301 ± 11b
0.346 ±
0.017b
2.939 ±
0.101b
0.647 ±
0.026b

1,4-dioxane vapor concentration (ppm)
0 (clean air)
187 ±5
0.402 ±0.013
2.353 ±0.081
0.647± 0.014
100
195 ±8
0.402 ±
0.015
2.338 ±
0.092
0.631 ±
0.019
200
174+ 10"
0.435 ±
0.018b
2.395±
0.092
0.668 ±
0.012
400
180 ±5
0.429 ±
0.029C
2.408 ±
0.066
0.662 ±
0.024
800
175±6b
0.430 ±
0.013b
2.513 ±
0.076b
0.679 ±
0.018b
1,600
173±8b
0.454 ±
0.018b
2.630 ±
0.139b
0.705 ±
0.028b
3,200
168 ±4b
0.457 ±
0.016b
2.828 ±
0.1 44b
0.749 ±
0.024b
     aData are presented for 10 sacrificed animals.
     bp < 0.01 by Dunnett's test.
     cp < 0.05 by Dunnett's test.
     Source: Kasai et al. (20081
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                   50

-------
Table 4-16  Hematology and clinical chemistry of F344/DuCrj rats exposed to 1,4-dioxane vapor
            by whole-body inhalation for 13 weeks
Males

Red blood cell (10b/Ml)
Hemoglobin (g/dl)
Hematocrit (%)
MCV (fl)
AST(IU/I)
ALT(IU/I)
Glucose (mg/dl)
Triglyceride (mg/dl)
Females

Red blood cell (10b/Ml)
Hemoglobin (g/dl)a
Hematocrit (%)a
MCV(fl)a
AST(IU/l)a
ALT(IU/l)a
Glucose (mg/dl) a
Triglyceride (mg/dl)



Males3



1,4-dioxane vapor concentration (ppm)
0 (clean air)
9.55 ±0.17
16.0 ±0.2
46.2 ± 1.2
48.4 ±0.7
73 ±8
27 ±3
197 ± 17
125± 17

100
9.53 ± 0.24
16.1 ±0.4
46.3 ± 1.3
48.6 ±0.7
75 ± 14
27 ±4
206 ± 13
148 ±37

200
9.54 ±0.1 8
15.9 ±0.2
46.3 ±0.9
48.6 ±0.4
73 ± 10
27 ±4
192 ±9
118±33

400
9.59 ± 0.26
16.1 ±0.3
46.3 ± 1.4
48.3 ±0.4
72 ±5
28 ± 1
190± 12
131 ±30
Females3
800
9.55 ±0.18
16.0 ±0.3
46.3± 1.1
48.5 ±0.6
72 ±3
27 ±2
187± 15
113±27

1,600
9.58 ±0.14
16.2 ±0.3
46.8 ±0.9
48.9 ±0.6
70 ±4
27 ±2
184± 12
106 ±24

3,200
9.57 ±0.37
16.4±0.4C
47.3 ± 1.7
49.4 ±0.5°
73 ±4
30 ±2
170± 11°
87 ± 22C

1,4-dioxane vapor concentration (ppm)
0 (clean air)
8.77 ±0.23
16.2 ±0.3
46.0 ± 1.5
52.5 ±0.7
64 ±6
23 ±3
143 ± 18
45±5
100
8.69 ±0.21
16.0 ±0.3
45.5 ± 1.2
52.3 ±0.7
65 ±3
21 ±2
144± 18
48 ±6
200
8.73 ±0.25
16.3 ±0.4
45.8 ± 1.7
52.4 ±0.7
74 ± 14C
26 ± 10
137 ±9
42 ±4
400
8.88 ±0.21
16.2 ±0.4
46.5 ± 1.5
52.4 ±0.8
69 ±5
25 ±3
140± 15
47 ±8
800
8.68 ± 0.69
16.2 ±0.6
45.4 ±3.6
52.3 ±0.6
68 ±6
24 ±4
141 ± 15
42 ±6
1,600
8.86 ±0.16
16.3 ±0.2
46.2 ±0.7
52.1 ±0.5
70 ±5
25 ±3
139± 11
39 ±7
3,200
9.15±0.12C
16.6±0.2C
47.5 ± 0.6 c
52.0 ±0.7
76 ±5°
30 ±3"
139± 18
42 ±7
"Data are presented for 10 sacrificed animals.
bp < 0.01 by Dunnett's test.
cp < 0.05 by Dunnett's test.
dData were reported for 9/10 female rats.

Source: Kasai et al. (20081
Table 4-17  Incidence data of histopathological lesions in F344/DuCrj rats exposed to 1,4-dioxane
            vapor by whole-body inhalation for 13 weeks
Males
Effect0
Nuclear enlargement; nasal
respiratory epithelium
Nuclear enlargement; nasal
olfactory epithelium
Nuclear enlargement; tracheal
epithelium
Nuclear enlargement; bronchial
epithelium
Vacuolic change; olfactory
epithelium
Vacuolic change; bronchial
epithelium



Males3



1,4-dioxane vapor concentration (ppm)
0 (clean air)
0/10
0/10
0/10
0/10
0/10
0/10
100
7/1 Oc
(7, 1+)
0/10
0/10
0/10
1/10
(1,1+)
0/10
200
9/1 Oc
(9, 1+)
5/10°
(5, 1+)
0/10
0/10
3/10
(3, 1+)
0/10
400
7/1 Oc
(7, 1+)
10/10C
(10, 1+)
0/10
0/10
6/1 Oa
(6, 1+)
0/10
800
10/10C
(10, 1+)
10/10C
(10, 1+)
1/10
(1, 1+)
0/10
10/10C
(10, 1+)
4/10
(4, 1+)
1,600
10/10C
(10,2+)
10/10C
(10,2+)
10/10C
(10,1+)
9/1 Oc
(9, 1+)
10/10C
(10, 1+)
6/10°
(6, 1+)
3,200
10/10C
(10,2+)
10/10C
(10,2+)
10/10C
(10,1+)
10/10C
(10,1+)
9/1 Oc
(10, 1+)
6/10°
(6, 1+)
Atrophy; olfactory epithelium8 - ...
Hepatocyte centrilobular swelling
Hepatocyte single-cell necrosis
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
1/10
(1,1+)
1/10
(1,1+)
10/10C
(10,1+)
8/1 Oc
(8, 1+)
Hydropic change; renal proximal
tubule8
Females



Females3



                                                                                            51
                               DRAFT - DO NOT CITE OR QUOTE

-------
1,4-dioxane vapor concentration (ppm)
Effect0
Nuclear enlargement; nasal
respiratory epithelium
Nuclear enlargement; nasal
olfactory epithelium
Nuclear enlargement; tracheal
epithelium
Nuclear enlargement; bronchial
epithelium
Vacuolic change; olfactory
epithelium
Vacuolic change; bronchial
epithelium
Atrophy; olfactory epithelium
Hepatocyte centrilobular swelling
Hepatocyte single-cell necrosis
Hydropic change; renal proximal
tubule
0 (clean air)
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
0/10
100
5/1 Oa
(5, 1+)
2/10
(2, 1+)
0/10
0/10
1/10
(1, 1+)
0/10
0/10
0/10
0/10
0/10
200
9/1 Oc
(9, 1+)
6/1 Od
(6, 1+)
0/10
0/10
2/10
(2, 1+)
0/10
2/10
(2, 1+)
0/10
0/10
0/10
400
10/10C
(10,1+)
10/10C
(9, 1+;
1, 2+)
0/10
0/10
3/10
(3, 1+)
1/10
(1,1+)
3/10
(3, 1+)
0/10
0/10
0/10
800
10/10C
(10,1+)
10/10C
(10, 1+)
2/10
(2, 1+)
0/10
7/1 Oc
(7, 1+)
1/10
(1, 1+)
5/10°
(5, 1+)
0/10
0/10
0/10
1,600
10/10C
(10,2+)
10/10C
(7, 1+;
3,2+)
7/1 Oc
(7, 1+)
0/10
9/1 Oc
(9, 1+)
3/10
(3, 1+)
5/10°
(5, 1+)
1/10
(1,1+)
0/10
0/10
3,200
10/10C
(10,2+)
10/10C
(10,2+)
10/10C
(10, 1+)
10/10C
(10, 1+)
10/10C
(10,1+)
4/10
(4, 1+)
4/10
(4, 1+)
8/1 Oc
(8, 1+)
3/10
(3, 1+)
6/1 Oa
(6, 1+)
      3Data are presented for sacrificed animals.
      bValues listed are the number of animals with the indicated lesion. Values in parentheses, are the number of lesion bearing animals for
         a given grade of lesion severity. Severity key: 1+, slight and , 2+, moderate.
      °p < 0.01 by x2 test.
      dp < 0.05 by x2 test.
      eData were not reported for male rats.
      Source: Kasai et al. (20081
      4.2.2.2   Chronic Inhalation Toxicity and Carcinogenicity

 1    4.2.2.2.1   Torkelson et al. Whole body exposures of male and female Wistar rats
 2    (288/sex) to 1,4-dioxane vapors (99.9% pure) at a concentration of 0.4 mg/L (111 ppm), were carried out
 3    7 hours/day, 5 days/week for 2 years (Torkelson et al.. 1974). The age of the animals at the beginning of
 4    the study was not provided. The concentration of 1,4-dioxane vapor during exposures was determined
 5    with infrared analyzers. Food and water were available ad libitum except during exposures. Endpoints
 6    examined included clinical signs, eye and nasal irritation, skin condition, respiratory distress, and tumor
 7    formation. BWs were determined weekly. Standard hematological parameters were determined on all
 8    surviving animals after 16 and 23 months of exposure. Blood collected at termination was used also for
 9    determination of clinical chemistry parameters (serum AST and ALP activities, blood urea nitrogen
10    [BUN], and total protein). Liver, kidneys, and spleen were weighed and the major tissues and organs were
11    processed for microscopic examination (lungs, trachea, thoracic lymph nodes, heart, liver, pancreas,
12    stomach, intestine, spleen, thyroid, mesenteric lymph nodes, kidneys, urinary bladder, pituitary, adrenals,
13    testes, ovaries, oviduct, uterus, mammary gland, lacrimal gland, lymph nodes, brain, vagina, and bone
14    marrow, and any abnormal growths). Nasal tissues were not obtained for histopathological evaluation.
15    Control and experimental groups were compared statistically using Student's t test, Yates corrected
16    %2 test, or Fisher's Exact test.
                                                                                                     52
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1           Exposure to 1,4-dioxane vapors had no significant effect on mortality or BW gain and induced no
 2    signs of eye or nasal irritation or respiratory distress. Slight, but statistically significant, changes in
 3    hematological and clinical chemistry parameters were within the normal physiological limits and were
 4    considered to be of no toxicological importance by the investigators. Altered hematological parameters
 5    included decreases in packed cell volume, RBC count, and hemoglobin, and an increase in WBC count in
 6    male rats. Clinical chemistry changes consisted of a slight decrease in both BUN (control—23 ± 9.9;
 7    111-ppm 1,4-dioxane—19.8 ±8.8) and ALP activity (control—34.4 ±  12.1; 111-ppm  1,4-dioxane—29.9
 8    ± 9.2) and a small increase in total protein (control—7.5 ± 0.37; 111-ppm 1,4-dioxane—7.9 ± 0.53) in
 9    male rats (values are mean ± standard deviation). Organ weights were not significantly affected.
10    Microscopic examination of organs and tissues did not reveal any treatment-related effects. Based on the
11    lack of significant effects on several endpoints, EPA identified the exposure concentration of 0.4 mg/L
12    (111 ppm) as a free standing NOAEL. The true NOAEL was likely to be higher.

13           Tumors, observed in all groups including controls, were characteristic of the rat strain used and
14    were considered unrelated to 1,4-dioxane inhalation. The most common tumors were reticulum cell
15    sarcomas and mammary tumors. Using Fisher's Exact test and a significance level of p <  0.05, no one
16    type of tumor occurred more frequently in treated rats than in controls. No hepatic or nasal cavity tumors
17    were seen in any rat.
                                                                                                    53
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    4.2.2.2.2  Kasai et al. Groups of male 6-week-old F344/DuCrj rats (50/group) weighing
 2    120 ± 5g (mean ± SD) at the beginning of the study were exposed via inhalation to nominal
 3    concentrations of 0 (clean air). 50. 250. and 1.250 ppm (0. 180. 900. and 4.500 mg/m3. respectively) of
 4    vaporized 1.4-dioxane (>99% pure) for 6 hours/day. 5 days/week, for 104 weeks (2 years') in whole body
 5    inhalation chambers (Kasai et al., 2009). Each inhalation chamber housed male rats individually in
 6    stainless-steel wire hanging cages. The authors stated female counterparts were not exposed given data
 7    illustrating the absence of induced mesotheliomas following exposure to 1.4-dioxane in drinking water
 8    (Yamazaki et al.. 1994). During exposure, the concentration of 1.4-dioxane vapor was determined every
 9    15 minutes by gas chromatography and animals received food and water ad libitum. In addition, during
10    the 2-year exposure period, clinical signs and mortality were recorded daily. BW and food intake were
11    measured once weekly for the first 14 weeks of exposure, and thereafter, every 4 weeks. At the end of the
12    2-year exposure period or at the time of an animal's death during exposure, all organs were collected.
13    weighed, and evaluated for macroscopic lesions. Additional examinations were completed on rats
14    sacrificed at the end of the 2-year exposure period. Endpoints examined included: 1) measurement of
15    hematological and clinical chemistry parameters using blood collected from the abdominal aorta of rats
16    following an overnight fasting at the end of the 2-year exposure period: 2) measurement of urinary
17    parameters using Ames reagent strips during the last week of the exposure period; and 3)
18    histopathological evaluations of organs and tissues outlined in the OECD test guideline which included
19    all tissues of the respiratory tract. For measured hematological and clinical chemistry parameters.
20    analyses included: red blood cell count, hemoglobin, hematocrit. MCV. mean corpuscular hemoglobin
21    (MCH). AST. ALT. ALP, and y-GTP. Organs and tissues collected for histopathological examination
22    were fixed in 10% neutral buffered formalin with the exception of nasal cavity samples.  Nasal tissue was
23    trimmed transversely at three levels after decalcification and Fixation in a formic acid-formalin solution.
24    The levels were demarcated at the following points: at the posterior  edge of the upper incisor teeth (level
25    1). at the incisive papilla (level 2). and at the anterior edge of the upper molar teeth (level 3). All tissue
26    samples were embedded in paraffin, and then sectioned (at 5_ [im thickness) and stained with hematoxylin
27    and eosin (H&E). Dunnett's test. y2test. and Fisher's exact test were used by study authors to determine
28    statistical differences (p-value of 0.05) between 1.4-dioxane exposed and clean air exposed group data.

 1           Deformity in the nose was the only clinical sign reported in  this study. This deformity was seen at
 2    exposure weeks 74 and 79 in one rat each, exposed to 250 ppm and  1.250 ppm of 1.4-dioxane.
 3    respectively. Both of these rats did not survive the 2-year exposure with deaths caused by malignant nasal
 4    tumors.

 5           Growth rates and survival rates were analyzed. Growth rates were not significantly affected by
 6    1.4-dioxane exposures, but a decreasing trend in growth was observed during the latter half of the 2-year
 7    exposure period for all exposure doses (i.e.. 50. 250. and 1.250 ppm). Survival rates were significantly
 8    decreased following 91 weeks of exposure to 1.250 ppm of 1.4-dioxane. The authors attributed these
 9    deaths to increased incidences of peritoneal mesotheliomas. but also noted that nasal tumors  could have
10    been a contributing factor. Terminal survival rates were 37/50. 37/50. 29/50. and 25/50 for 0. 50. 250. and
11    1.250 ppm exposed groups, respectively.
                                                                                                  54
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Exposure-related effects on final BWs. organ weights, and hematological and clinical chemistry
 2    parameters were reported. Changes in these effects, as compared to control are outlined in Table 4-18 and
 3    Table 4-19. Briefly, at 1.250 ppm terminal BWs were significantly decreased and relative liver and lung
 4    weights were significantly increased. It is of note that the observed change in terminal body weight was
 5    not an effect of food consumption, which was determined to be unaltered by the study authors. Altered
 6    hematological and clinical chemistry parameters were also observed with significant changes at
 7    1.250 ppm. Altered endpoints included decreased hemoglobin. MCV. and MCH. and increased AST.
 8    ALT. ALP, and y-GTP (p < 0.01) levels. In addition, urine pH was significantly decreased in 1.250 ppm
 9    exposed rats.

10           Histopathology Findings of pre- and nonneoplastic lesions associated with 1.4-dioxane treatment
11    were seen in the nasal cavity, liver, and kidneys (Table 4-20). At the highest concentration of 1.250 ppm.
12    all pre- and nonneoplastic lesions were significantly increased, as compared to controls, with the
13    exception of clear and mixed cell foci in the liver. At the lowest concentration of 50 ppm. nuclear
14    enlargement of the respiratory epithelium was the most sensitive lesion observed in the nasal cavity.
15    Based on this finding, the study authors identified a LOAEL of 50 ppm in male rats.

16           Tumor development was observed in the nasal cavity (squamous cell carcinoma), liver
17    (hepatocellular adenoma and carcinoma), peritoneum (peritoneal mesothelioma). kidney (renal cell
18    carcinoma), mammary gland (fibroadenoma and adenoma). Zvmbal gland (adenoma), and subcutaneous
19    tissue (subcutis fibroma). Tumor incidences with a dose-dependent, statistically significant positive trend
20    (Peto's test) included nasal squamous cell carcinoma, hepatocellular adenoma, peritoneal mesothelioma.
21    mammary gland fibroadenoma. and Zymbal gland adenoma. Renal cell carcinoma was also identified as
22    statistically significant with a positive dose-dependent trend; however, no tumor incidences were reported
23    at 50 and 250 ppm. At 1.250 ppm. significant increases in nasal  squamous cell carcinoma, hepatocellular
24    adenoma, and peritoneal mesothelioma were observed. At 250 ppm. significant increases in peritoneum
25    mesothelioma and subcutis fibroma were observed. Table 4-21 presents a summary of tumor incidences
26    found in this study.  Further characterizations of neoplasms revealed nasal squamous  cell carcinoma
27    occurred at the dorsal area of the nose (levels 1-3) marked by keratinization and the progression of growth
28    into surrounding tissue. Peritoneal mesotheliomas were characterized by complex branching structures
29    originating from the mesothelium of the scrotal sac. Invasive growth into surrounding tissues was
30    occasionally observed for peritoneal mesotheliomas.
                                                                                                 55
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-18  Terminal body and relative organ weights of F344/DuCrj male rats exposed to
           1,4-dioxane vapor by whole-body inhalation for 2 years
Males
1,4-dioxane vapor concentration (ppm)

Number of animals
examined
Body weight (g)
Lung (%)
Liver (%)
Kidneys (%)
0 (clean air)
37
383 ± 50
0.45 ±0.25
3.57 ±0.66
0.87 ±0.21
50
37
383 ± 53
0.49 ±0.27
3.86 ± 1.05
0.93 ±0.32

29
376 ± 38
0.45 ±0.1 8
3.58 ±0.52
0.81 ±0.13

25
359 ± 129"
0.46 ± 0.07a
4.53 ±0.71"
0.86 ±0.12
ap < 0.01 by Dunnett's test.
bp < 0.05 by Dunnett's test.
Source: Kasai et al. (20081
Table 4-19  Hematology and clinical chemistry of F344/DuCrj male rats exposed to 1,4-dioxane
           vapor by whole-body inhalation for 2 years
                                                   Males
1,4-dioxane vapor concentration (ppm)

Number of animals
examined
Red blood cell (10B/ul)
Hemoglobin (g/dl)
Hematocrit (%)
MCV (fl)
MCH (pg)
AST(IU/I)
ALT(IU/I)
ALP (IU/I)
Y-GTP (IU/I)
Urinary pH
0 (clean air)
35
7.4± 1.8
12.5 ±3.5
38.6 ±8.7
52.4 ±5.7
16.9 ±2.2
67 ±31
37 ± 12
185 ±288
6±3
7.1 ±0.6
50
35
6.8± 1.8
12.0 ±3.1
36.9 ±7.9
55.6 ±8.7
17.8 ±2.4
95 ±99
42 ±21
166 ±85
8±5
7.1 ±0.6

28
7.9± 1.0
13.4± 1.9
40.7 ±5.1
51.8±2.3
17.1 ± 1.2
95± 116
49 ±30
145 ± 171
10±8
7.1 ±0.6

25
7.0 ± 1.8
10.9 ±2.8"
34.3 ±7.6
49.4 ±4.0"
15.5± 1.3a
98 ± 52a
72± 36a
212 ± 109a
40 ± 26a
6.6 ±0.4°
ap < 0.01 by Dunnett's test.
bp < 0.05 by Dunnett's test.
Source: Kasai et al. (20081
                                                                                     56
                             DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-20  Incidence of pre-and nonneoplastic lesions in male F344/DuCrj rats exposed
             to 1,4-dioxane vapor by whole-body inhalation for 2 years
Effect
                                                                1,4-dioxane vapor concentration (ppm)
                                                               3 (clean air)
              50
           250
          1,250
Nuclear enlargement; nasal respiratory epithelium
 0/50
50/50a
48/50a
38/50a
Squamous cell metaplasia; nasal respiratory epithelium
 0/50
 0/50
 7/50"    44/50a
Squamous cell hyperplasia; nasal respiratory epithelium
 0/50
 0/50
 1/50
10/50a
Inflammation; nasal respiratory epithelium
13/50
 9/50
 7/50
39/50a
Nuclear enlargement; nasal olfactory epithelium
 0/50
48/50a
48/50a
45/50a
Respiratory metaplasia; nasal olfactory epithelium
11/50
34/50a
49/50a
48/50a
Atrophy; nasal olfactory epithelium
 0/50
40/50a
47/50a
48/50a
Inflammation; nasal olfactory epithelium
 0/50
 2/50
32/50a
34/50a
Hydropic change; lamina propria
 0/50
 2/50
36/50a
49/50a
Sclerosis; lamina propria
 0/50
 0/50
22/50a
40/50a
Proliferation; nasal gland
 0/50
 1/50
 0/50
 6/50"
Nuclear enlargement; liver centrilobular
 0/50
 0/50
 1/50
30/50a
Necrosis; liver centrilobular
 1/50
 3/50
 6/50
12/50a
Spongiosis hepatis; liver
7/50
 6/50
 13/50     19/50a
Clear cell foci; liver
15/50
 17/50
20/50
23/50
Basophilic cell foci; liver
17/50
20/50
 15/50    44/50a
Acidophilic cell foci; liver
5/50
 10/50
 12/50    25/50a
Mixed-cell foci; liver
5/50
 3/50
 4/50
 14/50
Nuclear enlargement; kidney proximal tubule
 0/50
 1/50
20/50a
47/50a
Hydropic change; kidney proximal tubule
 0/50
 0/50
 5/50
 6/50a
ap<0.01 by x" test.
bp < 0.05 by x2 test.

Source: Kasai et al. (2009).
                                   DRAFT - DO NOT CITE OR QUOTE
                                                                                                      57

-------
      Table 4-21  Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane vapor by
                 whole-body inhalation for 2 years
1 ,4-dioxane vapor concentration (ppm)
Effect
Nasal squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Mammary gland adenoma
Zymbal gland adenoma
Subcutis fibroma
0 (clean air)
0/50
1/50
0/50
0/50
2/50
1/50
0/50
0/50
1/50
50
0/50
2/50
0/50
0/50
4/50
2/50
0/50
0/50
4/50

1/50
3/50
1/50
0/50
14/50a
3/50
0/50
0/50
9/50a
1,250
6/50D'c
21/50a'c
2/50
4/50c
41/50a'c
5/50a
1/50
4/50c

     ap < 0.01 by Fisher's exact test.
     bp < 0.05 by Fisher's exact test.
     cp < 0.01 by Peto's test for dose-related trend.
     dp < 0.05 by Peto's test for dose-related trend.
     Source: Kasai et al. (2009).
     4.2.3  Initiation/Promotion Studies
     4.2.3.1  Bulletal.

 1           Bull et al. (1986) tested 1,4-dioxane as a cancer initiator in mice using oral, subcutaneous, and
 2   topical routes of exposure. A group of 40 female SENCAR mice (6-8 weeks old) was administered a
 3   single dose of 1,000 mg/kg 1,4-dioxane (purity >99%) by gavage, subcutaneous injection, or topical
 4   administration (vehicle was not specified). A group of rats was used as a vehicle control (number of
 5   animals not specified). Food and water were provided ad libitum. Two weeks after administration of
 6   1,4-dioxane, 12-O-tetradecanoylphorbol-13-acetate (TPA) (1.0 (ig in 0.2 mL of acetone) was applied to
 7   the shaved back of mice 3 times/week for a period of 20 weeks. The yield of papillomas at 24 weeks was
 8   selected as a potential predictor of carcinoma yields at 52 weeks following the start of the promotion
 9   schedule. Acetone was used instead of TPA in an additional group of 20 mice in order to determine
10   whether a single dose of 1,4-dioxane could induce tumors in the absence of TPA promotion.

11           1,4-Dioxane did not increase the  formation of papillomas compared to mice initiated with vehicle
12   and promoted with TPA, indicating lack of initiating activity under the conditions of the study. Negative
13   results were obtained for all three exposure routes. A single dose of 1,4-dioxane did not induce tumors in
14   the absence  of TPA promotion.
     4.2.3.2  Kingetal.
15           1,4-Dioxane was evaluated for complete carcinogenicity and tumor promotion activity in mouse
16   skin (King etal.. 1973). In the complete carcinogenicity study, 0.2 mL of a solution of 1,4-dioxane (purity
                                                                                                  58
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    not specified) in acetone was applied to the shaved skin of the back of Swiss Webster mice (30/sex)
 2    3 times/week for 78 weeks. Acetone was applied to the backs of control mice (30/sex) for the same time
 3    period. In the promotion study, each animal was treated with 50 ug of dimethylbenzanthracene 1 week
 4    prior to the topical application of the 1,4-dioxane solution described above (0.2 mL, 3 times/week,
 5    78 weeks) (30 mice/sex). Acetone vehicle was used in negative control mice (30/sex). Croton oil was
 6    used as a positive control in the promotion study (30/sex). Weekly counts of papillomas and suspect
 7    carcinomas were made by gross examination. 1,4-Dioxane was also administered in the drinking water
 8    (0.5 and 1%) to groups of Osborne-Mendel rats (35/sex/group) and B6C3Fi mice for 42 weeks (control
 9    findings were only reported for 34 weeks).

10           1,4-Dioxane was negative in the complete skin carcinogenicity test using dermal exposure. One
11    treated female mouse had malignant lymphoma; however, no papillomas were observed in male or female
12    mice by 60 weeks. Neoplastic lesions of the skin, lungs, and kidney were observed in mice given the
13    promotional treatment with 1,4-dioxane. In addition, the percentage of mice with skin tumors increased
14    sharply after approximately 10 weeks of promotion treatment. Significant mortality was observed when
15    1,4-dioxane was administered as a promoter (only 4 male and 5 female mice survived for 60 weeks), but
16    not as a complete carcinogen (22 male and 25 female mice survived until 60 weeks). The survival of
17    acetone-treated control mice in the promotion study was not affected (29 male and 26 female mice
18    survived until 60 weeks); however, the mice treated with croton oil as a positive control experienced
19    significant mortality (0 male and 1 female mouse survived for 60 weeks). The incidence of mice with
20    papillomas was similar for croton oil and 1,4-dioxane; however, the tumor multiplicity (i.e., number of
21    tumors/mouse) was higher for the croton oil treatment.

22           Oral administration of 1,4-dioxane  in drinking water caused appreciable mortality in rats, but not
23    mice, and increased weight gain in surviving rats and male mice. Histopathological lesions (i.e.,
24    unspecified liver and kidney effects) were also reported in exposed male and female rats; however, no
25    histopathological changes were indicated for mice.

26           1,4-Dioxane was demonstrated to be a tumor promoter, but not a complete carcinogen in mouse
27    skin, in this study. Topical administration for 78 weeks following initiation with dimethylbenzanthracene
28    caused an increase in the incidence and multiplicity of skin tumors in mice. Tumors were also observed at
29    remote sites (i.e., kidney and lung), and survival was affected. Topical application of 1,4-dioxane for
30    60 weeks in the absence of the  initiating treatment produced no effects on skin tumor formation or
31    mortality in mice.
      4.2.3.3   Lundberg et al.

32           Lundberg et al. (1987) evaluated the tumor promoting activity of 1,4-dioxane in rat liver. Male
33    Sprague Dawley rats (8/dose group, 19 for control group) weighing 200 g underwent a partial
34    hepatectomy followed 24 hours later by an i.p. injection of 30 mg/kg diethylnitrosamine (DEN) (initiation
35    treatment). 1,4-Dioxane (99.5% pure with 25 ppm butylated hydroxytoluene as a stabilizer) was then
36    administered daily by gavage (in saline vehicle) at doses of 0, 100, or 1,000 mg/kg-day, 5 days/week for

                                                                                                  59
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    7 weeks. Control rats were administered saline daily by gavage, following DEN initiation. 1,4-Dioxane
 2    was also administered to groups of rats that were not given the DEN initiating treatment (saline used
 3    instead of DEN). Ten days after the last dose, animals were sacrificed and liver sections were stained for
 4    GOT. The number and total volume of GGT-positive foci were determined.

 5           1,4-Dioxane did not increase the number or volume of GGT-foci in rats that were not given the
 6    DEN initiation treatment. The high dose of 1,4-dioxane (1,000 mg/kg-day) given as a promoting
 7    treatment (i.e., following DEN injection) produced an increase in the number of GGT-positive foci and
 8    the total foci volume. Histopathological changes were noted in the livers of high-dose rats. Enlarged,
 9    foamy hepatocytes were observed in the midzonal region of the liver, with the foamy appearance due to
10    the presence of numerous fat-containing cytoplasmic vacuoles. These results suggest that cytotoxic doses
11    of 1,4-dioxane may be associated with tumor promotion of 1,4-dioxane in rat liver.
      4.3  Reproductive/Developmental Studies—Oral  and Inhalation
      4.3.1   Giavini et al.

12           Pregnant female Sprague Dawley rats (18-20 per dose group) were given 1,4-dioxane (99% pure,
13    0.7% acetal) by gavage in water at concentrations of 0, 0.25, 0.5, or 1 mL/kg-day, corresponding to dose
14    estimates of 0, 250, 500, or 1,000 mg/kg-day (density of 1,4-dioxane is approximately 1.03 g/mL)
15    (Giavini et al., 1985). The chemical was administered at a constant volume of 3 mL/kg on days 6-15 of
16    gestation. Food consumption was determined daily and BWs were measured every 3 days. The dams were
17    sacrificed with chloroform on gestation day 21 and the numbers of corpora lutea, implantations,
18    resorptions, and live fetuses were recorded. Fetuses were weighed and examined for external
19    malformations prior to the evaluation of visceral and skeletal malformations (Wilson's free-hand section
20    method and staining with Alizarin red) and a determination of the degree of ossification.

21           Maternal weight gain was reduced by 10% in the high-dose group (1,000 mg/kg-day). Food
22    consumption for this group was 5% lower during the dosing period, but exceeded control levels for the
23    remainder of the study. No change from control was observed in the number of implantations, live
24    fetuses, or resorptions; however, fetal birth weight was 5% lower in the highest dose group (p < 0.01).
25    1,4-Dioxane exposure did not increase the frequency of major malformations or minor anomalies and
26    variants. Ossification of the sternebrae was reduced in the 1,000 mg/kg-day dose group (p < 0.05). The
27    study authors suggested that the observed delay in sternebrae ossification combined with the decrease in
28    fetal birth weight indicated a developmental delay related to 1,4-dioxane treatment. NOAEL and LOAEL
29    values of 500 and 1,000 mg/kg-day were identified from this study by EPA and based on delayed
30    ossification of the sternebrae and reduced fetal BWs.
                                                                                                60
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     4.4  Other  Duration  or E ndpoint Specific  Studies
     4.4.1   Acute and Short-term Toxicity
 1           The acute (< 24 hours) and short-term toxicity studies (<30 days) of 1,4-dioxane in laboratory
 2   animals are summarized in Table 4-22. Several exposure routes were employed in these studies, including
 3   dermal application, drinking water exposure, gavage, vapor inhalation, and i.v. or i.p. injection.
     4.4.1.1  Oral Toxicity

 4           Mortality was observed in many acute high-dose studies, and LD50 values for 1,4-dioxane were
 5   calculated for rats, mice, and guinea pigs (Pozzani etal., 1959; HF Jr et al., 1941; Laug et al., 1939).
 6   Clinical signs of CNS depression were observed, including staggered gait, narcosis, paralysis, coma, and
 7   death (Nelson. 1951; Laug et al.. 1939; Schrenk and Yant 1936; de Navasquez. 1935). Severe liver and
 8   kidney degeneration and necrosis were often seen in acute studies (JBRC. 1998; David. 1964; Kesten et
 9   al. 1939; Laugetal.. 1939; Schrenk and Yant. 1936; de Navasquez. 1935). JBRC (1998) additionally
10   reported histopathological lesions in the nasal cavity and the brain of rats following 2 weeks of exposure
11   to 1,4-dioxane in the drinking water.
     4.4.1.2  Inhalation Toxicity

12           Acute and short-term toxicity studies (all routes) are summarized in Table 4-18. Mortality
13   occurred in many high-concentration studies (Pozzani et al., 1959; Nelson. 1951; Wirth and Klimmer.
14   1936). Inhalation of 1,4-dioxane caused eye and nasal irritation, altered respiration, and pulmonary edema
15   and congestion (Yantet al.. 1930). Clinical signs of CNS depression were observed, including staggered
16   gait, narcosis, paralysis, coma, and death (Nelson. 1951; Wirth and Klimmer. 1936). Liver and kidney
17   degeneration and necrosis were also seen in acute and short-term inhalation studies (Drew et al.. 1978;
18   Fairlev et al.. 1934).
                                                                                                 61
                                    DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-22     Acute and short-term toxicity studies of 1,4-dioxane
Animal
Exposure
route
Test conditions
Results
Dose3
Reference
Oral studies
Rat (inbred strain
and gender
unspecified)
Rat (strain and
gender
unspecified)
F344/DuCrj rat
Female
Sprague Dawley
rat
Female Carworth
Farms-Nelson rat
Male Wistar rat,
guinea pig
Rat, mouse,
guinea pig
Rabbit
Rat, rabbit
Crj:BDF1 mouse
Dog
Oral via
drinking
water
Oral via
drinking
water
Oral via
drinking
water
Gavage
Gavage
Gavage
Gavage
Gavage
Gavage
Oral via
drinking
water
Drinking
water
ingestion
1-10 days of exposure
5-12 days of exposure
14-day exposure
0, 168, 840, 2550, or
4,200 mg/kg by
gavage, 21 and
4 hours prior to
sacrifice
Determination of a
single dose LDso
Single dose,
LDso determination
Single dose; several
dose groups
Single gavage dose of
0,207, 1,034, or
2,068 mg/kg-day
Single dose; mortality
after 2 weeks
14-day exposure
3-1 0 days of exposure
Ultrastructural changes
in the kidney,
degenerative nephrosis,
hyaline droplet
accumulation, crystal
formation in
mitochondria
Extensive degeneration
of the kidney, liver
damage, mortality in
8/10 animals by
12 days
Mortality, decreased
BWs, histopathological
lesions in the nasal
cavity, liver, kidney, and
brain
Increased ODC activity,
hepatic
CYP450 content, and
DMA single-strand
breaks
Lethality
Lethality
Clinical signs of CMS
depression, stomach
hemorrhage, kidney
enlargement, and liver
and kidney
degeneration
Clinical signs of CMS
depression, mortality at
2,068 mg/kg, renal
toxicity (polyuria
followed by anuria),
histopathological
changes in liver and
kidneys
Mortality and narcosis
Mortality, decreased
BWs, histopathological
lesions in the nasal
cavity, liver, kidney, and
brain
Clinical signs of CMS
depression, and liver
and kidney
degeneration
11,000 mg/kg-day
(5%)
11,000 mg/kg-day
(5%)
2,500 mg/kg-day
(nuclear enlargement
of olfactory epithelial
cells),
>7,500 mg/kg-day for
all other effects
840 mg/kg (ODC
activity only)
LDso =6, 400 mg/kg
(14,200ppm)
LDso (mg/kg):
rat = 7, 120
guinea pig - 3,150
LDso (mg/kg):
mouse = 5,900
rat = 5,400

1,034 mg/kg-day
3,160 mg/kg
10,800 mg/kg-day;
hepatocellular swelling
11,000 mg/kg-day
(5%)
David
(1964)
Kesten et
al. (1939)
JBRC
(1998)
Kitchin and
Brown
(1990)
Pozzani et
al. (1959)
Smyth et al.
(1941)
Laug et al.
(1939)
de
Navasquez
(1935)
Nelson
(1951)
JBRC
(1998)
Schrenk
and Yant
(1936)
Inhalation studies
Male CD1 rat
Vapor
inhalation
Serum enzymes
measured before and
after a single 4 hour
Increase in ALT, AST,
and OCT; no change in
G-6-Pase
1,000 ppm
Drew et al.
(1978)
                             exposure
                             DRAFT - DO NOT CITE OR QUOTE
                                                                                       62

-------
Rat
Female Carworth
Farms-Nelson rat

Mouse, cat


Guinea pig


Rabbit, guinea
pig, rat, mouse
Vapor
inhalation
Vapor
inhalation

Vapor
inhalation


Vapor
inhalation


Vapor
inhalation
5 hours of exposure
Determination of a
4-hour inhalation LCso

8 hours/day for
17 days


8-Hour exposure to
0.1-3% by volume


3 hours exposure, for
5 days; 1.5 hour
exposure for 1 day
Mortality and narcosis
Lethality

Paralysis and death
Eye and nasal irritation,
retching movements,
altered respiration,
narcosis, pulmonary
edema and congestion,
hyperemia of the brain
Degeneration and
necrosis in the kidney
and liver, vascular
congestion in the lungs
6,000 ppm
LCso = 51.3 mg/L

8,400 ppm


0.5% by volume


10,000 ppm
Nelson
(1951)
Pozzani et
al. (1959)
Wirth and
Klimmer
(1936)


Yant et al.
(193Q)


Fairley et
al.(1934)
Other routes


Male
COBS/Wistar rat


Rabbit, cat
Female
Sprague Dawley
rat
CBA/J mouse


Dermal


i.v.
injection
i.p.
injection
i.p.
injection
"Lowest effective dose for positive
ND = no data; OCT =
Nonoccluded
technique using
shaved areas of the
back and flank; single
application, 14-day
observation
Single injection of 0,
207, 1,034,
1,600 mg/kg-day
Single dose;
LDso values
determined 24 hours
and 14 days after
injection
Daily injection for
7 days, 0, 0.1, 1, 5,
and 10%


Negative; no effects
noted


Clinical signs of CNS
depression, narcosis at
1,034 mg/kg, mortality
at 1,600 mg/kg
Increased serum SDH
activity at 1/1 6th of the
LDso dose; no change
at higher or lower doses
Slightly lower
lymphocyte response to
mitogens


8,300 mg/kg


1,034 mg/kg-day
LDso (mg/kg):
24 hours = 4,848
14 days - 799
2,000 mg/kg-day
(10%)


Clark et al.
(1984)


de
Navasquez
(1935)
Lundberg et
al. (1986)
Thurman et
al. (1978)
results/ highest dose tested for negative results.
ornithine carbamyl transferase; ODC =
ornithine decarboxylase; SDH
= sorbitol dehydrogenase

     4.4.2  Neurotoxicity

 1           Clinical signs of CNS depression have been reported in humans and laboratory animals following
 2   high dose exposure to 1,4-dioxane (see Sections 4.1 and 4.2.1.1). Neurological symptoms were reported
 3   in the fatal case of a worker exposed to high concentrations of 1,4-dioxane through both inhalation and
 4   dermal exposure (Johnstone. 1959). These symptoms included headache, elevation in blood pressure,
 5   agitation and restlessness, and coma. Autopsy findings demonstrated perivascular widening in the brain,
 6   with small foci of demyelination in several regions (e.g., cortex, basal nuclei). It was suggested that these
 7   neurological changes may have been secondary to anoxia and cerebral edema. In laboratory animals, the
 8   neurological effects of acute high-dose exposure included staggered gait, narcosis, paralysis, coma, and
 9   death (Nelson. 1951: Laugetal. 1939: Schrenk and Yant. 1936: de Navasquez. 1935: Yantetal.. 1930).
10   The neurotoxicity of 1,4-dioxane was further investigated in several studies described below (Frantik et
11   al.. 1994: Kanadaetal.. 1994: Goldberg et al.. 1964: Knoefel. 1935V
                                                                                                  63
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      4.4.2.1   Frantiketal.

 1           The acute neurotoxicity of 1,4-dioxane was evaluated following a 4-hour inhalation exposure to
 2    male Wistar rats (four per dose group) and a 2-hour inhalation exposure to female H-strain mice (eight
 3    per dose group) (Frantiketal., 1994). Three exposure groups and a control group were used in this study.
 4    Exposure concentrations were not specified, but apparently were chosen from the linear portion of the
 5    concentration-effect curve. The neurotoxicity endpoint measured in this study was the inhibition of the
 6    propagation and maintenance of an electrically-evoked seizure discharge. This endpoint has been
 7    correlated with the behavioral effects and narcosis that occur following acute exposure to higher
 8    concentrations of organic solvents. Immediately following 1,4-dioxane exposure, a short electrical
 9    impulse was applied through ear electrodes (0.2 seconds, 50 hertz (Hz), 180 volts (V) in rats, 90 V in
10    mice). Several time characteristics of the response were recorded; the most sensitive and reproducible
11    measures of chemically-induced effects were determined to be the duration of tonic hind limb extension
12    in rats and the velocity of tonic extension in mice.

13           Linear regression analysis of the concentration-effect data was used to calculate an isoeffective
14    air concentration that corresponds to the concentration producing a 30% decrease in the maximal response
15    to an electrically-evoked seizure. The isoeffective air concentrations for 1,4-dioxane were 1,860 ±
16    200 ppm in rats and 2,400 ± 420 ppm in mice. A NOAEL value was not identified from this study.
      4.4.2.2   Goldberg etal.

17           Goldberg et al. (1964) evaluated the effect of solvent inhalation on pole climb performance in
18    rats. Female rats (Carworth Farms Elias strain) (eight per dose group) were exposed to 0, 1,500, 3,000, or
19    6,000 ppm of 1,4-dioxane in air for 4 hours/day, 5 days/weeks, for 10 exposure days. Conditioned
20    avoidance and escape behaviors were evaluated using a pole climb methodology. Prior to exposure, rats
21    were trained to respond to a buzzer or shock stimulus by using avoidance/escape behavior within
22    2 seconds. Behavioral criteria were the abolishment or significant deferment (>6 seconds) of the
23    avoidance response (conditioned or buzzer response) or the escape response (buzzer plus shock response).
24    Behavioral tests were administered on day 1, 2, 3, 4, 5, and 10 of the exposure period. Rat BWs were also
25    measured on test days.

26           1,4-Dioxane exposure produced a dose-related effect on conditioned avoidance behavior in
27    female rats, while escape behavior was generally not affected. In the 1,500 ppm group, only one of eight
28    rats had a decreased avoidance response, and this only occurred on days 2 and 5 of exposure. A larger
29    number of rats exposed to 3,000 ppm (two or three of eight) experienced a decrease  in the avoidance
30    response, and this response was observed on each day of the exposure period. The maximal decrease in
31    the avoidance response was observed in the 6,000 ppm group during the first 2 days of exposure
32    (75-100% of the animals were inhibited in this response). For exposure days 3-10, the percent of rats in
33    the 6,000 ppm group with significant inhibition of the avoidance response ranged from 37-62%. At the
34    end of the exposure period (day 10), the BWs for rats in the high exposure group were lower than
35    controls.
                                                                                                   64
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      4.4.2.3  Kanadaetal.

 1           Kanada et al. evaluated the effect of oral exposure to 1,4-dioxane on the regional neurochemistry
 2    of the rat brain (Kanadaetal.. 1994). 1,4-Dioxane was administered by gavage to male Sprague Dawley
 3    rats (5/group) at a dose of 1,050 mg/kg, approximately equal to one-fourth the oral LD50. Rats were
 4    sacrificed by microwave irradiation to the head 2 hours after dosing, and brains were dissected into small
 5    brain areas. Each brain region was analyzed for the content of biogenic amine neurotransmitters and their
 6    metabolites using high-performance liquid chromatography (HPLC) or GC methods. 1,4-Dioxane
 7    exposure was shown to reduce the dopamine and serotonin content of the hypothalamus. The
 8    neurochemical profile of all other brain regions in exposed rats was similar to control rats.
      4.4.2.4  Knoefel

 9           The narcotic potency of 1,4-dioxane was evaluated following i.p. injection in rats and gavage
10    administration in rabbits (Knoefel. 1935). Rats were given i.p. doses of 20, 30, or 50 mmol/kg. No
11    narcotic effect was seen at the lowest dose; however, rats given 30 mmol/kg were observed to sleep
12    approximately 8-10 minutes. Rats given the high dose of 50 mmol/kg died during the study. Rabbits were
13    given 1,4-dioxane at oral doses of 10, 20, 50, 75, or 100 mmol/kg. No effect on the normal erect animal
14    posture was observed in rabbits treated with less than 50 mmol/kg. At 50 and 75 mmol/kg, a semi-erect or
15    staggering posture was observed; lethality occurred at both the 75 and 100 mmol/kg doses.
      4.5  Mechanistic Data and Other Studies  in Support of the Mode of
           Action
      4.5.1   Genotoxicity

16           The genotoxicity data for 1,4-dioxane are presented in Table 4-23 and Table 4-24 for in vitro and
17    in vivo tests, respectively. 1,4-Dioxane has been tested for genotoxic potential using in vitro assay
18    systems with prokaryotic organisms, non-mammalian eukaryotic organisms, and mammalian cells, and in
19    vivo assay systems using several strains of rats and mice. In the large majority of in vitro systems,
20    1,4-dioxane was not genotoxic. Where a positive genotoxic response was observed, it was generally
21    observed in the presence of toxicity. Similarly, 1,4-dioxane was not genotoxic in the majority of available
22    in vivo studies. 1,4-Dioxane did not bind covalently to DNA in a single study with calf thymus DNA.
23    Several investigators have reported that 1,4-dioxane caused increased DNA synthesis indicative of cell
24    proliferation. Overall, the available literature indicates that 1,4-dioxane is nongenotoxic or weakly
25    genotoxic.

26           Negative findings were reported for mutagenicity in in vitro assays with the prokaryotic
27    organisms Salmonella typhimurium, Escherichia coli, and Photobacterium phosphoreum (Mutatox assay)
28    (Morita and Havashi. 1998; Hellmer and Bolcsfoldi.  1992; Kwanetal.. 1990; Khudolev et al.. 1987;
                                                                                                 65
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    Nestmann et al., 1984; Haworth et al., 1983; Stottet al., 1981). In in vitro assays with nonmammalian
 2    eukaryotic organisms, negative results were obtained for the induction of aneuploidy in yeast
 3    (Saccharomyces cerevisiae) and in the sex-linked recessive lethal test in Drosophila melanogaster (Yoon
 4    et al.. 1985; Zimmermann et al.. 1985). In the presence of toxicity, positive results were reported for
 5    meiotic nondisj unction in Drosophila (Munoz and Barnett. 2002).

 6           The ability of 1,4-dioxane to induce genotoxic effects in mammalian cells in vitro has been
 7    examined in model test systems with and without exogenous metabolic activation and in hepatocytes that
 8    retain their xenobiotic-metabolizing capabilities. 1,4-Dioxane was reported as negative in the mouse
 9    lymphoma cell forward mutation assay (Morita and Hayashi. 1998; McGregor et al..  1991). 1,4-Dioxane
10    did not produce chromosomal aberrations or micronucleus formation in Chinese hamster ovary (CHO)
11    cells (Morita and Hayashi. 1998; Galloway et al.. 1987). Results were negative in one assay for sister
12    chromatid exchange (SCE) in CHO (Morita and Hayashi. 1998) and were weakly positive in the absence
13    of metabolic activation in another (Galloway et al.. 1987). In rat hepatocytes, 1,4-dioxane exposure in
14    vitro caused single-strand breaks in DNA at concentrations also toxic to the hepatocytes (Sina et al..
15    1983) and produced a positive genotoxic response in a cell transformation assay with BALB/3T3 cells
16    also in the presence of toxicity (Sheu et al.. 1988).

17           1,4-Dioxane was not genotoxic in the majority of available in vivo mammalian assays. Studies of
18    micronucleus formation following in vivo exposure to 1,4-dioxane produced mostly negative results,
19    including studies of bone marrow micronucleus formation in B6C3Fi, BALB/c, CBA, and  C57BL6 mice
20    (McFee etal.. 1994; Mirkova. 1994; Tinwell and Ashby. 1994) and micronucleus formation in peripheral
21    blood of GDI mice (Morita and Hayashi. 1998; Morita. 1994). Mirkova (1994) reported a dose-related
22    increase in the incidence of bone marrow micronuclei in male and female C57BL6 mice 24 or 48 hours
23    after administration of 1,4-dioxane. At a sampling time of 24 hours, a dose of 450 mg/kg produced no
24    change relative to control, while doses of 900, 1,800, and 3,600 mg/kg increased the incidence of bone
25    marrow micronuclei by approximately two-, three-, and fourfold, respectively. A dose of 5,000 mg/kg
26    also increased the incidence of micronuclei by approximately fourfold at 48 hours. This compares with
27    the negative results for BALB/c male mice tested in the same study at a dose of 5,000 mg/kg and
28    sampling time of 24 hours. Tinwell and Ashby (1994) could not explain the difference in response in the
29    mouse bone marrow micronucleus assay with C57BL6 mice obtained in their laboratory (i.e.,
30    non-significant 1.6-fold  increase over control) with the dose-related positive findings reported by
31    Mirkova (Mirkova. 1994) using the same mouse strain, 1,4-dioxane dose (3,600 mg/kg) and sampling
32    time (24 hours).  Morita and Hayashi (1998) demonstrated an increase in micronucleus formation in
33    hepatocytes following 1,4-dioxane dosing and partial hepatectomy to induce cellular mitosis. DNA
34    single-strand breaks were demonstrated in hepatocytes following gavage exposure to female rats (Kitchin
35    and Brown. 1990).

36           Roy et al. (2005) examined micronucleus formation in male CD1 mice exposed to  1,4-dioxane to
37    confirm the mixed findings from earlier mouse micronucleus studies and to identify the origin of the
38    induced micronuclei. Mice were administered 1,4-dioxane by gavage at doses of 0, 1,500, 2,500, and
39    3,500 mg/kg-day for 5 days. The mice were also implanted with 5-bromo-2-deoxyuridine
40    (BrdU)-releasing osmotic pumps to measure cell proliferation in the liver and to increase the sensitivity of
                                                                                                  66
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    the hepatocyte assay. The frequency of micronuclei in the bone marrow erythrocytes and in the
 2    proliferating BrdU-labeled hepatocytes was determined 24 hours after the final dose. Significant
 3    dose-related increases in micronuclei were seen in the bone-marrow at all the tested doses (>
 4    1,500 mg/kg-day). In the high-dose (3,500-mg/kg) mice, the frequency of bone marrow erythrocyte
 5    micronuclei was about 10-fold greater than the control frequency. Significant dose-related increases in
 6    micronuclei were also observed at the two highest doses (> 2,500 mg/kg-day) in the liver.
 7    Antikinetochore (CREST) staining or pancentromeric fluorescence in situ hybridization (FISH) was used
 8    to determine the origin of the induced micronuclei. The investigators determined that 80-90% of the
 9    micronuclei in both tissues originated from chromosomal breakage; small increase in micronuclei
10    originating from chromosome loss was seen in hepatocytes. Dose-related statistically significant
11    decreases in the ratio of bone marrow polychromatic erythrocytes (PCE):normochromatic erythrocytes
12    (NCE), an indirect measure of bone marrow toxicity, were observed. Decreases in hepatocyte
13    proliferation were also observed. Based on these results, the authors concluded that at high doses
14    1,4-dioxane exerts genotoxic effects in both the mouse bone marrow and liver; the induced micronuclei
15    are formed primarily from chromosomal breakage; and 1,4-dioxane can interfere with cell proliferation in
16    both the liver and bone marrow. The authors noted that reasons for the discrepant micronucleus assay
17    results among various investigators was unclear, but could be related to the inherent variability present
18    when detecting moderate to weak responses using small numbers of animals, as well as differences in
19    strain, dosing regimen, or scoring criteria.

20            1,4-Dioxane did not affect in vitro or in vivo DNA repair in hepatocytes  or in vivo DNA repair in
21    the nasal cavity (Goldsworthy et al.. 1991; Stott etal.. 1981). but increased hepatocyte DNA  synthesis
22    indicative of cell proliferation in several  in vivo studies (Miyagawa et al.. 1999; Uno et al.. 1994;
23    Goldsworthy et al., 1991; Stott et al., 1981). 1,4-Dioxane caused a transient inhibition of RNA
24    polymerase A and B in the rat liver (Kurl et al.. 1981). indicating a negative impact on the synthesis of
25    ribosomal and messenger RNA (DNA transcription). Intravenous administration of 1,4-dioxane at doses
26    of 10 or 100 mg/rat produced inhibition of both polymerase enzymes, with a quicker and more complete
27    recovery of activity for RNA polymerase A, the polymerase for ribosomal RNA synthesis.

28            1,4-Dioxane did not covalently bind to DNA under in vitro study conditions  (Woo et al.. 1977b).
29    DNA alkylation was also not detected in the liver 4  hours following a single gavage exposure
30    (1,000 mg/kg) in male Sprague Dawley rats (Stott etal.. 1981).

31           Rosenkranz and Klopman (1992) analyzed  1,4-dioxane using the computer automated structure
32    evaluator (CASE) structure activity method to predict its potential genotoxicity and carcinogenicity. The
33    CASE analysis is based on information contained in the structures of approximately 3,000 chemicals
34    tested for endpoints related to mutagenic/genotoxic  and carcinogenic potential. CASE selects descriptors
35    (activating [biophore] or inactivating [biophobe] structural fragments) from a learning set of active and
36    inactive molecules. Using the CASE methodology, Rosenkranz and Klopman (1992) predicted that
37    1,4-dioxane would be inactive for mutagenicity in several in vitro systems, including Salmonella,
38    induction of chromosomal aberrations in CHO cells, and unscheduled DNA synthesis in rat hepatocytes.
39    1,4-Dioxane was predicted to induce  SCE in cultured CHO cells, micronuclei formation in rat bone
40    marrow, and carcinogenicity in rodents.
                                                                                                   67
                                     DRAFT - DO NOT CITE OR QUOTE

-------
1           Gene expression profiling in cultured human hepatoma HepG2 cells was performed using DNA
2    microarrays to discriminate between genotoxic and other carcinogens (van Delft et al.. 2004). Van Delft
3    et al. (2004) examined this method using a training set of 16 treatments (nine genotoxins and seven
4    nongenotoxins) and a validation set (three and three), with discrimination models based on Pearson
5    correlation analyses for the 20 most discriminating genes. As reported by the authors (van Delft et al..
6    2004). the gene expression profile for 1,4-dioxane indicated a classification of this chemical as a
7    "nongenotoxic" carcinogen, and thus, 1,4-dioxane was included in the training set as a "nongenotoxic"
8    carcinogen. The accuracy for carcinogen classification using this method ranged from 33 to 100%,
9    depending on which chemical data sets and gene expression signals were included in the analysis.
                                                                                                  68
                                    DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-23  Genotoxicity studies of 1,4-dioxane; in vitro
Test system
Endpoint
Results3
Test conditions Without With
activation activation
Doseb
Source
Prokaryotic organisms in vitro
S. typhimurium
strains TA98,
TA100, TA1535,
TA1537
S. typhimurium
strains TA98,
TA100, TA1530,
TA1535,
TA1537
S. typhimurium
strains TA98,
TA100, TA1535,
TA1537
S. typhimurium
strains TA100,
TA1535
S. typhimurium
strains TA98,
TA100, TA1535,
TA1537,
TA1538
£. CO//K-12
uvrB/recA
£. co//
WP2/WP2uvrA
P. phosphoreum
M169
Nonmammalian
S. cerevisiae
D61.M
D. melanogaster
D. melanogaster
Reverse
mutation
Reverse
mutation
Reverse
mutation
Reverse
mutation
Reverse
mutation
DMA repair
Reverse
mutation
Mutagenicity,
DMA damage
Plate incorporation assay
Plate incorporation assay - -
Plate incorporation and
preincubation assays
Preincubation assay - -
Plate incorporation assay
Host mediated assay
Plate incorporation and
preincubation assays
Mutatox assay - ND
10,000 ug/plate
ND
5,000 ug/plate
103 mg
103 mg
1,150mmol/L
5,000 ug/plate
ND
Haworth et
al. (1983)
Khudoley et
al. (1987)
Morita and
Hayashi
(1998)
Nestmann et
al. (1984)
Stott et al.
(1981)
Hellmer and
Bolcsfoldi
(1992)
Morita and
Hayashi
(1998)
Kwan et al.
(1990)
eukaryotic organisms in vitro
Aneuploidy
Meiotic
nondisjunction
Sex-linked
recessive lethal
test
Standard 16-hour
incubation or -T ND
cold-interruption regimen
Oocytes were obtained for
evaluation 24 and +TC NDd
48 hours after mating
Exposure by feeding and Nr.d
injection
4.75%
2% in sucrose
media
35,000 ppm in
feed, 7 days or
50,000 ppm
(5% in water)
by injection
Zimmerman
et al. (1985)
Munoz and
Barnett
(2002)
Yoon et al.
(1985)
Mammalian cells in vitro
Rat hepatocytes
Primary
hepatocyte
culture from
male F344 rats
L5178Y mouse
lymphoma cells
L5178Y mouse
lymphoma cells
DMA damage;
single-strand
breaks
measured by
alkaline elution
DMA repair
Forward
mutation assay
Forward
mutation assay
3-Hour exposure to
isolated primary +Te NDd
hepatocytes
Autoradiography - NDd
Thymidine kinase
mutagenicity assay
(trifluorothymidine
resistance)
Thymidine kinase
mutagenicity assay T
(trifluorothymidine
resistance)
0.3 mM
1 mM
5,000 ug/mL
5,000 ug/mL
Sina et al.
(1983)
Goldsworthy
et al. (1991)
McGregor et
al. (1991)
Morita and
Hayashi
(1998)
                             DRAFT - DO NOT CITE OR QUOTE
                                                                                      69

-------
BALB/3T3 cells
     Cell
transformation
48-Hour exposure
followed by 4 weeks
incubation;  13 day
exposure followed by
2.5 weeks incubation
+Tf
NDa
0.5 mg/mL
Sheu et al.
  (1988)
                                   BrdU was added 2 hours
CHO cells
CHO cells
CHO cells
SCE
Chromosomal
aberration
SCE
auer i,t-uioxane auumon,
chemical treatment was ±g
2 hours with S9 and
25 hours without S9
Cells were harvested 8-
12 hours or 18-26 hours
after treatment (time of
first mitosis)
3 hour pulse treatment;
followed by continuous
treatment of BrdU for
23 or 26 hours
1 0,520 ug/ml_
1 0,520 ug/ml_
5,000 ug/mL
Galloway et
al. (1987)
Galloway et
al. (1987)
Morita and
Hayashi
(1998)
CHO cells
Chromosomal
  aberration
5 hour pulse treatment,
20 hour pulse and
continuous treatments, or
44 hour continuous
treatment; cells were
harvested 20 or 44 hours
following exposure	
                     5,000 ug/mL
                          Morita and
                           Hayashi
                            (1998)
CHO cells
 Micronucleus
   formation
5 hour pulse treatment or
44 hour continuous
treatment; cells were
harvested 42 hours
following exposure	
                     5,000 ug/mL
                          Morita and
                           Hayashi
                            (1998)
Calf thymus
DMA
   Covalent
binding to DMA
Incubation with
microsomes from
3-methylcholanthrene
treated rats
                    0.04 pmol/mg
                     DMA (bound)
                          Woo et al.
                           (1977b)
a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity. Endogenous metabolic activation is not
   applicable for in vivo studies.
Yowest effective dose for positive results/highest dose tested for negative results; ND = no data.
°Rats were given doses of 0, 168, 840, 2,550, or 4,200 mg/kg at 4 and 21 hours prior to sacrifice. A 43 and 50% increase
   in the fraction of DMA eluted was observed for doses of 2,550 and 4,200 mg/kg, respectively. Alkaline elution of DMA
   was not significantly different from control in the two lowest dose groups (168 and 840 mg/kg).
dA dose-related increase in the incidence of bone marrow micronuclei was observed in male and female C57BL6 mice 24
   or 48 hours after administration of 1,4-dioxane. A dose of 450 mg/kg produced no change relative to control, while
   doses of 900, 1,800, 3,600, and 5,000 mg/kg increased the incidence of bone marrow micronuclei by approximately
   two-,three-, four- and fourfold, respectively.
eA dose-related increase in the incidence of hepatocyte micronuclei was observed in partially hepatectomized mice 6 days
   after administration of 1,4-dioxane. A dose of 1,000 mg/kg produced no change  relative to control, while doses of
   2,000 and 3,000 mg/kg increased the incidence of hepatocyte micronuclei by 2.4- and 3.4-fold, respectively.
'Significant increases in the frequency of micronucleated erythrocytes were observed at each test dose of 1,4-dioxane (1,500, 2,500
   and 3,500 mg/kg-day, 5 days/week).
9A dose-related increase in the frequency of micronuclei was observed in proliferating cells with micronuclei at 2,500 and
   3,500 mg/kg-day, 5 days/week. No increase in the frequency of micronuclei was seen in the non-proliferating cells.
hNo increase in the hepatocyte labeling index was observed 24 or 48 hours following a single gavage exposure of
   1,000 mg/kg. Continuous administration of 1 % 1,4-dioxane in the drinking water for up to 2 weeks produced a twofold
   increase in the hepatocyte labeling index.
A similar pattern of RNA polymerase inhibition was observed at doses of 10 and 100 mg/rat. Inhibition was more
   pronounced at the higher dose.
'Hepatocyte viability was 86, 89, 87, 88, 78, and 86% 24 hours following exposure to 0,  1,000, 1,500, 2,000, or
   4,000 mg/kg. The incidence (%) of replicative DMA synthesis was increased by 2.5-fold (1,000 mg/kg) or 4.5-fold
   (1,500 and 2,000 mg/kg). No increase in replicative DNA synthesis was observed at the  highest dose (4,000 mg/kg).
kReplicative DNA synthesis was measured 24, 39, and 48 hours following a single dose of 0, 1,000, or 2,000 mg/kg.
   Hepatocyte viability ranged from 71 to 82%. The only increase in replicative DNA synthesis was observed 24 hours
   after administration of 2,000 mg/kg (threefold increase). Cell viability for this group was 79%.
'Replicative DNA synthesis was increased 1.5-fold in rats given 1,000 mg/kg of 1,4-dioxane for 11 weeks. No change from
   control was observed in rats exposed to 10 mg/kg for 11 weeks or rats acutely exposed to 10, 100, or 1,000 mg/kg.
                                      DRAFT - DO NOT CITE OR QUOTE
                                                                                                                70

-------
Table 4-24  Genotoxicity studies of 1,4-dioxane; mammalian in vivo
Test system
Female
Sprague Dawley
Rat
Male
Sprague Dawley
Rat
Male
B6C3F-,
Mouse
Male and female
C57BL6
Mouse;
male BALB/c
Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
CBA and
C57BL6 Mouse
Male
CD1
Mouse
Male
CD1
Mouse
Male
Sprague Dawley
Rat
Test system
Male
F344
Rat
Male
F344
Rat
Endpoint
DMA damage;
single-strand breaks
measured by alkaline
elution
DMA alkylation in
hepatocytes
Micronucleus
formation in bone
marrow
Micronucleus
formation in bone
marrow
Micronucleus
formation in
peripheral blood
Micronucleus
formation in
hepatocytes
Micronucleus
formation in
peripheral blood
Micronucleus
formation in bone
marrow
Micronuclei formation
in bone marrow
Micronuclei formation
in hepatocytes
DMA repair in
hepatocytes
Endpoint
DMA repair in
hepatocytes
(autoradiography)
DMA repair in nasal
epithelial cells from
the nasoturbinate or
maxilloturbinate
Test Conditions Results3
Two gavage doses given 21
and 4 hours prior to +c
sacrifice
Gavage; DMA isolation and
HPLC analysis 4 hours after
dosing
i.p. injection; analysis of
polychromatic erythrocytes -
24 or 48 hours after dosing
Gavage; analysis of + /C57BL6xd
polychromatic erythrocytes /DAI R/ (
24 or 48 hours after dosing ~(bALb/c)
Two i.p. injections (1/day);
micronucleated
reticulocytes measured 24, -
48, and 72 hours after the
2nd dose
Gavage, partial
hepatectomy 24 hours after
dosing, hepatocytes +e
analyzed 5 days after
hepatectomy
Gavage, partial
hepatectomy 24 hours after
dosing, peripheral blood -
obtained from tail vein
24 hours after hepatectomy
Gavage; analysis of
polychromatic erythrocytes
from specimens prepared
24 hours after dosing
Gavage; analysis for
micronucleated erythrocytes +f
24 hours after dosing
Gavage; analysis for
micronuclei 24 hours after +9
dosing
Drinking water; thymidine
incorporation with
hydroxyurea to repress
normal DMA synthesis
Test Conditions Results3
Gavage and drinking water
exposure; thymidine
incorporation
Gavage and drinking water
exposure; thymidine
incorporation
Dose0
2,550 mg/kg
1,000 mg/kg
Single dose of
4,000 mg/kg;
3 daily doses of
2,000
900 mg/kg
(C57BL6);
5,000 mg/kg
(BALB/c)
3,200 mg/kg
2,000 mg/kg
3,000 mg/kg
3,600 mg/kg
1,500 mg/kg-day
for 5 days
2,500 mg/kg-day
for 5 days
1,000 mg/kg-day
for 1 1 weeks
Dose0
1,000 mg/kg for
2 or 12 hours;
1,500 mg/kg-day
for 2 weeks or
3,000 mg/kg-day
for 1 week
1,500 mg/kg-day
for 8 days +
1,000 mg/kg
gavage dose
12 hours prior to
sacrifice
Source
Kitchin and
Brown (1990)
Stott et al.
(1981)
McFee et al.
(1994)
Mirkova
(1994)
Morita (1994)
Morita and
Hayashi
(1998)
Morita and
Hayashi
(1998)
Tinwell and
Ashby (1994)
Roy et al.
(2005)
Roy et
al.(2005)
Stott et al.
(1981)
Source
Goldsworthy
et al. (1991)
Goldsworthy
et al. (1991)
                            DRAFT - DO NOT CITE OR QUOTE
                                                                                    71

-------
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Male
F344
Rat
Male
F344
Rat
Male
Sprague Dawley
Rat
Replicative DMA
synthesis (i.e., cell
proliferation) in
hepatocytes
Replicative DMA
synthesis (i.e., cell
proliferation) in nasal
epithelial cells
RNA synthesis;
inhibition of RNA
polymerase A and B
DMA synthesis in
hepatocytes
DMA synthesis in
hepatocytes
DMA synthesis in
hepatocytes
Gavage and drinking water +h
exposure; thymidine (1-2-week
incorporation exposure)
Drinking water exposure;
thymidine incorporation
i.v. injection; activity
measured in isolated +'
hepatocytes
Gavage; thymidine and +j
BrdU incorporation
Thymidine incorporation ±k
Drinking water; thymidine i
incorporation
1,000 mg/kg for
24 or 48 hours;
1,500 mg/kg-day
for 1 or 2 weeks
1,500 mg/kg-day
for 2 weeks
10 mg/rat
1,000 mg/kg
2,000 mg/kg
1,000 mg/kg-day
for 1 1 weeks
Goldsworthy
et al. (1991)
Goldsworthy
et al. (1991)
Kurl et al.
(1981)
Miyagawa
(1999)
Uno et al.
(1994)
Stott et al.
(1981)
      a+ = positive, ± = equivocal or weak positive, - = negative, T = toxicity. Endogenous metabolic activation is not
         applicable for in vivo studies.
      Yowest effective dose for positive results/highest dose tested for negative results; ND = no data.
      °Rats were given doses of 0, 168, 840, 2,550, or 4,200 mg/kg at 4 and 21 hours prior to sacrifice. A 43 and 50%
         increase in the fraction of DMA eluted was observed for doses of 2,550 and 4,200 mg/kg, respectively. Alkaline
         elution of DMA was  not significantly different from control in the two lowest dose groups (168 and 840 mg/kg).
      dA dose-related increase in the incidence of bone marrow micronuclei was observed in male and female C57BL6 mice
         24 or 48 hours after administration of 1,4-dioxane. A dose of 450 mg/kg produced no change relative to control,
         while doses of 900,  1,800, 3,600, and 5,000 mg/kg increased the incidence of bone marrow micronuclei by
         approximately two-,three-, four- and fourfold, respectively.
      eA dose-related increase in the incidence of hepatocyte micronuclei was observed in partially hepatectomized mice
         6 days after administration of 1,4-dioxane. A dose of 1,000 mg/kg produced no change relative to control, while
         doses of 2,000 and  3,000 mg/kg increased the incidence of hepatocyte micronuclei by 2.4- and 3.4-fold,
         respectively.
      'Significant increases in the frequency of micronucleated erythrocytes were observed at each test dose of 1,4-dioxane (1,500,
         2,500 and 3,500 mg/kg-day, 5 days/week).
      9A dose-related increase in the frequency of micronuclei was observed in proliferating  cells with micronuclei at 2,500 and
         3,500 mg/kg-day, 5  days/week. No increase in the frequency of micronuclei was seen  in the non-proliferating cells.
      hNo increase in the  hepatocyte labeling index was observed 24 or 48 hours following a single gavage exposure of
         1,000 mg/kg. Continuous administration of 1 % 1,4-dioxane in the drinking water for up to 2 weeks produced a
         twofold increase in the hepatocyte labeling index.
      A similar pattern of RNA polymerase inhibition was observed at doses of 10 and 100 mg/rat. Inhibition was more
         pronounced at the higher dose.
      'Hepatocyte viability was 86,  89, 87, 88, 78, and 86% 24 hours following exposure to 0, 1,000, 1,500, 2,000, or
         4,000 mg/kg. The incidence (%) of replicative DNA synthesis was increased by 2.5-fold (1,000 mg/kg) or 4.5-fold
         (1,500 and 2,000 mg/kg). No increase  in replicative DNA synthesis was observed at the highest dose (4,000 mg/kg).
      kReplicative  DNA synthesis was measured 24, 39, and 48 hours following a single dose of 0, 1,000, or 2,000 mg/kg.
         Hepatocyte viability ranged from 71 to  82%.  The only increase in replicative DNA synthesis was observed 24 hours
         after administration  of 2,000 mg/kg (threefold increase). Cell viability for this group was 79%.
      'Replicative DNA synthesis was increased  1.5-fold in rats given 1,000 mg/kg of 1,4-dioxane for  11 weeks. No change
         from control was observed in rats exposed to 10 mg/kg for 11 weeks or rats acutely exposed to 10, 100, or
         1,000 mg/kg.
      4.5.2  Mechanistic Studies
      4.5.2.1   Free Radical Generation

1             Burmistrov et al. (2001) investigated the effect of 1,4-dioxane inhalation on free radical processes
2     in the rat ovary and brain. Female rats (6-9/group, unspecified strain) were exposed to 0, 10, or
3     100 mg/m3 of 1,4-dioxane vapor for 4 hours/day, 5 days/week, for 1 month. Rats were sacrificed during
                                                                                                                     72
                                           DRAFT - DO NOT CITE OR QUOTE

-------
 1    the morning or evening following exposure and the ovaries and brain cortex were removed and frozen.
 2    Tissue preparations were analyzed for catalase activity, glutathione peroxidase activity, and protein
 3    peroxidation. Inhalation of 100 mg/m3 of 1,4-dioxane resulted in a significant increase (p < 0.05) in
 4    glutathione peroxidase activity, and activation of free radical processes were apparent in both the rat
 5    ovary and brain cortex. No change in catalase activity or protein peroxidation was observed at either
 6    concentration. A circadian rhythm for glutathione peroxidase activity was absent in control rats, but
 7    occurred in rat brain and ovary following 1,4-dioxane exposure.
      4.5.2.2   Induction of Metabolism

 8           The metabolism of 1,4-dioxane is discussed in detail in Section 3.3. 1,4-Dioxane has been shown
 9    to induce its own metabolism (Young et al., 1978a; 1978b). Nannelli et al. (2005) (study details provided
10    in Section 3.3) characterized the CYP450 isozymes that were induced by 1,4-dioxane in the liver, kidney,
11    and nasal mucosa of the rat. In the liver, the activities of several CYP450 isozymes were increased (i.e.,
12    CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was inducible in the kidney and nasal mucosa.
13    CYP2E1 mRNA was increased approximately two- to threefold in the kidney and nasal mucosa, but
14    mRNA levels were not increased in the liver, suggesting that regulation of CYP2E1 is organ-specific.
15    Induction of hepatic CYPB1/2 and CYP2E1 levels by phenobarbital or fasting did not increase the liver
16    toxicity of 1,4-dioxane, as measured by hepatic glutathione content or serum ALT activity. This result
17    suggested that highly reactive and toxic intermediates did not play a large role in the liver toxicity of
18    1,4-dioxane, even under conditions where metabolism was enhanced. This finding is similar to an earlier
19    conclusion by Kociba et al. (1975) who evaluated toxicity from a chronic drinking water study alongside
20    data providing a pharmacokinetic profile for 1,4-dioxane. Kociba et al. (1975) concluded that liver
21    toxicity and eventual tumor formation occurred only at doses where clearance pathways were saturated
22    and elimination of 1,4-dioxane from the blood was reduced. Nannelli et al. (2005)  further suggested that a
23    sustained induction of CYP2E1 may lead to generation of reactive oxygen species contributing to target
24    organ toxicity and regenerative cell proliferation; however, no data were provided  to support this
25    hypothesis.
      4.5.2.3   Mechanisms of Tumor Induction

26           Several studies have been performed to evaluate potential mechanisms for the carcinogenicity of
27    1,4-dioxane (Goldsworthy et al.. 1991; Kitchin and Brown. 1990; Stottetal. 1981). Stott et al. (1981)
28    evaluated 1,4-dioxane in several test systems, including salmonella mutagenicity in vitro, rat hepatocyte
29    DNA repair activity in vitro, DNA synthesis determination in male Sprague Dawley rats following acute
30    gavage dosing or an 11-week drinking water exposure (described in Section 4.2.1), and hepatocyte DNA
31    alkylation and DNA repair following a single gavage dose. This study used doses of 0, 10, 100, or
32    1,000 mg/kg-day, with the highest dose considered to be a tumorigenic dose level. Liver histopathology
33    and liver to BW ratios were also evaluated in rats from acute gavage or repeated dose drinking water
34    experiments.
                                                                                                  73
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           The histopathology evaluation indicated that liver cytotoxicity (i.e., centrilobular hepatocyte
 2    swelling) was present in rats from the 1,000 mg/kg-day dose group that received 1,4-dioxane in the
 3    drinking water for 11 weeks (Stott et al., 1981). An increase in the liver to BW ratio accompanied by an
 4    increase in hepatic DNA synthesis was also seen in this group of animals. No effect on histopathology,
 5    liver weight, or DNA synthesis was observed in acutely exposed rats or rats that were exposed to a lower
 6    dose of 10 mg/kg-day for 11 weeks. 1,4-Dioxane produced negative findings in the remaining
 7    genotoxicity assays conducted as part of this study (i.e., Salmonella mutagenicity, in vitro and  in vivo rat
 8    hepatocyte DNA repair, and DNA alkylation in rat liver). The study authors suggested that the observed
 9    lack of genotoxicity at tumorigenic and cytotoxic dose levels  indicates an epigenetic mechanism for
10    1,4-dioxane hepatocellular carcinoma in rats.

11           Goldsworthy et al. (1991) evaluated potential mechanisms for the nasal and liver carcinogenicity
12    of 1,4-dioxane in the rat. DNA repair activity was evaluated as a measure of DNA reactivity and DNA
13    synthesis was measured as an indicator of cell proliferation or promotional activity. In vitro DNA repair
14    was evaluated in primary hepatocyte cultures from control and 1,4-dioxane-treated rats (1 or 2% in the
15    drinking water for 1 week). DNA repair and DNA synthesis were also measured in vivo following a
16    single gavage dose of 1,000 mg/kg,  a drinking water exposure of 1% (1,500 mg/kg-day) for  1 week, or a
17    drinking water exposure of 2% (3,000 mg/kg-day) for 2 weeks. Liver to BW ratios and palmitoyl CoA
18    oxidase activity were measured in the rat liver to determine whether peroxisome proliferation played a
19    role in the liver carcinogenesis of 1,4-dioxane.  In vivo DNA repair was evaluated in rat nasal epithelial
20    cells derived from either the nasoturbinate or the maxilloturbinate of 1,4-dioxane-treated rats. These rats
21    received 1% 1,4-dioxane (1,500 mg/kg-day) in the drinking water for 8 days, followed by a single gavage
22    dose of 10,  100, or 1,000 mg/kg 12 hours prior to sacrifice. Archived tissues from the NCI (1978)
23    bioassay were reexamined to determine the primary sites for tumor formation in the nasal cavity
24    following chronic exposure in rats. Histopathology  and cell proliferation were determined for specific
25    sites in the nasal cavity that were related to tumor formation.  This evaluation was performed in rats that
26    were exposed to drinking water containing 1% 1,4-dioxane (1,500 mg/kg-day)  for 2 weeks.

27           1,4-Dioxane and its metabolite l,4-dioxane-2-one did not affect in vitro DNA repair in primary
28    hepatocyte cultures (Goldsworthy et al., 1991). In vivo DNA  repair was also unaffected by acute gavage
29    exposure or ingestion of 1,4-dioxane in the drinking water for a 1- or 2-week period. Hepatocyte cell
30    proliferation was not affected by acute gavage exposure, but was increased approximately twofold
31    following a 1-2-week drinking water exposure. A 5-day drinking water exposure to 1% 1,4-dioxane
32    (1,500 mg/kg-day) did not increase the activity of palmitoyl coenzyme A or the liver to BW  ratio,
33    suggesting that peroxisome proliferation did not play a role in the hepatocarcinogenesis of 1,4-dioxane.
34    Nannelli et al. (2005) also reported a lack of hepatic palmitoyl CoA induction following 10 days of
35    exposure to 1.5% 1,4-dioxane in the drinking water (2,100 mg/kg-day).

36           Treatment of rats with 1% (1,500 mg/kg-day) 1,4-dioxane for 8 days did not alter DNA repair in
37    nasal epithelial cells (Goldsworthy et al.,  1991). The addition of a single gavage dose of up to
38    1,000  mg/kg 12 hours prior to sacrifice also did not induce DNA repair. Reexamination of tissue sections
39    from the NCI (1978) bioassay suggested that the majority of nasal tumors were located in the dorsal nasal
40    septum or the nasoturbinate of the anterior portion of the dorsal meatus (Goldsworthy et al..  1991). No
                                                                                                    74
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    histopathological lesions were observed in nasal section of rats exposed to drinking water containing 1%
 2    1,4-dioxane (1,500 mg/kg-day) for 2 weeks and no increase was observed in cell proliferation at the sites
 3    of highest tumor formation in the nasal cavity.

 4           Female Sprague Dawley rats (three to nine per group) were given 0, 168, 840, 2,550, or
 5    4,200 mg/kg 1,4-dioxane (99% purity) by corn oil gavage in two doses at 21 and 4 hours prior to sacrifice
 6    (Kitchin and Brown. 1990). DNA damage (single-strand breaks measured by alkaline elution), ODC
 7    activity, reduced glutathione content, and CYP450 content were measured in the liver. Serum ALT
 8    activity and liver histopathology were also evaluated. No changes were observed in hepatic reduced
 9    glutathione content or ALT activity. Light microscopy revealed minimal to mild vacuolar degeneration in
10    the cytoplasm of hepatocytes from three of five rats from the 2,550 mg/kg dose group. No
11    histopathological lesions were seen in any other dose group, including rats given a higher dose of
12    4,200 mg/kg. 1,4-Dioxane caused 43 and 50% increases in DNA single-strand breaks at dose levels of
13    2,550 and 4,200 mg/kg, respectively. CYP450 content was also increased at the two highest dose levels
14    (25 and 66% respectively). ODC activity was increased approximately two-, five-, and eightfold above
15    control values at doses of 840, 2,550, and 4,200 mg/kg, respectively. The results of this study
16    demonstrated that hepatic DNA damage can occur in the absence of significant cytotoxicity. Parameters
17    associated  with tumor promotion (i.e., ODC activity,  CYP450 content) were also elevated, suggesting that
18    promotion may play a role in the carcinogenesis of 1,4-dioxane.
      4.6   Synthesis  of Major  Noncancer Effects

19           Liver, kidney, and nasal toxicity were the primary noncancer health effects associated with
20    exposure to 1,4-dioxane. In humans, several fatal cases of hemorrhagic nephritis and centrilobular
21    necrosis of the liver were related to occupational exposure (i.e., inhalation and dermal contact) to
22    1,4-dioxane (Johnstone. 1959; Barber. 1934). Neurological changes were also reported in one case;
23    including, headache, elevation in blood pressure, agitation and restlessness, and coma (Johnstone.  1959).
24    Perivascular widening was observed in the brain of this worker, with small foci of demyelination in
25    several regions (e.g., cortex, basal nuclei). In laboratory animals, following oral and inhalation exposure
26    to 1.4-dioxane. liver and kidney degeneration and necrosis were observed(JBRC. 1998; Drew et al. 1978;
27    David. 1964; Kesten et al., 1939; Laugetal.. 1939; Schrenk and Yant. 1936; de Navasquez. 1935; Fairley
28    etal..  1934). m addition to changes in the nasal epithelium (JBRC. 1998)(Kano et al.. 2008)(Kano et al..
29    2009)(Kasai et al.. 2008)(Kasai et al.. 2009). The results of subchronic and chronic studies are discussed
30    below.
      4.6.1   Oral

31           Table 4-25 presents a summary of the noncancer results for the subchronic and chronic oral
32    studies of 1,4-dioxane toxicity in experimental animals. Liver and kidney toxicity were the primary
33    noncancer health effects of oral exposure to 1,4-dioxane in animals. Kidney damage at high doses was

                                                                                                  75
                                     DRAFT - DO NOT CITE OR QUOTE

-------
1    characterized by degeneration of the cortical tubule cells, necrosis with hemorrhage, and
2    glomerulonephritis (NCI. 1978; Kocibaetal. 1974; Argus etal.. 1965; Fairley et al.. 1934). Renal cell
3    degeneration generally began with cloudy swelling of cells in the cortex (Fairley et al.. 1934). Nuclear
4    enlargement of proximal tubule cells was observed at doses below those producing renal necrosis  (Kano
5    et al.. 2008; JBRC. 1998). but is of uncertain toxicological significance. The lowest dose reported to
6    produce kidney damage was 94 mg/kg-day, which produced renal degeneration and necrosis of tubule
7    epithelial cells in male rats in the Kociba et al. (1974) study.  Cortical tubule degeneration was seen at
8    higher doses in the NCI (1978) bioassay (240 mg/kg-day, male rats), and glomerulonephritis was reported
9    for rats given doses of > 430 mg/kg-day (Argus etal.. 1973; Argus et al.. 1965).
                                                                                                  76
                                    DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-25  Oral toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration , ^ , , L°AEHL ,
(mg/kg-day) (mg/kg-day)
Effect
Reference
Subchronic studies
Rat and Mouse
(6/species);
unknown strain
Male
Sprague Dawley
Rat
(4-6/group)
F344/DuCrj Rat
(10/sex/group)
Crj:BDF1 Mouse
(10/sex/group)
Rats
0 or 1,900 mg/kg-day;
Mice NA J'lnn •
0 or 3,300 mg/kg-day 3,300 mice
for 67 days
Rats
0, 10, or 1,000 mg/kg-day 10 1,000
for 1 1 weeks
Rats
Males 0, 52, 126,274,
657, or 1,554 mg/kg-day; „ ,9R
Females 0, 83, 185, 427,
756, or 1,614 mg/kg-day
for 13 weeks
Mice
Males 0, 86,231, 585,
882, or 1,570 mg/kg-day;
Females 0, 170, 387, 170 387
898, 1,620, or
2,669 mg/kg-day
for 13 weeks
Renal cortical degeneration
and necrosis, hemorrhage;
hepatocellular degeneration
Minimal centrilobular
hepatocyte swelling;
increased DMA synthesis
Nuclear enlargement of
nasal respiratory
epithelium; hepatocyte
swelling
Nuclear enlargement of
bronchial epithelium
Fairley et al.
(1934)
Stott et al.
(1981)
Kano et al.
(2008)
Kano et al.
(2008)
Chronic studies
Male
Wistar
Rat (26 treated,
9 controls)
Male
Sprague Dawley
Rat (30/group)
Sherman Rat
(60/sex/dose
group)
Osborne-Mendel
Rat
(35/sex/dose
level)
B6C3F-I Mouse
(50/sex/dose
level)
F344/DuCrj Rat
(50/sex/dose
level)
Rats
0 or 640 mg/kg-day NA 640
for 63 weeks
Rats
0, 430, 574, 803, or
1,032 mg/kg-day NA ^JU
for 13 months
Rats
Males 0, 9.6, 94, or
1,01 5 mg/kg-day; qfi q.
Females 0, 19, 148, or
1,599 mg/kg-day
for 2 years
Rats
Males 0, 240, or
530 mg/kg-day;
Females 0, 350, or
640 mg/kg-day
for 110 weeks
Mice
Males 0, 720, or
830 mg/kg-day;
Females 0, 380, or NA dOU
860 mg/kg-day
for 90 weeks
Rats
Males 0, 11, 55, or
274 mg/kg-day; „ „.
Females 0, 18, 83, or
429 mg/kg-day
for 2 years
Hepatocytes with enlarged
hyperchromic nuclei;
glomerulonephritis
Hepatocytomegaly;
glomerulonephritis
Degeneration and necrosis
of renal tubular cells and
hepatocytes
Pneumonia, gastric ulcers,
and cortical tubular
degeneration in the kidney
Pneumonia and rhinitis
Atrophy of nasal olfactory
epithelium; nasal adhesion
and inflammation
Argus et al.
(1965)
Argus et al.
(1973)
Kociba et al.
(1974)
NCI (1978)
NCI (1978)
JBRC (1998);
Kano et al.
(2009)
                                                                                      77
                             DRAFT - DO NOT CITE OR QUOTE

-------
Rats
F344/DuCri Rat MaleS °' 1 1 ' 55' Or
(50/sx/doe P T'lr^P, H
LvPh Females 0, 18, 83, or
; 429 mg/kg-day
for 2 years
Rats
F344/DuCri Rat MaleS °' 1 1- 55' °r
/en, M 274 mg/kg-day;
50/sex/dose Fema|L g, 18y'83] or 55
' 429 mg/kg-day
for 2 years
Mice
Crj:BDF1 Mouse Males 049 191 or
(50/sex/dose f7 ™9/knlay;07a 66
LvPh Females 0, 66, 278, or
; 964 mg/kg-day
for 2 years
Mice
Crj:BDF1 Mouse Males 049 191 or
.'. . , 677 mg/kg-day; ....
50/sex/dose Fema|L g, 66y 278, or 49
ve ' 964 mg/kg-day
for 2 years
55 Liver hyperplasia
Increases in serum liver
274 enzymes (GOT, GPT, LDH,
and ALP)
278 Nasal inflammation
Increases in serum liver
191 enzymes (GOT, GPT, LDH,
and ALP)
JBRC (1998):
Kano et al.
(2009)
JBRC (1998):
Kano et al.
(2009)
JBRC (1998):
Kano et al.
(2009)
JBRC (1998):
Kano et al.
(2009)
Developmental studies
Dpf c
SpragueDawley Pregnant dams 0] 250]
/HO on/ x 500, or 1,000 mg/kg-day
(18-20/group) on gestation dayy£ 6y_15Y
Delayed ossification of the
1,000 sternebrae and reduced
fetal BWs
Giavini et al.
(1985)
 1           Liver effects included degeneration and necrosis, hepatocyte swelling, cells with hyperchromic
 2    nuclei, spongiosis hepatis, hyperplasia, and clear and mixed cell foci of the liver (Kano et al., 2008; NCI.
 3    1978: Kocibaetal..  1974: Argus etal.. 1973: Argus etal.. 1965: Fairlev et al.. 1934).  Hepatocellular
 4    degeneration and necrosis were seen at high  doses in a subchronic study (1,900 mg/kg-day in rats)
 5    (Fairlev et al.. 1934) and at lower doses in a  chronic study (94 mg/kg-day, male rats) (Kociba et al..
 6    1974). Argus et al. (1973) described a progression of preneoplastic effects in the liver of rats exposed to a
 7    dose of 575 mg/kg-day. Early changes (8 months exposure) were  described as an increased nuclear size of
 8    hepatocytes, disorganization of the rough endoplasmic reticulum, an increase in smooth endoplasmic
 9    reticulum, a decrease in glycogen, an increase in lipid droplets in  hepatocytes, and formation of liver
10    nodules. Spongiosis hepatis, hyperplasia, and clear and mixed-cell foci were also observed in the liver of
11    rats (doses >55 mg/kg-day in male rats) (Kano et al.. 2009: JBRC. 1998). Clear and mixed-cell foci are
12    commonly considered preneoplastic changes and would not be considered evidence of noncancer toxicity
13    when observed in conjunction with tumor formation. If exposure to  1,4-dioxane had not resulted in tumor
14    formation, these lesions could represent potential noncancer toxicity. The nature of spongiosis hepatis as a
15    preneoplastic change is less well understood (Bannasch. 2003: Karbe and Kerlin. 2002: Stroebel et al..
16    1995). Spongiosis hepatis is a cyst-like lesion that arises from the perisinusoidal Ito cells of the liver. This
17    change is sometimes associated with hepatocellular hypertrophy and liver toxicity (Karbe and Kerlin.
18    2002). but may also occur in combination with preneoplastic foci, or hepatocellular adenoma or
19    carcinoma (Bannasch. 2003: Stroebel  et al..  1995). In the case of the JBRC (1998) study, spongiosis
20    hepatis was associated with other preneoplastic changes in the liver (hyperplasia, clear and mixed-cell
21    foci). No other lesions indicative of liver toxicity were seen in this study; therefore, spongiosis hepatis
                                                                                                    78
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    was not considered indicative of noncancer effects.  The activity of serum enzymes (i.e., AST, ALT,
 2    LDH, and ALP) was increased in rats and mice exposed to 1,4-dioxane, although only in groups with
 3    high incidence of liver tumors. Blood samples were collected only at the end of the 2-year study, so
 4    altered  serum chemistry may be associated with the tumorigenic changes in the liver.

 5           Hematological changes were reported in the JBRC (1998) study only. Mean doses are reported
 6    based on information provided in Kano et al. (2009). Observed increases  in RBCs, hematocrit,
 7    hemoglobin in high-dose male mice (677 mg/kg-day) may be related to lower drinking water
 8    consumption (74% of control drinking water intake). Hematological effects noted in male rats given
 9    55 mg/kg-day (decreased RBCs, hemoglobin, hematocrit, increased platelets) were within 20% of control
10    values.  A reference range database for hematological effects in laboratory animals (Wolford et al., 1986)
11    indicates that a 20% change in these parameters may fall within a normal range (10th-90th percentile
12    values) and may not represent a treatment-related effect of concern.

13           Rhinitis and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
14    dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in mice). The
15    JBRC (1998) study also demonstrates atrophy of the nasal epithelium and adhesion in rats and mice.
16    Nasal inflammation may be a response to direct contact of the nasal mucosa with drinking water
17    containing 1,4-dioxane (Sweeney et al., 2008; Goldsworthy et al.,  1991) or could result from systemic
18    exposure. Regardless, inflammation may indicate toxicity due to 1,4-dioxane exposure.  A significant
19    increase in the incidence of pneumonia was reported in mice from the NCI (1978) study. The significance
20    of this effect is unclear, as it was not observed in other studies that evaluated lung histopathology (Kano
21    et al.. 2008; JBRC. 1998; Kociba et al.. 1974). No studies were available  regarding the potential for
22    1,4-dioxane to cause immunological effects. Metaplasia and hyperplasia of the nasal epithelium were also
23    observed in high-dose male and female rats (JBRC. 1998): however, these effects are likely to be
24    associated with the formation of nasal cavity tumors in these dose  groups. Nuclear enlargement of the
25    nasal olfactory epithelium was observed at a dose of 83 mg/kg-day in female rats (Kano et al.. 2009):
26    however,  it is unclear whether this alteration represents an adverse toxicological effect. Nuclear
27    enlargement of the tracheal and bronchial epithelium and an accumulation of foamy cells in the lung were
28    also seen in male and female mice give 1,4-dioxane at doses of > 278 mg/kg for 2 years (JBRC. 1998).
      4.6.2  Inhalation

29           Two subchronic (Kasai et al., 2008: Fairley et al., 1934) and two chronic inhalation studies (Kasai
30    et al.. 2009: Torkelson et al.. 1974) were identified. Nasal, liver, and kidney toxicity were the primary
31    noncancer health effects of inhalation exposure to 1.4-dioxane in animals. Table 4-26 presents a summary
32    of the noncancer results for the subchronic and chronic inhalation studies of 1.4-dioxane toxicity in
33    laboratory animals.

34           Of the inhalation studies, nasal tissue was only collected in rat studies conducted by Kasai et al.
35    (2009: 2008). Damage to nasal tissue was reported frequently in these studies and statistically significant
36    observations were  noted as low as 50 ppm. Nasal effects included deformity of the nose and

                                                                                                   79
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    histopathological lesions characterized by enlarged epithelial nuclei (respiratory epithelium, olfactory
 2    epithelium, trachea, and bronchus), atrophy (olfactory epithelium), vacuolic change (olfactory epithelium
 3    and bronchial epithelium), squamous cell metaplasia and hyperplasia (respiratory epithelium), respiratory
 4    metaplasia (olfactory epithelium), inflammation (respiratory and olfactory epithelium), hydropic change
 5    (lamina propria). and sclerosis (lamina propria). In both studies, a concentration-dependent, statistically
 6    significant change in enlarged nuclei of the respiratory epithelium was considered the most sensitive nasal
 7    effect by the study authors; however, the toxicological significance of nuclear enlargement is uncertain.

 8           At high doses, liver damage was characterized by cell degeneration which varied from swelling
 9    (Kasai et al.. 2008: Fairlev et al.. 1934) to necrosis (Kasai et al.. 2009: Kasai et al.. 2008: Fairlev et al..
10    1934). spongiosis hepatis (Kasai et al.. 2009). nuclear enlargement of centrilobular cells (Kasai et al..
11    2009) and basophilic and acidophilic cell foci (Kasai et al.. 2009). Altered cell foci are commonly
12    considered preneoplastic changes and would not be considered evidence of noncancer toxicity when
13    observed in conjunction with tumor formation (Bannasch et al.. 1982). Since exposure to 1.4-dioxane
14    resulted in tumor formation in the liver, these lesions are not considered as potential noncancer toxicity.

15           At concentrations ranging from 200 ppm to 3.200 ppm. altered liver enzymes (i.e.. AST. ALT.
16    ALP, and y-GTP). increased liver weights, and induction of GST-P was also observed (Kasai et al.. 2009;
17    Kasai et al.. 2008). Changes in the activity of serum enzymes were mostly observed in exposed rat groups
18    of high 1.4-dioxane concentrations (Kasai et al.. 2009; Kasai et al.. 2008). Induction of GST-P positive
19    hepatocytes was observed in female rats at 1.600 ppm and male and female rats at 3.200 ppm following
20    13 weeks of exposure to  1.4-dioxane. GST-P is considered a good enzymatic marker for early detection of
21    chemical hepatocarcinogenesis (Sato. 1989). Although. GST-P positive liver foci were not observed in the
22    2 year bioassay. the focally and proliferating GST-P positive hepatocytes noted in the H week study
23    suggests eventual progression to hepatocellular tumors after 2 years of exposure and therefore would not
24    be a potential noncancer  effect.

25           The lowest concentration reported to produce liver lesions was 1.250 ppm. characterized by
26    necrosis of centrilobular  cells, spongiosis hepatis. and nuclear enlargement in the Kasai et al.  (2009)
27    study. However, as previously stated, the toxicological significance of nuclear enlargement lesions is
28    uncertain.

29           Kidney effects were reported less frequently in these inhalation studies and were generally
30    observed at higher exposure concentrations than nasal and liver effects. Kidney damage was described as
31    patchy degeneration of cortical tubules with vascular congestion and hemorrhage (Fairlev et al.. 1934).
32    hydropic change of proximal tubules (Kasai et al.. 2009; Kasai et al.. 2008). and as nuclear enlargement
33    of proximal tubules cells (Kasai et al.. 2009).  Changes in serum chemistry and urinalysis variables were
34    also noted as evidence of renal damage. In a 13 week inhalation study of male and female rats (Kasai et
35    al.. 2008) kidney toxicity was only observed in female rats  exposed to 3.200 ppm of 1.4-dioxane (i.e.
36    hydropic change in the renal proximal tubules), which suggests a possible increased susceptibility of
37    female rats to renal damage following inhalation exposure to 1.4-dioxane.

38           Other noted noncancer effects in laboratory animals included acute vascular congestion of the
39    lungs (Fairlev et al.. 1934); changes in relative lung weights (Kasai et al.. 2008); and decrease in body
                                                                                                    80
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1   weight gain (Kasai et al.. 2009; Kasai et al.. 2008). Following a 13-week exposure, higher 1.4-dioxane
 2   plasma levels were found in female rats as compared to male rats (Kasai et al.. 2008). 1.4-Dioxane was
 3   observed in plasma along with systemic effects following subchronic inhalation exposure to 1.4-dioxane
 4   in rats.

     Table 4-26  Inhalation toxicity studies (noncancer effects) for 1,4-dioxane
Species
Dose/duration N^
™
Reference
Subchronic studies
Rat, mouse, rabbit,
and guinea pig
(3-6/species/group);
unknown strains
F344/DuCrj rat
(10/sex/group)
0, 1,000,2,000, 5,000,
or 10,000 ppmfor7
days. Days 1-5, two 1.5 .,,
hour exposures; day 6,
one 1.5 hour exposure;
and day 7, no exposure
0, 100,200,400, 800,
1,600, 3,200, or ...
6,400 ppm 6 hours/day 5
days/wk, for 13 wk
Renal cortical
degeneration and
1,000 hemorrhage;
hepatocellular
degeneration and necrosis
Respiratory epithelium:
1 00 nuclear enlargement of
epithelial cells
Fairley et al.
(1934)
Kasai et al.
(20081
Chronic studies
Wistar rat (288/sex)
F344/DuCrj male
rat
(50/group)
111 ppm for /hours/day, 111 (free
5days/wk, for 2 years standing)
0, 50, 250, or 1,250 ppm
for 6 hours/day, 5 N/A
days/wk for 2 years
No significant effects were
observed on BWs,
NA survival, organ weights,
hematology, clinical
chemistry, or
histopathology
Respiratory epithelium:
„ nuclear enlargement of
epithelial cells, atrophy,
and metaplasia
Torkelson et
al. (19741
Kasai et al.
(20091
     4.6.2.1  Mode of Action Information

 5           The metabolism of 1,4-dioxane in humans was extensive at low doses (<50 ppm). The linear
 6   elimination of 1,4-dioxane in both plasma and urine indicated that 1,4-dioxane metabolism was a
 7   nonsaturated, first-order process at this exposure level (Young et al.. 1977; 1976). Like humans, rats
 8   extensively metabolized inhaled 1,4-dioxane; however, plasma data from rats given single i.v. doses of 3,
 9   10, 30, 100, or 1,000 mg [14C]-l,4-dioxane/kg demonstrated a dose-related shift from linear, first-order to
10   nonlinear, saturable metabolism of 1,4-dioxane (Young et al.. 1978a; 1978b). Conversely, using the
11   Young et al. (1978b; 1978a) rat model, the metabolism of 1.4-dioxane in rats that were exposed to 400.
12   800. 1.600. and 3.200 ppm via inhalation for 13 weeks could not be accurately depicted due to a lack of
13   knowledge on needed model parameters and biological processes (See Section 3.5.3 and Appendix B). It
14   appears, following prolonged inhalation exposure to 1.4-dioxane at concentrations up to 3.200 ppm.  that
15   metabolism is induced (Appendix B).
                                                                                                  81
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           1,4-Dioxane oxidation appeared to be CYP450-mediated, as CYP450 induction with
 2    phenobarbital or Aroclor 1254 and suppression with 2,4-dichloro-6-phenylphenoxy ethylamine or
 3    cobaltous chloride was effective in significantly increasing and decreasing, respectively, the appearance
 4    of HEAA in the urine of rats (Wooetal.. 1978. 1977c). 1,4-Dioxane itself induced CYP450-mediated
 5    metabolism of several barbiturates in Hindustan mice given i.p. injections of 25 and 50 mg/kg of
 6    1,4-dioxane (Mungikar and Pawar. 1978). The differences between single and multiple doses in urinary
 7    and expired radiolabel support the notion that 1,4-dioxane may induce its own metabolism. 1,4-Dioxane
 8    has been shown to induce several isoforms of CYP450 in various tissues following acute oral
 9    administration by gavage or drinking water (Nannelli et al.. 2005). In the liver, the activity of several
10    CYP450 isozymes was increased (i.e., CYP2B1/2, CYP2E1, CYPC11); however, only CYP2E1 was
11    inducible in the kidney and nasal mucosa. CYP2E1 mRNA was increased approximately two- to threefold
12    in the kidney and nasal mucosa, but mRNA levels were not increased in the liver, suggesting that
13    regulation of CYP2E1 was organ-specific.

14           Nannelli et al. (2005) investigated the role of CYP450 isozymes in the liver toxicity of
15    1,4-dioxane. Hepatic CYPB1/2 and CYP2E1 levels were induced by phenobarbital or fasting and liver
16    toxicity was measured as hepatic glutathione content or serum ALT activity. No increase in glutathione
17    content or ALT activity was observed, suggesting that highly reactive and toxic intermediates  did not play
18    a large role in the liver toxicity of 1,4-dioxane, even under conditions where metabolism was enhanced.
19    Pretreatment with inducers of mixed-function oxidases also did not significantly change the extent of
20    covalent binding in subcellular fractions (Woo etal.. 1977b). Covalent binding was measured  in liver,
21    kidney, spleen, lung, colon, and skeletal muscle 1-12 hours after i.p. dosing with 1,4-dioxane. Covalent
22    binding was highest in liver, spleen,  and colon. Within hepatocytes, 1,4-dioxane distribution was greatest
23    in the cytosolic fraction, followed by the microsomal, mitochondrial, and nuclear fractions.

24           The absence of an increase in toxicity following an increase in metabolism suggests that
25    accumulation of the parent compound may be related to 1,4-dioxane toxicity. This hypothesis  is supported
26    by a comparison of the pharmacokinetic profile of 1,4-dioxane with the toxicology data from a chronic
27    drinking water study (Kociba et al.. 1975). This analysis indicated that liver toxicity did not occur unless
28    clearance pathways were saturated and elimination of 1,4-dioxane from the blood was reduced. A
29    dose-dependent increase of 1.4-dioxane accumulation in the blood was seen, which correlated to the
30    observed dose-dependent increase in incidences of nasal liver, and kidney toxicities (Kasai et al., 2008).
31    Alternative metabolic pathways (i.e., not CYP450 mediated) may be present at high doses of 1,4-dioxane;
32    however, the available studies have not characterized these pathways or identified any possible reactive
33    intermediates. Thus, the mechanism  by which 1,4-dioxane  induces tissue damage is not known, nor is it
34    known whether the toxic moiety is 1,4-dioxane or a transient or terminal metabolite.
                                                                                                  82
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      4.7   Evaluation of Carcinogenicity
      4.7.1   S ummary of Overall Weight of Evidence

 1           Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is "likely
 2    to be carcinogenic to humans" based on evidence of carcinogenicity in several 2-year bioassays
 3    conducted in four strains of rats, two strains of mice, and in guinea pigs (Kano et al.. 2009; Kasai et al..
 4    2009: JBRC. 1998: Yamazaki et al. 1994: NCI. 1978: Kocibaetal.. 1974: Argus etal.. 1973: Hoch-
 5    Ligeti and Argus. 1970: Hoch-Ligeti et al., 1970: Argus et al., 1965). Tissue sites where tumors have been
 6    observed in these laboratory animals due to exposure to 1.4-dioxane include, peritoneum (Kano et al..
 7    2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). mammary gland (Kano et al.. 2009: Kasai
 8    et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). liver (Kano et al.. 2009: Kasai et al.. 2009). kidney
 9    (Kasai et al.. 2009). Zymbal gland (Kasai et al.. 2009). subcutaneous (Kasai et al.. 2009). nasal tissue
10    (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kociba et al.. 1974:
11    Argus etal.. 1973: Hoch-Ligeti et al.. 1970). and lung (Hoch-Ligeti and Argus. 1970). Studies in humans
12    are inconclusive regarding evidence for a causal link between occupational  exposure to  1,4-dioxane and
13    increased risk for cancer; however, only two studies were available and these were limited by small
14    cohort size and a small number of reported cancer cases (Buffler et al.. 1978: Thiess etal.. 1976).

15           The available evidence is inadequate to establish a mode of action (MOA) by which 1,4-dioxane
16    or a transient or terminal metabolite induces liver tumors in rats and mice. A MOA hypothesis involving
17    sustained proliferation of spontaneously transformed liver cells has some support from data indicating
18    that 1,4-dioxane acts as a tumor promoter in mouse skin and rat liver bioassays (Lundberg et al.. 1987:
19    King etal.. 1973). Dose-response and temporal data support the occurrence of cell proliferation and
20    hyperplasia prior to the development of liver tumors (JBRC. 1998: Kocibaetal.. 1974) in the rat model.
21    However, the dose-response relationship for induction of hepatic cell proliferation has not been
22    characterized, and it is unknown if it would reflect the dose-response relationship for liver tumors in the
23    2-year rat and mouse studies. Conflicting data from rat and mouse bioassays (JBRC. 1998: Kociba et al..
24    1974) suggest that cytotoxicity may not be a required precursor event for 1,4-dioxane-induced cell
25    proliferation. Data regarding a plausible dose response and temporal progression (see Table 4-21) from
26    cytotoxicity and cell proliferation to eventual liver tumor formation are not  available.

27           For nasal tumors, there is no known MOA. There is a hypothesized MOA that includes metabolic
28    induction, cytotoxicity. and regenerative cell proliferation (Kasai et al.. 2009). The induction of CYP450
29    has some support from data illustrating that following acute oral administration of 1.4-dioxane by gavage
30    or drinking water. CYP2E1 was inducible in nasal mucosa (Nannelli et al.. 2005). CYP2E1 mRNA was
31    increased approximately two- to threefold in nasal mucosa (and in the kidney, see section 3.3) in the
32    Nannelli et al. (2005) study. While cell proliferation was observed following 1.4-dioxane exposure in
33    both a 2-year inhalation study in male rats (1.250 ppm) (Kasai et al.. 2009) and a 2-year drinking water
34    study in male (274 mg/kg-day) and female rats (429 mg/kg-day). no evidence of cvtotoxicity in the nasal
35    cavity was observed (Kasai et al.. 2009): therefore, cytotoxicity. as a key event, is not supported.

                                                                                                  83
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    Following a 13-week inhalation study in rats, a concentration-dependent accumulation of 1.4-dioxane in
 2    the blood was observed (Kasai et al.. 2008). Studies have shown that water-soluble, gaseous irritants
 3    cause nasal injuries such as squamous cell carcinomas (Morgan et al., 1986). Similarly. 1.4-dioxane.
 4    which has been reported as a miscible compound (Hawley and Lewis. 2001). also caused nasal injuries
 5    that were concentration-dependent, including nasal tumors (Kasai et al., 2009). Additionally, it has been
 6    suggested that in vivo genotoxicity may contribute to the carcinogenic MOA for  1.4-Dioxane (Kasai et
 7    al., 2009) (see Section 4.7.3.6 for further discussion). Collectively, these data are insufficient to support
 8    the hypothesized MOAs.

 9           The MOA by which  1,4-dioxane produces kidneyi lung, peritoneal (mesotheliomas), mammary
10    gland, Zymbal gland, and subcutis tumors is also unknown, and there are no available data regarding any
11    hypothesized carcinogenic MOA for 1,4-dioxane in these tissues.

12           U.S. EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) indicate that for
13    tumors occurring at a site other than the initial point of contact, the weight of evidence for carcinogenic
14    potential may apply to all routes of exposure that have not been adequately tested at sufficient doses. An
15    exception occurs when there  is convincing information (e.g., toxicokinetic data) that absorption does not
16    occur by other routes. Information available on the carcinogenic effects of 1,4-dioxane via the oral route
17    demonstrates that tumors occur in tissues remote from the site of absorption. In addition, information on
18    the carcinogenic effects of 1.4-dioxane via the inhalation route in animals also demonstrates that tumors
19    occur at tissue sites distant from the portal of entry. Information on the carcinogenic effects of
20    1,4-dioxane via the inhalation and  dermal routes in humans and via the dermal route in animals is absent.
21    Based on the observance of systemic tumors following oral and inhalation exposure, it is assumed that an
22    internal dose will be achieved regardless of the route of exposure. Therefore, 1,4-dioxane is "likely to be
23    carcinogenic to humans" by all routes of exposure.
      4.7.2  Synthesis of Human, Animal, and Other Supporting Evidence

24           Human studies of occupational exposure to 1,4-dioxane were inconclusive; in each case, the
25    cohort size was limited and number of reported cases was small (Buffler et al.. 1978; Thiess et al.. 1976).

26           Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
27    pigs (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kociba et al..
28    1974; Torkelson et al.. 1974; Argus etal.. 1973; Hoch-Ligeti and Argus. 1970; Hoch-Ligeti et al.. 1970;
29    Argus etal.. 1965). Liver tumors have been observed following drinking water exposure in male Wistar
30    rats (Argus et al.. 1965). male guinea pigs (Hoch-Ligeti and Argus. 1970). male Sprague Dawley rats
31    (Argus etal.. 1973: Hoch-Ligeti et al.. 1970). male and female Sherman rats (Kociba et al.. 1974). female
32    Osborne-Mendel rats (NCI. 1978). male and female F344/DuCrj rats (Kano et al.. 2009: JBRC.  1998:
33    Yamazaki et al.. 1994). male and female B6C3Fi mice (NCI. 1978). and male and female Crj:BDFl  mice
34    (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994); and following inhalation exposure in male F344
35    rats (Kasai et al.. 2009). In the earliest cancer bioassays, the liver tumors were described as hepatomas
36    (Argus etal.. 1973: Hoch-Ligeti and Argus. 1970:  Hoch-Ligeti et al.. 1970: Argus etal.. 1965):  however,

                                                                                                 84
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    later studies made a distinction between hepatocellular carcinoma and hepatocellular adenoma (Kano et
 2    al. 2009: Kasai et al. 2009: JBRC. 1998: Yamazaki et al.. 1994: NCI. 1978: Kocibaetal. 1974). Both
 3    tumor types have been seen in rats and mice exposed to 1,4-dioxane via drinking water and inhalation.

 4           Kociba et al. (1974) noted evidence of liver toxicity at or below the dose levels that produced
 5    liver tumors but did not report incidence data for these effects. Hepatocellular degeneration and necrosis
 6    were observed in the mid- and high-dose groups of male and female Sherman rats exposed to 1,4-dioxane,
 7    while tumors were only observed at the highest dose. Hepatic regeneration was indicated in the mid- and
 8    high-dose groups by the formation of hepatocellular hyperplastic nodules. Kano et al., (2009) also
 9    provided evidence of liver hyperplasia in male F344/DuCrj rats at a dose level below the dose that
10    induced a statistically significant increase in tumor formation. Kasai et al (2009) noted evidence of liver
11    toxicity and tumor incidences (i.e. hepatocellular adenoma) in male F344/DuCrj rats following inhalation
12    exposures to 1.250 ppm. Increased liver toxicities included hepatocellular necrosis, spongiosis hepatis.
13    and acidophilic and basophilic cell foci.

14           Nasal cavity tumors were also  observed in Sprague Dawley rats (Argus et al.. 1973: Hoch-Ligeti
15    etal.. 1970). Osborne-Mendel rats (NCI. 1978). Sherman rats (Kocibaetal.. 1974). and F344/DuCrj rats
16    (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994). Most tumors were
17    characterized as squamous cell carcinomas. Nasal tumors were not elevated in B6C3F] or Crj:BDFl mice.
18    Kano et aL (2009) and Kasai et al (2009) were the only studies that evaluated nonneoplastic changes in
19    nasal cavity tissue following prolonged exposure to 1,4-dioxane via oral and inhalation routes.
20    respectively.

21           Histopathological lesions in female F344/DuCrj  rats following oral exposure to 1,4-dioxane were
22    suggestive of toxicity and regeneration in nasal tissue (i.e., atrophy, adhesion, inflammation, nuclear
23    enlargement, and hyperplasia and metaplasia of respiratory and olfactory epithelium). Some of these
24    effects occurred at a lower dose (83 mg/kg-day) than that shown to produce nasal cavity tumors
25    (429 mg/kg-day)  in female rats. Re-examination of tissue sections from the NCI (1978) bioassay
26    suggested that the majority of nasal tumors were located in the dorsal nasal septum or the nasoturbinate of
27    the anterior portion of the dorsal meatus.

28           Histopathological lesions in male F344/DuCrj rats following exposure to 1.4-dioxane via
29    inhalation were also suggestive of toxicity and regeneration in nasal tissue (i.e. atrophy, inflammation.
30    nuclear enlargement, hyperplasia and metaplasia of the respiratory and olfactory epithelium,  and
31    inflammation). Some of these effects occurred at lower concentrations (50 ppm and 250 ppm) than those
32    shown to produce nasal cavity tumors (1.250 ppm) in male rats. Nasal squamous cell carcinomas were
33    observed in the dorsal area of levels 1-3 of the nasal cavity and were characterized as we 11-differentiated
34    and keratinized. In two cases, invasive  growth into adjacent tissue was noted, marked by carcinoma
35    growth out of the nose and through a destroyed nasal bone.

36           In addition to the liver and nasal tumors observed in several studies, a statistically significant
37    increase in mesotheliomas of the peritoneum was seen in male rats from the Kano et al. (2009) study
38    (JBRC. 1998: Yamazaki et al.. 1994) and the Kasai et al (2009) study. Female rats dosed with
39    429 mg/kg-day in drinking water for 2  years also showed a statistically significant increase in mammary
                                                                                                   85
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1   gland adenomas (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). In male rats, exposed via
 2   inhalation, a statistically significant positive trend of mammary gland adenomas was observed by Kasai et
 3   al. (2009). A statistically significant increase and/or trend of subcutis fibroma. Zvmbal gland adenoma.
 4   and renal cell carcinoma incidences was also observed in male rats exposed for 2 years via inhalation
 5   (Kasai etal.. 2009). A significant increase in the incidence of these tumors was not observed in other
 6   chronic oral or inhalation bioassays of 1,4-dioxane (NCI. 1978; Kocibaet al.. 1974; Torkelson et al..
 7   1974).
     4.7.3   Mode of Action Information

 8           The MOA by which 1,4-dioxane produces liver, nasal, kidney, peritoneal (mesotheliomas),
 9   mammary glandi Zymbal gland, and subcutis tumors is unknown, and the available data do not support
10   any hypothesized mode of carcinogenic action for 1,4-dioxane. Available data also do not clearly identify
11   whether 1,4-dioxane or one of its metabolites is responsible for the observed effects. Furthermore, tumor
12   initiation and promotion studies in mouse skin and rat liver suggested that 1.4-dioxane exposure does not
13   initiate the carcinogenic process, but instead may act as a tumor promoter (Lundberg et al.. 1987; Bull et
14   al.. 1986; King et al.. 1973) (see Section 4.2.3).

15           The hypothesized MOAs for 1,4-dioxane carcinogenicity are discussed below within the context
16   of the modified Hill criteria of causality as recommended in the most recent Agency guidelines (U.S.
17   EPA. 2005a). MOA analyses were not conducted for kidney, peritoneal, mammary gland, Zymbal gland.
18   or subcutis tumors due to the absence of any chemical specific information for these tumor types.
                                                                                                 86
                                    DRAFT - DO NOT CITE OR QUOTE

-------
      4.7.3.1   Identification of Key Events for Carcinogenicity

 1    4.7.3.7.7  Liver. A key event in this MOA hypothesis is sustained proliferation of
 2    spontaneously transformed liver cells, resulting in the eventual formation of liver tumors. Precursor
 3    events in which 1,4-dioxane may promote proliferation of transformed liver cells are uncertain. One study
 4    suggests that induced liver cytotoxicity may be a key precursor event to cell proliferation leading to the
 5    formation of liver tumors (Kocibaet al.. 1974). however, this study did not report incidence data for these
 6    effects. Other studies suggest that cell proliferation can occur in the absence of liver cytotoxicity. Liver
 7    tumors were observed in female rats and female mice in the absence of lesions indicative of cytotoxicity
 8    (Kano et al.. 2008; JBRC. 1998; NCI. 1978). Figure 4-1 presents a schematic representation of possible
 9    key events in the MOA for 1,4-dioxane liver carcinogenicity. These include: (1) oxidation by CYP2E1
10    and CYP2B1/2 (i.e., detoxification pathway for 1,4-dioxane), (2) saturation of metabolism/clearance
11    leading to accumulation of the parent  1,4-dioxane, (3) liver damage followed by regenerative cell
12    proliferation, or (4) cell proliferation in the absence of cytotoxicity (i.e., mitogenesis), (5) hyperplasia,
13    and (6) tumor formation. It is suggested that liver toxicity is related to the accumulation of the parent
14    compound following metabolic saturation at high doses (Kociba et al.. 1975): however, since no in vivo
15    or in vitro assays have examined the toxic moiety resulting from 1.4-dioxane exposure, liver toxicity due
16    to metabolites cannot be ruled out. Therefore, this hypothesis is not supported. Nannelli et al. (2005)
17    demonstrated that an increase in the oxidative metabolism of 1,4-dioxane via CYP450 induction using
18    phenobarbital or fasting does not result in an increase in liver toxicity. This result suggested that highly
19    reactive  and toxic intermediates did not play a large role in the liver toxicity of 1,4-dioxane, even under
20    conditions where metabolism was enhanced. Alternative metabolic pathways (e.g., not CYP450
21    mediated) may be present at high doses of 1,4-dioxane; although the available studies have not
22    characterized these pathways nor identified any possible reactive intermediates. Tumor promotion studies
23    in mouse skin and rat liver suggest that 1,4-dioxane may enhance the growth of previously initiated cells
24    (Lundberg et al.. 1987; King et al.. 1973). This is consistent with the increase in hepatocyte cell
25    proliferation observed in several studies (Mivagawa et al.. 1999; Uno etal. 1994; Goldsworthy et al..
26    1991; Stott etal.. 1981). These mechanistic studies provide evidence of cell proliferation but do not
27    indicate  whether mitogenesis or cytotoxicity is responsible for increased cell turnover.
                                                                                                    87
                                      DRAFT - DO NOT CITE OR QUOTE

-------

Toxicokinetics*

Oral-absorption- of
1.4-dioxane'
1
Metabolism- by
CYP2El-and
CYP2B1/21
1
HEAA-elimination-
in-the-urine1



Hypothesized- MOA-for-Liver- Tumors'"

Metabolic1
saturation- and
accumulation- of
l=4-dioxane-in-the
blood





*
H epato cellular-
cvtotoxicitv1
I
Regenerative- cell-
proliferation'
1
Hvperplasia1
1
Tumor- formation1


t
Cell-proliferation- in-
absence- of
cvtotoxicitV
1
Hvperplasia'
1
Tumor- promotion*

Figure 4-1  A schematic representation of the possible key events in the delivery of
           1,4-dioxane to the liver and the hypothesized MOA(s) for liver
           carcinogenicity
                       DRAFT - DO NOT CITE OR QUOTE

-------
 1    4.7.3.1.2  Nasal cavity. A possible key event in the MOA hypothesis for nasal tumors is
 2    sustained proliferation of spontaneously transformed nasal epithelial cells, resulting in the eventual
 3    formation of nasal cavity tumors (Kasai et al.. 2009). Cell proliferation was observed following
 4    1.4-dioxane exposure in both a 2-year inhalation study in male rats (1.250 ppm) (Kasai et al.. 2009) and a
 5    2-year drinking water study in male (274 mg/kg-day) and female rats (429 mg/kg-day) (Kano. et al.
 6    2009). However, neither study reported evidence of cytotoxicity in the nasal cavity (Kasai et al.. 2009)
 7    therefore, cytotoxicity as a key event is not supported. Kasai et al. (2009; 2008) suggest that nasal
 8    toxicity is related to the accumulation of the parent compound following metabolic induction at high
 9    doses up to 3.200 ppm; however, since no in vivo or in vitro assays have examined the toxic moiety
10    resulting from 1.4-dioxane exposure, nasal toxicity due to metabolites cannot be ruled out. Nannelli et al.
11    (2005) demonstrated that CYP2E1 was inducible in nasal mucosa following acute oral administration of
12    1.4-dioxane by gavage and drinking water, which could potentially lead to an increase in the oxidative
13    metabolism of 1.4-dioxane and nasal toxicity. However. Nannelli et al. (2005) did not characterize this
14    pathway nor identify any possible reactive intermediates or nasal toxicities.
      4.7.3.2  Strength, Consistency, Specificity of Association

 1    4.7.3.2.1   Liver. The plausibility of a MOA that would include liver cytotoxicity, with
 2    subsequent reparative cell proliferation, as precursor events to liver tumor formation is minimally
 3    supported by findings that nonneoplastic liver lesions occurred at exposure levels lower than those
 4    resulting in significantly increased incidences of hepatocellular tumors (Kociba et al..  1974) and the
 5    demonstration of nonneoplastic liver lesions in subchronic (Kano et al.. 2008) and acute and  short-term
 6    oral studies (see Table 4-18). Because the incidence of nonneoplastic lesions was not reported by Kociba
 7    et al. (1974). it is difficult to know whether the incidence of liver lesions increased with increasing
 8    1,4-dioxane concentration. Contradicting the observations by Kociba et al. (1974). liver tumors were
 9    observed in female rats and female mice in the absence of lesions indicative of cytotoxicity (Kano et al..
10    2008; JBRC. 1998; NCI. 1978). This suggests that cytotoxicity may not be a requisite step in the MOA
11    for liver cancer. Mechanistic and tumor promotion studies suggest that enhanced cell proliferation without
12    cytotoxicity may be a key event; however, data showing a plausible dose response and temporal
13    progression from cell proliferation to eventual liver tumor formation are not available  (see Sections
14    4.7.3.3 and 4.7.3.4). Mechanistic studies that demonstrated cell proliferation after short-term exposure did
15    not evaluate liver cytotoxicity (Miyagawa et al.. 1999; Uno et al.. 1994; Goldsworthy et al.. 1991).
16    Studies have not investigated possible precursor events that may lead to cell proliferation in the absence
17    of cytotoxicity (i.e., genetic regulation of mitogenesis).
                                                                                                    89
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1   4.7.3.2.2   Nasal cavity. Nasal cavity tumors have been demonstrated in several rat strains
 2   (Kano et al.. 2009: Kasai et al.. 2009: JBRC. 1998: Yamazaki et al. 1994: NCI. 1978: Kociba et al..
 3   1974). but were not elevated in two strains of mice (Kano et al.. 2009: JBRC. 1998: Yamazaki et al..
 4   1994: NCI. 1978). Irritation of the nasal cavity of rats was indicated in studies by the observation of
 5   inflammation (Kasai et al (2009: 2008) and in one study, also rhinitis (JBRC. 1998). The Kasai et al
 6   (2009: 2008) studies also showed atrophy of the nasal epithelium in rats, and the JRBC (1998) study also
 7   observed atrophy of the nasal epithelium as well as adhesion in rats. Regeneration of the nasal epithelium
 8   is demonstrated by metaplasia and hyperplasia observed in rats exposed to 1,4-dioxane (Kano et al.. 2009:
 9   Kasai et al.. 2009: JBRC. 1998: Yamazaki et al..  1994). Oxidation of 1,4-dioxane metabolism by
10   CYP450s is not supported as a key event in the MOA hypothesis of nasal tumors. Although Nannelli et
11   al. (2005) demonstrated that CYP2E1 was inducible in nasal mucosa following acute oral administration
12   of 1.4-dioxane by gavage and drinking water, the study lacked details regarding the toxic moiety (e.g.
13   parent compound or reactive intermediate) and resulting nasal toxicity. Accumulation of 1.4-dioxane in
14   blood, as a precursor event of nasal tumor formation is also not supported because the parent compound
15   1.4-dioxane was only measured in one  subchronic study (Kasai et al.. 2008) and in this study no evidence
16   of nasal cvtotoxicitv. cell proliferation, or incidence of nasal tumors were reported.
     4.7.3.3  Dose-Response Relationship

 1   4.7.3.3.1   Liver. Table 4-27 presents the temporal sequence and dose-response
 2   relationship for possible key events in the liver carcinogenesis of 1,4-dioxane. Dose-response information
 3   provides some support for enhanced cell proliferation as a key event in the liver tumorigenesis of
 4   1,4-dioxane; however, the role of cytotoxicity as a required precursor event is not supported by data from
 5   more than one study. Kociba et al. (1974) demonstrated that liver toxicity and hepatocellular regeneration
 6   occurred at a lower dose level than tumor formation. Hepatocellular degeneration and necrosis were
 7   observed in the mid- and high-dose groups of Sherman rats exposed to 1,4-dioxane, although it is not
 8   possible to discern whether this effect was observed in both genders due to the lack of incidence data
 9   (Kociba et al.. 1974). Hepatic tumors were only observed at the highest dose  (Kociba et al.. 1974).
10   Hepatic regeneration was indicated in the mid- and high-dose group by the formation of hepatocellular
11   hyperplastic nodules. Liver hyperplasia was also seen in rats from the JBRC (1998) study, at or below the
12   dose level that resulted in tumor formation (Kano et al.. 2009): however, hepatocellular degeneration and
13   necrosis were not observed. These results suggest that hepatic cell proliferation and hyperplasia may
14   occur in the absence of significant cytotoxicity. Liver angiectasis (i.e., dilation of blood or lymphatic
15   vessels) was observed in male mice at the same dose that produced liver tumors; however, the
16   relationship between this vascular abnormality and tumor formation is unclear.
                                                                                                  90
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table 4-27  Temporal sequence and dose-response relationship for possible key events and liver
           tumors in rats and mice
Key event (time — >)
Dose (mg/kg-day) or ......
Exposure (ppm) Metabohsm
1,4-dioxane
Cell
Liver damage pro|iferation
Hyperplasia
Adenomas
and/or
carcinomas
Kociba et al., (1974) — Sherman rats (male and female combined)
0 mg/kg-day — a
14 mg/kg-day +b
121 mg/kg-day +b
1,307 mg/kg-day +b
a
a
+c
+c
a
a
a
a
a
a
+c
+c
a
a
a
+c
NCI, (1978)— female Osborne-Mendel rats
0 mg/kg-day — a
350 mg/kg-day +b
640 mg/kg-day +b
a
a
a
a
a
a
a
a
a
a
+c
+c
NCI, (1978)— male B6C3Fi mice
0 mg/kg-day — a
720 mg/kg-day +b
830 mg/kg-day +b
a
a
a
a
a
a
a
a
a
a
+c
+c
NCI. (1978)— female B6C3Fi mice
0 mg/kg-day — a
380 mg/kg-day +b
860 mg/kg-day +b
Kano et al., (2009); JBRC, (1998)— male
0 mg/kg-day — a
1 1 mg/kg-day +b
55 mg/kg-day +b
274 mg/kg-day +b
a
a
a
F344/DuCrj rats
a
a
a
+c
-------
          250 ppm
         1,250 ppm
      a— No evidence demonstrating key event.
      b+ 1,4-dioxane metabolism was not evaluated as part of the chronic bioassays. Data from pharmacokinetic studies suggest that
         metabolism of 1,4-dioxane by CYP2E1 and CYP2B2 occurs immediately and continues throughout the duration of exposure at
         all exposure levels.
      CD E \idence demonstrating key event.
      dD Single cellnecrosis was observed in a 13 weekbioassay formale rats(274 mg/kg-day), male mice (585 mg/kg-day), and
         female mice (898 mg/kg-day) exposed to 1,4-dioxane in drinking water (Kano et al.. 2008).
      e+ Kano et al. (2009) reported incidence rates for hepatocellular adenomas and carcinomas; however, information from JBRC
         (1998) on incidence of liver hyperplasia was used to create this table.
      f+ Kasai et al. (2008) reported significant incidence rates for single cell necrosis in female rats only (3,200 ppm) following a 2 year
         bioassay.
      9AII rats died during the first week of the 13-week bioassay (Kasai etal.. 2008).
      hKasai et al. (2009) reported incidence rates for centrilobular necrosis and hepatocellular adenomas in male rats (1,250 ppm).



      4.7.3.3.2   Nasal cavity.


 1            Table 4-28 presents the temporal sequence and dose-response relationship for possible key events

 2    in the nasal tissue carcinogenesis of 1.4-dioxane. Toxicity and regeneration in nasal epithelium (i.e.,

 3    atrophy, adhesion, inflammation, and hyperplasia and metaplasia of respiratory and olfactory epithelium)

 4    was evident in one study at the same dose levels that produced nasal cavity tumors (Kano et al.. 2009;

 5    JBRC. 1998). In another study, dose-response information provided some support for nasal toxicity and

 6    regeneration in nasal epithelium occurring before tumor development (Kasai et al.. 2009). However, the

 7    role of cvtotoxicity as a required precursor event is not supported by data from any of the reviewed

 8    studies. The accumulation of parent 1.4-dioxane as a key event has some support since

 9    concentration-dependent increases were noted for 1.4-dioxane in plasma concurrent with toxicities

10    observed that are possible precursor events (i.e.. regeneration in nasal epithelium) (Kasai et al.. 2008). In

11    a subsequent study by Kasai et al. (2009) some of these same possible precursor events were observed at

12    50. 250. and 1.250 ppm with evidence of nasal tumors at the highest concentration (1.250 ppm).
                                                                                                           92
                                        DRAFT - DO NOT CITE OR QUOTE

-------
 Table 4-28  Temporal sequence and dose-response relationship for possible key events and
            nasal tumors in rats and mice
                                                Key event (time —>)
 Dose (mg/kg-day)	*	J	'	—	
   or Exposure     Metabolism        Nasal            Cell                          Adenomas
      (ppm)        1,4-dioxane      cytotoxicity     proliferation     Hyperplasia        and/or
                                                                                  CdrCI nOiTldS
 Kociba et al.. (1974)—Sherman rats (male and female combined)	
      0 mg/kg-day	—a_	^	^	^	^	
     14 mg/kg-day       +h             —a             —a             —a             —a
    121  mg/kg-day       +b             —a             —a             —a             —a
   1,307 mg/kg-day       +b             —a             —a             —a             —a
 NCI. (1978)—female Osborne-Mendel rats	
      0 mg/kg-day	—a	^	^	^	^	
    350 mg/kg-day       +b             —a             —a             —a             —a
    640 mg/kg-day       +b             —a             —a             —a             —a
 NCI. (1978)—male B6C3Fi mice	
      0 mg/kg-day	—a_	^	^	^	^	
    720 mg/kg-day       +"             —a             —a             —a             —a
    830 mg/kg-day       +b             —a             —a             —a             —a
 NCI. (1978)—female B6C3Fi mice	
      0 mg/kg-day       —a	^	^	^	—a
    380 mg/kg-day	£	—*	—*	—*	—*	
    860 mg/kg-day       +b             —a             —a             —a             —a
 Kano et al., (2009); JBRC. (1998)—male F344/DuCrj rats	
      0 mg/kg-day       —a	-?	-?	-?	—a
     11  mg/kg-day       +b             —a             —a             —a             —a
     55 mg/kg-day       +b             —a             —a             —a             —a
    274 mg/kg-day       +b             —a             —a             +c'd             +c'd
 Kano et al., (2009): JBRC. (1998)—female F344/DuCrj rats	
      0 mg/kg-day       —a	-?	-?	-?	—a
     18 mg/kg-day       +b             —a             —a             —a             —a
     83 mg/kg-day       +b             —a             —a             —a             —a
    429 mg/kg-day       +b             —a             —a             +c'd             +c'd
 Kano et al.. (2009): JBRC. (1998)—male Crj:BDF1 mice	
      0 mg/kg-day       —a	^	^	^	—a
     49 mg/kg-day       +b             —a             —a             —a             —a
    191  mg/kg-day       +h             —a             —a             —a             —a
    677 mg/kg-day       +b             —a             —a             —a             —a
 Kano et al., (2009); JBRC. (1998)—female Crj:BDF1 mice	
      0 mg/kg-day       —a	^	^	^	—a
     66 mg/kg-day       +"             —a             —a             —a             —a
    278 mg/kg-day       +b             —a             —a             —a             —a
    964 mg/kg-day       +b             —a             —a             —a             —a
 Kasai et al. (2008)—F344 rats (male and female combined)	
	Oppm	—a_	-^	-^	-^	—a
      100 ppm          +b             —a             —a             —a             —a
      200 ppm          +b             —a             —a             —a             —a
      400 ppm          +c             —a             —a             —a             —a
      800 ppm	+^	—^	—^	—^	—^	
     1,600 ppm          +c             —a             —a             —a             —a
     3,200 ppm          +c             —a             —a             —a             —a
     6,400 ppm          +aAt            -aj             -aj             -aj             -aj
 Kasai et al. (2009)—male F344 rats	
      Oppm	—a_	^	^	^	—a
      50 ppm           +b             —a             —a             —a             —a
     250 ppm           ?             ^             ^             ^             ^

                                                                                              93
                                DRAFT - DO NOT CITE OR QUOTE

-------
         1,250 ppm
      a— No evidence demonstrating key event.
      b+ 1,4-dioxane metabolism was not evaluated as part of these studies. Data from pharmacokinetic studies suggest that metabolism
         of 1,4-dioxane by CYP2E1 and CYP2B2 occurs immediately and continues throughout the duration of exposure at all exposure
         levels.
      CD Evidence demonstrating key event.
      d+ Kano et al. (2009) reported incidence rates for squamous cell hyperplasia (respiratory epithelium) and squamous cell
         carcinomas (nasal cavity); however, information from JBRC (1998) on significant incidence of squamous cell hyperplasia was
         used to create this table.
      e+Kasai et al. (2009) reported incidence rates for squamous cell hyperplasia in male rats (1,250 ppm) following a 2 year bioassay.
      f+ All rats died during the first week of the 13 week bioassay (Kasai et al.. 2008).
      4.7.3.4  Temporal Relationship

 1    4.7.3.4.1    L iver. Available information regarding temporal relationships between the key
 2    event (sustained proliferation of spontaneously transformed liver cells) and the eventual formation of liver
 3    tumors is limited. A comparison of 13-week and 2-year studies conducted in F344/DuCrj rats and
 4    Crj :BDF1 mice at the same laboratory revealed that tumorigenic doses of 1,4-dioxane produced liver
 5    toxicity by 13 weeks of exposure (Kano etal.. 2009; Kano et al.. 2008; JBRC. 1998). Hepatocyte swelling
 6    of the centrilobular area of the liver, vacuolar changes in the liver, granular changes in the liver, and
 7    single cell necrosis in the liver were observed in mice and rats given 1,4-dioxane in the drinking water for
 8    13 weeks. Sustained liver damage may lead to regenerative cell  proliferation and tumor formation
 9    following chronic exposure. As discussed above, histopathological evidence of regenerative cell
10    proliferation has been seen following long-term exposure to 1,4-dioxane (JBRC. 1998; Kociba et al..
11    1974). Tumors occurred earlier at high doses in both mice and rats from this study (Yamazaki. 2006);
12    however, temporal information regarding hyperplasia or other possible key events was not available (i.e.,
13    interim blood samples not collected, interim sacrifices were not  performed). Argus et al. (1973) studied
14    the progression of tumorigenesis by electron microscopy of liver tissues obtained following interim
15    sacrifices at 8 and 13 months of exposure (five rats/group, 574 mg/kg-day). The first change observed
16    was an increase in the size of the nuclei of the hepatocytes, mostly in the periportal area. Precancerous
17    changes were characterized by disorganization of the rough endoplasmic reticulum, increase in smooth
18    endoplasmic  reticulum, and decrease in glycogen and increase in lipid droplets in hepatocytes. These
19    changes increased in severity in the hepatocellular carcinomas in rats exposed to 1,4-dioxane for
20    13 months.

 1           Three types of liver nodules were observed in exposed rats at  13-16 months.  The first consisted
 2    of groups of these cells with reduced cytoplasmic basophilia and a slightly nodular appearance as viewed
 3    by light microscopy. The second type of nodule was described consisting of large cells, apparently filled
 4    and distended with fat. The third type of nodule was described as finger-like strands, 2-3 cells thick, of
 5    smaller hepatocytes with large hyperchromic nuclei and dense cytoplasm. This third type of nodule was
 6    designated as an incipient hepatoma, since it showed all the histological characteristics of a fully
 7    developed hepatoma. All three types of nodules were generally present in the same liver.
                                                                                                       94
                                       DRAFT - DO NOT CITE OR QUOTE

-------
 1   4.7.3.4.2   Nasal cavity. No information was available regarding the temporal relationship
 2   between toxicity in the nasal epithelium and the formation of nasal cavity tumors. Sustained nasal damage
 3   may lead to regenerative cell proliferation and tumor formation following chronic exposure. As discussed
 4   above (Section 4.2.2.2.1). no evidence of cytotoxicity has been observed following exposure to
 5   1.4-dioxane. despite histopathological evidence of regenerative cell proliferation and nasal tumors at the
 6   highest exposure concentration (Kano et al.. 2009)(Kasai et al.. 2009) (See Table 4-28)^ Other incidences
 7   of nasal damage may have occurred before tumor formation; however, temporal information regarding
 8   these events was not available (i.e.. interim sacrifices were not performed).
      4.7.3.5  Biological Plausibility and Coherence

 1    4.7.3.5.1   Liver. The hypothesis that sustained proliferation of spontaneously transformed
 2    liver cells is a key event within a MOA is possible based on supporting evidence indicating that
 3    1,4-dioxane is a tumor promoter of mouse skin and rat liver tumors (Lundberg et al.. 1987; Bull et al..
 4    1986; King etal.. 1973). Further support for this hypothesis is provided by studies demonstrating that
 5    1,4-dioxane increased hepatocyte DNA synthesis, indicative of cell proliferation (Mivagawa et al.. 1999;
 6    Uno et al.. 1994; Goldsworthy et al.. 1991; Stott etal.. 1981). In addition, the generally negative results
 7    for 1,4-dioxane in a number of genotoxicity assays indicates the carcinogenicity of 1,4-dioxane may not
 8    be mediated by a mutagenic MOA. The importance of cytotoxicity as a necessary precursor to sustained
 9    cell proliferation is biologically plausible, but is not supported by the dose-response in the majority of
10    studies of 1,4-dioxane carcinogenicity.

 1    4.7.3.5.2   Nasal cavity. Sustained cell proliferation in response to cell death from toxicity
 2    may be related to the formation of nasal cavity tumors; however, this MOA is also not established. Nasal
 3    carcinogens are generally characterized as potent genotoxins (Ashby. 1994); however, other MOAs have
 4    been proposed for nasal carcinogens that induce effects through other mechanisms (Rasper et al.. 2007;
 5    Green et al.. 2000).

 1           The National Toxicological Program (NTP) database identified 12 chemicals from approximately
 2    500 bioassays as nasal carcinogens and 1,4-dioxane was the only identified nasal carcinogen that showed
 3    little evidence of genotoxicity (Haseman and Hailey.  1997). Nasal tumors were not observed in an
 4    inhalation study in Wistar rats exposed to 111 ppm for 5 days/week for 2 years (Torkelson et al.. 1974).
 5    but were observed in an inhalation study  in F344 rats exposed to 1.250 ppm for 5 days/week for 2 years.
 6    Two human studies of occupational exposure, ranging from 0.06 ppm to 75 ppm for Imonth up to 41
 7    years, reported inconclusive Findings regarding increased tumor risk (Buffler et al.. 1978; Thiess et al..
 8    1976). It is important to note, that nasal tumors were not evaluated in the human studies and genotoxicity
 9    was not assessed in  either the human or animal studies.

10           While there is no known MOA for 1.4-dioxane and the human studies are inconclusive regarding
11    tumor risk, the noted nasal tumors in rats are considered biologically plausible and relevant to humans.
12    since similar cell types considered to be at risk are prevalent throughout the respiratory tract of rats and
13    humans. Differences in the anatomy of the upper respiratory tract and resulting differences in absorption
                                                                                                   95
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    or in local respiratory system effects between humans and rats are acknowledged and considered sources
 2    of uncertainty.
      4.7.3.6  Other  Possible Modes of Action

 3           An alternate MOA could be hypothesized that 1,4-dioxane alters DNA, either directly or
 4    indirectly (Kasai et al. 2009). which causes mutations in critical genes for tumor initiation, such as
 5    oncogenes or tumor suppressor genes. Following these events, tumor growth may be promoted by a
 6    number of molecular processes leading to enhanced cell proliferation or inhibition of programmed cell
 7    death. The results from in vitro and in vivo assays do not provide overwhelming support for the
 8    hypothesis of a genotoxic MOA for 1,4-dioxane carcinogenicity. The genotoxicity data for 1,4-dioxane
 9    were reviewed in Section 4.5.1 and were summarized in Table 4-23. Negative findings were reported for
10    mutagenicity in Salmonella typhimurium, Escherichia coll, and Photobacterium phosphoreum (Mutatox
11    assay) (Morita and Havashi. 1998; Hellmer and Bolcsfoldi. 1992; Kwanetal..  1990; Khudolev et al..
12    1987; Nestmann et al.. 1984; Haworth et al.. 1983; Stottetal.. 1981). Negative results were also indicated
13    for the induction of aneuploidy in yeast (Saccharomyces cerevisiae) and the sex-linked recessive lethal
14    test in Drosophila melanogaster (Zimmermann et al.. 1985). In contrast, positive results were reported in
15    assays for sister chromatid exchange (Galloway et al.. 1987). DNA damage (Kitchin and Brown. 1990).
16    and in in vivo micronucleus formation in bone marrow (Roy et al.. 2005; Mirkova. 1994). and liver (Roy
17    etal.. 2005; Morita and Havashi. 1998). Lastly, in the presence of toxicity, positive results were reported
18    for meiotic nondisjunction in drosophila (Munoz and Barnett. 2002). DNA damage (Sinaet al..  1983).
19    and cell transformation (Sheu etal.. 1988).

20           Additionally,  1,4-dioxane metabolism did not produce reactive intermediates that  covalently
21    bound to DNA  (Stottetal..  1981; Woo etal.. 1977b) and DNA repair assays were generally negative
22    (Goldsworthy et al.. 1991; Stottetal.. 1981). No studies were available to assess the  ability of
23    1,4-dioxane or its metabolites to induce oxidative damage to DNA.
                                                                                                  96
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     4.7.3.7  Conclusions About the Hypothesized Mode of Action

 1   4.7.3.7.1   Liver. The MOA by which 1,4-dioxane produces liver tumors is unknown, and
 2   available evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is
 3   inconclusive. A MOA hypothesis involving 1,4-dioxane induced cell proliferation is possible but data are
 4   not available to support this hypothesis. Pharmacokinetic data suggest that clearance pathways were
 5   saturable and target organ toxicity occurs after metabolic saturation. Liver toxicity preceded tumor
 6   formation in one study (Kociba et al.. 1974) and a regenerative response to tissue injury was demonstrated
 7   by histopathology. Liver hyperplasia and tumor formation have also been observed in the absence of
 8   cytotoxicity (Kano et al.. 2009; JBRC. 1998). Cell proliferation and tumor promotion have been shown to
 9   occur after prolonged exposure to 1,4-dioxane (Mivagawa et al.. 1999; Uno et al.. 1994; Goldsworthy et
10   al.. 1991;Lundbergetal.. 1987; Bull et al.. 1986; Stottet al.. 1981; King et al.. 1973).

 1   4.7.3.7.2   Nasal cavity. The MOA forthe formation of nasal cavity tumors is unknown,
 2   and evidence in support of any hypothetical mode of carcinogenic action for 1,4-dioxane is inconclusive.
 3   Nasal carcinogens are generally characterized as potent genotoxins (Ashby.  1994); however, other MOAs
 4   have been proposed for nasal carcinogens that induce effects through other mechanisms (Kasper et al..
 5   2007; Green et al.. 2000). Neither nasal tumors in the human studies nor genotoxicity in human or animal
 6   studies following exposure to 1.4-dioxane was evaluated, so the role of genotoxicity cannot be ruled out.
 7   A MOA hypothesis involving nasal damage, cell proliferation, and hvperplasia is possible, but data are
 8   not available to support this hypothesis. In studies that examined nasal effects after exposure to
 9   1.4-dioxane. at least one of these events is missing. More specifically, nasal cavity tumors have been
10   reported by Kasai et al.  (2009) in the absence of cytotoxicity and in Kano et al. (2009) in the absence of
11   hvperplasia. Therefore,  as per EPA's Cancer Guidelines (U.S. EPA. 2005a). there is insufficient
12   biological support for potential key events and to have reasonable confidence in the sequence of events
13   and how they relate to the development of nasal tumors following exposure to 1.4-dioxane.  Using the
14   modified Hill criteria, exposure-response and temporal relationships have not been established in support
15   of any hypothetical mode of carcinogenic action for 1.4-dioxane.  Thus, the MOA cannot be established.
     4.7.3.8  Relevance of the Mode of Action to Humans

 1           Several hypothesized MOAs for 1,4-dioxane induced tumors in laboratory animals have been
 2   discussed along with the supporting evidence for each. As was stated, the MOA by which 1,4-dioxane
 3   produces liver, nasal, peritoneal, and mammary gland tumors is unknown. Some mechanistic information
 4   is available to inform the MOA of the liver and nasal tumors but no information exists to inform the
 5   MOA of the observed peritoneal or mammary gland tumors (Kano et al.. 2009; JBRC. 1998; Yamazaki et
 6   al.. 1994).
                                                                                                 97
                                    DRAFT - DO NOT CITE OR QUOTE

-------
      4.8  Susceptible Populations  and  Life Stages

 1           There is no direct evidence to establish that certain populations and lifestages may be susceptible
 2    to 1,4-dioxane. Changes in susceptibility with lifestage as a function of the presence of microsomal
 3    enzymes that metabolize and detoxify this compound (i.e., CYP2E1 present in liver, kidney, and nasal
 4    mucosa can be hypothesized). Vieira et al. (1996) reported that large increases in hepatic CYP2E1 protein
 5    occur postnatally between 1 and 3 months in humans. Adult hepatic concentrations of CYP2E1 are
 6    achieved sometime between 1 and 10 years. To the extent that hepatic CYP2E1 levels are lower, children
 7    may be more susceptible to liver toxicity from 1,4-dioxane than adults. CYP2E1 has been shown to be
 8    inducible in the rat fetus. The level of CYP2E1 protein was increased by 1.4-fold in the maternal liver and
 9    2.4-fold in the fetal liver following ethanol treatment, as compared to the untreated or pair-fed groups
10    (Carpenter et al.. 1996). Pre- and postnatal induction of microsomal enzymes resulting from exposure to
11    1,4-dioxane or other drugs or chemicals may reduce overall toxicity following sustained exposure to
12    1,4-dioxane.

13           Genetic polymorphisms have been identified for the human CYP2E1 gene CWatanabe et al..
14    1994; Havashi et al..  1991) and were considered to be possible factors in the abnormal liver function seen
15    in workers exposed to vinyl chloride (Huang et al.. 1997). Individuals with a CYP2E1 genetic
16    polymorphism resulting in increased expression of this enzyme may be less susceptible to toxicity
17    following exposure to 1,4-dioxane.

18           Gender differences were noted in subchronic and chronic toxicity studies of 1,4-dioxane in mice
19    and rats (see Sections 4.6 and 4.7). No consistent pattern of gender sensitivity was  identified across
20    studies. In a 13 week inhalation study of male and female rats (Kasai et al., 2008) kidney toxicity. as
21    evidenced by hydropic change in the renal proximal tubules, was observed in female rats exposed to
22    3.200 ppm of 1.4-dioxane. but not male rats. This suggests a possible increased susceptibility of female
23    rats to renal damage following inhalation  exposure to 1.4-dioxane.
                                                                                                 98
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      5   DOSE-RESPONSE  ASSESSMENTS
      5.1  Oral Reference Dose (RfD)
      5.1.1   Choice of Principal Studies  and Critical  Effect with Rationale and
             J ustification

 1           Liver and kidney toxicity were the primary noncancer health effects associated with exposure to
 2    1,4-dioxane in humans and laboratory animals. Occupational exposure to 1,4-dioxane has resulted in
 3    hemorrhagic nephritis and centrilobular necrosis of the liver (Johnstone. 1959; Barber. 1934). In animals,
 4    liver and kidney degeneration and necrosis were observed frequently in acute  oral and inhalation studies
 5    (JBRC. 1998: Drewetal.. 1978: David. 1964: Kestenetal.. 1939: Laugetal.. 1939: Schrenk and Yant.
 6    1936: de Navasquez. 1935: Fairley et al.,  1934). Liver and kidney effects were also observed following
 7    chronic oral exposure to 1,4-dioxane in animals (Kano et al.. 2009: JBRC. 1998: Yamazaki et al.. 1994:
 8    NCI. 1978: Kociba et al.. 1974: Argus etal.. 1973: Argus etal.. 1965) (see Table 4-25).

 9           Liver toxicity in the available chronic studies was characterized by necrosis, spongiosis hepatic,
10    hyperplasia, cyst formation, clear foci, and mixed cell foci. Kociba et al. (1974) demonstrated
11    hepatocellular degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats) or greater. The
12    NOAEL for liver toxicity was 9.6 mg/kg-day and 19 mg/kg-day in male and female rats, respectively. No
13    quantitative incidence data were provided in this study. Argus et al. (1973) described early preneoplastic
14    changes in the liver and JBRC (1998) demonstrated liver lesions that are primarily associated with the
15    carcinogenic process. Clear and mixed-cell foci in the liver are commonly considered preneoplastic
16    changes and would not be considered evidence of noncancer toxicity. In the JBRC (1998) study,
17    spongiosis hepatis was associated with other preneoplastic changes in the liver (clear and mixed-cell foci)
18    and no other lesions indicative of liver toxicity were seen. Spongiosis hepatis was therefore not
19    considered indicative of noncancer effects in this study. The activity of serum enzymes (i.e., AST, ALT,
20    LDH, and ALP) was increased in mice and rats chronically exposed to 1,4-dioxane (JBRC. 1998):
21    however, these increases were seen only at tumorigenic dose levels. Blood samples were collected at
22    study termination and elevated serum enzymes may reflect changes associated with tumor formation.
23    Histopathological evidence of liver toxicity was not seen in rats from the JBRC (1998) study. The highest
24    non-tumorigenic dose levels for this study approximated the LOAEL derived from the Kociba et al.
25    (1974) study (94 and 148 mg/kg-day for male and female rats, respectively).

26           Kidney damage in chronic toxicity studies was characterized by degeneration of the cortical
27    tubule cells, necrosis with hemorrhage, and glomerulonephritis (NCL 1978: Kociba etal.. 1974: Argus et
28    al.. 1973: Argus etal.. 1965: Fairley et al.. 1934). Kociba et al. (1974) described renal tubule epithelial
29    cell degeneration and necrosis at doses of 94 mg/kg-day (LOAEL in male rats) or greater, with a NOAEL
30    of 9.6 mg/kg-day. No quantitative incidence data were provided in this study (Kociba et al.. 1974). Doses
31    of > 430 mg/kg-day 1,4-dioxane induced marked kidney alterations (Argus etal., 1973). The observed

                                                                                                 99
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    changes included glomerulonephritis and pyelonephritis, with characteristic epithelial proliferation of
 2    Bowman's capsule, periglomerular fibrosis, and distension of tubules. Quantitative incidence data were
 3    not provided in this study. In the NCI (1978) study, kidney lesions in rats consisted of vacuolar
 4    degeneration and/or focal tubular epithelial regeneration in the proximal cortical tubules and occasional
 5    hyaline casts. Kidney toxicity was not seen in rats from the JBRC (1998) study at any dose level (highest
 6    dose was 274 mg/kg-day in male rats and 429 mg/kg-day in female rats).

 7           Kociba et al. (1974) was chosen as the principal study for derivation of the RfD because the liver
 8    and kidney effects in this study are considered adverse and represent the most sensitive effects identified
 9    in the database (NOAEL 9.6 mg/kg-day, LOAEL 94 mg/kg-day in male rats). Kociba et al. (1974)
10    reported degenerative effects in the liver, while liver lesions reported in other studies (JBRC. 1998; Argus
11    et al.. 1973) appeared to be related to the carcinogenic process. Kociba et al. (1974) also reported
12    degenerative changes in the kidney. NCI (1978) and Argus et al. (1973) provided supporting data for this
13    endpoint; however, kidney toxicity was observed in these studies at higher doses. JBRC (1998) reported
14    nasal inflammation in rats (NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and mice (NOAEL
15    66 mg/kg-day, LOAEL 278 mg/kg-day).

16           Even though the study reported by Kociba et al. (1974) had one noteworthy weakness, it had
17    several noted strengths, including: (1) two-year study duration; (2) use of both male and female rats and
18    three dose levels, 10-fold apart, plus a control group; (3) a sufficient number of animals per dose group
19    (60 animals/sex/dose group; and (4) the authors conducted a comprehensive evaluation of the animals
20    including body weights and clinical observations, blood samples, organ weights of all the major tissues,
21    and a complete histopathological examination of all rats. The authors did not report individual incidence
22    data that would have allowed for a BMD analysis of this robust dataset.
      5.1.2  Methods of Analysis—Including Models (PBPK, BMD, etc.)

23           Several procedures were applied to the human PBPK model to determine if an adequate fit of the
24    model to the empirical model output or experimental observations could be attained using biologically
25    plausible values for the model parameters. The re-calibrated model predictions for blood 1,4-dioxane
26    levels did not come within 10-fold of the experimental values using measured tissue: air partition
27    coefficients of Leung and Paustenbach (1990) or Sweeney et al. (2008) (Figure B-8 and Figure B-9). The
28    utilization of a slowly perfused tissue:air partition coefficient 10-fold lower than measured values
29    produces exposure-phase predictions that are much closer to  observations, but does not replicate the
30    elimination kinetics (Figure B-10). Re-calibration of the model with upper bounds on the tissue:air
31    partition coefficients results in predictions that are still six- to sevenfold lower than empirical model
32    prediction or observations (Figure B-12 and Figure B-13).  Exploration of the model space using an
33    assumption of zero-order metabolism (valid for the 50 ppm inhalation exposure) showed that an adequate
34    fit to the exposure and elimination data can be achieved only when unrealistically low values are assumed
35    for the slowly perfused tissue:air partition coefficient (Figure B-16). Artificially low values for the other
36    tissue:air partition coefficients are not expected to improve the model fit, as these parameters are shown

                                                                                                  100
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    in the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km. This suggests
 2    that the model structure is insufficient to capture the apparent 10-fold species difference in the blood
 3    1,4-dioxane between rats and humans. In the absence of actual measurements for the human slowly
 4    perfused tissue:air partition coefficient, high uncertainty exists for this model parameter value.
 5    Differences in the ability of rat and human blood to bind 1,4-dioxane may contribute to the difference in
 6    Vd. However, this is expected to be evident in very different values for rat and human blood:air partition
 7    coefficients, which is not the  case (Table B-l). Therefore, some other, as yet unknown, modification to
 8    model structure  may be necessary.

 9           Kociba et al. (1974) did not provide quantitative incidence or severity data for liver and kidney
10    degeneration and necrosis. Benchmark dose (BMD) modeling could not be performed for this study and
11    the NOAEL for liver and kidney degeneration (9.6 mg/kg-day in male rats) was used as the point of
12    departure (POD) in deriving the RfD  for 1,4-dioxane.

13           Alternative PODs were calculated using incidence data reported for cortical tubule degeneration
14    in male and female rats (NCI. 1978) and liver hyperplasia (JBRC. 1998). The incidence data for cortical
15    tubule cell degeneration in male and female rats  exposed to 1,4-dioxane in the drinking water for 2 years
16    are presented in Table 5-1. Details of the BMD analysis of these data are presented in Appendix C. Male
17    rats were more sensitive to the kidney effects of 1,4-dioxane than females and the male rat data provided
18    the lowest POD for cortical tubule degeneration  in the NCI (1978) study (BMDL10 of 22.3 mg/kg-day)
19    (Table 5-2). Incidence data (Kano et al.. 2009; JBRC.  1998) for liver hyperplasia in male and female rats
20    exposed to 1,4-dioxane in the drinking water for 2 years are presented in Table 5-3. Details of the BMD
21    analysis of these data are presented in Appendix C. Male rats were more sensitive to developing liver
22    hyperplasia due  to exposure to 1,4-dioxane than  females and the male rat data provided the lowest POD
23    for hyperplasia in the JBRC (1998) study (BMDL10 of 23.8 mg/kg-day) (Table 5-4). The BMDL10 values
24    of 22.3 mg/kg-day and 23.8 mg/kg-day from the NCI (1978) and JBRC (1998) studies, respectively, are
25    about double the NOAEL (9.6 mg/kg-day) observed by Kociba et al. (1974).
                                                                                                  101
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table 5-1    Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed to
             1,4-dioxane in drinking water for 2 years

0
0/3 1a
Males (mg/kg-day)
240
20/3 1b
Females (mg/kg-day)
530
27/33b
0
0/3 1a
350
0/34
640
10/32b
Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this review.
"Incidence significantly elevated compared to control by Fisher's Exact test (p < 0.001) performed for this review.

Source: NCI (1978).
Table 5-2    BMD and BMDL values derived from BMD modeling of cortical tubule degeneration in
             male and female Osborne-Mendel rats exposed to 1,4-dioxane in drinking water for 2
             years
                                    BMDio (mg/kg-day)                    BMDL10 (mg/kg-day)

         Male rats                                 28.8                                 22.3

       Female rats                               596.4                                452.4

Source: NCI (1978).
Table 5-3    Incidence of liver hyperplasia in F344/DuCrj rats exposed to 1,4-dioxane in drinking
             water for 2 years
Males (mg/kg-day)a
0
3/40
11
2/45
55
9/35b
274
12/22C
0
0/38b
Females (mg/kg-day)a
18
0/37
83
1/38
429
14/24C
 Dose information from Kano et al. (2009) and incidence data for sacrificed animals from JBRC (1998).
"Statistically significant compared to controls by the Dunnett's test (p < .05).
""Incidence significantly elevated compared to control by x2 test (p < 0.01).

Sources: Kano et al. (2009): JBRC (1998).
                                                                                                 102
                                 DRAFT - DO NOT CITE OR QUOTE

-------
      Table 5-4   BMD and BMDL values derived from BMD modeling of liver hyperplasia in male and
                 female F344/DuCrj rats exposed to 1,4-dioxane in drinking water for 2 years
                                       BMDio (mg/kg-day)                   BMDLio (mg/kg-day)
              Male rats                              35.9                                23.8
             Female rats                            137.3                                88.5
      Source: Kano et al. (2009): JBRC (1998).
      5.1.3  RfD Derivation - Including Application of Uncertainty Factors (UFs)

 1           The RfD of 3 x 1CT2 mg/kg-day is based on liver and kidney toxicity in rats exposed to
 2    1,4-dioxane in the drinking water for 2 years (Kociba et al., 1974). The Kociba et al. (1974) study was
 3    chosen as the principal study because it provides the most sensitive measure of adverse effects by
 4    1,4-dioxane. The incidence of liver and kidney lesions was not reported for each dose group. Therefore,
 5    BMD modeling could not be used to derive a POD. The RfD for 1,4-dioxane is derived by dividing the
 6    NOAEL of 9.6 mg/kg-day (Kociba etal., 1974) by a composite UF of 300, as follows:
 7                         RfD    =       NOAEL/UF
 8                                =9.6 mg/kg-day / 300
 9                                =       0.03 or 3 x 10~2 mg/kg-day

10           The composite UF of 300 includes factors of 10 for animal-to-human extrapolation and for
11    interindividual variability, and an UF of 3 for database deficiencies.

12           A default interspecies UF of 10 was used to account for pharmacokinetic and pharmacodynamic
13    differences across species. Existing PBPK models could not be used to derive an oral RfD for 1,4-dioxane
14    (Appendix B).

15           A default interindividual variability UF of 10 was used to account for variation in sensitivity
16    within human populations because there  is limited information on the degree to which humans of varying
17    gender, age, health status, or genetic makeup might vary in the disposition of, or response to, 1,4-dioxane.

18           An UF of 3 for database deficiencies was applied due to the lack of a multigeneration
19    reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
20    1,4-dioxane (Giavini et al., 1985). This developmental study indicates that the  developing fetus may be a
21    target of toxicity.

22           An UF to extrapolate from a subchronic to a chronic exposure duration was not necessary
23    because the RfD was derived from a study using a chronic exposure protocol.

24           An UF to extrapolate from a LOAEL to a NOAEL was not necessary because the RfD was based
25    on a NOAEL. Kociba et al. (1974) was a well-conducted, chronic drinking water study with an adequate
26    number of animals. Histopathological examination was performed for many organs and tissues, but

                                                                                               103
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    clinical chemistry analysis was not performed. NOAEL and LOAEL values were derived by the study
 2    authors based on liver and kidney toxicity; however quantitative incidence data was not reported. Several
 3    additional oral studies (acute/short-term, subchronic, and chronic durations) were available that support
 4    liver and kidney toxicity as the critical effect (Kano etal.. 2008: JBRC. 1998: NCI.  1978: Argus et al..
 5    1973) (Table 4-15 and Table 4-17). Although degenerative liver and kidney toxicity was not observed in
 6    rats from the JBRC (1998) study at doses at or below the LOAEL in the Kociba et al. (1974) study, other
 7    endpoints such as metaplasia and hyperplasia of the nasal epithelium, nuclear enlargement, and
 8    hematological effects, were noted.
      5.1.4  RfD Comparison Information

 9           PODs and sample oral RfDs based on selected studies included in Table 4-18 are arrayed in
10    Figure 5-1 to Figure 5-3, and provide perspective on the RfD supported by Kociba et al. (1974). These
11    figures should be interpreted with caution because the PODs across studies are not necessarily
12    comparable, nor is the confidence in the data sets from which the PODs were derived the same. PODs in
13    these figures may be based on a NOAEL, LOAEL, or BMDL (as indicated), and the nature, severity, and
14    incidence of effects occurring at a LOAEL are likely to vary. To some extent, the confidence associated
15    with the resulting sample RfD is reflected in the magnitude of the total UF applied to the POD (i.e., the
16    size of the bar); however, the text of Sections 5.1.1 and 5.1.2 should be consulted for a more complete
17    understanding of the issues associated with each data set and the rationale for the selection of the critical
18    effect and principal study used to derive the RfD.

19           The predominant noncancer effect of chronic oral exposure to 1,4-dioxane is degenerative effects
20    in the liver and kidney. Figure 5-1 provides a graphical display of effects that were observed in the liver
21    following chronic oral exposure to 1,4-dioxane. Information presented includes the PODs and UFs that
22    could be considered in deriving the oral RfD. As discussed in Sections 5.1.1 and 5.1.2, among those
23    studies that demonstrated liver toxicity, the study by Kociba et al. (1974) provided the data set most
24    appropriate for deriving the RfD. For degenerative liver effects resulting from 1,4-dioxane exposure, the
25    Kociba et al. (1974) study represents the most sensitive effect and dataset observed in a chronic bioassay
26    (Figure 5-1).

27           Kidney toxicity as evidenced by glomerulonephritis (Argus etal.. 1973: Argus etal.. 1965) and
28    degeneration of the cortical tubule (NCI. 1978: Kociba etal.. 1974) has also been observed in response to
29    chronic exposure to 1,4-dioxane.  As was discussed in Sections 5.1 and 5.2, degenerative effects were
30    observed in the kidney at the same dose level as effects in the liver (Kociba et al.. 1974). A comparison of
31    the available datasets from which an RfD could potentially be derived is presented in Figure 5-2.

32           Rhinitis  and inflammation of the nasal cavity were reported in both the NCI (1978) (mice only,
33    dose > 380 mg/kg-day) and JBRC (1998) studies (> 274 mg/kg-day in rats, >278 mg/kg-day in  mice).
34    JBRC (1998) reported nasal inflammation in rats (NOAEL 55 mg/kg-day, LOAEL 274 mg/kg-day) and
35    mice (NOAEL 66 mg/kg-day, LOAEL 278 mg/kg-day). A comparison of the available datasets from
36    which an RfD could potentially be derived is presented in Figure 5-3.

                                                                                                 104
                                     DRAFT - DO NOT CITE OR QUOTE

-------
1            Figure 5-4 displays PODs for the major targets of toxicity associated with oral exposure to
2    1,4-dioxane. Studies in experimental animals have also found that relatively high doses of 1,4-dioxane
3    (1,000 mg/kg-day) during gestation can produce delayed ossification of the sternebrae and reduced fetal
4    BWs (Giavini et al.. 1985). This graphical display (Figure 5-4) compares organ specific toxicity for
5    1,4-dioxane, including a single developmental study. The most sensitive measures of degenerative liver
6    are and kidney effects. The sample RfDs for degenerative liver and kidney effects are identical since they
7    were derived from the same study and dataset (Kocibaetal.. 1974) and are presented for completeness.
          100
                    Rat
                                   Rat
                                                 Mouse
                                                                  Rat
                                                                                 Rat
           10
        o
        Q
           0.1
          0.01
                                                   ffi
I
• POD
[HAnimal-to-human
nHuman variation
0LOAELto NOAEL
DSubchronic to Chronic
BDatabase deficiencies
ORfD
                Liver hyperplasia;     Hepatocellular    Increase in serum liver Increase in serum liver   Liver hyperplasia;
                 NOAEL; 2 yr rat     degeneration and   enzymes; NOAEL; 2 yr enzymes; NOAEL; 2 yr   BMDL10; 2 yr rat
              drinking water study    necrosis; NOAEL; 2 yr  mouse drinking water  rat drinking water study drinking water study
                             rat drinking water study    study
             Figure 5-1  Potential points of departure (POD) for liver toxicity endpoints with
                          corresponding applied uncertainty factors and derived RfDs following
                          oral exposure to 1,4-dioxane.
                                                                                                          105
                                       DRAFT - DO NOT CITE OR QUOTE

-------
                Rat
                                         Rat
                                                                  Rat
1000
 100
  10
 0.1 -
0.01
 • POD

[HAnimal-to-human

QHuman variation

E2LOAELIONOAEL

dSubchronic to Chronic

• Database deficiencies

 ORfD
     Glomerulonephritis; LOAEL; 13 month  Degeneration and necrosis of tubular  Cortical tubule degeneration; BMDL10;
        rat drinking water study         epithelium; NOAEL; 2 yr rat drinking   2 yr rat drinking water study
                                    water study


  Figure 5-2   Potential points of departure  (POD) for kidney toxicity endpoints with

                corresponding applied uncertainty factors and derived RfDs following

                oral exposure to  1,4-dioxane.
                                                                                                      106
                              DRAFT - DO NOT CITE OR QUOTE

-------
                      Mouse
                                                             Rat
 100
  10
o
Q
  0.1
                      I
                       O
I
 O
                        • POD
                        [jAnimal-to-human
                        QHuman variation
                        0LOAELto NOAEL
                        Dsubchronic to Chronic
                        B Database deficiencies
                        ORfD
     Nasal inflammation; NOAEL; 2 yr mouse drinking water study Nasal inflammation; NOAEL; 2 yr rat drinking water study

     Figure 5-3  Potential points of departure (POD) for nasal inflammation with
                  corresponding applied uncertainty factors and derived sample RfDs
                  following oral exposure to 1,4-dioxane.
                                                                                                   107
                                DRAFT - DO NOT CITE OR QUOTE

-------
         1000
          100 -
                     Rat
                                      Rat
                                                        Rat
                                                                        Mouse
           10
       m
       8
       Q
         0.01
  POD
OAnimal-to-human
QHuman variation
^LOAEL to NOAEL
QSubchronic to Chronic
^Database deficiencies
ORfD
              Degeneration and necrosis     Hepatocellular       Delayed ossification of     Nasal inflammation;
               of tubular epithelium;   degeneration and necrosis;  sternebrae and reduced    NOAEL; 2 yr mouse
              NOAEL; 2 yr rat drinking   NOAEL; 2 yr rat drinking fetal body weight; NOAEL;  drinking water study
                water study           water sduy        rat study gestation days 6-
                                                      15

            Figure 5-4  Potential points of departure (POD) for organ specific toxicity
                        endpoints with corresponding applied uncertainty factors and derived
                        sample RfDs following oral exposure to 1,4-dioxane.
     5.1.5  Previous RfD Assessment
                   An assessment for 1,4-dioxane was previously posted on the IRIS database in 1988. An
                   oral RfD was not developed as part of the 1988 assessment.
     5.2   Inhalation  Reference Concentration (RfC)
     5.2.1   Choice of Principal Study and Candidate Critical  Effect(s) with
                    Rationale and J ustification

3           Two human studies of occupational exposure to 1.4-dioxane have been published (Buffler et al.
4    1978; Thiess et al..  1976); however, neither study provides sufficient information and data to quantify
5    subchronic or chronic noncancer effects. In each study. Findings were inconclusive and the cohort size
6    and number of reported cases were limited (Buffler et al.. 1978; Thiess et al.. 1976).
                                                                                                 108
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1           Four inhalation studies in animals were identified in the literature; two, 13-week subchronic
 2    studies in laboratory animals (Kasai et al.. 2008; Fairley et al.. 1934) and two. 2-year chronic studies in
 3    rats (Kasai et al.. 2009; Torkelson et al.. 1974).

 4           In the subchronic study by Fairley et al. (1934) rabbits, guinea pigs, rats, and mice
 5    (3-6/species/group) were exposed to 1.000. 2.000. 5.000. or  10.000 ppm of 1.4-dioxane vapor for
 6    1.5 hours two times a day for 5 days. 1.5 hours for one day, and no exposure on the seventh day. Animals
 7    were exposed until death occurred or were sacrificed after various durations of exposure (3-202.5 hours).
 8    Detailed dose-response information was not provided; however, severe kidney and liver damage and
 9    acute vascular congestion of the lungs were  observed at concentrations > 1.000 ppm. Kidney damage was
10    described as patchy degeneration of cortical tubules with vascular congestion and hemorrhage. Liver
11    lesions varied from cloudy hepatocyte swelling to large areas of necrosis. In this study, a LOAEL of
12    1.000 ppm for liver and kidney degeneration in rats, mice, rabbits, and guinea pigs was identified by EPA.

13           In the subchronic study by Kasai et al. (2008) male and female rats (10/group/sex) were exposed
14    to 0. 100. 200. 400. 800. 1.600. 3.200. and 6.400 ppm of 1.4-dioxane for 6 hours/day. 5 days/week for  13
15    weeks. This study observed a range of 1.4-dioxane induced nonneoplastic effects across several organ
16    systems including the liver and respiratory tract (from the nose to the bronchus region) in both sexes and
17    the kidney in females. Detailed dose-response information was provided, illustrating a
18    concentration-dependent increase of nuclear enlargement of nasal  (respiratory and olfactory), trachea, and
19    bronchus epithelial cells (both sexes): vacuolic change of nasal and bronchial epithelial cells (both sexes).
20    necrosis and centrilobular swelling of hepatocytes (both sexes); and hydropic change in the proximal
21    tubules of the kidney (females). The study authors determined nuclear enlargement of the nasal
22    respiratory epithelium as the most sensitive lesion and a LOAEL of 100 ppm was identified based on this
23    effect.

24           Torkelson et al (1974) performed a chronic inhalation study in which male and female Wistar
25    rats (288/sex) were exposed to 111 ppm 1.4-dioxane vapor for 7 hours/day. 5 days/week for 2 years.
26    Control rats (192/sex) were exposed to filtered air. No significant effects were observed on BWs.
27    survival organ weights, hematology. clinical chemistry, or histopathology. A free standing NOAEL of
28    111 ppm was identified in this study by EPA.

29           Kasai et al. (2009) reported data for groups of male F344 rats (50/group) exposed to 0. 50. 250.
30    and 1.250 ppm of 1.4-dioxane for 6 hours/day. 5 days/week, for 2  years. In contrast to the subchronic
31    Kasai et al. (2008) study, this 2-year bioassay reported more nonneoplastic effects in multiple organ
32    systems. Additional noted incidences included: (1) inflammation of nasal respiratory and olfactory
33    epithelium. (2) squamous cell metaplasia and hyperplasia of nasal  respiratory epithelium. (3) atrophy and
34    respiratory metaplasia of olfactory epithelium. (4) hydropic change and sclerosis in the lamina propria  of
35    nasal cavity. (5) nuclear enlargement in proximal tubules of the kidney and in the centrilobular region of
36    the liver. (6) centrilobular necrosis in the liver, and (7) spongiosis  hepatis. Some of these
37    histopathological lesions were significantly increased compared to controls at the lowest exposure level
38    (50 ppm). including nuclear enlargement of respiratory and olfactory epithelium; and atrophy and
                                                                                                   109
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    respiratory metaplasia of olfactory epithelium. Many of these histopathological lesions were increased in
 2    a concentration-dependent manner.

 3           The Fairley et al. (1934) study was insufficient to characterize the inhalation risks of 1.4-dioxane
 4    because control animals were not used, thus limiting the ability to perform statistical analysis;
 5    additionally, no data for low dose exposure were reported. Because Torkelson et al. (1974) identified a
 6    free-standing NOAEL only, this study was also insufficient to characterize the inhalation risks of
 7    1.4-dioxane. A route extrapolation from oral toxicity data was not performed because 1,4-dioxane
 8    inhalation causes direct effects on the respiratory tract (i.e., respiratory irritation in humans, pulmonary
 9    congestion in animals) (Wirth and Klimmer. 1936; Fairley et al.. 1934; Yant et al.. 1930). which would
10    not be accounted for in a cross-route extrapolation. In addition, available kinetic models are not
11    suitable for this purpose (Appendix B).

12           The chronic Kasai et al. (2009) study was selected as the principal study for the derivation of the
13    RfC. The Kasai et al.  (2009) 2-year bioassay utilized 50 animals per exposure group, a range of exposure
14    concentrations which were based on the results of the subchronic study (Kasai et al.. 2008) and
15    thoroughly examined toxicity of l-4.dioxane in multiple organ systems. Based on the noncancer database
16    for 1.4-dioxane. this study demonstrated exposure concentration-related effects for histopathological
17    lesions at a lower concentration (50 ppm) compared to the subchronic Kasai et al. study (2008). The
18    2-year bioassay (Kasai et al.. 2009) did not observe effects in both sexes, but the use of only male rats
19    was proposed by the study authors as justified by data illustrating the  absence of induced mesotheliomas
20    in female rats following exposure to 1.4-dioxane in drinking water (Yamazaki et al.. 1994). Additionally.
21    a similar pattern of effects was observed after oral exposure to 1.4-dioxane (Kano et al.. 2009; JBRC.
22    1998) as observed in the Kasai et al (2009) 2-year inhalation study.

23           Nonneoplastic lesions from the Kasai et al. (2009) study that were statistically increased as
24    compared to control were considered candidates for the critical effect. The candidate endpoints included
25    centrilobular necrosis of the liver, spongiosis hepatis. squamous cell metaplasia of nasal respiratory
26    epithelium, squamous cell hyperplasia of nasal respiratory epithelium, respiratory metaplasia of nasal
27    olfactory epithelium,  sclerosis in lamina propria of nasal cavity, and two degenerative nasal lesions, that
28    is.  atrophy of nasal olfactory epithelium and hydropic change in the lamina propria (Table 5-5). Despite
29    statistical increases at the low- and mid exposure concentrations (50 and 250 ppm. respectively).
30    incidences of nuclear enlargement of respiratory epithelium (nasal cavity), olfactory epithelium (nasal
31    cavity), and proximal tubule (kidney) were not considered candidates for the  critical effect given that the
32    toxicological significance of nuclear enlargement is uncertain (See Section 4.6.2 and Table 4-22).
                                                                                                    110
                                      DRAFT - DO NOT CITE OR QUOTE

-------
     Table 5-5   Incidences of nonneoplastic lesions resulting from chronic exposure (ppm) to
                 1,4-dioxane considered for identification of a critical effect.
Species/Strain
Rat/ F344 (male)


Centrilobular necrosis
Spongiosis hepatis
Squamous cell metaplasia;
respiratory epithelium
Squamous cell hyperplasia;
respiratory epithelium
Respiratory metaplasia;
asa olfactory epithelium
Atrophy; olfactory epithelium
Hydropic change;
lamina propria
Sclerosis; lamina propria
Concentration (ppm)
0
1/50
7/50
0/50
0/50
11/50
0/50
0/50
0/50
50
3/50
6/50
0/50
0/50
34/50a
40/50a
2/50
0/50
250
6/50
13/50
7/50b
1/50
49/50a
47/50a
36/50a
22/50a
1,250
12/50a
19/50a
44/50a
10/50a
48/50a
48/50a
49/50a
40/50a
ap<0.01 by x^ test.
bp < 0.05 by x2 test.
     Source: Kasai et al. (2009).
 5.2.2  Methods  of Analysis

 1           Benchmark dose (BMP) modeling methodology (U.S. EPA. 2000a) was used to analyze the
 2   candidate endpoints identified for 1.4-dioxane. Use of BMP methods involves fitting mathematical
 3   models to the observed dose-response data and provides a BMP and its 95% lower confidence limit
 4   (BMDL) associated with a predetermined benchmark response (BMR). For 1.4-dioxane. the selected
 5   datasets in Table 5-5 were considered as candidate critical effects and analyzed using BMP modeling to
 6   determine potential POPs. Information regarding the degree of change in the selected endpoints that is
 7   considered biologically significant was not available. Therefore, a BMR of 10% extra risk was selected
 8   under the assumption that it represents a minimally biologically significant response level (U.S. EPA.
 9   2000a).
10           The BMDs and BMDLs for centrilobular necrosis, spongiosis hepatis. squamous cell metaplasia
11   of the respiratory epithelium, and hydropic change of lamina propria are presented in Table 5-6. Due to
12   poor fit or substantial model uncertainty. BMP model results were inadequate for the following nasal
13   lesions: atrophy (olfactory epithelium), respiratory metaplasia (olfactory epithelium), and sclerosis
14   (lamina propria). Consequently, the NOAEL/LOAEL approach was used to determine potential POPs for
15   these endpoints. The detailed results of the BMP analysis are provided in Appendix F.
 5.2.3  Exposure Duration and Dosimetric Adjustments

16           Because an RFC is a measure that assumes continuous human exposure over a lifetime, data
17   derived from animal studies need to be adjusted to account for the noncontinuous exposure protocols used
18   in animal studies. In the Kasai et al. (2009) study, rats were exposed to 1.4-dioxane for 6 hours/day. 5
                                                                                                111
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    days/week for 2 years. Therefore, the duration-adjusted POPs for liver and nasal lesions in rats were
 2    calculated as follows:
                     PODADJ(ppm) = POD(ppm):
                                                   hours exposed per day   days exposed per week
                                                          24hours
                                                                               Vdays
 4           RfCs are typically expressed in units of mg/m3; so POD^m (ppm) values were converted using

 5    the chemical specific conversion factor of 1 ppm = 3.6 mg/m3 for 1.4-dioxane (Table 2-1). The following

 6    calculation was used:
                     POD ADJ (mg/m3) =
                                                   3.6 mg/m3
                                                      Ippm
 8           The calculated POD/\DJ (mg/m3) values for all considered endpoints are presented in the last

 9    column of Table 5 -6^

10
      Table 5-6    Duration adjusted POD estimates for BMDLs (from best fitting BMDS models) or
                  NOAELs/LOAELs from chronic exposure to 1,4-dioxane
11


12

13
Endpoint . .
L,OAEITD Model
(ppm)
BMR
BMD
(ppm)
BMDL
(ppm)
PODADJ
(mg/m3)
Liver Effects
Centrilobular necrosis;
Liver
Spongiosis hepatis; Liver
Dichotomous-Hill
Log-logistic0
10
10
220
314
60
172
38.6
111
Nasal Effects
Squamous cell metaplasia;
respiratory epithelium
Squamous cell
hyperplasia; respiratory
epithelium
Respiratory metaplasia;
olfactory epithelium
Atrophy; olfactory
epithelium
Hydropic change;
lamina propria
Sclerosis; lamina propria 50
Log-probit
Log-probit
50
50
Log-logistic
-c
10
10
-
-
10

218
756
-
-
69

160
561
-
-
47

103
361
32.2
32.2
30.2
32.2e
      aNOAEL is identified in this assessment as the highest tested exposure dose at which there is no statistically significant effect in the
         exposed group as compared to control.
      bLOAEL is identified in this assessment as the lowest tested exposure dose at which there is a statistically significant effect in the
         exposed group as compared to control.
      CBMDS model results are not adequate for use to derive a POD. Therefore, the NOAEL/LOAEL approach is used to determine a POD
         for these endpoints. BMDS analysis for these endpoints is included in Appendix F.
      dDichotomous Hill model had lowest BMDL, but model output warned that the BMDL estimate was "imprecise at best".
      eBased on the NOAEL of 50 ppm.
        Based on analysis of data in Table 5-6a the liver effects (i.e.. centrilobular necrosis and spongiosis
hepatis) were shown to be less sensitive than the nasal effects and were not considered further as
                                                                                                    112
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    candidate critical effects. Similarly, the squamous cell metaplasia and hyperplasia of the respiratory
 2    epithelium yielded potential POPs that were 3-fold or greater than the remaining nasal effects: thus, these
 3    effects were not considered further as candidate critical effects. The POPs adjusted for continuous
 4    exposure for sclerosis of the lamina propria. atrophy of the olfactory epithelium, and respiratory
 5    metaplasia of the olfactory epithelium were identical (32.2 mg/m3) and similar to the POD^m for hydropic
 6    change of the lamina propria (30.2 mg/m3). Although the POD ADJ estimates were either identical or
 7    similar, the responses (i.e.. increased incidence of effect) associated with the POD Am for these effects.
 8    (i.e.. 0% for sclerosis. 10% for hydropic change. 59% for respiratory metaplasia. 80% for atrophy) varied.

 9           As  shown in Table 5-5. atrophy and respiratory metaplasia of the olfactory epithelium were the
10    most sensitive effects based on the responses of 80 and 59%. respectively, observed in animals exposed at
11    the lowest concentration (50 ppm). Increased incidences of the other nasal effects and liver effects were
12    observed at either 50 ppm or greater; however, not to the extent that was observed for atrophy and
13    respiratory metaplasia of the olfactory epithelium. Typically, chemical-induced nasal effects include
14    atrophy and/or necrosis, cell proliferation/hvperplasia. and metaplasia depending on the nature of the
15    tissue damage and exposure (Harkema et al.. 2006; Boorman et al.. 1990; Gaskell. 1990). However the
16    pathological progression of these events is uncertain and often accompanied by an inflammatory
17    response. Since the data do not support a continuum of pathological events associated with respiratory
18    tract effects, both atrophy and respiratory metaplasia of the olfactory epithelium are  selected as co-critical
19    effects in this assessment.

20           For the derivation of a RfC based upon an animal study, the selected POD must be adjusted to
21    reflect the human equivalent concentration (HEC). The HEC was calculated by the application of a
22    dosimetric adjustment factor (DAF). in accordance with the U.S. EPA Methods for Derivation of
23    Inhalation Reference Concentrations and Application of Inhalation Dosimetry (hereafter referred to as the
24    RfC methodology) (U.S. EPA. 1994). DAFs are ratios of animal and human physiologic parameters, and
25    are dependent on the nature of the contaminant (particle or gas) and the target site (e.g.. respiratory tract
26    or remote to the portal-of-entry)  (U.S. EPA. 1994).

27           1.4-Dioxane is miscible  with water and has a high blood:air partition coefficient. Typically.
28    highly water-soluble and directly reactive chemicals (i.e. Category 1 gases) partition greatly into the
29    upper respiratory tract, induce portal-of-entry effects, and do not accumulate significantly in the blood.
30    1.4-Dioxane induces effects throughout the respiratory tract, liver, and kidneys, and  it has been measured
31    in the blood after inhalation exposure (Kasai et al.. 2008). The observations of systemic (i.e..
32    nonrespiratory) effects and measured blood levels resulting from 1.4-dioxane exposure indicate that this
33    compound is absorbed into the bloodstream and distributed throughout the body. Furthermore, the lack of
34    an anterior to posterior gradient for the nasal effects induced by  1.4-dioxane is not typical of chemicals
35    which are predominantly directly reactive. Thus. 1.4-dioxane might be best described as a water-soluble
36    and non-directly reactive gas. Gases such as these are readily taken up into respiratory tract tissues and
37    can also diffuse  into the blood capillaries (Medinsky and Bond. 2001). The effects in the olfactory
38    epithelium may be the result of the metabolism of 1.4-dioxane to an acid metabolite: however, for the
39    reasons stated above it is unclear whether or not these effects are solely the result of portal-of-entry or
                                                                                                   113
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    systemic delivery. A similar pattern of effects was observed after oral exposure to 1.4-dioxane (JBRC.
 2    1998: KanoetaL 2009).

 3           In consideration of the evidence described above, the human equivalent concentration (FffiC) for
 4    1.4-dioxane was calculated by the application of the appropriate dosimetric adjustment factor (DAF) for
 5    systemic acting gases (i.e.. Category 3 gases), in accordance with the U.S. EPA RfC methodology (U.S.
 6    EPA. 1994). However, since 1.4-dioxane is water soluble and might induce portal-of-entry effects, an
 7    alternative calculation of the FffiC for 1.4-dioxane. based on the application of the corresponding DAF for
 8    portal-of-entry acting gases (i.e.. Category 1) is provided in Appendix G.

 9           The calculation of the HEC used in this assessment is as follows:

10                  DAF = (Hb/g)A/(Hb/g)H

11                  DAF= 1.861/1.666

12                  DAF= 1.12

13           where:

14                  (FIb/g)A = the animal blood:air partition coefficient =1.861 (Sweeney et al.. 2008)

15                  (FIb/g)H = the human blood:air partition coefficient =1.666 (Sweeney et al.. 2008)

16           Given that the animal blood:air partition coefficient is higher than the human value resulting in a
17    DAF>1. a default value of 1 is substituted in accordance  with the U.S. EPA RfC methodology (U.S. EPA.
18    1994). Analysis of the existing inhalation dosimetry modeling database supports the application of a DAF
19    of 1 (U.S. EPA. 2009c).  Application of these models to gases that have similar physicochemical
20    properties and induce similar nasal effects as 1.4-dioxane estimate DAFs > j^

21           Utilizing a DAF of 1. the FffiC for atrophy and respiratory metaplasia of the olfactory epithelium
22    in male F344/DuCrj rats is calculated as follows:
23                                      POD^ (mg/m3) = POD^m (mg/m3) x DAF

24                                                      = PODAm (mg/m3) x. LO
25                                                      =32. 2 mg/m3 x l.Q

26                                                      = 32.2 mg/m3

27           Therefore, the P()DHEC of 32.2 mg/m3 for the co-critical effects of atrophy and respiratory
28    metaplasia of the olfactory epithelium is used for the derivation of a RfC for 1.4-dioxane.
                                                                                                 114
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 5.2.4  RfC Derivation- Including Application of Uncertainty Factors  (UFs)

 1           The RfC of 3 x ICT2 mg/m3 is based on atrophy and respiratory metaplasia of the olfactory
 2    epithelium in male rats exposed to 1.4-dioxane via inhalation for 2 years (Kasai et al.. 2009). The RfC for
 3    1.4-dioxane is derived by dividing the POPART by a composite UF of 1.000.

 4                                RfQ = PODHEc/UF
 5                                    =32.2 mg/m3/1.000
 6                                    = 0.0322 or 3 x 10~2 mg/m3 (rounded to 1 significant figure)

 7           An UF of 10 was used to extrapolate from a LOAEL to a NOAEL because a LOAEL was used as
 8    the POD for critical effects. A NOAEL for atrophy and respiratory metaplasia of the olfactory epithelium
 9    was not identified in the study by Kasai et al. (2009).

10           A default interindividual variability UF of 10 was used to account for variation in sensitivity
11    within human populations because there  is limited information on the degree to which humans of varying
12    gender, age, health status, or genetic makeup might vary in the disposition of. or response to. 1.4-dioxane.

13           An interspecies UF of 3 was used for animal-to-human extrapolation to account for
14    pharmacodynamic differences between species. This uncertainty factor is comprised of two separate areas
15    of uncertainty to account for differences  in the toxicokinetics and toxicodynamics of animals and humans.
16    In this assessment, the toxicokinetic uncertainty was accounted for by the calculation of a HEC and
17    application of a dosimetric adjustment factor as outlined in the RfC methodology  (U.S. EPA. 1994). As
18    the toxicokinetic differences are thus accounted for, only the toxicodynamic uncertainties remain, and an
19    UF of 3 is retained to account for this uncertainty.

20           An UF of 3 for database deficiencies was applied due to the lack of a multigeneration
21    reproductive toxicity study. The oral toxicity database included a single prenatal developmental study that
22    indicated the developing fetus may be a target of toxicity (Giavini et al..  1985).

23           An UF of 1 was used to extrapolate from a subchronic to a chronic exposure duration because the
24    RfC was derived from a study using a chronic exposure protocol.
 5.2.5  RfC Comparison Information

25           Figure 5-5 presents POPs, applied UFs. and derived sample RfCs for possible endpoints from the
26    chronic inhalation Kasai et al (2009) in male rats. The POPs are based on the BMDLjn. NOAEL. or
27    LOAEL and appropriate unit conversion, duration, and dosimetric adjustments were applied before
28    applications of UFs. The predominant noncancer effects of chronic inhalation exposure to 1.4-dioxane
29    include nasal and liver effects.  Figure 5-5 provides a graphical display of effects that were observed in
30    the Kasai et al. (2009) study.  Information presented includes the POPs and UFs that could be considered
31    in deriving the inhalation RfC.  As discussed in Sections 5.2.1 and 5.2.3. the Kasai et al. (2009) study
                                                                                                115
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    provided the data set for deriving the RfC.  The nasal effects of the olfactory epithelium represent the
 2    most sensitive effects.
        I
        s.
        '•••-
        o
        c
        o
        u
        £
        I
               Squarnous cell Squamous cell Respiratory Atrophy in the   Hydropic
               metaplasia in hyperplasia in metaplasia in nasal olfactory   change in
               the respiratory the respiratory the olfactory  epithelium,  lamina propria;
                epithelium;   epithelium;  epithelium;   LOAEL    BMDLIO
                 BMDL10    BMDL1D     LOAEL
                            • POD
                           UJ Animal- to-human
                           ^]Human variation
                            LOAEL to NOAEL
                           dSubchromc to Chronic
                           • Database deficiencies
                            oRfC
Sclerosis of  Centrilobular  Spongiosis
the lamina  necrosis in the hepatismthe
 propria;   liver; BMDL10 liver; BMDL10
 NOAEL
             Figure 5-5  Potential points of departure (POD) for candidate endpoints with
                         corresponding applied uncertainty factors and derived sample RfCs
                         following inhalation exposure to 1,4-dioxane.
             Source: Kasai et al. (2009)
      5.2.6  Previous  RfC  Assessment
             An RfC for 1.4-dioxane was not previously available on the IRIS database.
      5.3   Uncertainties in  the Oral  Reference  Dose and  Inhalation
            Reference Concentration

 4           Risk assessments need to portray associated uncertainty. The following discussion identifies
 5    uncertainties associated with the RfD and RfC for 1,4-dioxane. As presented earlier in this section (see
 6    Sections 5.1.2,5.1.3 for the RfD and Sections 5.2.2. and 5.2.3  for the RfC). the uncertainty factor
 7    approach (U.S. EPA. 2002a. 1994) was used to derive the RfD and RfC for 1.4-dioxane. Using this
 8    approach, the POD was divided by a set of factors to account for uncertainties associated with a number
 9    of steps in the analysis^ including extrapolation from LOAEL to NOAEL. extrapolation from animals to
10    humans, a diverse population of varying  susceptibilities, and to account for database deficiencies.
                                                                                                     116
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1    Because information specific to 1.4-dioxane was unavailable to fully inform these extrapolations, default
 2    factors were generally applied.

 3           An adequate range of animal toxicology data are available for the hazard assessment of
 4    1,4-dioxane, as described throughout the previous section (Section 4). The database of oral toxicity
 5    studies includes chronic drinking water studies in rats and mice, multiple subchronic drinking water
 6    studies conducted in rats and mice, and a developmental study in rats. Toxicity associated with oral
 7    exposure to 1,4-dioxane is observed predominately in the liver and kidney. The database of inhalation
 8    toxicity studies in animals includes two subchronic bioassays in rabbits, guinea pigs, mice, and rats, and
 9    two chronic inhalation bioassays in rats. Toxicity associated with inhalation exposure to 1.4-dioxane was
10    observed predominately in the liver and nasal cavity. In addition to oral and inhalation data, there are
11    PBPK models and genotoxicity studies of 1,4-dioxane. Critical data gaps have been identified and
12    uncertainties associated with data deficiencies of 1,4-dioxane are more fully discussed below.

13           Consideration of the available dose-response data led to the selection of the two-year drinking
14    water bioassay in Sherman rats (Kocibaet al.. 1974) as the principal study and increased liver and kidney
15    degeneration as the  critical effects for deriving the RfD for 1,4-dioxane. The dose-response relationship
16    for oral exposure to 1,4-dioxane and cortical tubule degeneration in Osborne-Mendel rats (NCI. 1978)
17    was also suitable for deriving a RfD, but it is associated with higher a POD and potential RfD compared
18    toKocibaetal. (1974).

19           The RfD was derived by applying UFs to a NOAEL  for degenerative liver and kidney effects.
20    The incidence data for the observed effects were not reported in the principal study (Kociba et al.. 1974).
21    precluding modeling of the dose-response. However confidence in the NOAEL can be derived from
22    additional studies (JBRC.  1998: NCI. 1978: Argus etal.. 1973: Argus etal.. 1965) that observed effects
23    on the same organs  at comparable dose levels and by the BMDL generated by modeling of the kidney
24    dose-response data from the  chronic NCI (1978) study.

25           The RfC was derived by applying UFs to a LOAEL for atrophy and respiratory metaplasia of the
26    olfactory epithelium. The incidence data for the observed effects were not amenable to BMP modeling
27    (see Appendix F). The LOAEL  for these effects was less than or equal to the LOAEL or NOAEL for
28    other effects observed in the Kasai et al. (2009) study.

29           Extrapolating from animals to humans embodies further issues and uncertainties. The effect and
30    the magnitude associated with the dose at the POD in rodents are extrapolated to human response.
31    Pharmacokinetic models are useful to examine species differences in pharmacokinetic processing;
32    however, it was determined that dosimetric adjustment using pharmacokinetic modeling to reduce
33    uncertainty following oral exposure to 1,4-dioxane was not supported. Insufficient information was
34    available to quantitatively assess toxicokinetic or toxicodynamic differences between animals and
35    humans, so a 10-fold UF was used to account for uncertainty in extrapolating from laboratory animals to
36    humans in the derivation of the RfD. A DAF was used to account for pharmacokinetic differences
37    between rodents and humans in the derivation of the RfC: however, there was no information to inform
38    pharmacodynamic differences between species, so an UF of 3 was used in derivation of the RfC to
39    account for these uncertainties.
                                                                                                  117
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Heterogeneity among humans is another uncertainty associated with extrapolating doses from
 2    animals to humans. Uncertainty related to human variation needs consideration. In the absence of
 3    1,4-dioxane-specific data on human variation, a factor of 10 was used to account for uncertainty
 4    associated with human variation in the derivation of the RfD and RfC. Human variation may be larger or
 5    smaller; however, 1,4-dioxane-specific data to examine the potential magnitude of over- or
 6    under-estimation are unavailable.

 7           Uncertainties in the assessment of the health hazards of 1,4-dioxane are associated with
 8    deficiencies in reproductive toxicity information. The oral and inhalation databases lack a multigeneration
 9    reproductive toxicity study. A single oral prenatal developmental toxicity study in rats was available for
10    1,4-dioxane (Giavini et al.. 1985). This developmental study indicates that the developing fetus may be a
11    target of toxicity. The database  of inhalation studies also lacks a developmental toxicity study.
      5.4  Cancer Assessment
      5.4.1   Choice of Study/Data - with Rationale and J ustification
      5.4.1.1  Oral Study/Data

12           Three chronic drinking water bioassays provided incidence data for liver tumors in rats and mice,
13    and nasal cavity, peritoneal, and mammary gland tumors in rats only (Kano et al.. 2009; JBRC. 1998;
14    Yamazaki et al.. 1994; NCI. 1978; Kocibaet al., 1974). The dose-response data from each of these studies
15    are summarized in Table 5-7. With the exception of the NCI (1978) study, the incidence of nasal cavity
16    tumors was generally lower than the incidence of liver tumors in exposed rats. The Kano et al. (2009)
17    drinking water study was chosen as the principal study for derivation of an oral cancer slope  factor (CSF)
18    for 1,4-dioxane. This study used three dose groups in addition to controls and characterized the
19    dose-response relationship at lower exposure levels, as compared to the high doses employed in the NCI
20    (1978) bioassay (Table 5-7). The Kociba et al. (1974)  study also used three dose groups and low
21    exposures; however, the study authors only reported the incidence of hepatocellular carcinoma, which
22    may underestimate the combined incidence of rats with adenoma or carcinoma. In addition to increased
23    incidence of liver tumors, chosen as the most sensitive target organ for tumor formation, the Kano et al.
24    (2009) study also noted increased incidence of peritoneal and mammary gland tumors. Nasal cavity
25    tumors were also seen in high-dose male and female rats; however, the incidence of nasal tumors was
26    much lower than the incidence of liver tumors in both rats  and mice.

27           In a personal communication, Dr. Yamazaki (2006) provided that the survival of mice was low in
28    all male groups (31/50, 33/50, 25/50 and 26/50 in control, low-, mid-, and high-dose groups,  respectively)
29    and particularly low in high-dose females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-, and
30    high-dose groups, respectively). These deaths occurred primarily during the second year of the study.

                                                                                                 118
                                    DRAFT - DO NOT  CITE OR QUOTE

-------
1    Survival at 12 months in male mice was 50/50, 48/50, 50/50, and 48/50 in control, low-, mid-, and
2    high-dose groups, respectively. Female mouse survival at 12 months was 50/50, 50/50, 48/50, and 48/50
3    in control, low-, mid-, and high-dose groups, respectively (Yamazaki. 2006). Furthermore, these deaths
4    were primarily tumor related. Liver tumors were listed as the cause of death for 31 of the 45
5    pretermination deaths in high-dose female Crj:BDFl mice (Yamazaki. 2006). Thus, the high mortality
6    rates in the female mice were still considered to be relevant for this analysis.
                                                                                                119
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     Table 5-7   Incidence of liver, nasal cavity, peritoneal, and mammary gland tumors in rats and
                  mice exposed to 1,4-dioxane in drinking water for 2 years (based on survival to
                  12 months)
Study Species/strain/gender

Kociba et al. Sherman rats, male and
(1974) female combined3'13


rats"


Female
M^l 1 1 O7O\

Male B6C3F-I miced


Female B6C3F-I miced


Male F344/DuCrj
ratsd'e'f'g


Female F344/DuCrj
ratsd'e'f'g
Kanoetal. (2009)








Animal dose
(mg/kg-day)
0
14
121
1,307
0
240
530
0
350
640
0
720
830
0
380
860
0
11
55
274
0
18
83
429
0
49
191
677
0
66
278
964
Tumor Incidence
Liver
1/106"
0/110
1/106
10/66'
NA
NA
NA
0/31"
10/30'
11/29'
8/49"
19/50'
28/47'
0/50"
21/48'
35/37'
3/50
4/50
7/50
39/50J'K
3/50
1/50
6/50
48/50J'K
23/50
31/50
37/50'
40/50J'K
5/50
35/50J
41/50J
46/50J'K
Nasal
cavity
0/106"
0/110
0/106
3/66
0/33"
12/26
16/33'
0/34"
10/30'
8/29'
NA
NA
NA
NA
NA
NA
0/50
0/50
0/50
7/50K
0/50
0/50
0/50
8/50J'K
0/50
0/50
0/50
1/50
0/50
0/50
0/50
1/50
Peritoneal
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
2/50
2/50
5/50
28/50J'K
1/50
0/50
0/50
0/50
NA
NA
NA
NA
NA
NA
NA
NA
Mammary
gland
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
1/50
2/50
2/50
6/50K
8/50
8/50
11/50
18/50''K
NA
NA
NA
NA
NA
NA
NA
NA
     "Incidence of hepatocellular carcinoma.
     ""Incidence of nasal squamous cell carcinoma.
     Incidence of hepatocellular adenoma.
     dlncidence of hepatocellular adenoma or carcinoma.
     Incidence (sum) of all nasal tumors including squamous cell carcinoma, sarcoma, rhabdomyosarcoma, and
        esthesioneuroepithelioma.
     'incidence of peritoneal tumors (mesothelioma).
     Incidence of mammary gland tumors (fibroadenoma or adenoma)
     hp < 0.05; positive dose-related trend (Cochran-Armitage or Peto's test).
     'Significantly different from control at p < 0.05 by Fisher's Exact test.
     'Significantly different from control at p < 0.01 by Fisher's Exact test.
     kp < 0.01; positive dose-related trend (Peto's test).
     NA = data were not available for modeling (no significant change from controls)
     5.4.1.2   Inhalation Study/Data

1            Epidemiological studies of populations exposed to 1.4-dioxane are not adequate for
2    dose-response analysis and derivation of an inhalation unit risk (IUR). However, two chronic inhalation
                                                                                                          120
                                        DRAFT - DO NOT CITE OR QUOTE

-------
 1   studies in animals are available and were evaluated for the potential to estimate an IUR (Table 5-8^. The
 2   chronic inhalation study conducted by Torkelson et al. (1974) in rats did not find any treatment-related
 3   tumors; however, only a single exposure concentration was used (111 ppm 1.4-dioxane vapor for
 4   7 hours/day. 5 days/week for 2 years'). A chronic bioassay of 1.4-dioxane by the inhalation route reported
 5   by Kasai et al. (2009) provides data adequate for dose-response modeling and was subsequently chosen as
 6   the study for the derivation of an IUR for 1.4-dioxane. In this bioassay. groups of 50 male F344 rats were
 7   exposed to either 0. 50. 250 or 1.250  ppm 1.4-dioxane. 6 hours/day. 5 days/week, for 2 years
 8   (104-weeks). In male F344 rats. 1.4-dioxane produced a statistically significant increase in incidence
 9   and/or a statistically significant dose-response trend for the following tumor types: hepatomas. nasal
10   squamous cell carcinomas, renal cell  carcinomas, peritoneal mesotheliomas. mammary gland
11   fibroadenomas. Zvmbal gland adenomas, and subcutis fibromas (Kasai et al.. 2009). The incidence of
12   adenomas and carcinomas were combined in this assessment in accordance with EPA's Guidelines on
13   Carcinogen Risk Assessment which notes that etiologically similar tumor types, i.e.. benign and malignant
14   tumors of the same cell type, can be combined due to the possiblity that benign tumors could progress to
15   the malignant form (U.S. EPA. 2005a; McConnell et al., 1986). Consistent with the oral cancer
16   assessment (Appendix D). the incidence of hepatic adenomas and carcinomas (combined) and was used to
17   calculate an IUR in rodents (See Table 5-8^.
                                                                                                 121
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     Table 5-8    Incidence of liver, nasal cavity, kidney, peritoneal, and mammary gland, Zymbal
                  gland, and subcutis tumors in rats exposed to 1,4-dioxane vapors for 2 years.
Study
Torkelson
etal.
(1974)a
Kasai et al.
(2009)b
Species/ Animal
strain/ Exposure
gender (ppm)
Male
Wistar
rats
Female
Wistar
rats

Male
F344 rats

0
111
0
111
0
50
250
1,250
Tumor Incidence
Liverc
0/150
0/206
0/139
0/217
1/50
2/50
4/50
22/50
Nasal
cavityd
0/150
0/206
0/139
0/217
0/50
0/50
1/50
6/50m
Kidney6
0/150'
1/2061
1/1 39J
0/2 17j
0/50
0/50
0/50
4/50
Peritoneal'
NA
NA
NA
NA
2/50
4/50
14/50"
41/50"
Mammary
gland
NA
NA
11/139"
29/2 17k
1/50'
2/50'
3/50'
5/50'
Zymbal
gland9
NA
NA
NA
NA
0/50
0/50
0/50
4/50
Subcutis11
0/150
2/206
0/139
0/217
1/50
4/50
9/50"
5/50
     Incidence reported based on survival to 9 months.
     blncidence reported based on survival to 12 months.
     Incidence of hepatocellular adenoma or carcinoma. For Kasai et al. (2009) incidence data was provided via personal communication
        from Dr. Tatsuya Kasai to Dr. Reeder Sams on 12/23/2008 (2008). Statistics were not reported. Individual incidence rates for
        adenomas and carcinomas are in Table 5-10.
     Incidence of nasal squamous cell carcinoma.
     Incidence of renal cell carcinoma.
     'incidence of peritoneal mesothelioma.
     Incidence of Zymbal gland adenoma.
     Incidence of subcutis fibroma.
     'Incidence of kidney fibroma.
     'Incidence of kidney adenocarcinoma
     Incidence of mammary gland adenoma.
     'incidence of mammary gland fibroadenoma.
     "Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.05).
     "Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p< 0.01).
     NA = data are not available
     5.4.2   Dose-Response Data
     5.4.2.1  Oral Data

1            Table 5-9 summarizes the incidence of hepatocellular adenoma or carcinoma in rats and mice

2    from the Kano et al. (2009) 2-year drinking water study. There were statistically significant increasing

3    trends in tumorigenic response for males and females of both species. The dose-response curve for female

4    mice is steep, with 70% incidence of liver tumors occurring in the low-dose group (66 mg/kg-day).

5    Exposure to 1,4-dioxane increased the incidence of these tumors in a dose-related manner.

6            A significant increase in the incidence of peritoneal mesothelioma was observed in high-dose

7    male rats only (28/50 rats, Table 5-7). The incidence of peritoneal mesothelioma was lower than the

8    observed incidence of hepatocellular adenoma or carcinoma in male rats (Table 5-9); therefore,

9    hepatocellular adenoma or carcinoma data were used to derive an oral CSF for 1,4-dioxane.
                                                                                                      122
                                      DRAFT - DO NOT CITE OR QUOTE

-------
      Table 5-9    Incidence of hepatocellular adenoma or carcinoma in rats and mice exposed to
                  1,4-dioxane in drinking water for 2 years
Species/strain/gender
Male F344/DuCrj rats
Female F344/DuCrj rats
Male Crj:BDF1 mice
Female Crj:BDF1 mice
Animal dose incidence of liver tumors3
(mg/kg-day)
0
11
55
274
0
18
83
429
0
49
191
677
0
66
278
964
3/50
4/50
7/50
39/50c'c
3/50
1/50
6/50
48/50c'c
23/50
31/50
37/50"
40/50C'C
5/50
35/50c
41/50C
46/50c'c
      "Incidence of either hepatocellular adenoma or carcinoma.
      bp < 0.05; positive dose-related trend (Peto's test).
      °Significantly different from control at p < 0.01 by Fisher's Exact test.
      dSignificantly different from control at p < 0.01 by Fisher's Exact test.
      Source: Reprinted with permission of Elsevier, Ltd., Kano et al. (2009).
      5.4.2.2   Inhalation Data

 1           Multi-tumor dose-response modeling was performed for all tumor responses from the Kasai et al.
 2    (2009) bioassay. Kasai et al. (2009) reported tumor incidence data for male F344 rats exposed via
 3    inhalation to 0. 50. 250. or 1.250 ppm 1.4-dioxane for 6 hours/day. Sclavs/week, for 2 years (104-weeks).
 4    Statistically significant dose-response trends for the increase in tumors with increasing dose was observed
 5    for the nasal cavity squamous cell carcinomas, hepatomas. renal cell carcinomas, peritoneal
 6    mesotheliomas. mammary gland fibroadenomas. and Zymbal gland adenomas. Following 250 ppm
 7    1.4-dioxane exposure, statistically elevated tumor incidences were found in two tissue types (peritoneal
 8    mesothelioma and subcutis fibroma) compared to controls. It is important to note, for observations of
 9    subcutis fibroma, the incidence was increased compared to controls at all concentrations but a decrease in
10    incidence, compared to the mid-concentration, was noted at the highest concentration (1.250 ppm).
11    However, a significantly decreased survival rate was noted in this exposure group by the  study authors.
12    Interim sacrifices were not performed. Tumor incidences following 1.250 ppm inhalation exposure to
13    1.4-dioxane were statistically elevated compared to controls in three tissues  (nasal cavity squamous cell
14    carcinoma, hepatomas. and peritoneal mesothelioma). Incidence data for the tumor types reported by
15    Kasai et al. (2009) are summarized in Table 5-10.
                                                                                                   123
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     Table 5-10  Incidence of tumors in F344 male rats exposed to  1,4-dioxane for 104 weeks (6
                 hours/day, 5 days/week)
Tumor Type
Nasal cavity squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma8
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Mammary gland adenoma
Zymbal gland adenoma
Subcutis fibroma
Animal Exposure (ppm)
0
0/50
1/50
0/50
1/50
0/50
2/50
1/50
0/50
0/50
1/50
















0/50
14/50C


0/50
9/50c
1,250
6/50a'°
21/50a'c
2/50
22/50a'c
4/50a
41/50a'c
5/50a
1/50
4/50a

     "Statistically significant trend for increased tumor incidence by Peto's test (p < 0.01).
     "Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.05).
     Tumor incidence significantly elevated compared with that in controls by Fisher's exact test (p < 0.01).
     Statistically significant trend for increased tumor incidence by Peto's test (p S 0.05).
     eProvided via personal communication from Dr. Tatsuya Kasai to Dr. Reeder Sams on 12/23/2008 (2008). Statistics were not
        reported for these data by study authors, so statistical analyses were conducted by EPA.

     Source: Kasai et al. (2009) and Kasai personal communication (2008)
     5.4.3   Dose Adjustments  and Extrapolation Method(s)
     5.4.3.1  Oral

1           Human equivalent doses (HEDs) were calculated from the administered animal doses using a BW
2    scaling factor (BW°75) (U.S. EPA. 201 Ib). This was accomplished using the following equation:
                            HED = animal dose (mg/kg) x
animal BW (kg)
human BW (kg)
4           For all calculations, a human BW of 70 kg was used. HEDs for the principal study (Kano et al..

5    2009) are given in Table 5-11. HEDs were also calculated for supporting studies (NCI. 1978; Kociba et

6    al.. 1974) and are also shown in Table 5-11.
                                                                                                     124
                                      DRAFT - DO NOT CITE OR QUOTE

-------
      Table 5-11   Calculated HEDs for the tumor incidence data used for dose-response modeling
ft
Study Species/strain/gender

Male F344/DuCrj rats


Female F344/DuCrj rats


Male Crj:BDF1 mice


Female Crj:BDF1 mice


Kociba et al. (1974) 0, .
old man ra ^



Mf^l /-I O7O\
NU (ISJ/o)



mimal BW (g)
432a
432a
432a
267a
267a
267a
47.9a
47.9a
47.9a
35.9a
35.9a
35.9a
325°
325°
285C
470°
470°
310°
310°
32°
32°
30°
30°
Animal dose HED
(mg/kg-day) (mg/kg-day)d
11
81
398
18
83
429
49
191
677
66
278
964
14
121
1,307
240
530
350
640
720
830
380
860
3.1
23
112
4.5
21
107
7.9
31
110
10
42
145
3.7
32
330
69
152
90
165
105
121
55
124
      a TWA BWs were determined from BW growth curves provided for each species and gender.
      bTWA BWs were determined from BW curve provided for control animals.
      °BWs of high dose male and female rats were significantly lower than controls throughout the study. TWA represents the mean of
         TWA for male and females (calculated separately from growth curves).
      dHEDs are calculated as HED = (animal dose) x (animal BW/ human BW)025.
      Sources: Kano et al. (2009): Kociba et al. (1974): and NCI (1978).

 1           The U.S. EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) recommend that
 2    the method used to characterize and quantify cancer risk from a chemical is determined by what is known
 3    about the mode of action of the carcinogen and the shape of the cancer dose-response curve. The linear
 4    approach is recommended if the mode of action of carcinogenicity is not understood (U.S. EPA. 2005a).
 5    In the case of 1,4-dioxane, the mode of carcinogenic action for peritoneal, mammary, nasal, and liver
 6    tumors is unknown. Therefore, a linear low-dose extrapolation approach was used to estimate human
 7    carcinogenic risk associated with 1,4-dioxane exposure.

 8           However, several of the external peer review panel members (Appendix A: Summary of External
 9    Peer Review and Public Comments and Disposition) recommended that the mode of action data support
10    the use of a nonlinear extrapolation approach to estimate human carcinogenic risk associated with
11    exposure to  1,4-dioxane and that such an approach should be presented in the Toxicological Review. As
12    discussed in Section 4.5.1, numerous short-term in vitro and a few in vivo tests were nonpositive for
13    1,4-dioxane-induced genotoxicity. Results from two-stage mouse  skin tumor bioassays demonstrated that
14    1,4-dioxane does not initiate mouse skin tumors, but it is a promoter of skin tumors initiated by DMBA
15    (King et al.. 1973). These data suggest that a potential mode of action for 1,4-dioxane-induced tumors
16    may involve proliferation of cells initiated spontaneously,  or by some other agent, to become tumors
                                                                                                   125
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    (Mivagawa et al.. 1999; Uno et al. 1994; Goldsworthy et al.. 1991; Lundberg et al.. 1987; Bull et al.
 2    1986; Stott et al.. 1981; King et al.. 1973). However, key events related to the promotion of tumor
 3    formation by 1,4-dioxane are unknown. Therefore, under the U.S. EPA Guidelines for Carcinogen Risk
 4    Assessment (U.S. EPA. 2005a). EPA concluded that the available information does not establish a
 5    plausible mode of action for 1,4-dioxane and data are insufficient to establish significant biological
 6    support for a nonlinear approach. EPA determined that there are no data available to inform the low-dose
 7    region of the dose response, and thus, a nonlinear approach was not included.

 8           Accordingly, the CSF for 1,4-dioxane was derived via a linear extrapolation from the POD
 9    calculated by curve fitting the experimental dose-response data. The POD is the 95% lower confidence
10    limit on the dose associated with a benchmark response (BMR) near the lower end of the observed data.
11    The BMD modeling analysis used to estimate the POD is described in detail in Appendix D and is
12    summarized below in Section 5.4.4.

13           Model estimates were derived for all available bioassays and tumor endpoints (Appendix D);
14    however, the POD used to derive the CSF is based on the most sensitive species and target organ in the
15    principal study (Kano etal. 2009).

16           The oral CSF was calculated using the  following equation:

             BMR
17    CSF = •
             BMDL
      5.4.3.2  Inhalation

18           In accordance with the U.S. EPA (1994) RfC methodology, the FffiC values were calculated by
19    the application of DAFs. As discussed in Section 5.2.3. since 1.4-dioxane is miscible with water, has a
20    high partition coefficient, and induces effects throughout the body of the rat. a DAF of 1.0 was applied.
21    The lifetime continuous inhalation risk for humans is defined as the slope of the line from the POD, the
22    lower 95% bound on the exposure associated with a level of extra risk near the low end of the data range.

23           All POPs were converted to equivalent continuous exposure levels by multiplying by [(6
24    hours)/(24 hours)] x[(5 days)/(7 days)1. under the assumption of equal cumulative exposures leading to
25    equivalent outcomes.

26           Given the multiplicity of tumor sites, basing the IUR on one tumor site may underestimate the
27    carcinogenic potential of 1.4-dioxane. Also, simply pooling the counts of animals with one or more
28    tumors (i.e.. counts of tumor bearing animals) would tend to underestimate the overall risk for tumors
29    observed at independent sites and ignores potential differences in the dose-response relationships across
30    the sites (NRC. 1994; Bogen.  1990). NRC (1994) also noted that the assumption of independence across
31    tumor types is not likely to produce substantial error in the risk estimates unless tumors are known to be
32    biologically dependent.
                                                                                                 126
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Kopylev et al. (2009) describe a Markov Chain Monte Caro (MCMC) computational approach to
 2    calculating the dose associated with a specified composite risk under assumption of independence of
 3    tumors. The Guidelines for Carcinogen Risk Assessment recommend calculation of an upper bound to
 4    account for uncertainty in the estimate (U.S. EPA. 2005a). For uncertainty characterization. MCMC
 5    methods have the advantage of providing information about the full distribution of risk and/or benchmark
 6    dose, which can be used in generating a confidence bound. This MCMC approach which builds on the
 7    re-sampling approach recommended by Bogen (1990). also provides a distribution of the combined
 8    potency across sites. The Bayesian MCMC computations were conducted using WinBugs (Spiegelhalter
 9    et al., 2003) and additional details of this analysis are included in Appendix G. In addition, the best fitting
10    BMDS multistage model was determined for each individual tumor type as shown in Section 5.4.4.2 and
11    Appendix H.

12           The carcinogenic MOA(s) by which 1.4-dioxane produces liver, nasal, kidney, peritoneal
13    (mesotheliomas). mammary gland. Zymbal gland, and subcutis tumors is unknown. Several hypothesized
14    MOA(s) have been proposed for liver and nasal tumors although these MOA(s) are not supported by the
15    available data (see Sections 4.7.3.3 and 4.7.3.4). Specifically, tumors occur in rodent models in the
16    absence of data to identify hypothesized key events (e.g.. cytotoxicity). Furthermore, studies evaluating
17    the kinetics of 1.4-dioxane suggest that liver carcinogenicity is related to the accumulation of the parent
18    compound following metabolic saturation: however, the in vivo metabolism of 1.4-dioxane is  unknown
19    (Section 3.3). nor are data available to determine the toxic moiety (i.e.. parent compound and/or
20    metabolite(s)) (see Section 4.7.3.1.1  and 3.3.). For kidney, lung, peritoneal (mesotheliomasX mammary
21    gland. Zvmbal gland, and subcutis tumors there are no available data regarding any hypothesized
22    carcinogenic MOA(s) for 1.4-dioxane.
23
24           The EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). recommend that the
25    method used to characterize and quantify cancer risk from a chemical is determined by what is known
26    about the MOA of the carcinogen and the shape of the cancer dose-response curve. The linear
27    extrapolation approach is used as a default option if the mode of carcinogenic action is unknown. A
28    nonlinear extrapolation approach can be used for cases with sufficient data to ascertain the mode of action
29    and to conclude that it is not linear at low doses.  Also, nonlinear extrapolation having significant
30    biological support may be presented in addition to a linear approach when the available data and weight
31    of evidence support a nonlinear approach. In the case of 1.4-dioxane. there is insufficient biological
32    support to identify key events and to have reasonable confidence in the sequence of events and how they
33    relate to the development of tumors following exposure to 1.4-dioxane; thus, the data are not strong
34    enough to ascertain the mode of action applying the Agency's mode of action framework (U.S. EPA.
35    2005a. Therefore. EPA concluded that a default linear extrapolation should be utilized to estimate the
36    cancer risk estimates for inhalation or oral exposure to 1.4-dioxane.

37                            IUR estimates were calculated using the following equation:

38                                                  IUR = BMR / HEC
                                                                                                 127
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     5.4.4  Oral Slope Factor and Inhalation Unit Risk
     5.4.4.1   Oral Slope Factor

 1          The dichotomous models available in the Benchmark Dose Software (BMDS, version 2.1.1) were
 2   fit to the incidence data for "either hepatocellular carcinoma or adenoma" in rats and mice, as well as
 3   mammary and peritoneal tumors in rats exposed to 1,4-dioxane in the drinking water (Kano et al.. 2009;
 4   NCI. 1978; Kociba et al..  1974) (Table 5-7). Animal doses are used for BMD modeling and RED BMD
 5   and BMDL values are calculated using the animal TWAs (Table 5-12) and a human BW of 70kg. Doses
 6   associated with a BMR of 10% extra risk were calculated. BMDs and BMDLs from all models are
 7   reported, and the output and plots corresponding to the best-fitting model are shown (Appendix D). When
 8   the best-fitting model is not a multistage model, the multistage model output and plot are also provided
 9   (Appendix D). A summary of the BMDS model predictions for the Kano et al. (2009). NCI (1978). and
10   Kociba et al. (1974) studies is shown in Table 5-12.
                                                                                             128
                                   DRAFT - DO NOT CITE OR QUOTE

-------
      Table 5-12   BMD HED and BMDLHED values from models fit to tumor incidence data for rats and
                  mice exposed to 1,4-dioxane in drinking water for 2 years and corresponding oral
                  CSFs
Study
Kano et al.
(2009)
Kociba et al.
(1974)
NCI (1978)
Gender/strain/species
Male F344/DuCrj rats'
Female F344/DuCrj ratsc
Male Crj:BDF1 micea
Female Crj:BDF1 micea
Female Crj:BDF1 mice0'8
Female Crj:BDF1 micea'T
Female F344/DuCrj ratsg
Male F344/DuCrj ratsg
Male F344/DuCrj ratsb
Female F344/DuCrj ratsd
Male and female (combined)
Sherman ratsg
Male and female (combined)
Sherman ratsb
Male Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female Osborne Mendel ratsd
Female B6C3F-I micec
Male B6C3Fi miceh
Tumor type BMDHE°a BMDLHEDa Oral CSF
'^ (mg/kg-day) (mg/kg-day) (mg/kg-day)


Hepalocelluldi



Nasal
squamous cell
carcinoma
Peritoneal
mesothelioma
Mammary
gland adenoma
Nasal
squamous cell
carcinomas
Hepatocellular
carcinoma
Nasal
squamous cell
carcinomas
Hepatocellular
adenoma
Hepatocellular
adenoma or
carcinoma
17
19
5
0
3.
7,
94
91
26
40
448
290
16
40
28
23
87
.43
.84
.63
.83
22e
,51T
.84
.97
.09
.01
.24
.78
.10
.07
.75
.12
.98
14
14
2
0
2.
4.
70
68
21
20
340
240
10
25
18
9
35
.33
.43
.68
.55
12e
95T
.23
.85
.39
.35
.99
.31
.66
.82
.68
.75
.67
7.
6.
3.



1.
1.
4.
4.
2.
4.
9.
3.
5.
1.
2.
0
9
,7
0
0
0
,4
5
7
,9
,9
2
,4
,9
,4
,0
8
x 10
x 10
x 10
.18
.14
.10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
x 10
-6
-3
~2



-'3
-3
-3
-3
-4
-4
-'6
-3
-3
-A
-3
      aValues associated with a BMR of 10% unless otherwise noted.
      bProbit model, slope parameter not restricted.
      °Multistage model, degree of polynomial = 2.
      dLog-logistic model, slope restricted > 1.
      eValues associated with a BMR of 30%.
      Values associated with a BMR of 50%.
      9Multistage model, degree of polynomial =3.
      hGamma model.
 1           The multistage model did not provide an adequate fit (as determined by AIC, p-value < 0.1, and
 2    %2 p > 10.11) to the data for the incidence of hepatocellular adenoma or carcinoma in female mice
 3    (Appendix D). The high dose was dropped for the female mouse liver tumor dataset in an attempt to
 4    achieve an adequate fit; however, an adequate fit was still not achieved. Because the female mice were
 5    clearly the most sensitive group tested, other BMD models were applied to the female mouse liver tumor
 6    dataset to achieve an adequate fit. The log-logistic model was the only model that provided adequate fit
 7    for this data set due to the steep rise in the dose-response curve (70% incidence at the low dose) followed
 8    by a plateau at near maximal tumor incidence in the mid- and high-dose regions (82 and 92% incidence,
 9    respectively). The predicted BMD10 and BMDL10 for the female mouse data are presented in Table 5-12,
10    as well as BMDnED and BMDLHED values associated with BMRs of 30 and 50% .

11           The multistage model also did not provide an adequate fit to mammary tumor incidence data for
12    the female rat or male rat peritoneal tumors. The predicted BMD10 and BMDL10 for female rat mammary
                                                                                                  129
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    tumors and male peritoneal tumors obtained from the log-logistic and probit models, respectively, are
 2    presented in Table 5-12.
 3           A comparison of the model estimates derived for rats and mice from the Kano et al. (2009). NCI
 4    (1978). and Kociba et al. (1974) studies (Table 5-12) indicates that female mice are more sensitive to liver
 5    carcinogenicity induced by 1,4-dioxane compared to other species or tumor types. The BMDL50 HED for
 6    the female mouse data was chosen as the POD and the CSF of 0.10 (mg/kg-day)"1 was calculated as
 7    follows:

 8                  CSF=	—	:	= 0.10 (mg/kg-day)-1
                           4.95 mg/kg - day (BMDL50HED for female mice)

 9           Calculation of a CSF for 1,4-dioxane is based upon the dose-response data for the most sensitive
10    species and gender.
      5.4.4.2  Inhalation Unit Risk

11           As stated in Section 5.4.2.2. multiple tumor types have been observed in rats following inhalation
12    exposure to 1.4-dioxane. These data have been used to develop IUR estimates for 1.4-dioxane. The
13    multistage cancer models available in the BMDS (version 2.1.1) were fit to the incidence data for each
14    tumor type observed in rats exposed to 1.4-dioxane via inhalation (Kasai et al.. 2009) to determine the
15    degree (e.g.. 1st. 2nd. or 3rd) of the multistage model that best fit the data (details in Appendix H). A
16    Bayesian MCMC analysis was performed using WinBUGS to calculate the total tumor risk. For
17    comparative purposes only, a total tumor analysis was also performed with the BMDS (version 2.2Beta)
18    MSCombo model and yielded similar results (See Appendix FD. MSCombo is a new addition to BMDS
19    that allows for multi-tumor analysis. A summary of the BMDS model predictions for the Kasai et al.
20    (2009) study is shown in Table  5-13. Experimental exposure concentrations were used for BMP
21    modeling and continuous human equivalent exposures were calculated by adjusting for duration of
22    exposure (Table 5-13) and applying an appropriate DAF (see Section 5.2.3). In accordance with the U.S.
23    EPA Guidelines for Carcinogen Risk Assessment (2005a). the BMCLjn (lower bound on the concentration
24    estimated to produce a 10% increase in tumor incidence over background) was estimated for the
25    dichotomous incidence data and the results of the model that best characterized the cancer incidences
26    were selected. BMCs and BMCLs from all models are reported, and the output and plots corresponding to
27    the best-fitting model are shown (Appendix H).

28           The IUR estimates are provided in Table 5-13. Human equivalent risks estimated from the
29    individual rat tumor sites ranged from 2  x 1Q"7 to 2 x 1Q"6 (^g/m3)"1 (rounded to one significant figure).
30    The highest IUR (2 * 1Q"6 (f^g/m3)"1) corresponded to peritoneal mesotheliomas in male rats, and the
31    lowest IUR (2 x 10"7 (fig/m3)"1) corresponded to renal cell carcinoma and Zymbal gland adenomas in male
32    rats.
                                                                                                130
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      Table 5-13  Dose-response modeling summary results for male rat tumors associated with
inhalation exposure to 1,4-dioxane for

Tumor Type3
Nasal cavity squamous cell
carcinoma
Hepatocellular adenoma or
carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland
fibroadenoma
Zymbal gland adenoma

Bayesian Total Tumor Analysis'

Multistage
Model
Degree13
1
1
3
1
1
3



2 years




Point of Departure0
Bioassay Exposure
Concentration (ppm)
BMCio
1107
252.8
1355
82.21
1635
1355
141.8

BMCLio
629.9
182.3
1016
64.38
703.0
1016
81.91

HEC
(mg/m3)d
Ei
i.


IUR
stimate6
BMCio BMCLio »•»•""''
712.3
162.7
872
52.89
1052
872
91.21

405.3
117.3
653.7
41.42
452.4
653.7


2
8,
1
2
2
1
1

5
5
5
4
2
5


x10-7
x10-7
X10"'
xiQ-B
x10-7
x10"
x1Q-B

      "Tumor incidence data from Kasai et al. (2009).
      ""Best-fitting multistage model degree (p>0.1, lowest AIC). See Appendix G for modeling details.
      CBMC = Concentration at specified extra risk (benchmark dose); BMCL = 95% lower bound on concentration at specified extra risk.
      dHuman continuous equivalent estimated by multiplying exposures by [(6 hours)/(24 hours) x (5 days)/(7 days) x molecular weight of
         1,4-dioxane]/24.45.
      The inhalation unit risk (|jg/m3)-1 was derived from the BMCL10, the 95% lower bound on the concentration associated with a 10%
         extra cancer risk. Specifically, by dividing the BMR (0.10) by the BMCL10. Thus, representing an upper bound, continuous lifetime
         exposure estimate of cancer potency.
      'Results in this Table are from the Bayesian analysis using WinBUGS. Additionally, for comparative purposes only, total tumor analysis
         was performed with the draft BMDS (version 2.2Beta) MSCombo model and yielded similar results (See Appendix H).
 1
 2           The carcinogenic MOA(s) by which 1.4-dioxane produces liver, nasal, kidney, peritoneal
 3    (mesotheliomas). mammary gland. Zymbal gland, and  subcutis tumors is unknown. Several hypothesized
 4    MOA(s) have been proposed for liver and nasal tumors although these MOA(s) are not supported by the
 5    available data (see Sections 4.7.3.3 and 4.7.3.4). Specifically, tumors occur in rodent models in the
 6    absence of data to identify hypothesized key events (e.g.. cytotoxicity).  Furthermore, studies evaluating
 7    the kinetics of 1.4-dioxane suggest that liver carcinogenicity is related to the accumulation of the parent
 8    compound following metabolic saturation: however,  the in vivo metabolism of 1.4-dioxane is unknown
 9    (Section 3.3). nor are  data available to determine the toxic moiety (i.e.. parent compound and/or
10    metabolite^)) (see Section 4.7.3.1.1 and  3.3.). For kidney, lung, peritoneal (mesotheliomasX mammary
11    gland. Zvmbal gland,  and subcutis tumors there are no  available data regarding any hypothesized
12    carcinogenic MOA(s) for 1.4-dioxane.
13
14           The EPA Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). recommend that the
15    method used to characterize and quantify cancer risk from a chemical is determined by what is known
16    about the MOA of the carcinogen and the shape  of the  cancer dose-response curve.  The linear
17    extrapolation approach is used as a default option if the mode of carcinogenic action is unknown. A
18    nonlinear extrapolation approach can be used for cases with sufficient data to ascertain the mode of action
19    and to conclude that it is not linear at low doses. Also, nonlinear extrapolation having significant
20    biological support may be presented in addition to a linear approach when the available data and weight
21    of evidence support a nonlinear approach. In the case of 1.4-dioxane. there is insufficient biological
22    support to identify key events and to have reasonable confidence in the sequence of events and how they

                                                                                                       131
                                       DRAFT - DO NOT CITE OR QUOTE

-------
 1    relate to the development of tumors following exposure to 1.4-dioxane; thus, the data are not strong
 2    enough to ascertain the mode of action applying the Agency's mode of action framework (U.S. EPA.
 3    2005a. Therefore. EPA concluded that a default linear extrapolation should be utilized to estimate the
 4    cancer risk estimates for inhalation or oral exposure to 1.4-dioxane.

 5           Given the multiplicity of tumor sites, basing the inhalation unit risk on one tumor site may
 6    underestimate the carcinogenic potential of 1.4-dioxane. Consistent with recommendations of the NRC
 7    (1994) and the EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a) the total risk and
 8    upper bound risk for all tumor sites in male F344 rats was estimated. This estimate of total risk describes
 9    the risk of developing any combination of the tumor types considered. As shown in Table 5-13. the
10    resulting inhalation unit risk for all tumor types in male F344 rats was 5 x  1Q"6 (f^g/m3)"1 . Consideration of
11    all tumor sites approximately doubled the unit risk compared to the highest unit risk associated with any
12    individual tumor type. 2 x 10"6 (fig/m3)"1 for male peritoneal mesotheliomas.

13           The HEC BMCLui for the combined tumor estimate in male rats was chosen as the POD and the
14    lURof 5 x 1Q-6 (^g/nvV was calculated as follows:

                    IUR (mg/m3)-1 =     °'10,  3 = 0.005 (mg/m3)-1
                                     20.2 mg/m
15                  lURCwg/m3)-1  =0.005 (mg/m3)-1 x  *fg  = 5 xlO'6(^g/m3)-
                                                      10 mg
                    IUR Cwg/m3)-1 = 5
16           Based on the analysis discussed above, the recommended upper bound estimate on human extra
17    cancer risk from continuous lifetime exposure to 1.4-dioxane is 5 x IP"6 (fig/m3)"1. The IUR reflects the
18    exposure-response relationships for the multiple tumor sites in male F344 rats.
      5.4.5  Previous Cancer Assessment

19           A previous cancer assessment was posted for 1,4-dioxane on IRIS in 1988. 1,4-Dioxane was
20    classified as a Group B2 Carcinogen (probable human carcinogen; sufficient evidence from animal
21    studies and inadequate evidence or no data from human epidemiology studies (U.S. EPA. 1986aV) based
22    on the induction of nasal cavity and liver carcinomas in multiple strains of rats, liver carcinomas in mice,
23    and gall bladder carcinomas in guinea pigs. An oral CSF of 0.011 (mg/kg-day)"1 was derived from the
24    tumor incidence data for nasal squamous cell carcinoma in male rats exposed to 1,4-dioxane in drinking
25    water for 2 years (NCI. 1978). The linearized multistage extra risk procedure was used for linear low dose
26    extrapolation. An inhalation unit risk was not previously derived.
                                                                                               132
                                    DRAFT - DO NOT CITE OR QUOTE

-------
      5.5  Uncertainties  in Cancer Risk Values

 1           As in most risk assessments, extrapolation of study data to estimate potential risks to human
 2    populations from exposure to 1,4-dioxane has engendered some uncertainty in the results. Several types
 3    of uncertainty may be considered quantitatively, but other important uncertainties cannot be considered
 4    quantitatively. Thus an overall integrated quantitative uncertainty analysis is not presented. However, the
 5    sources of uncertainty and assumptions are described below and in Table 5-14.
      5.5.1   Sources of Uncertainty
      5.5.1.1  Choice of Low-Dose Extrapolation Approach

 6           The range of possibilities for the low-dose extrapolation of tumor risk for exposure to
 7    1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible MOA(s)
 8    for the observed tumors. The MOA is a key consideration in clarifying how risks should be estimated for
 9    low-dose exposure. Exposure to 1,4-dioxane has been observed in animal models to induce multiple
10    tumor types, including liver adenomas and carcinomas, nasal carcinomas, mammary adenomas and
11    fibroadenomas, and mesotheliomas of the peritoneal cavity (Kano et al.. 2009; Kasai et al.. 2009; JBRC.
12    1998; NCI. 1978; Kociba et al.. 1974). MOA information that is available for the carcinogenicity of
13    1,4-dioxane has largely focused on liver adenomas and carcinomas,  with little or no MOA information
14    available for the remaining tumor types. In Section 4.7.3, hypothesized MOAs were explored for
15    1,4-dioxane. Information that would provide sufficient support for any MOA is not available. In the
16    absence of a MOA(s) for the observed tumor types, a linear low-dose extrapolation approach was used to
17    estimate human carcinogenic risk associated with 1,4-dioxane exposure.

18           It is not possible to predict how additional MOA information would impact the dose-response
19    assessment for 1,4-dioxane because of the variety of tumors observed and the lack of data on how
20    1,4-dioxane or a metabolite thereof, interacts with cells starting the progression to the observed tumors.

21           In general, the Agency has preferred to use the multistage model for analyses of tumor incidence
22    and related endpoints because they have a generic biological motivation based on long-established
23    mathematical models such as the Moolgavkar-Venzon-Knudsen (MVK) model.

24           The MVK model does not necessarily characterize all modes of tumor formation, but it is a
25    starting point for most investigations and, much more often than not, has provided at least an adequate
26    description of tumor incidence data.

27           The multistage cancer model provided adequate fits for the tumor incidence data following a
28    2-year inhalation exposure to 1.4-dioxane by male rats (Kasai et al..  2009). In the studies evaluated for the
29    oral cancer assessment (Kano et al.. 2009; NCI. 1978; Kocibaetal.  1974). the multistage model provided
30    good descriptions of the incidence of a few tumor types in male (nasal cavity) and female (hepatocellular

                                                                                                133
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    and nasal cavity) rats and in male mice (hepatocellular) exposed to 1,4-dioxane (Appendix D for details).
 2    The multistage model did not provide an adequate fit for the female mouse liver tumor dataset based upon
 3    the following (U.S. EPA. 2000a):

 4                      •   Goodness-of-fit/>-value was not greater than 0.10;
 5                      •   Akaike's Information Criterion (AIC) was larger than other acceptable models;
 6                      •   Data deviated from the fitted model, as measured by their %2 residuals (values
 7                         were greater than an absolute value of one).
 8           BMDS software typically implements the guidance in the external peer review draft BMD
 9    technical guidance document (U.S. EPA. 2000a) by imposing constraints on the values of certain
10    parameters of the models. When these constraints were imposed, the multistage model and most other
11    models did not fit the incidence data for female mouse liver adenomas or carcinomas.

12           The log-logistic model was selected because it provides an adequate fit for the female mouse data
13    (Kano et al.. 2009). A BMR of 50% was used because it is proximate to the response at the lowest dose
14    tested and the BMDL50 HED was derived by applying appropriate parameter constraints, consistent with
15    recommended use of BMDS in the BMD technical guidance document (U.S. EPA. 2000a).

16           The human equivalent oral CSFs estimated from tumor datasets with statistically significant
17    increases ranged from 4.2 x 10"4to 0.18 permg/kg-day (Table 5-12), arange of aboutthree orders of
18    magnitude, with the extremes coming from the combined male and female rat data for hepatocellular
19    carcinomas (Kocibaet al., 1974) and the female mouse combined liver adenoma and carcinomas (Kano et
20    al.. 2009).
      5.5.1.2  Dose Metric

21           1,4-Dioxane is known to be metabolized in vivo. However, it is unknown whether a metabolite or
22    the parent compound, or some combination of parent compound and metabolites, is responsible for the
23    observed toxicity. If the actual carcinogenic moiety is proportional to administered exposure, then use of
24    administered exposure as the dose metric is the least biased choice.  On the other hand, if this is not the
25    correct dose metric, then the impact on the CSF is unknown.
      5.5.1.3  Cross-Species  Scaling

26           For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied to
27    address toxicological equivalence of internal doses between each rodent species and humans, consistent
28    with the 2005 Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). It is assumed that equal
29    risks result from equivalent constant lifetime exposures.

30           Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
31    in local respiratory system effects are sources of uncertainty in the inhalation cancer assessment.
                                                                                                134
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    However, since similar cell types are prevalent throughout the respiratory tract of both rats and humans.
 2    the tumors are considered biologically plausible and relevant to humans.
      5.5.1.4  Statistical Uncertainty at the POD
 3           Parameter uncertainty can be assessed through confidence intervals. Each description of
 4    parameter uncertainty assumes that the underlying model and associated assumptions are valid. For the
 5    log-logistic model applied to the female mouse data following oral exposure, there is a reasonably small
 6    degree of uncertainty at the  10% excess incidence level (the POD for linear low-dose extrapolation). For
 7    the multistage model applied for the male rat inhalation dataset. there is a reasonably small degree of
 8    uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
      5.5.1.5  Bioassay Selection

 9           The study by Kano et al. (2009) was used for development of an oral CSF. This was a
10    well-designed study, conducted in both sexes in two species (rats and mice) with a sufficient number
11    (N=50) of animals per dose group. The number of test animals allocated among three dose levels and an
12    untreated control group was adequate, with examination of appropriate toxicological endpoints in both
13    sexes of rats and mice. Alternative bioassays (NCI.  1978; Kocibaetal.. 1974) were available and were
14    fully considered for the derivation of the oral CSF.

15           The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
16    well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
17    Three dose levels plus an untreated control group were examined following exposure to 1.4-dioxane via
18    inhalation for 2 years.
      5.5.1.6  Choice of S pecies/Gender

19           The oral CSF for 1,4-dioxane was quantified using the tumor incidence data for the female
20    mouse, which was shown to be more sensitive than male mice or either sex of rats to the carcinogenicity
21    of 1,4-dioxane. While all data, both species and sexes reported from the Kano et al. (2009) study, were
22    suitable for deriving an oral CSF, the female mouse data represented the most sensitive indicator of
23    carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day or 10 mg/kg-day [HED])
24    resulted in a considerable and significant increase in combined liver adenomas and carcinomas observed.
25    Additional testing of doses within the range of control and the lowest dose (66 mg/kg-day or
26    10 mg/kg-day [HED]) could refine and reduce uncertainty for the oral CSF.

27           A personal communication from Dr. Yamazaki (2006) provided that the survival of mice was
28    particularly low in high-dose females (29/50, 29/50, 17/50, and 5/50 in control, low-, mid-,  and high-dose
29    groups, respectively). These deaths occurred primarily during the second year of the study.  Female mouse
                                                                                                 135
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    survival at 12 months was 50/50, 50/50, 48/50, and 48/50 in control, low-, mid-, and high-dose groups,
 2    respectively (Yamazaki. 2006). Furthermore, these deaths were primarily tumor related. Liver tumors
 3    were listed as the cause of death for 1/21, 2/21, 8/33, and 31/45 of the pretermination deaths in control,
 4    low-, mid- and, high-dose female Crj:BDFl mice (Yamazaki. 2006). Therefore, because a number of the
 5    deaths in female mice were attributed to liver tumors, this endpoint and species was still considered to be
 6    relevant for this analysis; however, the high mortality rate does contribute uncertainty.

 7           Additionally, the incidence of hepatocellular adenomas and carcinomas in historical controls was
 8    evaluated with the data from Kano et al. (2009). Katagiri et al. (1998) summarized the incidence of
 9    hepatocellular adenomas and carcinomas in control male and female BDF1 mice from ten 2-year
10    bioassays at the JBRC. For female mice, out of 499 control mice, the incidence rates were 4.4% for
11    hepatocellular adenomas and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported a 10%
12    incidence rate for hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in
13    control female BDF1. These incidence rates are near the historical control values and thus are appropriate
14    for consideration in this assessment.

15           Male F344 rat data were used to estimate risk following inhalation of 1,4-dioxane. Kasai et al.
16    (2008) showed that male rats were more sensitive than female  rats to the effects of 1,4-dioxane following
17    inhalation; therefore, male rats were chosen to be studies in the 2-year bioassay conducted by the same
18    laboratory (Kasai et al.. 2009).
      5.5.1.7   Relevance to Humans

19           The derivation of the oral CSF is derived using the tumor incidence in the liver of female mice. A
20    thorough review of the available toxicological data available for 1,4-dioxane provides no scientific
21    justification to propose that the liver adenomas and carcinomas observed in animal models due to
22    exposure to 1,4-dioxane are not relevant to humans. As such, liver adenomas and carcinomas were
23    considered relevant to humans due to exposure to  1,4-dioxane.

24           The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
25    rats. There is no information on 1,4-dioxane to indicate that the observed rodent tumors are not relevant to
26    humans. Further, no data exist to  guide quantitative adjustment for differences in sensitivity among
27    rodents and humans. In the absence of information to indicate otherwise and considering similar cell types
28    are prevalent throughout the respiratory tract of rats and humans, the nasal liver, renal peritoneal
29    mammary gland. Zymbal gland and subcutis tumors were considered relevant to humans.
      5.5.1.8   Human Population Variability

30           The extent of inter-individual variability in 1,4-dioxane metabolism has not been characterized. A
31    separate issue is that the human variability in response to 1,4-dioxane is also unknown. Data exploring
32    whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life stages are unavailable.

                                                                                                  136
                                     DRAFT - DO NOT CITE OR QUOTE

-------
1    This lack of understanding about potential differences in metabolism and susceptibility across exposed
2    human populations thus represents a source of uncertainty. Also, the lack of information linking a MOA
3    for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.

     Table 5-14  Summary of uncertainty in the 1,4-dioxane cancer risk estimation
Consideration/
approach
Low-dose
extrapolation
procedure
Dose metric
Cross-species
scaling
Bioassay
Species /gender
combination
Human
relevance of
mouse tumor
data
Human
population
variability in
metabolism and
response/
sensitive
subpopulations
Potential Impact

Departure from
EPA's Guidelines for
Carcinogen Risk
Assessment POD
paradigm, if justified,
could | or t unit risk
an unknown extent
Alternatives could f
or | CSF by an
unknown extent
Alternatives could J,
orf CSF [e.g.,
3.5-fold I (scaling by
BW) or } twofold
(scaling by BW067 )]
Alternatives could f
or 1 cancer potency
by an unknown
extent
Human risk could J, or
f, depending on
relative sensitivity
If rodent tumors
proved not to be
relevant to humans,
unit risk would not
apply i.e., could J,
CSF
Low-dose risk f or J,
to an unknown extent
Decision
Log-logistic model
to determine POD,
for CSF; Bavesian
multistage modeling
for IUR: linear
low-dose
extrapolation from
POD
Used administered
exposure
BW075 (default
approach)
CSF (Kano et al..
2009): IUR (Kasai
et al., 2009)
Female mouse
Mouse liver
adenomas and
carcinomas are
relevant to humans
(basis for CSF). Rat
tumors at multiple
sites are relevant to
humans (basis for
IUR)
Considered
qualitatively
Justification
A linear low-dose extrapolation approach was used to
estimate human carcinogenic risk associated with
1,4-dioxane exposure. Where data are insufficient to
ascertain the MOA, EPA's 2005 Guidelines for
Carcinogen Risk Assessment recommend application
of a linear low-dose extrapolation approach.
Experimental evidence supports a role for metabolism
in toxicity, but it is unclear if the parent compound,
metabolite or both contribute to 1,4-dioxane toxicity.
There are no data to support alternatives. BWU 'b
scaling was used to calculate equivalent cumulative
exposures for estimating equivalent human risks.
PBPK modeling was conducted but not deemed
suitable for interspecies extrapolation.
Alternative bioassays were available and considered
for derivation of oral CSF and inhalation IUR.
There are no MOA data to guide extrapolation
approach for any choice. It was assumed that
humans are as sensitive as the most sensitive rodent
gender/species tested; true correspondence is
unknown. Calculation of the CSF for 1,4-dioxane was
based on dose-response data from the most sensitive
species and gender. The carcinogenic response
occurs across species.
1,4-dioxane is a multi-site carcinogen in rodents and
the MOA(s) is unknown; carcinogenicity observed in
the rodent studies is considered relevant to human
exposure.
No data to support range of human
variability/sensitivity, including whether children are
more sensitive.
                                                                                                137
                                    DRAFT - DO NOT CITE OR QUOTE

-------
      6   MAJOR  CONCLUSIONS   IN  THE
          CHARACTERIZATION  OF  HAZARD  AND  DOSE
          RES PONS E


      6.1   Human Hazard Potential

 1           1,4-Dioxane is absorbed rapidly following oral and inhalation exposure, with much less
 2    absorption occurring from the dermal route. 1,4-Dioxane is primarily metabolized to HEAA, which is
 3    excreted in the urine.  Liver, kidneyi and nasal toxicity are the primary noncancer health effects
 4    associated with exposure to 1,4-dioxane in humans and laboratory animals. Several fatal cases of
 5    hemorrhagic nephritis and centrilobular necrosis of the liver were related to occupational exposure  (i.e.,
 6    inhalation and dermal contact) to 1,4-dioxane (Johnstone. 1959; Barber. 1934). Neurological changes
 7    were also reported in one case, including headache, elevation in blood pressure, agitation and restlessness,
 8    and coma (Johnstone. 1959). Perivascular widening was observed in the brain of this worker, with small
 9    foci of demyelination in several regions (e.g., cortex, basal nuclei).  Severe liver and kidney degeneration
10    and necrosis were observed frequently in acute oral and inhalation studies (> 1,000 mg/kg-day oral, >
11    1,000 ppm inhalation) (JBRC. 1998: Drewetal.. 1978: David. 1964: Kestenetal.. 1939: Laug et al..
12    1939:  Schrenk and Yant. 1936: deNavasquez. 1935: Fairlev et al.. 1934).

13           Liver and kidney toxicity were the primary noncancer health effects of subchronic and chronic
14    oral exposure to 1,4-dioxane in animals. Hepatocellular degeneration and necrosis were observed
15    (Kociba et al.. 1974) and preneoplastic changes were noted in the liver following chronic administration
16    of 1,4-dioxane in drinking water (Kano et al.. 2008: JBRC. 1998: Argus et al.. 1973) Liver and kidney
17    toxicity appear to be related to saturation of clearance pathways and an increase in the 1,4-dioxane
18    concentration in the blood (Kociba. etal..  1974). Kidney damage was characterized by degeneration of
19    the cortical tubule cells, necrosis with hemorrhage, and glomerulonephritis (Argus, etal.. 1965: Argus, et
20    al.. 1973: Fairlev. etal.. 1934: Kociba. et al.. 1974: NCI. 1978). In chronic inhalation studies conducted in
21    rats, nasal and liver toxicity were the primary noncancer health effects. Degeneration of nasal tissue (i.e.
22    metaplasia, hyperplasia. atrophy, hydropic change, and vacuolic change) and preneoplastic cell
23    proliferation were observed in the nasal cavity following inhalation exposure to 1.4-dioxane for 2 years
24    (Kasai. et al.. 2009). Liver toxicity was described as necrosis of the centrilobular region and preneoplastic
25    changes were noted as well.

26           Several carcinogenicity bioassays have been conducted for 1,4-dioxane in mice, rats, and guinea
27    pigs (Argus, et al.. 1965: Argus, etal.. 1973: Hoch-Ligeti & Argus.  1970: Hoch-Ligeti. et al.. 1970:
28    JBRC. 1998: Kano. et al.. 2009: Kasai. et al.. 2009: Kociba. et al.. 1974: NCI. 1978: Torkelson. et al..
29    1974). Liver tumors (hepatocellular adenomas and carcinomas) have been observed following drinking
30    water exposure in several species and strains of rats, mice, and guinea pigs and following inhalation
31    exposure in rats. Nasal (squamous cell carcinomas), peritoneal, mammary. Zvmbal gland, and
32    subcutaneous tumors were also observed in rats, but were not seen in mice. With the exception of the NCI
                                                                                                  138
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    (1978) study, the incidence of nasal cavity tumors was generally lower than that of tumors observed in
 2    other tissues of the same study population.

 3           Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is "likely
 4    to be carcinogenic to humans" based on evidence of multiple tissue carcinogenicity in several 2-year
 5    bioassays conducted in three strains of rats, two strains of mice, and in guinea pigs (Argus, etal.. 1965;
 6    Argus, etal.. 1973; Hoch-Ligeti & Argus. 1970; Hoch-Ligeti. et al.. 1970; JBRC. 1998; Kano. et al..
 7    2009; Kasai. et al.. 2009; Kociba. et al..  1974; NCI. 1978).  Studies in humans found no conclusive
 8    evidence for a causal link between occupational exposure to 1,4-dioxane and increased risk for cancer;
 9    however, only two studies were available and these were limited by small cohort size and a small number
10    of reported cancer cases (Buffler. etal..  1978; Thiess. etal.. 1976).

11           The available evidence is inadequate to establish a MOA by which 1,4-dioxane induces tumors in
12    rats and mice. The genotoxicity data for 1,4-dioxane is generally characterized as negative, although
13    several studies may suggest the possibility of genotoxic effects (Galloway, et al.. 1987; Kitchin & Brown.
14    1990; Mirkova. 1994; Morita & Hayashi. 1998; Roy, et al.. 2005).  A MOA hypothesis for liver tumors
15    involving sustained proliferation of spontaneously transformed liver cells has some support by evidence
16    that suggests 1,4-dioxane is a tumor promoter in mouse skin and rat liver bioassays (King, etal.. 1973;
17    Lundberg. et al.. 1987). Some dose-response and temporal evidence support the occurrence of cell
18    proliferation and hyperplasia prior to the development of liver tumors  (JBRC. 1998; Kociba.  et al.. 1974).
19    However, the dose-response relationship for the induction of hepatic cell proliferation has not been
20    characterized, and it is unknown if it would reflect the dose-response relationship for liver tumors in the
21    2-year rat and mouse studies.  Conflicting data from rat and mouse bioassays (JBRC. 1998; Kociba. et al..
22    1974) suggest that cytotoxicity is not a required precursor event for 1,4-dioxane-induced cell
23    proliferation. Liver tumors were observed in female rats and female mice in the absence of lesions
24    indicative of cytotoxicity (JBRC. 1998;  Kano. et al.. 2009;  NCI.  1978). Data regarding a plausible dose
25    response and temporal progression from cytotoxicity to cell proliferation and eventual liver tumor
26    formation are not available. Hypothesized MOAs by which 1.4-dioxane induces tumors in other organ
27    systems such as the respiratory system are uncertain (See Section 4.7.3).

      6.2    DOSE  RESPONSE

      6.2.1  Noncancer/Oral

28           The RfD of 3 x 10"2 mg/kg-day  was derived based  on liver and kidney toxicity in rats exposed to
29    1,4-dioxane in the drinking water for 2 years (Kociba. et al.. 1974).  This study was chosen as the
30    principal study because it provides the most sensitive measure of adverse effects by 1,4-dioxane. The
31    incidence of liver and kidney lesions was not reported for each dose group. Therefore, BMD modeling
32    could not be used to derive a POD. Instead, the RfD is derived by dividing the  NOAEL of 9.6 mg/kg-day
33    by a composite UF of 300  (factors of 10 for animal-to-human extrapolation and interindividual
34    variability, and an UF of 3 for database  deficiencies).  Information was unavailable to quantitatively
35    assess toxicokinetic or toxicodynamic differences between animals and humans and the potential
36    variability in human susceptibility; thus, the interspecies and intraspecies uncertainty factors  of 10 were

                                                                                                 139
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    applied. In addition, a threefold database uncertainty factor was applied due to the lack of information
 2    addressing the potential reproductive toxicity associated with 1,4-dioxane.

 3           The overall confidence in the RfD is medium.  Confidence in the principal study (Kociba. et al..
 4    1974) is medium. Confidence in the database is medium due to the lack of a multigeneration reproductive
 5    toxicity study. Reflecting medium confidence in the principal study and medium confidence in the
 6    database, confidence in the RfD is medium.

      6.2.2  Noncancer/lnhalation
 7           The RfC of 3 x 1Q"2 mg/m3 was derived based on co-critical effects of olfactory epithelium
 8    atrophy and respiratory metaplasia in rats exposed for 2 years to 1.4-dioxane via inhalation (Kasai. et al..
 9    2009). This study was chosen as the principal study because it provides an adequate study design and the
10    most sensitive measure of adverse effects by 1.4-dioxane. The POD was derived using the LOAEL for
11    olfactory epithelium atrophy and respiratory metaplasia in male rats ( Kasai et al. 2009). A composite UF
12    of 1.000 was applied, consisting of factors of 10 for a LOAEL-to NOAEL extrapolation. 10 for
13    interindividual variability. 3 for animal-to-human extrapolation, and 3 for database deficiencies.

14           The overall confidence in the RfC is medium. Confidence in the principal study (Kasai. et al..
15    2009) is medium. Confidence in the database is medium due to the lack of supporting studies and a
16    multigeneration reproductive toxicity study. Reflecting medium confidence in the principal study and
17    medium confidence in the database, the confidence in the RfC is medium.

      6.2.3  Cancer
18           Under EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-dioxane is
19    "likely to be carcinogenic to humans" by all routes of exposure. This descriptor is based on evidence of
20    carcinogenicity from animal studies.

      6.2.3.1 Oral
21           An oral CSF for 1,4-dioxane of 0.10 (mg/kg-day)"1 was based on liver tumors in female mice
22    from a chronic study (Kano. et al.. 2009).  The available data indicate that the MOA(s) by which
23    1,4-dioxane induces peritoneal, mammary, or nasal tumors in rats and liver tumors in rats and mice is
24    unknown (see Section 4.7.3 for a more detailed discussion of 1,4-dioxane's hypothesized MOAs).
25    Therefore, based on the U.S.  EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a). a
26    linear low dose extrapolation was used.  The POD was calculated by curve fitting the animal experimental
27    dose-response data from the range of observation and converting it to a HED  (BMDL50 HED of
28    4.95 mg/kg-day).

29           The uncertainties associated with the quantitation of the oral CSF are discussed below.

      6.2.3.2 Inhalation
30           The IUR for 1.4-dioxane of 5 x 10"6 (fig/m3)"1 was based on a chronic inhalation study conducted
31    by Kasai et al. (2009). Statistically significant increases in tumor incidence and positive dose-response

                                                                                                 140
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    trends were observed at multiple sites in the male rat including the nasal cavity (squamous cell
 2    carcinoma), liver (adenoma), peritoneal (mesothelioma). and the subcutis (fibroma). Statistically
 3    significant dose-response trends were also observed in the kidney (carcinoma), mammary gland
 4    (fibroadenoma). and the Zymbal gland (adenoma). The available data indicate that the MOA(s) by which
 5    1.4-dioxane induces tumors in rats is unknown (see Section 4.7.3 for a more detailed discussion of
 6    1.4-dioxane's hypothesized MOAs). Therefore, based on the EPA's Guidelines for Carcinogen Risk
 1    Assessment (U.S. EPA. 2005a). a linear low dose extrapolation was used. A Bayesian approach (see
 8    Section 5.4.3.2 and Appendix G for details) was used to calculate the POD for the total tumor risk
 9    following inhalation of 1.4-dioxane. The POD was calculated by curve fitting the animal experimental
10    dose-response data from the range of observation and converting it to a continuous human equivalent
11    exposure.

12           The uncertainties associated with the quantitation of the IUR are discussed below.

      6.2.3.3 Choice of Low-Dose Extrapolation Approach
13           The range  of possibilities for the low-dose extrapolation of tumor risk for exposure to
14    1,4-dioxane, or any chemical, ranges from linear to nonlinear, but is dependent upon a plausible MOA(s)
15    for the observed tumors.  The MOA is a key consideration in clarifying how risks should be estimated for
16    low-dose exposure. Exposure to  1,4-dioxane has been observed in animal models to induce multiple
17    tumor types, including liver adenomas and carcinomas, nasal carcinomas, mammary adenomas and
18    fibroadenomas, and mesotheliomas of the peritoneal cavity (Kano. et al., 2009). MOA information that is
19    available for the carcinogenicity of 1,4-dioxane has largely focused on liver adenomas and carcinomas,
20    with little or no MOA information available for the remaining tumor types. In Section 4.7.3,
21    hypothesized MOAs were explored for 1,4-dioxane.  Data are not available to support a carcinogenic
22    MOA for 1,4-dioxane.  In the absence of a MOA(s) for the observed tumor types associated with
23    exposure to 1,4-dioxane, a linear low-dose extrapolation approach was used to estimate  human
24    carcinogenic  risk associated with  1,4-dioxane exposure.

25           In general, the Agency has preferred to use the multistage model for analyses of tumor incidence
26    and related endpoints because they have a generic biological motivation based on long-established
27    mathematical models such as the MVK model. The MVK model does not necessarily characterize all
28    modes of tumor formation, but it is a starting point for most investigations and, much more often than not,
29    has provided  at least an adequate  description of tumor incidence data.

30           The multistage cancer model provided adequate fits for the tumor incidence data following a 2-
31    year inhalation exposure to 1.4-dioxane by male rats (Kasai. et al.. 2009). However,  in the studies
32    evaluated for the oral cancer assessment (Kano. et al., 2009; Kociba. et al., 1974; NCI. 1978) the
33    multistage model provided good descriptions of the incidence of a few tumor types in male (nasal cavity)
34    and female (hepatocellular and nasal cavity) rats and in male mice (hepatocellular) exposed to
35    1,4-dioxane (see Appendix D for  details).  However, the multistage model did not provide an adequate fit
36    for female mouse liver tumor dataset based upon the following (U.S. EPA.  2000a):

         •   Goodness-of-fit />-value was not greater than 0.10;

                                                                                                 141
                                     DRAFT - DO NOT CITE OR QUOTE

-------
         •   AIC was larger than other acceptable models;
         •   Data deviated from the fitted model, as measured by their %2 residuals (values were greater than an absolute
             value of one).
 1           BMDS software typically implements the guidance in the BMD technical guidance document
 2    (U.S. EPA. 2000a) by imposing constraints on the values of certain parameters of the models. When
 3    these constraints were imposed, the multistage model and most other models did not fit the incidence data
 4    for female mouse liver adenomas or carcinomas.

 5           The log-logistic model was selected because it provides an adequate fit for the female mouse data
 6    (Kano. et al.. 2009). A BMR of 50% was used because it is proximate to the response at the lowest dose
 7    tested and the BMDL50 was derived by applying appropriate parameter constraints, consistent with
 8    recommended use of BMDS in the BMD technical guidance document (U.S. EPA. 2000a).

 9           The human equivalent oral CSF estimated from liver tumor datasets with statistically significant
10    increases ranged from 4.2 x 10"4to 1.0 x  10"1 per mg/kg-day, a range of about three orders of magnitude,
11    with the extremes coming from the combined male and female data for hepatocellular carcinomas
12    (Kociba. et al..  1974) and the female mouse liver adenoma and carcinoma dataset (Kano. et al.. 2009).

      6.2.3.4 Dose  Metric
13           1,4-Dioxane is known to be metabolized in vivo. However, evidence does not exist to determine
14    whether the parent compound, metabolite(s), or a combination of the parent compound and metabolites is
15    responsible for the observed toxicity following exposure to 1,4-dioxane.  If the actual carcinogenic moiety
16    is proportional to administered exposure, then use of administered exposure as the dose metric is the least
17    biased choice.  On the other hand, if this is not the correct dose metric, then the impact on the CSF is
18    unknown.

      6.2.3.5 Cross-Species  Scaling
19           For the oral cancer assessment, an adjustment for cross-species scaling (BW°75) was applied to
20    address toxicological equivalence of internal doses between each rodent species and humans, consistent
21    with the Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a).  It is assumed that equal risks
22    result from equivalent constant lifetime exposures.

23           Differences in the anatomy of the upper respiratory tract and resulting differences in absorption or
24    in local respiratory system effects are  sources of uncertainty in the inhalation cancer assessment.

      6.2.3.6 Statistical Uncertainty at the POD
25           Parameter uncertainty can be assessed through confidence intervals. Each description of
26    parameter uncertainty assumes that the underlying model and associated assumptions are valid.  For the
27    log-logistic model applied to the female mouse data following oral exposure, there is a reasonably small
28    degree of uncertainty at the  50% excess incidence level (the POD for linear low-dose extrapolation).  For
29    the multistage model applied for the male rat inhalation dataset. there is a reasonably small degree of
30    uncertainty at the 10% extra risk level (the POD for linear low-dose extrapolation).
                                                                                                 142
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      6.2.3.7 Bioassay Selection
 1           The study by Kano et al. (2009) was used for development of an oral CSF.  This was a well-
 2    designed study, conducted in both sexes in two species (rats and mice) with a sufficient number (N=50)
 3    of animals per dose group. The number of test animals allocated among three dose levels and an
 4    untreated control group was adequate, with examination of appropriate toxicological endpoints in both
 5    sexes of rats and mice.  Alternative bioassays (Kociba. et al.. 1974; NCI. 1978) were available and were
 6    fully considered for the derivation of the oral CSF.

 7           The study by Kasai et al. (2009) was used for derivation of an inhalation unit risk. This was a
 8    well-designed study, conducted in male rats with a sufficient number (N=50) of animals per dose group.
 9    Three dose levels plus an untreated control group were examined following exposure to 1.4-dioxane via
10    inhalation for 2 years.

      6.2.3.8 Choice of S pecies/Gender
11           The oral CSF for 1,4-dioxane was derived using the tumor incidence data for the female mouse,
12    which was thought to be more sensitive than male mice or either sex of rats to the carcinogenicity of
13    1,4-dioxane.  While all data, from both species and sexes reported from the Kano et al. (2009) study, were
14    suitable for deriving an oral CSF, the female mouse data represented the most sensitive indicator of
15    carcinogenicity in the rodent model. The lowest exposure level (66 mg/kg-day [animal dose] or
16    10 mg/kg-day [HED]) observed a considerable and significant increase in combined liver adenomas and
17    carcinomas. Additional testing of doses within the range of control and the lowest dose (66 mg/kg-day
18    [animal dose] or 10 mg/kg-day [FiED]) could refine and reduce uncertainty for the oral CSF.

19           Male F344 rat data were used to estimate risk following inhalation of 1.4-dioxane. Kasai et al.
20    (2008) showed that male rats were more sensitive than female rats to the effects of 1.4-dioxane following
21    inhalation; therefore, male rats were studied in the 2-year bioassay conducted by the same laboratory
22    (Kasai. et al.. 2009).

      6.2.3.9 Relevance to Humans
23           The oral CSF was  derived using the tumor incidence in the liver of female mice.  A thorough
24    review of the available toxicological data available for 1,4-dioxane provides no scientific justification to
25    propose that the liver adenomas and carcinomas observed in animal models following exposure to
26    1,4-dioxane are not plausible in humans.  Liver adenomas and carcinomas were  considered plausible
27    outcomes in humans due to exposure to 1,4-dioxane.

28           The derivation of the inhalation unit risk is based on the tumor incidence at multiple sites in male
29    rats. There is no information on 1.4-dioxane to indicate that the observed rodent tumors are not relevant to
30    humans. Further, no data exist to guide quantitative adjustment for differences in sensitivity among
31    rodents and humans.
                                                                                                 143
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     6.2.3.10       Human  Population Variability
1           The extent of inter-individual variability in 1,4-dioxane metabolism has not been characterized.
2    A separate issue is that the human variability in response to 1,4-dioxane is also unknown.  Data exploring
3    whether there is differential sensitivity to 1,4-dioxane carcinogenicity across life stages is unavailable.
4    This lack of understanding about potential differences in metabolism and susceptibility across exposed
5    human populations thus represents a source of uncertainty. Also, the lack of information linking a MOA
6    for 1,4-dioxane to the observed carcinogenicity is a source of uncertainty.
                                                                                                  144
                                    DRAFT - DO NOT CITE OR QUOTE

-------
REFERENCES
Agrawal. A and Shapiro. B. (2000). Differential expression of gender-dependent hepatic isoforms of cvtochrome P-
        450 by pulse signals in the circulating masculine episodic growth hormone profile of the rat. J Pharmacol
        Exp Ther 292: 228-237.
Amendments to the Clean Air Act. Sec. 604. Phase-out of production and consumption of class I substances. 42.
        U.S.C. §76710(1990).
Andersen. M: Clewell. H. Ill: Gargas. M: Smith. F: Reitz. R. (1987). Physiologically based pharmacokinetics and
        the risk assessment process for methylene chloride. Toxicol Appl Pharmacol 87: 185-205.
        http://dx.doi.org/10.1016/0041-008X(87)90281-X.
Argus. M: Arcos. J: Hoch-Ligeti. C. (1965). Studies on the carcinogenic activity of protein-denaturing agents:
        Hepatocarcinogenicity of dioxane. J Natl Cancer Inst 35: 949-958.
Argus. M: Sohal. R: Bryant. G: Hoch-Ligeti. C: Arcos. J. (1973). Dose-response and ultrastructural alterations in
        dioxane carcinogenesis. Influence of methylcholanthrene on acute toxicitv. Eur J Cancer 9: 237-243.
        http://dx.doi.org/10.1016/0014-2964(73)90088-1.
Ashbv. J. (1994). The genotoxicitv of 1.4-dioxane. MutatRes 322: 141-142. http://dx.doi.org/10.1016/0165-
        1218(94)00022-0.
Atkinson. R. (1989). Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic
        compounds. JPhys ChemRef Data 1: 1-246.
ATSDR. (Agency for Toxic Substances and Disease Registry). (2007). Toxicological profile for 1.4 dioxane. Draft
        for public comment.  Atlanta. GA. http://www.atsdr.cdc.gov/toxprofiles/tpl87.pdf.
Bannasch. P: Moore. M: Klimek. F: Zerban. H. (1982). Biological markers of preneoplastic foci and neoplastic
        nodules in rodent liver. Toxicol Pathol 10: 19-34. http://dx.doi.org/10.1177/019262338201000204.
Bannasch. P. (2003). Comments on R. Karbe and R.L. Kerlin (2002) Cystic degeneration/spongiosis hepatis
        (Toxicol Pathol 30 (2). 216-227). Toxicol Pathol 31: 566-570.
        http://dx.doi.org/10.1080/01926230390224700.
Barber. H. (1934). Haemorrhagic nephritis and necrosis of the liver from dioxan poisoning. Guv's Hosp Rep 84:
        267-280.
Bogen. K. (1990). Uncertainty in environmental health risk assessment. New York: Garlend Publishing Company.
Boorman.  G: Morgan. K: Uriah. L. (1990). Nose, larynx and trachea Pathology of the Fischer Rat Reference and
        Atlas (pp. 315-337).  San Diego. CA: Academic Press. San Diego.
Braun. W and Young. J. (1977). Identification of beta-hydroxyethoxyacetic acid as the major urinary metabolite of
        1.4-dioxane in the rat. Toxicol Appl Pharmacol 39: 33-38. http://dx.doi.org/10.1016/0041-008X(77)90174-
        
-------
Derosa. C: Wilbur. S: Holler. J: Richter. P: Stevens. Y. (1996). Health evaluation of 1.4-dioxane. Toxicol Ind Health
        12: 1-43. http://dx.doi.org/10.1177/074823379601200101.
Drew. R: Patel J: Lin. F. (1978). Changes in serum enzymes in rats after inhalation of organic solvents singly and in
        combination. Toxicol Appl Pharmacol 45: 809-819. http://dx.doi.org/10.1016/0041-008X(78')90172-2.
Ernstgard. L: Iregren. A: Sjogren. B: Johanson. G. (2006). Acute effects of exposure to vapours of dioxane in
        humans. HumExp Toxicol 25: 723-729. http://dx.doi.org/10.1177/0960327106073805.
Fairlev. A: Linton. E: Ford-Moore. A. (1934). The toxicitv to animals of 1:4 dioxan. J Hyg 34: 486-501.
        http://dx.doi.org/10.1017/S0022172400043266.
Fisher. J: Mahle. D: Bankston. L: Greene. R: Gearhart J. (1997). Lactational transfer of volatile chemicals in breast
        milk. AmlndHyg Assoc J 58: 425-431. http://dx.doi.org/10.1080/15428119791012667.
Franke. C: Studinger. G: Berger. G: Bohling. S: Bruckmann. U: Cohors-Fresenborg. D: Johncke. U. (1994). The
        assessment of bioaccumulation. Chemosphere 29: 1501-1514. http://dx.doi.org/10.1016/0045-
        6535(94)90281-X.
Frantik. E: Hornychova. M: Horvath. M. (1994). Relative acute neurotoxicitv of solvents: Isoeffective air
        concentrations of 48 compounds evaluated in rats and mice. Environ Res 66: 173-185.
        http://dx.doi.org/10.1006/enrs. 1994.1053.
Galloway. S: Armstrong. M: Reuben. C:  Colman. S: Brown. B: Cannon.  C: Bloom. A: Nakamura. F: Ahmed. M:
        Duk. S: Rimpo. J: Margolin. B: Resnick. M:  Anderson. B: Zeiger. E. (1987). Chromosome aberrations and
        sister chromatid exchanges in Chinese hamster ovary cells: Evaluations of 108 chemicals. Environ Mol
        Mutaeen 10: 1-175. http://dx.doi.org/10.1002/em.2850100502.
Gaskell B. (1990). Nonneoplastic changes in the olfactory epithelium- experimental studies. Environ Health
        Perspect 85: 275-289.
Giavinj E: Vismara. C: Broccia. M. (1985). Teratogenesis study of dioxane in rats. Toxicol Lett 26: 85-88.
        http://dx.doi.org/10.1016/0378-4274(85')90189-4.
Goldberg. M: Johnson. H: Pozzani. U:  Smyth. HJ. (1964). Effect of repeated inhalation of vapors of industrial
        solvents on animal behavior: I. Evaluation of nine solvent vapors on pole-climb performance in rats. Am
        Ind Hyg Assoc J 25: 369-375.
Goldsworthy. T: Monticello. T: Morgan. K: Bermudez. E: Wilson. D: Jackh. R: BE. B. (1991). Examination of
        potential mechanisms of carcinogenicitv of 1.4-dioxane in rat nasal epithelial cells and hepatocvtes. Arch
        Toxicol 65: 1-9. http://dx.doi.org/10.1007/BF01973495.
Green. T: Lee. R: Moore. R: Ashbv. J:  Willis. G: Lund. V: MJL. C. (2000). Acetochlor-induced rat nasal tumors:
        Further studies on the mode of action and relevance to humans. Regul Toxicol Pharmacol 32: 127-133.
        http://dx.doi.org/10.1006/rtph.200Q. 1413.
Grosjean. D. (1990). Atmospheric chemistry of toxic contaminants. 2. Saturated aliphatics: Acetaldehyde. dioxane.
        ethylene glycol ethers, propylene oxide. J Air Waste Manag Assoc 40:  1522-1531.
Guvatt G: Oxman. A: Kunz. R: Vist G: Falck-Ytter.  Y: Schiinemann. H. (2008aX GRADE: What is "quality of
        evidence" and why is it important to clinicians. BMJ 336: 995-998.
        http://dx.doi.org/10.1136/bmi.39490.551019.BE.
Guvatt G: Oxman. A: Vist. G: Kunz. R: Falck-Ytter.  Y: Alonso-Coello. P: Schiinemann. H. (2008bX GRADE: An
        emerging consensus on rating quality  of evidence and strength of recommendations. BMJ 336: 924-926.
        http://dx.doi.org/10.1136/bmi.39489.470347.AD.
Hansch. C: Leo. A: Hoekman. D. (1995). Exploring QSAR: Hydrophobia electronic, and steric constants.
        Washington. DC: American Chemical Society.
Harkema. J: Carey. S: Wagner. J. (2006). The nose revisited: A brief review of the comparative structure, function.
        and toxicologic pathology of the nasal epithelium. Toxicol Pathol 34: 252-269.
Haseman. J: Huff. J: Boorman. G. (1984). Use of historical control data in carcinogenicitv studies in rodents.
        Toxicol Pathol 12: 126-135. http://dx.doi.org/10.1177/019262338401200203.
Haseman. J and Hailev. J. (1997). An update of the National Toxicology Program database on nasal carcinogens.
        MutatRes 380: 3-11. http://dx.doi.org/10.1016/S0027-5107(97')00121-8.
Hawlev. G and Lewis. R. Sr (Eds.X (2001). Hawlev's  Condensed Chemical Dictionary (14 ed.X New York. NY:
        John Wiley & Sons. Inc.
Haworth. S: Lawlor. T: Mortelmans. K: Speck. W: Zeiger. E. (1983).  Salmonella mutagenicitv test results for 250
        chemicals. Environ Mutagen 5: 3-142. http://dx.doi.org/10.1002/em.2860050703.
Havashi. S: Watanabe. J: Kawajiri. K. (1991). Genetic polymorphisms in the 5'-flanking region change
        transcriptional regulation of the human cvtochrome P450HE1 gene. JBiochem 110: 559-565.
Hellmer. L and Bolcsfoldi. G. (1992). An evaluation of the E. coli K-12 uvrB/recA DNA repair host-mediated assay.
        I. In vitro sensitivity of the bacteria to 61 compounds. Mutat Res 272: 145-160.
        http://dx.doi.org/10.1016/0165-1161(92')90043-L.
                                                                                                    146
                                  DRAFT - DO NOT CITE OR QUOTE

-------
HF Jr. S: Seaton. J: Fischer. L. (1941). The single dose toxicitv of some glycols and derivatives. J Ind Hyg Toxicol
        23:259-268.
Hill. A. (1965). The environment and disease: Association or causation? Proc R Soc Med 58: 295-300.
Hoch-Ligeti. C and Argus. M. (1970). Effect of carcinogens on the lung of guinea pigs. In P Nettlesheim. MG
        Hanna Jr & JW Deatherage Jr (Eds.). Morphology of Experimental Respiratory  Carcinogenesis:
        Proceedings of a Biology Division. Oak Ridge National Laboratory.  Conference held in Gatlinburg.
        Tennessee. May 13-16. 1970 (Vol. 21. pp. 267-279). Oak Ridge. TN: United States Atomic Energy
        Co mission. Division of Technical Information.
Hoch-Ligeti. C: Argus. M: Arcos. J. (1970). Induction of carcinomas in the nasal cavity of rats by dioxane. Br J
        Cancer 24: 164-167.
HSDB. (Hazardous Substances Data Bank). (2007). 1.4-Dioxane. Bethesda. Maryland: National Library of
        Medicine.  National Toxicology Program. Hazardous Substances Data Bank.
Huang. C-Y: Huang. K-L: Cheng. T-J: Wang. J-D: Hsieh. L-L. (19971. The GST Tl and  CYP2E1 genotypes are
        possible factors causing vinyl chloride induced abnormal liver function. Arch Toxicol 71: 482-488.
        http://dx.doi.org/10.1007/s002040050416.
IARC. (International Agency for Research on Cancer). (1999). 1.4-Dioxane. In IARC Monographs on the
        Evaluation of Carciongenic Risks to Humans, pp. 589-602. Lyon. France: World Health Organization.
        http://monographs.iarc.fr/ENG/Monographs/vol71/mono71-25.pdf.
IARC. (International Agency for Research on Cancer). (2006). Preamble to the IARC monographs. Lyon. France.
        http://monographs.iarc.fr/ENG/Preamble/.
ICRP. (International Commission on Radiological Protection). (1975). Report of the task group on reference man:
        ICRP publication 23 (Vol. 23). New York. NY: International Commission of Radiological Protection.
        Pergamon Press.
JBRC. (Japan Bioassav Research Center). (1998). Two-year studies of 1.4-dioxane inF344 rats and BDF1 mice
        (drinking water). Kanagawa. Japan.
Johnstone. R. (1959). Death due to dioxane? AMA Arch Ind Health 20: 445-447.
Kanada. M: Miyagawa. M: Sato. M: Hasegawa. H: Honma. T. (1994). Neurochemical profile of effects of 28
        neurotoxic chemicals on the central nervous system in rats (1) Effects of oral administration on brain
        contents of biogenic amines and metabolites. Ind Health 32: 145-164.
        http://dx.doi.org/10.2486/indhealth.32.145.
Kano. H: Umeda. Y: Saito. M: Senoh. H: Ohbayashi H: Aiso. S: Yamazaki. K: Nagano.  K: Fukushima. S. (2008).
        Thirteen-week oral toxicitv of 1.4-dioxane in rats and mice. J Toxicol Sci 33: 141-153.
        http://dx.doi.org/10.2131/its.33.141.
Kano. H: Umeda. Y: Kasai T: Sasaki. T: Matsumoto. M: Yamazaki. K: Nagano. K: Arito. H: Fukushima. S. (2009).
        Carcinogenicity studies of 1.4-dioxane administered in drinking-water to rats and mice for 2 years. Food
        Chem Toxicol 47: 2776-2784. http://dx.doi.0rg/10.1016/i.fct.2009.08.012.
Karbe. E and Kerlin. R.  (2002). Cystic degeneration/spongiosis hepatis in rats. Toxicol Pathol 30: 216-227.
        http://dx.doi.org/10.1080/019262302753559551.
Kasai. T (2008)  ri.4-Dioxane toxicitv studies].
Kasai. T: Saito. M:  Senoh. H: Umeda. Y: Aiso. S: Ohbavashi. H: Nishizawa. T: Nagano.  K: Fukushima. S. (2008).
        Thirteen-week inhalation toxicitv of 1.4-dioxane in rats. Inhal Toxicol 20: 961-971.
        http://dx.doi.org/10.1080/08958370802105397.
Kasai. T: Kano. H:  Umeda. Y: Sasaki. T: Ikawa. N: Nishizawa. T: Nagano. K: Arito. H: Nagashima. H: Fukushima.
        S. (2009).  Two-year inhalation study of carcinogenicitv and chronic toxicitv of 1.4-dioxane in male rats.
        Inhal Toxicol 21: 889-897. http://dx.doi.org/10.1080/08958370802629610.
Kasper. P: Uno. Y:  Mauthe. R: Asano. N: Douglas. G: Matthews. E: Moore. M: Mueller. L: Nakaiima. M: Singer. T:
        Speit. G. (2007). Follow-up testing of rodent carcinogens not positive in the standard genotoxicitv testing
        battery: IWGT workgroup report. MutatRes 627: 106-116.
        http://dx.doi.0rg/10.1016/i.mrgentox.2006.10.007.
Katagiri. T: Nagano. K:  Aiso. S: Senoh. H: Sakura. Y: Takeuchi. T: Okudaira. M. (1998). A pathological study on
        spontaneous hepatic neoplasms inBDFl mice. J Toxicol Pathol 11: 21-25.
        http://dx.doi.org/10.1293/tox.ll.21.
Kesten. H: Mulinos. M:  Pomerantz. L. (1939). Pathologic effects of certain glycols and related compounds. Arch
        Pathol 27:  447-465.
Khudolev. V: Mizgireuv. I: Pliss. G. (1987). The study of mutagenic activity of carcinogens and other chemical
        agents with Salmonella tvphimurium assays: Testing of 126 compounds. Arch Geschwulstforsch 57: 453-
        462.
King. M: Shefner. A: Bates. R. (1973). Carcinogenesis bioassav of chlorinated dibenzodioxins and related
        chemicals. Environ Health Perspect 5: 163-170.

                                                                                                    147
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Kitchin. K and Brown. J. (1990). Is 1.4-dioxane a genotoxic carcinogen? Cancer Lett 53: 67-71.
        http://dx.doi.org/10.1016/0304-3835(90)90012-M.
Knoefel. P. (1935). Narcotic potency of some cyclic acetals. J Pharmacol Exp Ther 53: 440-444.
Kociba. R: McCollister. S: Park. C: Torkelson. T: Gehring. P. (1974). 1.4-dioxane. I. Results of a 2-year ingestion
        study in rats. Toxicol Appl Pharmacol 30: 275-286. http://dx.doi.org/10.1016/0041-008X(74')90099-4.
Kociba. R: Torkelson. T: Young. J: Gehring. P.  (1975). 1.4-Dioxane: Correlation of the results of chronic ingestion
        and inhalation studies with its dose-dependent fate in rats Proceedings of the 6th Annual Conference on
        Environmental Toxicology (pp. 345-354). Wright-Patterson Air Force Base. OH: Wright-Patterson Air
        Force Base. Air Force Systems Command. Aerospace Medical Division. Aerospace Medical Research
        Laboratory.
Kopylev. L: John Fox. J: Chen. C. (2009). Combining risks from several tumors using Markov Chain Monte Carlo.
        In RM Cooke (Ed.X Uncertainty Modeling in Dose Response (1 ed.. pp. 197-205). Hoboken. NJ: John
        Wiley & Sons.
Kurl R: Poellinger. L: Lund. J: Gustafsson. J-A. (1981). Effects of dioxane on RNA synthesis in the rat liver. Arch
        Toxicol 49: 29-33. http://dx.doi.org/10.1007/BF00352068.
Kwan. K: Dutka. B: Rao. S: Liu. D. (1990). Mutatox test: A new test for monitoring environmental genotoxic
        agents. Environ Pollut 65: 323-332. http://dx.doi.org/10.1016/0269-7491(90)90124-11.
Laug. E: Calvery. H: Morris. H: Woodard. G. (1939). The toxicology of some glycols and derivatives. J Ind Hyg
        Toxicol 21: 173-201.
Lesage. S: Jackson. R: Priddle. M: Riemann. P.  (1990). Occurrence and fate of organic solvent residues in anoxic
        groundwater at the Gloucester landfill Canada. Environ Sci Technol 24: 559-566.
        http://dx.doi.org/10.1021/es00074a016.
Leung. H-W and Paustenbach. D. (1990). Cancer risk assessment for dioxane based upon a physiologically-based
        pharmacokinetic approach. Toxicol Lett 51: 147-162.
Lewandowskj T and Rhomberg. L. (2005). A proposed methodology for selecting a trichloroethylene inhalation
        unit risk value for use in risk assessment. Regul Toxicol Pharmacol 41: 39-54.
        http://dx.doi.0rg/10.1016/i.vrtph.2004.09.003.
Lewis. R.  Sr. (2000). Sax's Dangerous Properties of Industrial Materials (10 ed.X New York. NY: John Wiley &
        Sons. Inc.
Lide. D. R.. (Ed.X  (2000). Handbook of Chemistry and Physics (81 ed.X Boca Raton. FL: CRC Press LLC.
Lundberg. I: Ekdahl. M: Kronevi. T: Lidums. V: Lundberg. S.  (1986). Relative hepatotoxicitv of some industrial
        solvents after intraperitoneal  injection or inhalation exposure in rats. Environ Res 40: 411-420.
        htto://dx.doi.org/10.1016/S0013-9351(86-)80116-5.
Lundberg. I: Hogberg. J: Kronevi. T: Holmberg. B. (1987). Three industrial solvents investigated for tumor
        promoting activity in the rat liver. Cancer Lett 36: 29-33. http://dx.doi.org/10.1016/0304-3835(87')90099-l.
Lyman. W: Reehl W: DH Rosenblatt (Eds.X (American Chemical Society). (1990). Handbook of Chemical
        Property Estimation Methods: Environmental Behavior of Organic Compounds. Washington. DC:
        American Chemical Society.
Marzulli. F: Anjo. D: Maibach. H. (1981). In vivo skin penetration studies of 2.4-toluenediamine. 2.4-
        diaminoanisole. 2-nitro-p-phenylenediamine. p-dioxane and N-nitrosodiethanolamine in cosmetics. Food
        Cosmet Toxicol 19: 743-747. htto://dx.doi.org/10.1016/0015-6264(81')90530-7.
McConnell E: Solleveld. H: Swenberg. J: Boorman. G. (1986). Guidelines for combining neoplasms for evaluation
        of rodent carcinogenesis studies. J Natl Cancer Inst 76: 283-289.
McFee. A: Abbott M: Gulati D: Shelby. M. (1994). Results of mouse bone marrow micronucleus studies on 1.4-
        dioxane. MutatRes 322: 145-148.
McGregor. D: Brown. A: Howgate. S: McBride. D: Riach. C: Caspary. W. (1991). Responses of the L5178Y  mouse
        lymphoma cell forward mutation assay. V: 27 coded chemicals. Environ Mol Mutagen 17: 196-219.
        http://dx.doi.org/10.1002/em.2850170309.
Medinsky. M and Bond. J. (2001). Sites and mechanisms for uptake of gases and vapors in the respiratory tract.
        Toxicology 160: 165-172.
Meylan. W: Howard. P: Boethling. R: Aronson. D: Printup. H: Gouchie. S. (1999). Improved method for estimating
        bioconcentration/bioaccumulation factor from octanol/water partition coefficient. Environ Toxicol Chem
        18: 664-672. http://dx.doi.org/10.1002/etc.5620180412.
Mikheev. M: Gorlinskava Ye. P: Solowova. T.  (1990). The body distribution and biological action of xenobiotics. J
        Hyg Epidemiol Microbiol Immunol 34: 329-336.
Mirkova. E. (1994). Activity of the rodent carcinogen 1.4-dioxane in the mouse bone marrow micronucleus assay.
        Mutat Res 322: 142-144.
                                                                                                    148
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Miyagawa. M: Shirotori. T: Tsuchitani. M: Yoshikawa. K. (1999). Repeat-assessment of 1.4-dioxane in a rat-
        hepatocvte replicative DNA synthesis (RDS) test: Evidence for stimulus of hepatocvte proliferation. Exp
        Toxicol Pathol 51: 555-558.
Morgan. K: Patterson. D: Gross. E. (1986). Responses of the nasal mucociliary apparatus of F-344 rats to
        formaldehyde gas. Toxicol Appl Pharmacol 82: 1-13. http://dx.doi.org/10.1016/0041-008X(86')90431-X.
Morita. T. (1994). No clastogenicitv of 1.4 dioxane as examined in the mouse peripheral blood micronucleus test.
        Mammalian Mutagenicitv Study Group Communications 2: 7-8.
Morita. T and Havashj M. (1998).  1.4-Dioxane is not mutagenic in five in vitro assays and mouse peripheral blood
        micronucleus assay, but is in mouse liver micronucleus assay.  Environ Mol Mutagen 32: 269-280.
        http://dx.doi.org/10.1002/(SICin098-2280(1998')32:3<269::AID-EM10>3.0.CO:2-8.
Mungikar. A and Pawar. S. (1978). Induction of the hepatic microsomal mixed function oxidase system in mice by
        p-dioxane. Bull Environ Contain Toxicol 20: 797-804. http://dx.doi.org/10.1007/BF01683603.
Munoz. E and Barnett B. (2002). The rodent carcinogens 1.4-dioxane and thiourea induce meiotic non-disjunction
        in Drosophila melanogaster females. MutatRes 517: 231-238. http://dx.doi.org/10.1016/S1383-
        5718(02)00083-9.
Nannelli A: De Rubertis. A: Longo. V: Gervasi P. (2005). Effects of dioxane on cvtochrome P450 enzymes in
        liver, kidney, lung and nasal mucosa of rat. Arch Toxicol 79: 74-82. http://dx.doi.org/10.1007/s00204-004-
        0590-z.
NCI. (National Institutes of Health. National Cancer Institute). (1978). Bioassav of 1.4-dioxane for possible
        carcinogenicitv. (78-1330 NCICGTR-80X Bethesda. MD: National Cancer Institute.
        http://ntp.niehs.nih.gov/ntp/htdocs/LT  rpts/trO80.pdf.
Nelson. N. (195IX Solvent toxicitv with particular reference to certain octyl alcohols and dioxanes. Med Bull 11:
        226-238.
Nestmann. E: Otson. R: Kowbel D: Bothwell P: Harrington. T. (1984). Mutagenicitv in a modified Salmonella
        assay of fabric-protecting products containing 1.1.1-trichloroethane. Environ Mol Mutagen 6: 71-80.
        http://dx.doi.org/10.1002/em.2860060109.
NRC. (National Research Council). (1994). Science and judgment in risk assessment. Washington. DC: National
        Academy Press.
NRC. (National Research Council). (2009). Science and decisions: Advancing risk assessment. Washington. DC:
        National Academies Press.
NRC. (National Research Council). (2011). Review of the Environmental Protection Agency's draft IRIS assessment
        of formaldehyde. Washington. DC: National Academies  Press, http://www.nap.edu/catalog/13142.html.
O'Neil MJ: Smith. A: Heckelmaa PE: Obenchain. JR: Gallipeau. JR: D'Arecca. MA (Eds.X (2001). The Merck
        Index:  An Encyclopedia of Chemicals. Drugs, and Biologicals (13th ed.X Whitehouse Station. NJ: Merck
        & Co.. Inc.
Park. J: Hussam A: Couasnon. P: Fritz. D: Carr. P. (1987). Experimental reexamination of selected partition
        coefficients from Rohrschneider's data set. Anal Chem 59:  1970-1976.
        http://dx.doi.org/10.1021/ac00142a016.
Platz. J: Sehested. J: Mogelberg. T: Nielsen. O:  Wallington. T. (1997). Atmospheric chemistry of 1.4-dioxane.
        Faraday Trans  1 93: 2855-2863. http://dx.doi.org/10.1039/a700598i.
Pozzani. U: Weil  C: Carpenter. C.  (1959). The toxicological basis of threshold limit values. 5: The experimental
        inhalation of vapor mixtures by  rats, with notes upon the relationship between single dose inhalation and
        single dose  oral data. Am Ind Hyg Assoc J 20: 364-369.  http://dx.doi.org/10.1080/00028895909343733.
Ramsey. J and Andersen. M. (1984). A physiologically based description of the inhalation pharmacokinetics of
        styrene in rats and humans. Toxicol Appl Pharmacol 73:  159-175. http://dx.doi.org/10.1016/0041-
        008X(84')90064-4.
Reitz. R: McCroskev. P: Park. C: Andersen. M: Gargas. M. (1990). Development of a physiologically based
        pharmacokinetic model for risk  assessment with 1.4-dioxane. Toxicol Appl Pharmacol  105: 37-54.
        http://dx.doi.org/10.1016/0041-008X(90')90357-Z.
Rosenkranz. H andKlopman. G. (1992).  1.4-dioxane: Prediction of in vivo clastogenicitv. MutatRes 280: 245-251.
        http://dx.doi.org/10.1016/0165-1218(92')90054-4.
Rothman. K and Greenland. S. (1998). Modern epidemiology (2nd ed.X Philadelphia. PA: Lippincott. Williams. &
        Wilkins.
Roy. S: Thilagar. A: Eastmond. D. (2005).  Chromosome breakage is primarily responsible for the micronuclei
        induced by  1.4-dioxane in the bone marrow and liver of young CD-I mice. Mutat Res 586: 28-37.
        http://dx.doi.0rg/10.1016/i.mrgentox.2005.05.007.
Sato. K. (1989). Glutathione transferases as markers of preneoplasia and neoplasia. Adv Cancer Res 52: 205-255.
Schrenk. H and Yant W. (1936). Toxicitv of dioxan. J Ind Hyg Toxicol 18: 448-460.
                                                                                                    149
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Sheu. C: Moreland. F: Lee. J: Dunkel. V. (1988). In vitro BALB/3T3 cell transformation assay of nonoxynol-9 and
        1.4-dioxane. Environ Mol Mutagen 11: 41-48. http://dx.doi.org/10.1002/em.2850110106.
Silverman. L: Schulte. H: First. M. (1946). Further studies on sensory response to certain industrial solvent vapors. J
        Ind Hyg Toxicol 28: 262-266.
Sina. J: Bean. C: Dysart. G: Taylor. V: Bradley. M. (1983). Evaluation of the alkaline elution/rat hepatocvte assay as
        a predictor of carcinogenic/mutagenic potential. MutatRes 113: 357-391. http://dx.doi.org/10.1016/0165-
        1161(83)90228-5.
Spiegelhalter. D: Thomas. A: Best. N: Lunn. D. (2003). WinBugs version 1.4 user manual. Cambridge. UK: MRC
        Biostatistics Unit.
Sticknev. J: Sager. S: Clarkson. J: Smith. L: Locev. B: Bock. M: Hartung. R: Olp. S. (2003). An updated evaluation
        of the carcinogenic potential of 1.4-dioxane. Regul Toxicol Pharmacol 38: 183-195.
        htto://dx.doi.org/10.1016/S0273-2300(03X)0090-4.
Stoner. G: Conran. P: Greisiger. E: Stober. J: Morgan. M: Pereira. M. (1986). Comparison of two routes of chemical
        administration on the lung adenoma response in strain A/J mice.  Toxicol Appl Pharmacol 82: 19-31.
        http://dx.doi.org/10.1016/0041-008X(86-)90433-3.
Stott W: Quasi J: Watanabe. P. (1981). Differentiation of the mechanisms of oncogenicitv of 1.4-dioxane and 1.3-
        hexachlorobutadiene in the rat.  Toxicol Appl Pharmacol 60: 287-300. http://dx.doi.org/10.1016/0041-
        008X(91)90232-4.
Stroebel. P: Mayer. F: Zerban. H: Bannasch. P. (1995). Spongiotic pericvtoma: A benign neoplasm deriving from
        the perisinusoidal (Ito) cells in rat liver. Am J Pathol 146: 903-913.
Surprenant K. (2002). Dioxane Ullmann's Encyclopedia of Industrial Chemistry (6th ed.. pp. 543). Weinheim.
        Germany: Wilev-VCH Verlag.
Sweeney. L: Thrall K: Poet T: Corlev. R: Weber. T: Locev. B: Clarkson. J: Sager. S: Gargas. M. (2008).
        Physiologically based pharmacokinetic modeling of 1.4-dioxane  in rats, mice,  and humans. Toxicol Sci
        101:  32-50. http://dx.doi.org/10.1093/toxsci/kfm251.
Thiess. A: Tress. E: Fleig. I.  (1976). Arbeitsmedizinische Untersuchungsergebnisse von Dioxan-exponierten
        Mitarbeitern [Industrial-medical investigation results in the case of workers exposed to dioxanel.
        Arbeitsmedizin. Sozialmedizin. Umweltmedizin 11: 35-46.
Thurman. G: Simms. B: Goldstein. A: Kilian. D. (1978). The effects of organic compounds used in the manufacture
        of plastics on the responsivitv of murine and human lymphocytes. Toxicol Appl Pharmacol 44: 617-641.
        http://dx.doi.org/10.1016/0041-008X(78')90269-7.
Tinwell. H and Ashbv. J. (1994). Activity of 1.4-dioxane in mouse bone marrow micronucleus assays. Mutat Res
        322:  148-150.
Torkelson. T: Leong. B: Kociba. R: Richter. W: Gehring. P. (1974). 1.4-Dioxane. II. Results of a 2-year inhalation
        study in rats. Toxicol Appl Pharmacol 30: 287-298.  http://dx.doi.org/10.1016/0041-008X(74)90100-8.
U.N. Environment Programme. (United Nations Environment Programme). (2000). The Montreal Protocol on
        substances that deplete the ozone layer. Nairobi. Kenya: United Nations Environment Programme. Ozone
        Secretariat.
        http://www.google.com/url?sa=t&source=web&cd=l&ved=OCBIQFjAA&url=http%3A%2F%2Fwww.une
        p.org%2Fozone%2Fpdfs%2Fmontreal-protocol2000.pdf&ei=-c89TPXON9PRngf-i-
        idDg&usg=AFOiCNH4OH15inPn5XFcYTvblPPRDZu-fO&sig2=aaSaM nuOlXlHc409kBvgw.
U.S. EPA. (U.S. Environmental Protection Agency). (1986a). Guidelines for carcinogen risk assessment.
        (EPA/630/R-00/004). pp. 38. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
        Forum. http://epa.gov/raf/publications/pdfs/CA%20GUIDELINES 1986.PDF.
U.S. EPA. (U.S. Environmental Protection Agency). (1986bX Guidelines for the health risk assessment of chemical
        mixtures. (EPA/630/R-98/002X  pp. 38. Washington. DC: U.S. Environmental Protection Agency. Risk
        Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=22567.
U.S. EPA. (U.S. Environmental Protection Agency). (1986cX Guidelines for mutagenicitv risk assessment.
        (EPA/630/R-98/003V pp. 23. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
        Forum, http://www.epa.gov/iris/backgrd.html.
U.S. EPA. (U.S. Environmental Protection Agency). (1988). Recommendations for and documentation of biological
        values for use in risk assessment. (EPA/600/6-87/008'), pp. 395.  Cincinnati. OH: U.S. Environmental
        Protection Agency. Environmental Criteria and Assessment Office.
        http://cfpub.epa. gov/ncea/cfm/recordisplav.cfm?deid=34855.
U.S. EPA. (U.S. Environmental Protection Agency). (1991).  Guidelines for developmental toxicitv risk assessment.
        (EPA/600/FR-91/00IX  pp. 83. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
        Forum, http://www.epa.gov/iris/backgrd.html.
U.S. EPA. (U.S. Environmental Protection Agency). (1994). Methods for derivation of inhalation reference
        concentrations and application of inhalation dosimetry. (EPA/600/8-90/066FX pp. 409. Research Triangle

                                                                                                   150
                                  DRAFT - DO NOT CITE OR QUOTE

-------
        Park. NC: U.S. Environmental Protection Agency. Office of Research and Development. Office of Health
        and Environmental Assessment Environmental Criteria and Assessment Office.
        http://cfpub.epa. gov/ncea/cfm/recordisplav.cfm?deid=71993.
U.S. EPA. (U.S. Environmental Protection Agency). (1996). Guidelines for reproductive toxicitv risk assessment.
        (EPA/630/R-96/009X pp. 143. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
        Forum. http://www.epa.gov/raf/publications/pdfs/REPRO51.PDF.
U.S. EPA. (U.S. Environmental Protection Agency). (1998). Guidelines for neurotoxicitv risk assessment.
        (EPA/630/R-95/001FX pp.  89. Washington. DC: U.S. Environmental Protection Agency. Risk Assessment
        Forum. http://www.epa.gov/raf/publications/pdfs/NEUROTOX.PDF.
U.S. EPA. (U.S. Environmental Protection Agency). (2000a). Benchmark dose technical guidance document
        [external review draft]. (EPA/630/R-00/001).  pp. 96. Washington. DC: U.S. Environmental Protection
        Agency. Risk Assessment Forum, http://www.epa.gov/raf/publications^enchmark-dose-doc-draft.htm.
U.S. EPA. (U.S. Environmental Protection Agency). (200010. Science policy council handbook: Risk
        characterization. (EPA 100-B-00-002). pp. 189. Washington. D.C.: U.S. Environmental Protection
        Agency. Office of Research and Development Office of Science Policy.
        http://www.epa.gov/osa/spc/pdfs/rchandbk.pdf.
U.S. EPA. (U.S. Environmental Protection Agency). (20000). Supplementary guidance for conducting health risk
        assessment of chemical mixtures. (EPA/630/R-00/002). pp. 209. Washington. DC: U.S. Environmental
        Protection Agency. Risk Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=20533.
U.S. EPA. (U.S. Environmental Protection Agency). (2002aX A review of the reference dose and reference
        concentration processes. (EPA/630/P-02/0002F). pp. 192. Washington. DC: U.S. Environmental  Protection
        Agency. Risk Assessment Forum. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=51717.
U.S. EPA. (U.S. Environmental Protection Agency). (2002b). Toxic Substances Control Act (TSCA) Inventory
        Update Database, from http://www.epa.gov/iur/
U.S. EPA. (U.S. Environmental Protection Agency). (2005a). Guidelines for carcinogen risk assessment.
        (EPA/630/P-03/001FX pp.  166. Washington. DC: U.S. Environmental Protection Agency. Risk
        Assessment Forum, http://www.epa.gov/cancerguidelines/.
U.S. EPA. (U.S. Environmental Protection Agency). (200510. Supplemental guidance for assessing susceptibility
        from early-life exposure to carcinogens. (EPA/630/R-03/003F). pp.  126. Washington. DC: U.S.
        Environmental Protection Agency. Risk Assessment Forum.
        http://www.epa.gov/cancerguidelines/guidelines-carcinogen-supplement.htm.
U.S. EPA. (U.S. Environmental Protection Agency). (2006). Peer review handbook (3rd edition). (EPA/100/B-
        06/002). pp. 190. Washington. DC: U.S. Environmental Protection Agency. Science Policy Council.
        http://www.epa.gov/peerreview/pdfs/peer  review handbook  2006.pdf.
U.S. EPA. (U.S. Environmental Protection Agency). (2009a). Integrated Risk Information System (IRIS) Update
        Project Announcement of 2009/2010 Agenda. Fed Reg 74: 54040-54042.
U.S. EPA. (U.S. Environmental Protection Agency). (2009b). Toxicological review of 1.4-dioxane (CAS No.  123-
        91-1) in support of summary information on the Intergrated Risk Information System (IRIS) [External
        Review Draftl. (EPA/635/R-09/005).  pp. 1-276. Washington. DC: U.S. Environmental Protection Agency.
        Integrated Risk Information System. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=199330.
U.S. EPA. (U.S. Environmental Protection Agency). (2009cX Status report: Advances in inhalation dosimetry of
        gases and vapors with portal of entry effects in the upper respiratory tract. (EPA/600/R-09/072X pp.  100.
        Research Triangle Park. NC: U.S. Environmental Protection Agency. National Center for Environmental
        Assessment. http://cfpub.epa.gov/ncea/cfm/recordisplav.cfm?deid=212131.
U.S. EPA. (U.S. Environmental Protection Agency). (2010). Toxicological review of 1.4-Dioxane (CAS No. 123-
        91-1) in support of summary information on the Integrated Risk Information System (IRIS). (EPA-635/R-
        09-005-FX pp. 319. Washington. DC. www.epa.gov/iris/toxreviews/0326tr.pdf.
U.S. EPA. (U.S. Environmental Protection Agency). (201 la). IRIS  Process, from http://epa.gov/iris/process.htm
U.S. EPA. (U.S. Environmental Protection Agency). (201 IbX Recommended use of body weight 3/4 as the default
        method in derivation of the  oral reference dose. (EPA/100/R11/000IX pp. 50. Washington. DC.
        http://www.epa.gov/raf/publications/interspecies-extrapolation.htm.
Uno. Y: Takasawa. H: Miyagawa. M: Inoue. Y: Murata. T: Yoshikawa. K. (1994). An in vivo-in vitro replicative
        DNA synthesis (RDS) test using rat hepatocvtes as an early prediction assay for nongenotoxic
        hepatocarcinogens screening of 22 known positives and 25 noncarcinogens. Mutat Res 320: 189-205.
        http://dx.doi.org/10.1016/0165-1218(94')90046-9.
van Delft. J: van Agen. E: van Breda. S: Herwijnen. M: Staal Y: Klemjans. J. (2004). Discrimination of genotoxic
        from non-genotoxic carcinogens by gene expression profiling. Carcinogenesis 25:  1265-1276.
        http://dx.doi.org/10.1093/carcin/bghl08.
                                                                                                   151
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Vieira. I: Sonnier. M: Cresteil. T. (1996). Developmental expression of CYP2E1 in the human liver:
        Hypermethylation control of gene expression during the neonatal period. Eur J Biochem 238: 476-483.
        http://dx.doi.org/10.1111/i. 1432-1033.1996.0476z.x.
Watanabe. J: Havashi. S: Kawajiri. K. (1994). Different regulation and expression of the human CYP2E1 gene due
        to the Rsal polymorphism in the 5'-flanking region. J Biochem 116: 321-326.
Waxman. D: Pampori. N: Ram. P: Agrawal A: Shapiro. B. (1991). Interpulse interval in circulating growth
        hormone patterns regulates sexually dimorphic expression of hepatic cvtochrome P450. PNAS 88: 6868-
        6872.
Wirth. W and Klimmer. O. (1936). [On the toxicology of organic solvents.  1.4 dioxane (diethylene dioxide)]. Archiv
        fuer Gewerbepathologie und Gewerbehygiene 17: 192-206.
Wolfe. N and Jeffers. P. (2000). Hydrolysis. InRS Boethling & D Mackav (Eds.). Handbook of Property Estimation
        Methods for Chemicals: Environmental and Health Sciences (pp. 311-333). Boca Raton. FL: Lewis
        Publishers.
Wolford. S: Schroer. R: Gohs. F: Gallo. P: Brodeck. M: Falk. H: Ruhrea R. (1986). Reference range database for
        serum chemistry and hematologv values in laboratory animals. J Toxicol Environ Health A 18: 161-188.
        http://dx.doi.org/10.1080/15287398609530859.
Woo. Y-T: Arcos. J: Argus. M: Griffin. G: K. N. (1911 a). Structural identification of p-dioxane-2-one as the major
        urinary metabolite of p-dioxane. Naunyn Schmiedebergs Arch Pharmacol 299: 283-287.
        http://dx.doi.org/10.1007/BF00500322.
Woo. Y-T: Argus. M: Arcos. J. (1977b). Tissue and subcellular distribution of 3H-dioxane in the rat and apparent
        lack of microsome-catalyzed covalent binding in the target tissue.  Life Sci 21: 1447-1456.
        http://dx.doi.org/10.1016/0024-3205(77')90199-0.
Woo. Y-T: Argus. M: Arcos. J. (1977cX Metabolism in vivo of dioxane: Effect of inducers and inhibitors of hepatic
        mixed-function oxidases. Biochem Pharmacol 26: 1539-1542. http://dx.doi.org/10.1016/0006-
        2952(77)90431-2.
Woo. Y-T: Argus. M: Arcos. J. (1978). Effect of mixed-function oxidase modifiers on metabolism and toxicitv of
        the oncogen dioxane. Cancer Res 38: 1621-1625.
Yamamoto. S: Urano. K: Koizumi. H: Wakana. S: Hioki. K: Mitsumori. K: Kurokawa. Y: Havashi. Y: T. N.
        (1998aX Validation of transgenic mice carrying the human prototype c-Ha-ras gene as a bioassav model for
        rapid carcinogenicitv testing. Environ Health Perspect 106: 57-69.
Yamamoto. S: Urano. K: Nomura. T. (1998bX Validation of transgenic mice harboring the human prototype c-Ha-
        ras gene as a bioassav model for rapid carcinogenicitv testing. Toxicol Lett 102-103: 473-478.
        http://dx.doi.org/10.1016/S0378-4274(98X)0341-5.
Yamamoto. S: Ohsawa. M: Nishizawa. T: Saito. A: Kasai. T: Noguchi. T: Nagano. K: Matsushima. T. (2000). Long-
        term toxicology study of 1.4-Dioxane in the F344 rats by multiple-route exposure (drinking water and
        inhalation). J Toxicol Sci 25: 347.
Yamazaki. K: Ohno. H: Asakura. M: Narumi. A: Ohbavashi. H: Fujita. H: Ohnishi. M: Katagiri. T: Senoh. H:
        Yamanouchi. K: Nakavama. E: Yamamoto. S: Noguchi. T: Nagano. K: Enomoto. M: Sakabe. H.  (1994).
        Two-year toxicological and carcinogenesis studies of 1.4-dioxane  in F344 rats and BDF1 mice. In K
        Sumino. S Sato & NG Shinkokai (Eds.X Proceedings: Second Asia-Pacific Symposium on Environmental
        and Occupational Health 22-24 July. 1993: Kobe (pp. 193-198). Kobe. Japan: Kobe University School of
        Medicine. International Center for Medical Research.
Yamazaki. K (2006) [Correspondence between Kazunori Yamazaki and Julie Sticknevl.
Yant W: Schrenk. H: Waite. C: Patty. F.  (1930). Acute response of guinea pigs to vapors of some new commercial
        organic compounds: VI. Dioxan. Public Health Rep 45: 2023-2032.
Yasuhara. A: Shiraishi. H: Nishikawa. M: Yamamoto. T: Uehiro. T: Nakasugi. O: Okumura. T: Kenmotsu. K:
        Fukuj H: Nagase. M: Ono. Y: Kawagoshj Y: Baba. K: Noma. Y. (1997). Determination of organic
        components in leachates from hazardous waste disposal sites  in Japan by gas  chromatography-mass
        spectrometry. J Chromatogr A 774: 321-332. http://dx.doi.org/10.1016/S0021-9673(97')00078-2.
Yasuhara. A: Tanaka. Y: Tanabe. A: Kawata. K: Katamj T. (2003). Elution of 1.4-dioxane from waste landfill sites.
        Bull Environ Contam Toxicol 71: 641-647. http://dx.doi.org/10.1007/s00128-003-8917-7.
Yoon. J: Mason. J: Valencia. R: Woodruff. R: Zimmering. S. (1985). Chemical mutagenesis testing in Drosophila.
        IV. Results of 45 coded compounds tested for the National Toxicology Program. Environ Mutagen 7: 349-
        367.  http://dx.doi.org/10.1002/em.286007Q310.
Young. J: Braun. W: Gehring. P: Horvath. B: Daniel R. (1976). 1.4-Dioxane andbeta-hydroxyethoxyacetic acid
        excretion in urine of humans exposed to  dioxane vapors.  Toxicol Appl Pharmacol 38: 643-646.
        http://dx.doi.org/10.1016/0041-008X(76-)90195-2.
Young. J: Braun. W: Rampy. L: Chenoweth. M: Blau. G. (1977). Pharmacokinetics of 1.4-dioxane in humans. J
        Toxicol Environ Health 3: 507-520. http://dx.doi.org/10.1080/15287397709529583.

                                                                                                   152
                                 DRAFT - DO NOT CITE OR QUOTE

-------
Young. J: Braun. W: Gehring. P. (1978aX Dose—dependent fate of 1.4-dioxane in rats(bX J Toxicol Environ
        Health A 4: 709-726. http://dx.doi.org/10.1080/15287397809529693.
Young. J: Braun. W: Gehring. P. (1978bX The dose-dependent fate of 1.4-dioxane in rats. J Environ Pathol Toxicol
        2: 263-282. http://dx.doi.org/10.1080/15287397809529693.
Zimmermann. F: Mayer. V: Scheel. I: Resnick. M. (1985). Acetone, methyl ethyl ketone. ethyl acetate, acetonitrile
        and other polar aprotic solvents are strong inducers of aneuploidv in Saccharomyces cerevisiae. Mutat Res
        149: 339-351. http://dx.doi.org/10.1016/0027-5107(85')90150-2.
                                                                                                    153
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      APPENDIX  A.     SUMMARY  OF   EXTERNAL  PEER
         REVIEW  AND   PUBLIC   COMMENTS  AND
         DIS  POS  ITION

 1    Note: The comments and responses in this appendix were in regards to the oral assessment previously
 2    reviewed. A summary of external peer review and public comments and disposition following review of
 3    the inhalation assessment for 1.4-dioxane will be included when they become available.

 4           The Toxicological Review of 1,4-Dioxane has undergone formal external peer review performed
 5    by scientists in accordance with EPA guidance on peer review (U.S. EPA. 2006. 2000b). The external
 6    peer reviewers were tasked with providing written answers to general questions on the overall assessment
 7    and on chemical-specific questions in areas of scientific controversy or uncertainty. A summary of
 8    significant comments made by the external reviewers and EPA's responses to these comments arranged
 9    by charge question follow. In many cases the comments of the individual  reviewers have been
10    synthesized and paraphrased for development of Appendix A. The majority of the specific observations
11    (in addition to EPA's charge questions) made by the peer reviewers were  incorporated into the document
12    and are not discussed further in this Appendix. Public comments that were received are summarized and
13    addressed following the peer-reviewers' comments and disposition.
      A.1   External Peer Review Panel Comments

14           The reviewers made several editorial suggestions to clarify portions of the text. These changes
15    were incorporated in the document as appropriate and are not discussed further.

16           In addition, the external peer reviewers commented on decisions and analyses in the
17    Toxicological Review ofl,4-Dioxane under multiple charge questions, and these comments were
18    organized and summarized under the most appropriate charge  question.
      A.1.1   General Charge Questions

19    1.  Is the Toxicological Review logical, clear and concise? Has EPA accurately, clearly and objectively
20       represented and synthesized the scientific evidence for noncancer and cancer hazards?

21           Comment. All reviewers found the Toxicological Review to be logical, clear, and concise. One
22           reviewer remarked that it was an accurate, open-minded and balanced analysis of the literature.
23           Most reviewers found that the scientific evidence was presented objectively and transparently;
24           however, one reviewer suggested two things to improve the objectivity and transparency (1)
25           provide a clear description of the mode of action and how it feeds into the choice of the
26           extrapolation for the cancer endpoint and (2) provide a presentation of the outcome if internal
27           dose was used in the cancer and noncancer assessments.

                                                                                                    A-l
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1                  One reviewer commented that conclusions could not be evaluated in a few places where
 2           dose information was not provided (Sections 3.2, 3.3 and 4.5.2.2). The same reviewer found the
 3           MOA schematics, key event temporal sequence/dose-response table, and the POD plots to be
 4           very helpful in following the logic employed in the assessment.
 6           Response. The mode of action analysis and how conclusions from that analysis fed into the
 7           choice of extrapolation method for the cancer assessment are discussed further under charge
 8           questions C2 and C5. Because of the decision not to utilize the PBPK models, internal doses were
 9           not calculated and thus were not included as alternatives to using the external dose as the POD for
10           the cancer and noncancer assessments.

11           In the sections noted by the reviewer (3.2, 3.3 and 4.5.2.2) dose information was added as
12           available. In Section 3.2, Mikheev et al. (1990) did not report actual doses, which is noted in this
13           section. All other dose information in this section was found to be present after further review by
14           the Agency. In Section 3.3, dose information for Woo et al. (1978.  1977c) was added to the
15           paragraph. In Section 4.5.2.2, study details for Nannelli et al. (2005) were provided earlier in
16           Section 3.3 and a statement referring the reader to this section was added.

17

18    2.  Please identify any additional studies that should be considered in the assessment of the noncancer
19       and cancer health effects of 1,4-dioxane.

20           Comment. Five reviewers stated they were unaware of any additional studies available to add to
21           the oral toxicity evaluation of 1,4-dioxane. These reviewers also acknowledged the Kasai et al.
22           (2009; 2008) publications that may be of use to derive toxicity values following inhalation of
23           1,4-dioxane.

24               a.   Kasai T; Saito H; Senoh Y; et al. (2008) Thirteen-week inhalation toxicity of 1,4-dioxane
25                   in rats. Inhal Toxicol 20: 961 -971.

26               b.   Kasai T; Kano Y; Umeda T; et al. (2009) Two-year inhalation study of carcinogenicity
27                   and chronic toxicity of 1,4-dioxane in male rats. Inhal Toxicol in press.

28           Other references suggested by reviewers include:

29               c.   California Department of Health Services (1989) Risk Specific Intake Levels for the
30                   Proposition 65 Carcinogen 1, 4-dioxane. Reproductive and Cancer Hazard Assessment
31                   Section. Office of Environmental Health Hazard Assessment

32               d.   National Research Council (2009) Science and Decisions: Advancing Risk Assessment.
                                             \	/                               O
33                   Committee on Improving Risk Analysis Approaches Used by the U.S. EPA. Washington,
34                   D.C., National Academy Press.
                                                                                                        A-2
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1               e.  ATSDR (2007) Toxicological Profile for 1,4-dioxane. Agency for Toxic Substances and
 2                  Disease Registry. Atlanta, GA.

 3               f.  Stickney JA; Sager SL; Clarkson JR; et al. (2003) An updated evaluation of the
 4                  carcinogenic potential of 1,4-dioxane. Regul Toxicol Pharmacol 38:  183-195.

 5               g.  Yamamoto S; Ohsawa M; Nishizawa T; et al. (2000) Long-term toxicology study of
 6                  1,4-dioxane in R344 rats by multiple-route exposure (drinking water and inhalation). J
 7                  Toxicol Sci 25: 347.
 9           Response. The references a-b above will be evaluated for derivation of an RfC and IUR, which
10           will follow as an update to this oral assessment. References c and e noted above were considered
11           during development of this assessment as to the value they added to the cancer and noncancer
12           analyses. Reference g listed above is an abstract from conference proceedings from the 27th
13           Annual Meeting of the Japanese Society of Toxicology; abstracts are not generally considered in
14           the development of an IRIS assessment. Reference d reviews EPA's current risk assessment
15           procedures and provides no specific information regarding 1,4-dioxane. The Stickney et al.
16           (2003) reference was a review article and no new data were presented, thus it was not referenced
17           in this Toxicological Review but the data were considered during the development of this
18           assessment.

19           Following external peer review (as noted above) Kano et al. (2009) was added to the assessment,
20           which was an update and peer-reviewed published manuscript of the JBRC (1998) report.

21

22    3.  Please discuss research that you think would be likely to increase confidence in the database for
23       future assessments of 1,4-dioxane.

24           Comment. All reviewers provided suggestions for additional research that would strengthen the
25           assessment and reduce uncertainty in several areas. The following is a brief list of questions that
26           were identified that could benefit from further research. What are the mechanisms responsible for
27           the acute and chronic nephrotoxicity? Is the acute  kidney injury (AKI) multifactorial? Are there
28           both tubular and glomerular/vascular toxicities that result in cortical tubule degeneration and
29           evidence for glomerulonephritis? What are the functional correlates of the histologic changes in
30           terms of assessment of renal function? What is the exposure in utero and risk to the fetus and
31           newborn? What are the concentrations in breast milk  following maternal exposure to
32           1,4-dioxane? What is the risk for use of contaminated drinking water to reconstitute infant
33           formula? What are the exposures during early human development? What is the pharmacokinetic
34           and metabolic profile of 1,4-dioxane during development? What are the susceptible populations
35           (e.g., individuals with decreased renal function or chronic renal disease, obese individuals,
36           gender, age)?
                                                                                                        A-3
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Additional suggestions for future research include: evaluation of potential epigenetic mechanisms
 2           of carcinogenicity, additional information on sources of exposure and biological concentrations as
 3           well as human toxicokinetic data for derivation of parameter to refine PBPK model, studies to
 4           determine toxic moiety, focused studies to inform mode of action, additional inhalation studies
 5           and a multigeneration reproductive toxicity study.

 6           One reviewer suggested additional analyses of the existing data including a combined analysis of
 7           the multiple datasets and outcomes for cancer and non-cancer endpoints, evaluation of the dose
 8           metrics relevant to the MOA to improve confidence in extrapolation approach and uncertainty
 9           factors, and complete a Bayesian analysis of human pharmacokinetic data to estimate human
10           variability in key determinants of toxicity (e.g., metabolic rates and partition coefficients).

11

12           Response. A number of research suggestions were provided for further research that may enhance
13           future health assessments of 1,4-dioxane. Regarding the suggested additional analyses for the
14           existing data, EPA did not identify a MOA in this assessment, thus combined analysis of the
15           cancer and non-cancer endpoints as well as application of various  dose metrics to  a MOA is not
16           applicable. Because the human PBPK model was not implemented in this assessment for oral
17           exposure to 1,4-dioxane a Bayesian analysis was not completed. No additional changes to the
18           Toxicological Review of 1,4-Dioxane were made in response to these research recommendations.

19

20    4.  Please comment on  the identification and characterization of sources of uncertainty in Sections 5 and
21       6 of the assessment  document. Please comment on whether the key sources of uncertainty have been
22       adequately discussed. Have the choices and assumptions made in the discussion of uncertainty been
23       transparently and objectively described? Has the impact of the uncertainty on the  assessment been
24       transparently and objectively described?

25           Comment. Six reviewers stated Sections 5 and 6 adequately discussed and characterized
26           uncertainty, in a succinct, and transparent manner. One reviewer suggested adding additional
27           discussion of uncertainty relating to the critical study used in the cancer assessment and another
28           reviewer suggested adding more discussion around the uncertainty of the toxic moiety.

29           One reviewer made specific comments on uncertainty surrounding the Kociba et al. (1974) study
30           as used for derivation of the RfD, choice of the non-cancer dose metric, and use of a 10%BMR as
31           the basis for the CSF derivation. These comments and responses are summarized below under
32           their appropriate charge question.

33

34           Response. The majority of the reviewers thought the amount of uncertainty discussion was
35           appropriate. Since the external review, Kano et al. (2009) was published and this assessment was
                                                                                                        A-4
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           updated accordingly (previously JBRC (1998). It is assumed the uncertainty referred to by the
 2           reviewer was addressed by the published Kano et al. (2009) paper.

 3           Clarification regarding the uncertainty surrounding the identification of the toxic moiety was
 4           added to Section 4.6.2.1 stating that the mechanism by which 1,4-dioxane induces tissue damage
 5           is not known, nor is it known whether the toxic moiety is 1,4-dioxane or a metabolite of
 6           1,4-dioxane. Additional text was added to Section 4.7.3 clarifying that available data also do not
 7           clearly identify whether 1,4-dioxane or one of its metabolites is responsible for the observed
 8           effects. The impact of the lack of evidence to clearly identify a toxic moiety related to
 9           1,4-dioxane exposure was summarized in Sections 5.5.1.2 and 6.2.3.2.
      A.1.2    Oral reference dose (RfD) for 1,4-dioxane

10    1.  A chronic RfD for 1,4-dioxane has been derived from a 2-year drinking water study (Kocibaet al.,
11       1974) in rats and mice. Please comment on whether the selection of this study as the principal study
12       has been scientifically justified. Has the selection of this study been transparently and objectively
13       described in the document? Are the criteria and rationale for this selection transparently and
14       objectively described in the document? Please identify and provide the rationale for any other studies
15       that should be selected as the principal study.

16           Comment. Seven of the reviewers agreed that the use of the Kociba et al. (1974) study was the
17           best choice for the principal study.

18           One reviewer stated that Kociba et al. (1974) was not the best choice because it reported only
19           NOAEL and LOAELs without providing incidence data for the endpoints. This reviewer also
20           stated that the study should not have been selected based on sensitivity of the endpoints, but
21           rather study design and adequacy of reporting of the study results. Additionally, this reviewer
22           suggested a better principal study would be either the NCI (1978) or JBRC (1998) study.

23           Response. The reviewer is correct that Kociba et al. (1974) did not provide incidence data;
24           however, Kociba et al. (1974) identified a NOAEL (9.6 mg/kg-day) and LOAEL (94 mg/kg-day)
25           within the text of the manuscript.  Kociba et al. (1974) was a well conducted chronic bioassay
26           (four dose levels, including controls, with 60 rats/sex/group) and seven of the peer reviewers
27           found this study to be appropriate as the basis for the RfD. Further support for the selection of the
28           Kociba et al. (1974) as the principal study comes from comparison of the liver and kidney
29           toxicity data reported by JBRC (1998) and NCI (1978). which was presented in Section 5.1. The
30           effects reported by JBRC  (1998) and NCI (1978) were consistent with what was observed by
31           Kociba et al. (1974) and within a  similar dose range. Derivation of an RfD from these datasets
32           resulted in a similar value (Section 5.1.).

33    2.  Degenerative liver and kidney effects were selected as the critical effect. Please  comment on whether
34       the rationale for the selection of this critical effect has been  scientifically justified. Are the criteria and
35       rationale for this selection transparently and objectively described in the document? Please provide a

                                                                                                        A-5
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1       detailed explanation. Please comment on whether EPA's rationale regarding adversity of the critical
 2       effect for the RfD has been adequately and transparently described and is scientifically supported by
 3       the available data. Please identify and provide the rationale for any other endpoints that should be
 4       considered in the selection of the critical effect.

 5           Comment. Five of the reviewers agreed with the selection of liver and kidney effects as the
 6           critical effect. One of these reviewers suggested analyzing all datasets following dose adjustment
 7           (e-g-, body weight scaling or PBPK model based) to provide a better rationale for selection of a
 8           critical effect.

 9           One  reviewer stated that 1,4-dioxane causing liver and kidney organ specific effects is logical;
10           however, with regards to nephrotoxicity, the models  and limited human data have not addressed
11           the mechanisms of injury or the clinical correlates to the histologic data. Also, advances in the
12           field of biomarkers have not yet been used for the study of 1,4-dioxane.

13           One  reviewer found the selection of these endpoints to be 'without merit' because of the lack of
14           incidence data to justify the NOAEL and LOAEL values identified in the study. This reviewer
15           suggested selecting the most sensitive  endpoint(s) from the NCI (NCI.  1978) or JBRC (1998)
16           studies for the basis of the RfD, but did not provide a suggestion as to what effect should be
17           selected.

18           Response. The liver and kidney effects from  Kociba et al. (1974) was supported as the critical
19           effect by most of the reviewers. PBPK model adjustment was not performed because the PBPK
20           model was found to be inadequate for  use in the assessment. EPA acknowledges that neither the
21           mechanisms of injury nor the clinical correlates to histologic data exist for 1,4-dioxane. This type
22           of information could improve future health assessments of 1,4-dioxane.

23           As stated above, Kociba et al. (1974) identified a NOAEL (9.6 mg/kg-day) and LOAEL
24           (94 mg/kg-day)  within the text of the manuscript  and was a well conducted chronic bioassay (four
25           dose levels, including controls, with 60 rats/sex/group).

26    3.  Kociba et al. (1974) derived a NOAEL based upon the observation of degenerative liver and kidney
27       effects and these data were utilized to derive the point of departure (POD) for the RfD. Please provide
28       comments with regard to whether the NOAEL approach is the best approach for determining the
29       POD. Has the approach been appropriately conducted and objectively and transparently described?
30       Please identify and provide rationales for any alternative approaches for the determination of the POD
31       and discuss whether such approaches are preferred to EPA's approach.

32           Comment. Seven reviewers agreed with the NOAEL approach described in the document. One  of
33           these reviewers  also questioned whether any  attempt was made to  "semi-qualitatively represent
34           the histopathological observations to facilitate a quantitative analysis".

35           One  reviewer stated that data were not used to derive the POD, but rather a claim by the authors
36           of Kociba et al.  (1974) of the NOAEL and LOAEL for the endpoints. This reviewer preferred the
                                                                                                       A-6
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           use of a BMD approach for which data include the reported incidence rather than a study reported
 2           NOAEL or LOAEL.

 3           Response. The suggestion to "semi-qualitatively represent the histopathological observations to
 4           facilitate a quantitative analysis" was not incorporated into the document because it is unclear
 5           how this would be conducted since Kociba et al. (1974) did not provide incidence data and the
 6           reviewer did not illustrate their suggested approach. See responses to B1 and B2 regarding the
 7           NOAEL and LOAEL approach. The Agency agrees that a Benchmark Dose approach is preferred
 8           over the use of a NOAEL or LOAEL for the POD if suitable data (e.g., reflecting the most
 9           sensitive sex, species, and endpoint identified) are available for modeling and, if suitable data are
10           not available, then NOAEL and LOAEL values are utilized. In this case, the data were not
11           suitable for BMD modeling and the LOAEL or NOAEL approach was used.

12    4.  EPA evaluated the PBPK and empirical models available to describe kinetics following inhalation of
13       1,4-dioxane (Reitzetal.. 1990; Young etal.. 1978a: Young etal. 1978b: Young etal.  1977). EPA
14       concluded that the use of existing, revised, and recalibrated PBPK models for 1,4-dioxane were not
15       superior to default approaches for the dose-extrapolation between species. Please comment on
16       whether EPA's rationale regarding the decision to not utilize existing or revised PBPK models has
17       been adequately and transparently described and is supported by the available data. Please identify
18       and provide the rationale for any alternative approaches that should be considered or preferred to the
19       approach presented in the toxicological review.

20           Comment. Six reviewers found the decision not to utilize the  available PBPK models to be
21           appropriate and supported by available data. One of these reviewers suggested presenting as part
22           of the uncertainty evaluation an adjustment of the experimental doses based on metabolic
23           saturation. Another reviewer stated Appendix B was hard to follow and that the main document
24           should include a more complete description of the model refinement effort performed by
25           Sweeney et al. (2008).

26           Two reviewers noted a complete evaluation of the models was evident; one of the reviewers
27           questioned the decision not to use the models  on the basis that they were unable to fit the human
28           blood PK data for  1,4-dioxane. This reviewer suggested the rat model might fit the  human blood
29           PK data, thus raising concern in the reliance on the human blood PK data to evaluate the PBPK
30           model for 1,4-dioxane. Instead, the reviewer suggested the human urinary metabolite data may be
31           sufficient to give confidence in the model. One other reviewer also questioned the accuracy of the
32           available human data. One reviewer commented that the rationale for not using the  PBPK model
33           to extrapolate from high to low dose was questioned. In addition, the reviewer suggested that two
34           aspects of the model code for Reitz et al. (1990) need to be verified:

35                  a.  In the  document, KLC is defined  as a first-order rate constant and is scaled by BW°7.
36                     This is inconsistent when multiplied by concentration does not result in units
37                     of mg/hr. However, if the parameter is actually considered a clearance constant
38                     (zero-order rate constant) then the scaling rule used,  as well as the interpretations
39                     provided, would be  acceptable.
                                                                                                      A-7
                                     DRAFT - DO NOT  CITE OR QUOTE

-------
 1                  b.  It is unclear as to why AM is calculated on the basis of RAM and not RMEX. RMEX
 2                      seems to represent the amount metabolized per unit time.

 3           Response. The U.S. EPA performed a rigorous evaluation of the PBPK models available for
 4           1,4-dioxane. This effort was extensively described in Section 3.5 and in Appendix B. In short,
 5           several procedures were applied to the human PBPK model to determine if an adequate fit of the
 6           model to the empirical model output or experimental observations could be attained using
 7           biologically plausible values for the model parameters. The re-calibrated model predictions for
 8           blood 1,4-dioxane levels did not come within 10-fold of the experimental values using measured
 9           tissue:air partition coefficients of (Leung and Paustenbach. 1990) or (Sweeney et al.. 2008)
10           (Figure B-8 and Figure B-9). The utilization of a slowly perfused tissue:air partition coefficient
11           10-fold lower than measured values produces exposure-phase predictions that are much closer to
12           observations, but does not replicate the elimination kinetics (Figure B-10). Re-calibration of the
13           model with upper bounds on the tissue:air partition coefficients results in predictions that are still
14           six- to sevenfold lower than empirical model prediction or observations (Figure B-12 and
15           Figure B-13). Exploration of the model space using an assumption of first-order metabolism
16           (valid for the 50 ppm inhalation exposure) showed that an adequate fit to the exposure  and
17           elimination data can be achieved only when unrealistically low values are assumed for the slowly
18           perfused tissue:air partition coefficient (Figure B-16). Artificially low values for the other
19           tissue:air partition coefficients are not expected to improve the model fit, as these parameters are
20           shown in the sensitivity analysis to exert less influence on blood 1,4-dioxane than VmaxC and Km.
21           In the absence of actual measurements for the human slowly perfused tissue:air partition
22           coefficient, high uncertainty exists for this model parameter value.  Differences in the ability of rat
23           and human blood to bind 1,4-dioxane may contribute to the difference in Vd. However, this is
24           expected to be evident in very different values for rat and human blood:air partition coefficients,
25           which is not the case (Table B-l). Therefore, some  other, as yet unknown, modification to model
26           structure may be necessary.

27           The results of U.S. EPA model evaluation were confirmed by other investigators (Sweeney et al..
28           2008). Sweeney et al. (2008) concluded that the available PBPK model with refinements resulted
29           in an under-prediction of human blood levels for 1,4-dioxane by six- to seven fold. It is
30           anticipated that the high uncertainty in predictions of the PBPK model for 1,4-dioxane would not
31           result in a more accurate derivation of human health toxicity values.

32           Because it is unknown whether the parent or the metabolite is the toxic moiety, analyses were not
33           conducted to adjust the experimental doses on the basis of metabolic saturation.

34           The discussion of Sweeney et al. (2008) was expanded in the main document in Section 3.5.3. In
35           the absence of evidence to the contrary, the Agency cannot discount the human blood kinetic data
36           published by Young et al. (1977). Even though the  PBPK model provided satisfactory  fits to the
37           rodent kinetic data, it was not used to extrapolate from high dose to low dose in the animal
38           because an internal dose metric was not identified and external doses were utilized in derivation
39           of the toxicity values.

                                                                                                        A-8
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           KLC was implemented by the U.S. EPA during the evaluation of the model and should have been
 2           described as a clearance constant (zero-order rate constant) with units of L/hr/kg070. These
 3           corrections have been made in the document; however, this does not impact the model predictions
 4           because it was in reference to the terminology used to describe this constant.

 5           The reviewer is correct that RMEX is the rate of metabolism of  1,4-dioxane per unit time;
 6           however an amount of 1,4-dioxane metabolized was not calculated in the Reitz et al. (1990)
 7           model code. Thus, AM is the amount of the metabolite (i.e., HEAA) in the body rather than the
 8           amount metabolized of 1,4-dioxane.  RAM was published by Reitz et al. (1990) as equation 2 for
 9           the change  in the amount of metabolite in the body per unit time. AMEX is the amount of the
10           metabolite  excreted in the urine. While the variables used are confusing, the code describes the
11           metabolism of 1,4-dioxane as published in the manuscripts. The comments in the model code
12           were updated to make this description more clear (Appendix B).

13    5.  Please comment on the selection of the uncertainty factors applied to the POD for the derivation of
14       the RfD. For instance, are they scientifically justified and transparently and objectively described in
15       the document?  If changes to the selected uncertainty factors are proposed, please identify and provide
16       a rationale(s). Please comment specifically on the following uncertainty factors:

17         •   An interspecies uncertainty factor of 10 was used to account for uncertainties in extrapolating
18             from laboratory animals to humans because a PBPK model to  support interspecies extrapolation
19             was not suitable.

20         •   An intraspecies (human variability) uncertainty factor of 10 was applied in deriving the RfD
21             because the available information on the variability in human response to 1,4-dioxane is
22             considered insufficient to move away from the default uncertainty factor of 10.

23         •   A database uncertainty factor of 3 was used to account for lack of adequate reproductive
24             toxicity data for 1,4-dioxane, and in particular absence of a multigeneration reproductive
25             toxicity study. Has the rationale for the selection of these uncertainty factors been transparently
26             and objectively described in the document? Please comment on whether the application of these
27             uncertainty factors has been scientifically justified.

28           Comment.  One reviewer noted the uncertainty factors appear to  be the standard default choices
29           and had no alternatives to suggest.

30             o   Five  reviewers agreed that the  use of an uncertainty factor of 10 for the interspecies
31                 extrapolation is fully supportable. One reviewer suggested using BW3/4 scaling rather than
32                 an uncertainty factor of 10 for  animal to human extrapolation. Along the same lines, one
33                 reviewer suggested a steady-state quantitative analysis to determine the importance of
34                 pulmonary clearance and hepatic clearance and stated that if hepatic clearance scales to
35                 body surface and pulmonary clearance is negligible, then an adjusted uncertainty factor
36                 based on body surface scaling would be more appropriate.
                                                                                                         A-9
                                      DRAFT - DO NOT CITE OR QUOTE

-------
 1             o   Seven reviewers stated that the uncertainty factor of 10 for interindividual variability
 2                 (intraspecies) is fully supportable.

 3             o   Six reviewers commented that the uncertainty factor of 3 for database deficiencies is fully
 4                justifiable. One reviewer suggested adding text to clearly articulate the science policy for
 5                the use of a factor of 3 for database deficiencies.

 6           Response. The preferred approach to interspecies scaling is the use of a PBPK model; however,
 7           the PBPK models available for 1,4-dioxane are not suitable for use in this health assessment as
 8           outlined elsewhere. Another approach that has been commonly implemented in the cancer
 9           assessments is the use of body weight scaling based on body surface area (BW3/4 scaling). It is not
10           standard practice to apply BW3/4 scaling in noncancer assessments at this time. The current
11           default approach used by the Agency when PBPK models are not available for extrapolation is
12           the application of an UFA of 10, which was implemented in this assessment.

13           The absence of a multigenerational reproductive study is why the uncertainty factor for database
14           deficiencies (UFD) was retained; however, it was reduced from 10 to 3. In the text in Section
15           5.1.3 text was included to clearly state that because of the absence of a multigenerational
16           reproductive study for 1,4-dioxane an uncertainty factor of 3 was used for database deficiencies.
17           No other changes regarding the use of the uncertainty factors were made to the document.
      A.1.3    Carcinogenicity of 1,4-dioxane
18    1.  Under the EPA's 2005 Guidelines for Carcinogen Risk Assessment
19       (www.epa.gov/iris/backgr-d.htm), the Agency concluded that 1,4-dioxane is likely to be carcinogenic
20       to humans. Please comment on the cancer weight of evidence characterization. Has the scientific
21       justification for the weight of evidence descriptor been sufficiently, transparently and objectively
22       described? Do the available data for both liver tumors  in rats and mice and nasal, mammary, and
23       peritoneal tumors in rats support the conclusion that 1,4-dioxane is a likely human carcinogen?

24           Comment. All reviewers agreed with the Agency's conclusion that 1,4-dioxane is "likely to be
25           carcinogenic to humans". However, two reviewers also thought 1,4-dioxane could be categorized
26           as a potential human carcinogen, since low-dose environmental exposures would be unlikely to
27           result in cancer. One reviewer also suggested providing  a brief recapitulation of the guidance
28           provided by the 2005 Guidelines for Carcinogen Risk Assessment regarding classification of a
29           compound as likely to be carcinogenic to humans and how a chemical falls into this category.

30           Response. The document includes a weight-of-evidence approach to categorize the carcinogenic
31           potential of 1,4-dioxane. This was included in Section 4.7.1 based upon U.S. EPA's Guidelines
32           for Carcinogen Risk Assessment (U.S. EPA. 2005a). 1,4-Dioxane can be described as likely to be
33           carcinogenic to humans based on  evidence of liver carcinogenicity in several 2-year bioassays
34           conducted in three strains of rats, two strains of mice,  and in guinea pigs. Additionally, tumors in
35           other organs and tissues have been observed in rats due to exposure to  1,4-dioxane.

                                                                                                       A-10
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    2.  Evidence indicating the mode of action of carcinogenicity of 1,4-dioxane was considered. Several
 2       hypothesized MOAs were evaluated within the Toxicological Review and EPA reached the
 3       conclusion that a MOA(s) could not be supported for any tumor types observed in animal models.
 4       Please comment on whether the weight of the scientific evidence supports this conclusion. Please
 5       comment on whether the rationale for this conclusion has been transparently and objectively
 6       described. Please comment on data available for 1,4-dioxane that may provide significant biological
 7       support for a MOA beyond what has been described in the Toxicological Review. Considerations
 8       should include the  scientific support regarding the plausibility for the hypothesized MOA(s), and the
 9       characterization of uncertainty regarding the MOA(s).

10           Comment. Three reviewers commented that the weight of evidence clearly supported the
11           conclusion that a mode of action could not be identified for any of the tumor sites. One reviewer
12           commented that there is inadequate evidence to support a specific MOA with any confidence and
13           low-dose linear extrapolation is necessary; this reviewer also pointed out that EPA should not rule
14           out a metabolite as the toxic moiety.

15           One reviewer stated this was outside of his/her area of expertise but indicated that the discussion
16           was too superficial and suggested adding statements as to what the Agency would consider
17           essential information to make a determination about a MOA.

18           Two reviewers commented that even though the MOA for 1,4-dioxane is not clear there is
19           substantial evidence that the MOA is non-genotoxic. One of these reviewers also suggested that a
20           nonlinear cancer risk assessment model should be utilized.

21           One reviewer suggested adding more text to the summary statement to fully reflect the available
22           MOA information which should be tied to the conclusion and choice  of an extrapolation model.

23           Response. The Agency agrees with the reviewer not to rule out a toxic metabolite as the toxic
24           moiety. In Section 5.5.1.2 text is included relating that there is not enough information to
25           determine whether the parent compound, its metabolite(s), or a combination is responsible for the
26           observed toxicities following exposure to 1,4-dioxane.

27           It is not feasible to describe the exact data that would be necessary to conclude that a particular
28           MOA was operating to induce the tumors observed following 1,4-dioxane exposure. In general,
29           the data would fit the general criteria described in the U.S. EPA's Guidelines for Carcinogen Risk
30           Assessment (U.S. EPA. 2005a). For 1,4-dioxane, several MOA hypotheses have  been proposed
31           and are explored for the observed liver tumors in Section 4.7.3. This analysis represents the extent
32           to which data could provide support for any particular MOA.

33           One reviewer suggested that the evidence indicating that 1,4-dioxane is not genotoxic supports a
34           nonlinear approach to low-dose extrapolation. In accordance with the U.S. EPA's Guidelines for
35           Carcinogen Risk Assessment (U.S. EPA, 2005a), the absence of evidence  for genotoxicity does
36           not invoke the  use of nonlinear low-dose extrapolation, nor does it define  a MOA. A nonlinear
37           low-dose extrapolation can be utilized when a MOA supporting a nonlinear dose response is
                                                                                                       A-ll
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           identified. For 1,4-dioxane this is not the case; a cancer MOA for any of the tumor types observed
 2           in animal models has not been elucidated. Therefore, as concluded in the Toxicological Review,
 3           the application of a nonlinear low-dose extrapolation approach was not supported.

 4           Additional text has been added to Section Error! Reference source not found, to relay the fact
 5           that several reviewers recommended that the MOA data support the use of a nonlinear
 6           extrapolation approach to estimate human carcinogenic risk associated with exposure to
 7           1,4-dioxane and that such an approach should be presented in the Toxicological Review.
 8           Additional text has also been added to the summary statement in Section 6.2.3 stating that the
 9           weight of evidence is inadequate to establish a MOA(s) by which 1,4-dioxane induces peritoneal,
10           mammary, or nasal tumors in rats and liver tumors in rats and mice (see Section 4.7.3 for a more
11           detailed discussion of 1,4-dioxane 's hypothesized MOAs).

12    3.  A two-year drinking water cancer bioassay (JBRC. 1998) was selected as the principal study for the
13       development of an oral slope factor (OSF). Please comment on the appropriateness of the selection of
14       the principal study. Has the rationale for this choice been transparently and objectively described?

15           Comment. Seven reviewers agreed with the choice of the JBRC (1998) study as the principal
16           study for the development of an OSF. However, two reviewers that agreed with the choice of
17           JBRC (1998) also commented on the description and evaluation of the study. One reviewer
18           commented the evaluation of the study should be separated from the evaluation/selection of
19           endpoints within the study. The other reviewer suggested that details on the following aspects
20           should be added to improve transparency of the study:  (1) rationale for selection of doses; (2)
21           temporal information on body weight for individual treatment groups;  (3) temporal information
22           on mortality rates; and (4) dosing details.

23           One reviewer thought that the complete rationale for selection of the JBRC (1998) study was not
24           provided because there was no indication of whether the  study was conducted under GLP
25           conditions, and the study was not peer reviewed or published. This reviewer noted the NCI
26           (1978) study was not appropriate for use, but that the Kociba et al. (1974) study may have
27           resulted in a lower POD had they employed both sexes of mice and combined benign and
28           malignant tumors.

29           Response. Since the External Peer Review draft of the Toxicological Review of 1,4-Dioxane was
30           released (U.S. EPA. 2009b). the cancer portion of the study conducted by the JBRC laboratory
31           was published in the peer-reviewed literature as Kano et al. (2009). This manuscript was
32           reviewed by EPA. EPA determined that the data published by Kano et al.  (2009) should be
33           included in the assessment of 1,4-dioxane for several reasons: (1) while the JBRC (1998) was a
34           detailed laboratory report, it was not peer-reviewed; (2) the JBRC improved the diagnosis of pre-
35           and neoplastic lesions in the liver according to the current diagnostic criteria and submitted the
36           manuscript based on this updated data; (3) the Kano et al. (2009) peer-reviewed manuscript
37           included additional information such as body weight growth curves and means and standard
38           deviations of estimated dose for both rats and mice of both sexes. Thus, the Toxicological Review
                                                                                                     A-12
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           was updated to reflect the inclusion of the data from Kano et al. (2009). and Appendix E was
 2           added for a clear and transparent display of the data included in the multiple reports.

 3           In response to the peer reviewers, dose information was updated throughout the assessment and
 4           are also provided in detail in Section 4.2.1.2.6, along with temporal information on body weights
 5           and mortality. Text was also added to Section 4.2.1.2.6 regarding the choice of high dose
 6           selection as included in the Kano et al. (2009) manuscript. Additional discussion regarding the
 7           mortality rates was also added to Section 5.4.1 in selection of the critical study for the oral cancer
 8           assessment. Documentation that the study was conducted in accordance with Organization for
 9           Economic Co-operation and Development (OECD) Principles of Good Laboratory Practice
10           (GLP) is provided in the manuscript (Kano et al.. 2009) and this was also added to the text in
11           Section 4.2.1.2.6.

12    4.  Combined liver tumors (adenomas and carcinomas) in female CjrBDFl mice from the JBRC  (1998)
13       study were chosen as the most sensitive species and gender for the derivation of the final OSF. Please
14       comment on the appropriateness of the selections of species and gender. Please comment on whether
15       the rationale for these selections is scientifically justified. Has the rationale for these choices been
16       transparently and objectively described?

17           Comment. Six reviewers agreed the female CjrBDFl mice should be used for the derivation of
18           the OSF. Five of these reviewers agreed with the rationale for the selection of the female
19           CjrBDFl  mouse as the most sensitive gender and  species. However, one reviewer suggested that
20           the specific rationale (i.e., that the final  OSF is determined by selecting the gender/species that
21           gives the greatest OSF value) be stated clearly in a paragraph separate from the other
22           considerations of study selection.

23           One reviewer was unsure of both the scientific justification for combining benign and malignant
24           liver tumors, as well as the  background  incidence of the observed liver tumors  in historical
25           control CjrBDFl male and female mice.

26           One reviewer commented that the scientific basis for the selection of female CjrBDFl mice was
27           unclear. This reviewer thought that the rationale for the choice of this strain/sex compared to all
28           others was not clearly articulated.

29           Response. Using the approach described in the Guidelines for Carcinogen Risk Assessment (U.S.
30           EPA. 2005a) studies were first evaluated based on their quality and suitability for inclusion in the
31           assessment. Once the studies were found to be of sufficient quality for inclusion in the
32           assessment, the dose-response analysis was performed with the goal of determining the  most
33           appropriate endpoint and species for use in the derivation of an OSF. These topics are discussed
34           in detail in Section 4.7 and  5.4.

35           Benign and malignant tumors that arise  from the same cell type (e.g., hepatocellular) may be
36           combined to more clearly identify the weight of evidence for a chemical. This is in accordance
37           with the U.S. EPA 2005 Guidelines for  Carcinogen Risk Assessment as referenced in the
                                                                                                      A-13
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           lexicological Review. In the absence of a MOA (MOA analysis described in detail in Section
 2           4.7.) for 1,4-dioxane carcinogenicity, it is not possible to determine which species may more
 3           closely resemble humans. Text in Section 5.4.4 indicates that the calculation of an OSF for
 4           1,4-dioxane is based upon the dose-response data for the most sensitive species and gender.

 5    5.  Has the scientific justification for deriving a quantitative cancer assessment been transparently and
 6       objectively described? Regarding liver cancer, a linear low-dose  extrapolation approach was utilized
 7       to derive the OSF. Please provide detailed comments on whether this approach to dose-response
 8       assessment is scientifically sound, appropriately conducted, and objectively and transparently
 9       described in the document. Please identify and provide the rationale for any alternative approaches for
10       the determination of the OSF and discuss whether such approaches are preferred to EPA's approach.

11           Comment. Four reviewers agreed with the approach for the dose-response assessment. One
12           reviewer commented that even if a nongenotoxic MOA were identified for 1,4-dioxane it may not
13           be best evaluated by threshold modeling. One reviewer commented the use of the female mouse
14           data provided an appropriate health protective and scientifically valid approach.

15           One reviewer commented that the basic adjustments and extrapolation method for derivation of
16           the OSF were clearly and adequately described, but disagreed with the linear low-dose
17           extrapolation. This reviewer suggested that the lack of certainty regarding the MOA was not a
18           sufficient cause to default to a linear extrapolation. Another reviewer commented that the
19           rationale for a linear low-dose extrapolation to derive the OSF was not clear, but may be in
20           accordance with current Agency policy in the absence of a known MOA. This reviewer also
21           commented that 1,4-dioxane appears to be non-genotoxic and nonlinear models should be tested
22           on the available data to determine if they provide a better fit  and are more appropriate.

23           One reviewer thought that the justification for a linear extrapolation was not clearly provided and
24           that a disconnect between the MOA summary and the choice of a linear extrapolation model
25           existed. In addition, this reviewer commented that the pharmacokinetic information did not
26           support the use of a linear extrapolation approach, but rather use of animal PBPK models to
27           extrapolate from high to low dose that would result in a mixture of linear and nonlinear
28           extrapolation models was warranted.

29           One reviewer suggested consideration of an integrated assessment of the cancer and noncancer
30           endpoints; however, if linear low-dose extrapolation remains the approach of choice by the
31           Agency, then the effect of choosing BMRs other than 10% was recommended to at least be
32           included in the uncertainty discussion. Using BMRs lower than  10% may allow for the
33           identification of a risk level for which the low-dose slope is 'best' estimated.

34           Response. The EPA conducted a cancer MOA analysis evaluating all of the available data for
35           1,4-dioxane. Application of the framework in the U.S. EPA Guidelines for Carcinogen Risk
36           Assessment (2005a) demonstrates that the available evidence to support any hypothesized MOA
37           for 1,4-dioxane-induced tumors does not exist. In the absence of a MOA, the U.S. EPA
38           Guidelines for Carcinogen Risk Assessment (2005a) indicate that a low dose linear extrapolation

                                                                                                      A-14
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           should be utilized for dose response analysis (see Section 5.4). Some of the potential uncertainty
 2           associated with this conclusion was characterized in Section 5.5. Note that there is no scientific
 3           basis to indicate that in the absence of evidence for genotoxicity a nonlinear low-dose
 4           extrapolation should be used. As concluded in the Toxicological Review, the application of a
 5           nonlinear low-dose extrapolation approach was not supported.

 6           With regards to the PBPK model available for 1,4-dioxane, it is clear that there currently exist
 7           deficiencies within the model and as such, the model was not utilized for interspecies
 8           extrapolation. Given the deficiencies and uncertainty in the 1,4-dioxane model it also does not
 9           provide support for a MOA.

10           Lastly, in the absence of a MOA for 1,4-dioxane carcinogenicity it is not possible to harmonize
11           the cancer and noncancer effects to assess the risk of health effects due to exposure. However, the
12           choice of the BMDLi0,which was more than 15-fold lower than the response at the  lowest dose
13           (66 mg/kg-day), was reconsidered in response to a public comment. BMDs and BMDLs were
14           calculated using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and BMDL50). A
15           BMR of 50% was used as it resulted in a BMDL closest to the response level at the lowest dose
16           tested in the bioassay.
      A.2  Public Comments

17           Comments on the Toxicological Review of 1,4-Dioxane submitted by the public are summarized
18           below in the following categories: Oral reference dose for 1,4-dioxane, carcinogenicity of
19           1,4-dioxane, PBPK modeling, and other comments.
      A.2.1    Oral reference dose (RfD) for  1,4-dioxane

20           Comment: An UF for database deficiencies is not necessary because of considerable evidence
21           showing no reproductive or developmental effects from 1,4-dioxane exposure.

22           Response: Due to the lack of a multigenerational reproductive study for 1,4-dioxane an UF of 3
23           was retained for database deficiencies. Without clear evidence showing a lack of reproductive or
24           developmental effects in a multigenerational reproductive study, there is still uncertainty in this
25           area.

26
      A.2.2    Carcinogenicity of 1,4-dioxane

27           Comment: Using liver tumors as the basis for the oral CSF is more appropriate than nasal tumors
28           (1988 IRIS assessment of 1,4-dioxane); however, the use of mouse liver tumor data is
                                                                                                     A-15
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1           inappropriate because it is inconsistent with other liver models both quantitatively and in the
 2           dose-response pattern. High mortality rates in the study are also a limitation. Liver tumor data
 3           from rats should be used instead, which represents a better animal model for 1,4-dioxane
 4           carcinogenicity assessment.

 5           Response: Even though the dose-response is different for mice and rats, the female mice were
 6           considered to be appropriate for the carcinogenicity assessment for several reasons. The female
 7           mouse liver tumors from the Kano et al. (2009) report were found to be the most sensitive species
 8           and endpoint. Section 4.2.1.2.6 was updated to include additional information on mortality rates.
 9           The majority of the animals lived past 52 weeks (only 4 females died prior to 52 weeks, 2 in each
10           the mid- and high-dose groups). The cause of death in the female mice that died between 1 and 2
11           years was attributed to liver tumors.

12           Comment: The OSF was based on the most sensitive group, Crj:BDFl mice; however BDF1
13           mice have a high background rate of liver tumors. The incidence of liver tumors in historical
14           controls for this gender/species should be considered in the assessment. Sensitivity of the test
15           species/gender as well as other criteria should be considered in the selection of the appropriate
16           study, including internal and external validity as outlined in Lewandowski and Rhomberg (2005).
17           The female Crj:BDFl mice had a low survival rate that should be considered in the selection of
18           the animal model for 1,4-dioxane carcinogenicity.

19           Response. Katagiri et al. (1998) summarized the incidence of hepatocellular adenomas and
20           carcinomas in control male and female BDF1 mice from ten 2-year bioassays at the JBRC. For
21           female mice, out of 499 control mice, the incidence rates were 4.4%  for hepatocellular adenomas
22           and 2.0% for hepatocellular carcinomas. Kano et al. (2009) reported  a 10% incidence rate for
23           hepatocellular adenomas and a 0% incidence rate for hepatocellular carcinomas in control female
24           BDF1. These incidence rates are near the historical control values and thus are appropriate for
25           consideration in this assessment. Additional text regarding these historical controls was added to
26           the study description in Section 4.2.1.2.6.

27           Comment: Low-dose linear extrapolation for the oral CSF is not appropriate nor justified by the
28           data. The weight of evidence supports a threshold (nonlinear) MOA when metabolic pathway is
29           saturated at high doses. Nonlinear extrapolations should be evaluated and presented for
30           1,4-dioxane. Oral CSFs should be derived and presented using both the BW3/4 scaling as well as
31           available PBPK models to extrapolate across species.

32           Response: The absence of evidence for genotoxicity/mutagenicity does not indicate the use of
33           nonlinear low-dose extrapolation. For 1,4-dioxane, a MOA to explain the induction of tumors
34           does not exist so the nature of the low-dose region of the dose-response is unknown. The oral
35           CSF for 1,4-dioxane was derived using BW3/4 scaling for interspecies extrapolation. The PBPK
36           and empirical models available for 1,4-dioxane were evaluated and found not to be adequate for
37           use in this assessment, described in detail in Appendix B.
                                                                                                      A-16
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Comment: The POD for the BDF1 female mouse is 15-fold lower than the lowest dose in the
 2           bioassay, thus the POD is far below the lower limit of the data and does not follow the U.S.
 3           EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA. 2005a).

 4           Response. The comment is correct that the animal BMDL10 was more than 15-fold lower than the
 5           response at the lowest dose (66 mg/kg-day) in the bioassay. BMDs and BMDLs were calculated
 6           using a BMR of 30 and 50% extra risk (BMD30, BMDL30, BMD50, and BMDL50). A BMR of
 7           50% was chosen as it resulted in a BMDL closest to the response level at the lowest dose tested in
 8           the bioassay.

 9           Comment. The geometric mean of the oral cancer slope factors (as done with B[a]P & DDT)
10           should have been used instead of relying on the female BDF1 mouse data, since a MOA could not
11           be determined for 1,4-dioxane.

12           Response. In accordance with the BMD technical guidance document (U.S. EPA. 2000a).
13           averaging tumor incidence is not  a standard or default approach. Averaging the tumor incidence
14           response diminishes the effect seen in the sensitive species/gender.

15           Comment. EPA should critically  reexamine the choice  of JBRC (1998) as the principal study
16           since it has not been published or peer-reviewed. A transcript of e-mail correspondence should be
17           provided.

18           Response. JBRC (1998) was published as conference proceedings as Yamazaki et al. (1994) and
19           recently in the peer-reviewed literature as Kano et al. (2009). Additional study information was
20           also gathered from the authors (Yamazaki. 2006) and is available upon request from the IRIS
21           Hotline. The peer-reviewed and published data from Kano et al. (2009) was incorporated into the
22           final version of the Toxicological Review ofl,4-Dioxane.

23           Comment. The WOE does not support a cancer descriptor of likely to be carcinogenic to humans
24           determination, but rather suggestive human carcinogen at the high dose levels used in rodent
25           studies seems more appropriate for the following reasons: 1) lack of conclusive human
26           epidemiological data; 2) 1,4-dioxane is not mutagenic;  and 3) evidence at high doses it would act
27           via cell proliferation MOA.

28           Response: A cancer classification of "likely, " based on evidence of liver carcinogenicity in
29           several two-year bioassays conducted in three strains of rats, two strains of mice, and in guinea
30           pigs was chosen. Also, mesotheliomas of the peritoneum, mammary, and nasal tumors have been
31           observed in rats. The Agency agrees that human epidemiological studies are inconclusive. The
32           evidence at any dose is insufficient to determine a MOA.

33
                                                                                                     A-17
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      A.2.3    PBPK  Modeling

 1           Comment. EPA should have used and considered PBPK models to derive the oral toxicity values
 2           (rat to human extrapolation) rather than relying on a default method. The draft did not consider
 3           the Sweeney et al. (2008) model. The PBPK model should be used for both noncancer and cancer
 4           dose extrapolation.

 5           Response: The Agency evaluated the Sweeney et al. (2008) publication and this was included in
 6           Appendix B of the document. Text was added to the main document in Section 3.5.2.4 and 3.5.3
 7           regarding the evaluation of Sweeney et al. (2008). This model was determined not to be
 8           appropriate for interspecies extrapolation. Additionally, see response to the external peer review
 9           panel comment B4.

10           Comment: EPA should use the modified inhalation inputs used in the Reitz et al. (1990) model
11           and the updated input parameters provided in Sweeney et al. (2008) and add a compartment for
12           the kidney

13           Response: See response to previous comment regarding evaluation of Sweeney et al.  (2008).
14           Modification of the model to add a kidney compartment is not within the scope of this
15           assessment.
      A.2.4    Other Comments

16           Comment: EPA should consider the Kasai et al. (2009; 2008) studies for inhalation and MOA
17           relevance.

18           Response: The 13 week and 2-year inhalation studies by Kasai et al. (2009; 2008) were published
19           late in the development stage of this assessment. The IRIS Program will evaluate these recently
20           published 1,4-dioxane inhalation data for the potential to derive an RfC in a separate assessment.

21           Comment: 1,4-Dioxane is not intentionally added to cosmetics and personal care products -
22           correct sentence on page 4.
23           Response: This oversight was corrected in the document.
                                                                                                    A-18
                                    DRAFT - DO NOT CITE OR QUOTE

-------
      APPENDIX   B.     EVALUATION   OF  EXISTING   PBPK
         MODE  LS   FOR   1,4-DIOXANE
      B.1   Background

 1           Several pharmacokinetic models have been developed to predict the absorption, distribution,
 2    metabolism, and elimination of 1,4-dioxane in rats and humans. Single compartment, empirical models
 3    for rats (Young et al., 1978a; 1978b) and humans (Young et al.. 1977) were developed to predict blood
 4    levels of 1,4-dioxane and urine levels of the primary metabolite, (3-hydroxyethoxy acetic acid (HEAA).
 5    Physiologically based pharmacokinetic (PBPK) models that describe the kinetics of 1,4-dioxane using
 6    biologically realistic flow rates, tissue volumes and affinities, metabolic processes, and elimination
 7    behaviors, were also developed (Fisher et al.. 1997; Leung and Paustenbach. 1990; Reitz etal.. 1990).

 8           In developing updated toxicity values for 1,4-dioxane, the available PBPK models were evaluated
 9    for their ability to predict observations made in experimental studies of rat and human exposures to
10    1,4-dioxane. The model of Reitz et al. (1990) was identified for further consideration to assist in the
11    derivation of toxicity values. Issues related to the biological plausibility of parameter values in the Reitz
12    et al. (1990) human model were identified. The model was able to predict the only available human
13    inhalation data set (Young et al.. 1977)  by increasing (i.e., doubling) parameter values for human alveolar
14    ventilation, cardiac output, and the blood:air partition coefficient above the measured values.
15    Furthermore, the measured value for the slowly perfused tissue:air partition coefficient (i.e., muscle) was
16    replaced with the measured liver value to  improve the fit. Analysis of the Young et al. (1977) human data
17    suggested that the apparent volume of distribution (Vd) for 1,4-dioxane was approximately 10-fold higher
18    in rats than humans, presumably due to species differences in tissue partitioning or other process not
19    represented in the model. Subsequent exercising of the model demonstrated that selecting a human slowly
20    perfused tissue:air partition coefficient much lower than the measured rat value resulted in better
21    agreement between model predictions of 1,4-dioxane in blood and experimental observations. Based upon
22    these observations, several model parameters (e.g., metabolism/elimination parameters) were
23    re-calibrated using biologically plausible values for flow rates and tissue:air partition coefficients.

24           This appendix describes activities conducted in the evaluation of the empirical models (Young et
25    al. (1978b; 1978a; 1977)). and re-calibration and exercising of the Reitz et al. (1990) PBPK model, and
26    evaluation of the Sweeney et al. (2008) model to determine the potential utility of the PBPK models for
27    1,4-dioxane for interspecies and route-to-route extrapolation.
      B.2  Scope

28           The scope of this effort consisted of implementation of the Young et al. (1978b; 1978a; 1977)
29    empirical rat and human models using the acslXtreme simulation software, re-calibration of the Reitz et
30    al. (1990) human PBPK model, and evaluation of model parameters published by Sweeney et al. (2008).

                                                                                                      B-l
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    Using the model descriptions and equations given in Young et al. (1978b; 1978a; 1977). model code was
 2    developed for the empirical models and executed, simulating the reported experimental conditions. The
 3    model output was then compared with the model output reported in Young et al. (1978b; 1978a; 1977).

 4           The PBPK model of Reitz et al. (1990) was re-calibrated using measured values for cardiac and
 5    alveolar flow rates and tissue:air partition coefficients. The predictions of blood and urine levels of
 6    1,4-dioxane and HEAA, respectively, from the re-calibrated model were compared with the empirical
 7    model predictions of the same dosimeters to determine whether the re-calibrated PBPK model could
 8    perform similarly to the empirical model. As part of the PBPK model evaluation, EPA performed a
 9    sensitivity analysis to identify the model parameters having the greatest influence on the primary
10    dosimeter of interest, the blood level of 1,4-dioxane. Variability data for the experimental measurements
11    of the tissue: air partition coefficients were incorporated to determine a range of model outputs bounded
12    by biologically plausible values for these parameters. Model parameters from Sweeney et al. (2008) were
13    also tested to evaluate the ability of the PBPK model to predict human data following exposure to
14    1,4-dioxane.
      B.3   Implementation of the Empirical Models  in aclsXtreme

15           The empirical models of Young et al. (1978b: 1978a: 1977) for 1,4-dioxane in rats and humans
16    were reproduced using acslXtreme, version 2.3 (Aegis Technologies, Huntsville, AL). Model code files
17    were developed using the equations described in the published papers. Additional files containing
18    experiment-specific information (i.e., BWs, exposure levels, and duration) were also generated.
      B.3.1    Model Descriptions

19           The empirical model of Young et al. (1978b: 1978a) for 1,4-dioxane in rats is shown in
20    Figure B-l. This is a single-compartment model that describes the absorption and metabolism kinetics of
21    1,4-dioxane in blood and urine. No information is reported describing pulmonary absorption or
22    intravenous (i.v.) injection/infusion of 1,4-dioxane. The metabolism of 1,4-dioxane and subsequent
23    appearance of HEAA is described by Michaelis-Menten kinetics governed by a maximum rate (Vmax,
24    ug/mL-hour) and affinity constant (Km, ug/mL) . Both 1,4-dioxane and HEAA are eliminated via the
25    first-order elimination rate constants, ke and kme, respectively (hour"1) by which 35% of 1,4-dioxane and
26    100% of HEAA appear in the urine,  while 65% of 1,4-dioxane is exhaled. Blood concentration of
27    1,4-dioxane is determined by dividing the instantaneous amount of 1,4-dioxane in blood by a Vd of 301
28    mL/kg BW.
                                                                                                    B-2
                                    DRAFT - DO NOT CITE OR QUOTE

-------
        Inhalation (kTNH)
            i.v. admin
dt
                                                      - - k, x Diox,,,
                                    dt
                                           „   „.
                                          K+DlOX
                                                    -~km
                                                  tody
                                                                          k x Diox
z+
Exhaled (65%)

  Urine (35%)
                                                                          k  xHEAA
                                                                                      Urine
             Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
             Figure B-l  Schematic representation of empirical model for 1,4-dioxane in rats.
 1           Figure B-2 illustrates the empirical model for 1,4-dioxane in humans as described in Young et al.
 2    (1977). Like the rat model, the human model predicts blood 1,4-dioxane and urinary 1,4-dioxane and
 3    HEAA levels using a single-compartment structure. However, the metabolism of 1,4-dioxane to HEAA in
 4    humans is modeled as a first-order process governed by a rate constant, KM (hour"1). Urinary deposition of
 5    1,4-dioxane and HEAA is described using the first order rate constants, ke(dlox) and kme(HEAA), respectively.
 6    Pulmonary absorption is described by a fixed rate of 76.1 mg/hour (k^). Blood concentrations of
 7    1,4-dioxane and HEAA are calculated as instantaneous amount (mg) divided by Vd(diOX) or
 8    respectively (104 and 480 mL/kg BW, respectively).
          Inhalation
                                            Dioxane
                                              HEAA
                                           "d(HEAA) X
                                                                           e (diox)
                                                                            (HEAA)
                                                       Urine
                                                    Cumulative
                                                    Dioxane and
                                                      HEAA
             Source: Reprinted with permission of Taylor & Francis, Young et al. (1977).
             Figure B-2  Schematic representation of empirical model for 1,4-dioxane in humans.
      B.3.2    Modifications to the Empirical  Models
10           Several modifications were made to the empirical models. The need for the modifications arose in
11    some cases from incomplete reporting of the Young et al. (1978b; 1978a; 1977) studies and in other cases
12    from the desire to add capabilities to the models to assist in the derivation of toxicity values.
13           For the rat model, no information was given by Young et al. (1978b: 1978a) regarding the
14    parameterization of pulmonary absorption (or exhalation) or i.v. administration of 1,4-dioxane. Therefore,
15    additional parameters were added to simulate these processes in the simplest form. To replicate

                                      DRAFT - DO NOT CITE OR QUOTE
                                                                    B-3

-------
 1    1,4-dioxane inhalation, a first-order rate constant, k^n (hour"1), was introduced. kINH was multiplied by
 2    the inhalation concentration and the respiratory minute volume of 0.238 L/minute (Young et al.. 1978b:
 3    1978a). The value for kWH was estimated by optimization against the blood time course data of Young et
 4    al. (1978b: 1978a). Intravenous (i.v.) administration was modeled as instantaneous appearance of the full
 5    dose at the start of the simulation. Rat urinary HEAA data were reported by Young et al. (1978b; 1978a)
 6    in units of concentration. To simulate urinary HEAA concentration, an estimate of urine volume was
 7    required. Since observed urinary volumes were not reported by Young et al. (1978b;  1978a). a standard
 8    rat urine production rate of 0.00145 L/hour was used.

 9           For humans, Young et al. (1977) used a fixed  1,4-dioxane inhalation uptake rate of 76.1 mg/hour,
10    which corresponded to observations during a 50 ppm exposure. In order to facilitate user-specified
11    inhalation concentrations, pulmonary absorption was modeled. The modeling was performed identically
12    to the rat model, but using a human minute volume of 7 L/minute. Urinary HEAA data were reported by
13    Young et al. (1977) as a cumulative amount (mg) of HEAA. Cumulative amount of HEAA in the urine is
14    readily calculated from the rate of transfer of HEAA from plasma to urine, so no modification was
15    necessary to simulate this dose metric for humans.

16           Neither empirical model of Young et al. (1978b: 1978a:  1977) described oral uptake of
17    1,4-dioxane. Adequate data to estimate  oral absorption parameters are not available for either rats or
18    humans; therefore, neither empirical model was modified to include oral uptake.
      B.3.3    Results

19           The acslXtreme implementation of the Young et al. (1978b; 1978a) rat empirical model simulates
20    the 1,4-dioxane blood levels from the i.v. experiments identically to the model output reported in the
21    published paper (Figure B-3). However, the acslXtreme version predicts urinary HEAA concentrations in
22    rats that are approximately threefold lower and reach a maximum sooner than the predicted levels
23    reported in the paper (Figure B-4). These discrepancies may be due, at least in part, to the reliance in the
24    acslXtreme implementation on a constant, standard, urine volume rather than experimental measurements,
25    which may have been different from the assumed value and may have varied over time. Unreported
26    model parameters (e.g., lag times for appearance of excreted HEAA in bladder urine) may also contribute
27    to the discrepancy.
                                                                                                      B-4
                                    DRAFT - DO NOT CITE OR QUOTE

-------
       Observations and predictions of 1,4-dioxane in rat blood
             following 3 to 1000 mg/kg IV injection
                          30    40     50
                          Time(hrs)
        Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
        Figure B-3  Output of 1,4-dioxane blood level data from the acslXtreme
                    implementation (left) and published (right) empirical rat model
                    simulations of i.v. administration experiments.
         Observations and predictions of HEAA in rat urine
            following 10 or 1000 mg/kg IV injection
                       20      30
                         Time (hrs)
        Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
        Figure B-4  Output of HEAA urine level data from acslXtreme implementation
                    (left) and published (right) empirical rat model simulations of i.v.
                    administration experiments.

        The Young et al. (1978b: 1978a) report did not provide model predictions for the 50-ppm
inhalation experiment. However, the acslXtreme implementation produces blood 1,4-dioxane predictions
that are quite similar to the reported observations (Figure B-5). As with the urine data from the i.v.
                                                                                                      B-5
                                 DRAFT - DO NOT CITE OR QUOTE

-------
1    experiment, the acslXtreme-predicted urinary HEAA concentrations are approximately threefold lower
2    than the observations, presumably for the same reasons discussed above for the i.v. predictions.
           Observations and predictions of 1,4-dioxane in rat blood
              following a 6 hour 50 ppm inhalation exposure
—
D
acsl version - Youn
(1978a, b) empirical
Young etal. (1978a
observations
etal.
model
b)
                             Time (hrs)
Observations and predictions of HEAA in rat urine
 following a 6 hour 50 ppm inhalation exposure
                                                             25.0
                                                             20.0-

                                                           3 1
                                                           Ul
                                                             5.0-
                                                             0.0


	 acsl version -Young et
al. (1978a, b) empirical
model
d Young etal. (1978a, b)
observations

[
X


[

/^~
                                                                       10
                                                                              20     30
                                                                               Time (hrs)
                                                                                             40
                                                                                                    50
             Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
             Figure B-5 acslXtreme predictions of blood 1,4-dioxane and urine HEAA levels
                         from the empirical rat model simulations of a 6-hour, 50-ppm inhalation
                         exposure.
3            Inhalation data for a single exposure level (50 ppm) are available for humans. The acslXtreme
4    predictions of the blood 1,4-dioxane observations are identical to the predictions reported in Young et al.
5    (1977) (Figure B-6). Limited blood HEAA data were  reported, and the specimen analysis was highly
6    problematic (e.g., an analytical interference was sometimes present from which HEAA could not be
7    separated). For this reason, Young et al. (1977) did not compare predictions of the blood HEAA data to
8    observations in their manuscript.
            Observations and predictions of 1,4-dioxane in human
            blood following a 6 hour 50 ppm inhalation exposure
D
acsl version - Young et
al (1977) empirical model
observed
                         5           10
                             Time (hrs)
             Source: Reprinted with permission of Taylor & Francis, Young et al. (1978b: 1978a).
                                       DRAFT - DO NOT CITE OR QUOTE
                                                                                                              B-6

-------
             Figure B-6  Output of 1,4-dioxane blood level data from the acslXtreme
                         implementation (left) and published (right) empirical human model
                         simulations of a 6-hour, 50-ppm inhalation exposure.
 1           Data for cumulative urinary HEAA amounts are provided in Young et al. (1977). and no
 2   analytical problems for these data were reported. Nevertheless, model predictions for urinary HEAA were
 3   not presented in the manuscript. The acslXtreme prediction of the HEAA kinetics profile is similar to the
 4   observations, although predicted values are approximately 1.5- to 2-fold lower than the observed values
 5   (Figure B-7). Unlike urinary HEAA observations in the rat, human observations were reported as
 6   cumulative amount produced, negating the need for urine volume data.  Therefore, discrepancies between
 7   model predictions and experimental observations for humans cannot be attributed to uncertainties in urine
 8   volumes in the subjects. Further evaluation of the Young et al (1977) empirical model was conducted
 9   against subchronic inhalation exposure data reported by Kasai et al. (2008). In the experimental  study.
10   male and female F344 rats were exposed to 0. 100. 200. 400. 800. 1.600. 3.200. or 6.400ppm 1.4-dioxane
11   in a 13-week inhalation study. The simulations of the Young et al. (1977) model did not provide an
12   adequate fit (Figure B-8) for the measured plasma levels at each exposure level of 1.4-dioxane as reported
13   bv Kasai etal. (2008).
                                  Observations and predictions of HEAA in human urine
                                    following a 6 hour 50 ppm inhalation exposure
                                 700.0
                                 600.0
                             >> ro 500.0 -
                               |
                             II
                             2 <
                               S
                             O I
400.0

300.0

200.0

100.0
D
acsl version - Young et al
(1977) empirical model
observed
                                                   10      15
                                                    Time (hrs)
                                                                  20
                                                                         25
             Source: Reprinted with permission of Taylor & Francis, Young et al. (1977).
             Figure B-7  Observations and acslXtreme predictions of cumulative HEAA in
                         human urine following a 6-hour, 50-ppm inhalation exposure.
                                                                                                       B-7
                                     DRAFT - DO NOT CITE OR QUOTE

-------
                             3000
                             2500
                             2000
                             1500
                           •*. 1000
                             500
 +  Male Rat Data
 +  Female Rat Data
-*—Simulated Data
                                     500
                                           1000
                                                 1500   2000
                                                    Dose
                                                              2500
                                                                    3000
                                                                          3500
             Figure B-8 EPA-modified Young et al. empirical model prediction (line) of plasma
                        1,4-dioxane levels in rats following exposure to 1,4-dioxane for 13 weeks
                        compared to data from Kasai et al. (2008).
      B.3.4    Conclusions for Empirical Model Implementation

 1           The empirical models described by Young et al. (1978b; 1978a; 1977) for rats and humans were
 2    implemented using acslXtreme. The models were modified to allow for user-defined inhalation levels by
 3    addition of a first-order rate constant for pulmonary uptake of 1,4-dioxane, fitted to the inhalation data.
 4    No modifications were made for oral absorption as adequate data are not available for parameter
 5    estimation. The acslXtreme predictions of 1,4-dioxane in the blood are identical to the published
 6    predictions for simulations of 6-hour, 50-ppm inhalation exposures in rats and humans and 3 to
 7    1,000 mg/kg i.v. doses in rats (Figure B-3, Figure B-5, and Figure  B-6). However, the acslXtreme version
 8    predicts lower urinary HEAA concentrations in rats appearing earlier than either the Young et al. (1978b:
 9    1978a) model predictions or the experimental observations. The lower predicted urinary HEAA levels in
10    the acslXtreme implementation for rats is likely due to use of default values for urine  volume in the
11    absence of measured volumes. The reason for differences in time-to-peak levels is unknown, but may be
12    the result of an unreported adjustment by Young et al. (1978b: 1978a) in model parameter values.
13    Additionally, the modified Young et al. (1978b; 1978a; 1977) model failed to provide adequate fit to
14    blood data reported following subchronic inhalation of 1.4-dioxane in rats (Kasai et al.. 2008). For
15    humans, Young et al. (1977) did not report model predictions of urinary HEAA  levels. The urinary
16    HEAA levels predicted by acslXtreme were low relative to the observations. However, unlike the
17    situation in rats, these data are not dependent on unreported urine volumes (observations were reported as
18    cumulative HEAA amount rather than HEAA concentration), but reflect the model parameter values
19    reported by Young et al. (1977). Presently, there is no explanation for the lack of fit of the reported
20    urinary HEAA elimination rate constant to the observations.
                                     DRAFT - DO NOT CITE OR QUOTE
                                                                                                      B-8

-------
      B.4  Initial Recalibration  of the PBPK  Model

 1           Concern regarding adjustments made to some of the parameter values in Reitz et al. (1990)
 2    prompted a re-calibration of the Reitz et al. (1990) human PBPK model using more biologically plausible
 3    values for all measured parameter values. Reitz et al. (1990) doubled the measured physiological flows
 4    and blood:air partition coefficient and substituted the slowly-perfused tissue:air partition coefficient with
 5    the liver:air value in order to attain an adequate fit to the observations. This approach increases
 6    uncertainty in these parameter values, and in the utilization of the model for cross-species dose
 7    extrapolation. Therefore, the model was re-calibrated using parameter values that are more biologically
 8    plausible to determine whether an adequate fit of the model to the available data can be attained.
      B.4.1    Sources of Values for Flow Rates

 9           The cardiac output of 30 L/hour/kg074 (Table B-l) reported by Reitz et al. (Reitz etal.. 1990) is
10    approximately double the mean resting value of 14 L/hour/kg0 74 reported in the widely accepted
11    compendium of Brown et al. (1997). Resting cardiac output was reported to be 5.2 L/minute (or 14
12    L/hour/kg074), while strenuous exercise resulted in a flow of 9.9 L/minute (or 26 L/hour/kg074) (Brown et
13    al.. 1997). Brown et al. (1997) also cite the ICRP (1975) as having a mean respiratory minute volume of
14    7.5 L/minute, which results in an alveolar ventilation rate of 5 L/minute (assuming 33% lung dead space),
15    or 13 L/minute/kg0 74. Again, this is roughly half the value of 30 L/hour/kg0 74 employed for this parameter
16    by Reitz et al. (1990). Young et al. (1977) reported that the human subjects exposed to 50 ppm for 6 hours
17    were resting inside a walk-in exposure chamber. Thus, use of cardiac output and alveolar ventilation rates
18    of 30 L/hour/kg0 74 is not consistent with the experimental conditions being simulated.
                                                                                                      B-9
                                     DRAFT - DO NOT CITE OR QUOTE

-------
      Table B-1   Human PBPK model parameter values for 1,4-dioxane
Parameter
Reitz et al. (1990)
Leung and
Paustenbach (1990)
Sweeney et al.
(2008)
EPAC
Physiological Flows
Cardiac output (QCCf
Alveolar ventilation (QPCf
30
30
—
—
—
—
17.0
17.7
Partition Coefficients (PCs)
Blood:air(PB)
Fatair(PFA)
Liverair (PLA)
Rapidly perfused tissue:air (PRA)
Slowly perfused tissue:air (PSA)
3,650
851
1,557
1,557
1,557
1,825 ±94
851 ± 118
1,557±114
—
997 ± 254
1,666 ±287
—
1,862 ±739"
—
1,348 ±290°
1,850
851
1,557
1,557
166
Metabolic Constants
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)d
Metabolic affinity constant (Km)e
HEAA urinary elimination rate
constant (kme)f
6.35
3.00
0.56
-
—
-
-
—
-
5.49
9.8
0.44
      aL/hour/kg BWU '4
      ""Measurement for rat tissue
      ""Biologically plausible values utilized by EPA in this assessment
      dmg/hour/kg BW075
      emg/L
      'hour'1
 2           Examination of the experimental data of Young et al. (1977) yields an estimated alveolar
 3   ventilation to be 7 L/minute (or 16 L/hour/kg°74) for volunteers having a mean BW of 84 kg. This rate is
 4   based on the Young et al. (1977) estimate of 76.1 mg/hour for 1,4-dioxane uptake. Based on these
 5   findings, the cardiac output and alveolar ventilation rates of 17.0 and 17.7 L/hour/kg°74 were biologically
 6   plausible for the experimental subjects. These rate estimates are based on calculations made using
 7   empirical data and are consistent with standard human values and the experimental conditions (i.e.,
 8   subject exertion level) reported by Young et al. (1977). Therefore, these flow values were chosen for the
 9   model re-calibration.
      B.4.2    Sources of Values  for  Partition Coefficients
10           Two data sources are available for the tissue:air equilibrium partition coefficients for 1,4-dioxane:
11    Leung and Paustenbach (1990) and Sweeney et al. (2008). Both investigators report mean values and
12    standard deviations for human blood:air, rat liverair, and rat muscle:air (e.g., slowly perfused tissue:air),
13    while Leung and Paustenbach et al. (1990) also reported values for rat fatair (Table B-1).
      B.4.3    Calibration Method
14           The PBPK model was twice re-calibrated using the physiological flow values suggested values
15    (current EPA assessment, see Table B-1) and the partition coefficients of Leung and Paustenbach (1990)

                                                                                                      B-10
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    and Sweeney et al. (2008) separately. For each calibration, the metabolic parameters VmaxC and Km, were
 2    simultaneously fit (using the parameter estimation tool provided in the acslXtreme software) to the output
 3    of 1,4-dioxane blood concentrations generated by the acslXtreme implementation of the Young et al.
 4    (1977) empirical human model for a 6 hour, 50 ppm inhalation exposure. Subsequently, the HEAA
 5    urinary elimination rate constant, kme, was fitted to the urine HEAA predictions from the empirical model.
 6    The empirical model predictions, rather than experimental observations, were used to provide a more
 7    robust data set for model fitting, as the empirical model simulation provided 240 data points (one
 8    prediction every 0.1 hour) compared with hourly experimental observations, and to avoid introducing
 9    error by calibrating the model to data digitally captured from Young et al. (1977).
      B.4.4    Results

10           Results of the model re-calibration are provided in Table B-2. The re-calibrated values for VmaxC
11    and kme associated with the Leung and Paustenbach (1990) or Sweeney et al. (2008) tissue:air partition
12    coefficients are very similar. However, the fitted value for Km using the Sweeney et al. (2008) partition
13    coefficients is far lower (0.0001 mg/L) than that resulting from use of the Leung and Paustenbach (1990)
14    partition coefficients (2.5 mg/L). This appears to be due to the higher slowly perfused tissue:air partition
15    coefficient determined by Sweeney et al. (2008) (1,348 vs. 997), resulting in a higher apparent Vd than if
16    the Leung and  Paustenbach (1990) value is used. Thus, the optimization algorithm selects a low Km,
17    artificially saturating metabolism in an effort to drive predicted blood 1,4-dioxane levels closer to the
18    empirical model output. Saturation of metabolism during a 50 ppm inhalation exposure is inconsistent
19    with the observed kinetics.
      Table B-2   PBPK metabolic and elimination parameter values resulting from re-calibration of the
                  human model using alternative values for physiological flow rates3 and tissue:air
                  partition coefficients
     	Source of Partition Coefficients	Leung and Paustenbach (1990)    Sweeney et al. (2008)
      Maximum rate for 1,4-dioxane metabolism (Vmaxc)	16.9	20.36
      Metabolic affinity constant (Km)c	2.5	0.0001
      HEAA urinary elimination rate constant (kme)                         0.18	0.17
      aCardiac output = 17.0 L/hour/kg BWU'4, alveolar ventilation = 17.7 L/hour/kg BWU '4
      bmg/hour/kg BW075
      >g/L
20           Plots of predicted and experimentally observed blood 1,4-dioxane and urinary HEAA levels are
21    shown in Figure B-9. Neither re-calibration resulted in an adequate fit to the blood 1,4-dioxane data from
22    the empirical model output or the experimental observations. Re-calibration using either the Leung and
23    Paustenbach (1990) or Sweeney et al. (2008) partition coefficients resulted in blood 1,4-dioxane
24    predictions that were at least 10-fold lower than empirical model predictions or observations.
                                                                                                        B-ll
                                      DRAFT - DO NOT CITE OR QUOTE

-------
          Observations and predictions of 1,4-dioxane in human blood from
            a 6-hour, 50 ppm exposure: VmaxC and Km fit while using PC
                 values from Leung and Paustenbach (1990)
          100.0
           0.1
                                      10    12    14
                              Time (hrs)
Observations and predictions of HEAA in human urine from a
6-hour, 50 ppm exposure: kme fit while using PC values from
          Leung and Paustenbach (1990)
         5       10       15

                 Time (hrs)
             Figure B-9  Predicted and observed blood 1,4-dioxane concentrations (left) and
                          urinary HEAA levels (right) following re-calibration of the human
                          PBPK model with tissue:air partition coefficient values.

             Source: Reprinted with permission of Elsevier, Ltd., Leung and Paustenbach (1990).

             The refitted values for kme resulted in HEAA levels in urine that were very similar to the

     empirical model output (compare Figure B-7, Figure B-9, and Figure B-10), which was not surprising,

     given the fitting of a single parameter to the data.
Observations and predictions of 1 ,4-dioxane in human blood from
a 6-hour, 50 ppm exposure: VmaxC and Km fit while using PC
values from Sweeney et al. (2008)
Blood 1,4-Dioxane Concentration
(mg/L)
p i
-»• o o c

PPPIY" H' t H
n observed
_ T jf_ J. JP • • • • empirical predricted
•II * T
i? *'|.
l.r
f" \ M--
\ I i-

0 5 10 15
Time (hrs)
                                                              Observations and predictions of HEAA in human urine from a
                                                              6-hour, 50 ppm exposure: kme fit while using PC values from
                                                                          Sweeney etal. (2008)
                                                             700.0
             Figure B-10 Predicted and observed blood 1,4-dioxane concentrations (left) and
                          urinary HEAA levels (right) following re-calibration of the human
                          PBPK model with tissue:air partition coefficient values.

             Source: Reprinted with permission of Oxford Journals, Sweeney et al. (2008).

4            Outputs of the blood 1,4-dioxane and urinary HEAA levels using the suggested (Table B-2)

5    parameters are shown in Figure B-l 1. These outputs rely on a very low value for the slowly perfused

6    tissue:air partition coefficient (166) that is six- to eightfold lower than the measured values reported in

7    Leung and Paustenbach (1990) and Sweeney et al. (2008). and 10-fold lower than the value used by Reitz

8    et al. (1990). While the predicted maximum blood 1,4-dioxane levels are much closer to the observations,
                                                                                                              B-12
                                       DRAFT - DO NOT CITE OR QUOTE

-------
 1    the elimination kinetics are markedly different, producing higher predicted elimination rates compared to
 2    observations during the post-exposure phase of the experiment.
          Observations and predictions of 1,4-dioxane in           Observations and predictions of HEAA in human
          human blood from a 6-hour, 50 ppm exposure:               urine from a 6-hour, 50 ppm exposure:
                EPA parameter estimates used                        EPA parameter estimates used
             Figure B-ll Predicted and observed blood 1,4-dioxane concentrations (left) and
                        urinary HEAA levels (right) using EPA estimated biologically plausible
                        parameters (Table B-l).
      B.4.5    Conclusions for PBPK Model Implementation

 3           Re-calibration of the human PBPK model was performed using experiment-specific values for
 4    cardiac output and alveolar ventilation (Young et al.. 1977) and measured mean tissue:air 1,4-dioxane
 5    partition coefficients reported by Leung and Paustenbach (1990) or Sweeney et al. (2008). The resulting
 6    predictions of 1,4-dioxane in blood following a 6-hour, 50-ppm inhalation exposure were 10-fold (or
 7    more) lower than either the observations or the empirical model predictions, while the predictions of
 8    urinary HEAA by the PBPK and empirical models were similar to each other, but lower than observed
 9    values (Figure B-9 and Figure B-10). Output from the model using biologically plausible parameter
10    values (Table B-l), Figure B-l 1 shows that application of a value for the  slowly perfused tissue:air
11    partition coefficient, which is 10-fold lower than the measured value reported by Leung and Paustenbach
12    (1990). results in closer agreement of the predictions to observations during the exposure phase, but not
13    during the elimination phase. Thus, model re-calibration using experiment-specific flow rates and mean
14    measured partition coefficients does not result in an adequate fit of the PBPK model to the available data.

15           The Sweeney et al. (2008) PBPK model consisted of compartments for fat, liver, slowly perfused.
16    and other well perfused tissues. Lung and stomach compartments were used to describe the route of
17    exposure, and an overall volume of distribution compartment was used for calculation of urinary
18    excretion levels of 1.4-dioxane and its metabolite. HEAA. Metabolic constants (VmaxC and Km) for the
19    rat PBPK model were derived by optimization data from an i.v. exposure of 1.000 mg/kg data (Young et
20    al.. 1978a; 1978b) for induced metabolism. For uninduced metabolism data generated by_ i.v. exposures to
21    3. 10. 30. and  100 mg/kg were used (Young et al.. 1978a: 1978bV Data generated from the 300 mg/kg i.v.
                                                                                                     B-13
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    exposure was not used to estimate VmaxC and Km. The best fitting values for VmaxC to estimate the
 2    blood data from the Young et al. (1978b: 1978a) study using the Sweeney et al. (2008) model resulted in
 3    VmaxC values of 12.7. 10.8. 7.4 mg/kg-hr; suggesting a gradual dose dependent increase in metabolic
 4    rate with dose. These estimates were for a range of doses between 3 and 1.000 mg/kg i.v. dose. Although
 5    the Sweeney et al. (2008) model utilized two values for VmaxC (induced and uninduced). the PBPK
 6    model does not include dose-dependent function description of the change of Vmax for i.v. doses between
 7    100 and 1.000 mg/kg. PBPK model outputs were compared with other data not used in fitting model
 8    parameters by visual inspection. The model predictions gave adequate match to the 1.4-dioxane
 9    exhalation data after a 1.000 mg/kg i.v. dose. 1.4-Dioxane exhalation was overpredicted by a factor of
10    about 3 for the 10 mg/kg i.v. dose.  Similarly, the simulations of exhaled 1.4-dioxane after oral dosing
11    were  adequate at 1.000 mg/kg. and 100 mg/kg (within 50%). but poor at 10 mg/kg (model overpredicted
12    by a factor of five). The fit of the model to the human data (Young et al.. 1977) was also problematic
13    (Sweeney et al.. 2008). Using physiological parameters of Brown et al. (1997) and measured partitioning
14    parameters (Sweeney et al..  2008; Leung and Paustenbach. 1990) with no metabolism, measured blood
15    1.4-dioxane  concentrations reported by Young et al. (1977)  could not be achieved unless the estimated
16    exposure concentration was increased from 53 to 100 ppm. Inclusion of any metabolism necessarily
17    decreased predicted blood concentrations. If estimated metabolism rates were used with the reported
18    exposure concentration, urinary metabolite excretion was underpredicted (Sweeney et al.. 2008). Thus.
19    the models were inadequate to use for rat to human extrapolation.
      B.4.6    Sensitivity Analysis
20           A sensitivity analysis of the Reitz et al. (1990) model was performed to determine which PBPK
21    model parameters exert the greatest influence on the outcome of dosimeters of interest—in this case, the
22    concentration of 1,4-dioxane in blood. Knowledge of model sensitivity is useful for guiding the choice of
23    parameter values to minimize model uncertainty.
      B.4.7    Method
24           A univariate sensitivity analysis was performed on all of the model parameters for two endpoints:
25    blood 1,4-dioxane concentrations after 1 and 4 hours of exposure. These time points were chosen to
26    assess sensitivity during periods of rapid uptake (1 hour) and as the model approached steady state
27    (4 hours) for blood 1,4-dioxane. Model parameters were perturbated 1% above and below nominal values
28    and sensitivity coefficients were calculated as follows:
                                                    Ax
                                                                                                     B-14
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1   where x is the model parameter, f(x) is the output variable, Ax is the perturbation of the parameter from
 2   the nominal value, and f (x) is the sensitivity coefficient. The sensitivity coefficients were scaled to the
 3   nominal value of x and f(x) to eliminate the potential effect of units of expression. As a result, the
 4   sensitivity coefficient is a measure of the proportional change in the blood 1,4-dioxane concentration
 5   produced by a proportional change in the parameter value, with a maximum value of 1.
      B.4.8   Results

 6           The sensitivity coefficients for the seven most influential model parameters at 1 and 4 hours of
 7    exposure are shown in Figure B-12. The three parameters with the highest sensitivity coefficients in
 8    descending order are alveolar ventilation (QPC) (1.0), the blood:air partition coefficient (PB) (0.65), and
 9    the slowly perfused tissue:air partition coefficient (PSA) (0.51). Not surprisingly, these were the
10    parameters that were doubled or given surrogate values in the Reitz et al. (1990) model in order to
11    achieve an adequate fit to the data. Because of the large  influence of these parameters on the model, it is
12    important to assign values to these parameters in which high confidence is placed, in order to reduce
13    model uncertainty.
0.

QPC
-
PB
-
fe PSA
X
0
a QSC
ro
°- QCC
-
V -

Km

Sensitivity Coefficients: CV - 1hr
31 0.10 1.



I

I

|

I

I

I

DO































0.

QPC
-
PB

0 PSA
X
0
E y
ro
^ ^

PRA

QSC

Sensitivity Coefficients: CV - 4 hr
31 0.10







I

I

I

I

1.















DO















             Figure B-12 The highest seven sensitivity coefficients (and associated parameters)
                        for blood 1,4-dioxane concentrations (CV) at 1 (left) and 4 (right) hours
                        of a 50-ppm inhalation exposure.
      B.5   PBPK Model Exercises  Using Biologically  Plausible  Paramter
            Boundaries

14           The PBPK model includes numerous physiological parameters whose values are typically taken
15    from experimental observations. In particular, values for the flow rates (cardiac output and alveolar
16    ventilation) and tissue:air partition coefficients (i.e., mean and standard deviations) are available from
17    multiple sources as means and variances. The PBPK model was exercised by varying the partition
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                    B-15

-------
 1    coefficients over the range of biological plausibility (parameter mean ± 2 standard deviations),
 2    re-calibrating the metabolism and elimination parameters, and exploring the resulting range of blood
 3    1,4-dioxane concentration time course predictions. Cardiac output and alveolar ventilation were not
 4    varied because the experiment-specific values used did not include any measure of inter-individual
 5    variation.
      B.5.1    Observations Regarding  the Volume of Distribution
 6           Young et al. (1978b: 1978a) used experimental observations to estimate a Vd for 1,4-dioxane in
 7    rats of 301 mL, or 1,204 mL/kg BW. For humans, the Vd was estimated to be 104 mL/kg BW (Young et
 8    al.. 1977). It is possible that a very large volume of the slowly perfused tissues in the body of rats and
 9    humans may be a significant contributor to the estimated 10-fold difference in distribution volumes for
10    the two species. This raises doubt regarding the appropriateness of using the measured rat slowly perfused
11    tissue:air partition coefficient as a surrogate values for humans in the PBPK model.
      B.5.2    Defining Boundaries for Parameter Values

12           Given the possible 10-fold species differences in the apparent Vd for 1,4-dioxane in rats and
13    humans, boundary values for the partition coefficients were chosen to exercise the PBPK model across its
14    performance range to either minimize or maximize the simulated Vd. This was accomplished by defining
15    biologically plausible values for the partition coefficients as the mean ± 2 standard deviations of the
16    measured values. Thus, to minimize the simulated Vd for 1,4-dioxane, the selected blood:air partition
17    coefficient was chosen to be the mean + 2 standard deviations, while all of the other tissue:air partition
18    coefficients were chosen to be the mean - 2 standard deviations. This created conditions that would
19    sequester 1,4-dioxane in the blood, away from other tissues. To maximize the simulated 1,4-dioxane Vd,
20    the opposite selections were made: blood and other tissue:air partition coefficients were chosen as the
21    mean - 2 standard deviations and mean + 2 standard deviations, respectively. Subsequently, VmaxC, Km,
22    and kme were optimized to the empirical model output data as described in Section B.4.3. This procedure
23    was performed for both the Leung and Paustenbach (1990) and Sweeney et al. (2008) partition
24    coefficients (Table B-l). The two predicted time courses resulting from the re-calibrated model with
25    partition coefficients chosen to minimize or maximize the 1,4-dioxane Vd represent the range of model
26    performance as bounded by biologically plausible parameter values.
      B.5.3    Results

27           The predicted time courses for a 6-hour, 50-ppm inhalation exposure for the re-calibrated human
28    PBPK model with mean (central tendency) and ± 2 standard deviations from the mean values for partition
29    coefficients are shown in Figure B-13 for the Leung and Paustenbach (1990) values and Figure B-14 for
30    the Sweeney et al. (2008) values. The resulting fitted values for VmaxC, Km, and kme, are given in

                                                                                                      B-16
                                     DRAFT - DO NOT CITE OR QUOTE

-------
1    Table B-3. By bounding the tissue:air partition coefficients with upper and lower limits on biologically
2    plausible values from Leung and Paustenbach (1990) or Sweeney et al. (2008). the model predictions are
3    still at least six- to sevenfold lower than either the empirical model output or the experimental
4    observations. The range of possible urinary HEAA predictions brackets the prediction of the empirical
5    model, but this agreement is not surprising, as the cumulative rate of excretion depends only on the rate of
6    metabolism of 1,4-dioxane,  and not on the apparent Vd for 1,4-dioxane. These data show that the PBPK
7    model cannot adequately reproduce the predictions of blood 1,4-dioxane concentrations of the Young et
8    al. (1977) human empirical model or the experimental observations when constrained by biologically
9    plausible values for physiological flow rates and tissue:air partition coefficients.
               1,4-Dioxane In human blood from a 6-hour, SO ppm
                             exposure
          1000
           100



„
	 Ltfung and PAualafibacfr ; 1 9iO| PC - UCi.
	 Lsuna and p*us!«ib*ai 1 1 SBQi PC - c*Mra
.....u. Lwng,jrwJP»Lalefibsch,19«JlPC-LCL
Q Voting *i il. J19TTJ obtervfiuOrt Hate
«n "°" \
' \
»° t.

'/" ~\\
/ \s
\\ " N
-• Vs





                                                            Cumulative HEAA In human urine from a 6-hour, 50 ppm
                                                                            exposure
                                                           700
                                                           600
                                                        £• 0)500
                                                        S E
S I
11
I?
D     2     4
                             Time(hrs)
                                                           400

                                                           300

                                                           200

                                                           100

                                                            0
                                                                   ~ VpungelBl 11977j •mpmol mwfct
                                                                  	Leung nwpausKfibaerj) 1990) PC-LJCL
                                                                  	 L«unfl«wIPaiistqtiBcml$»(PC -Cwlra
                                                                  ...	 LcL*ig srd PauileflofltM 13901 PC - L'JL
                                                                   D  Vwietgaal 119*71 olservabonflaifl
                                                                            10      15
                                                                             Time (hrs)
                                                                                                 25
             Source: Reprinted with permission of Elsevier, Ltd., Leung and Paustenbach (1990)
             Figure B-13 Comparisons of the range of PBPK model predictions from upper and
                         lower boundaries on partition coefficients with empirical model
                         predictions and experimental observations for blood 1,4-dioxane
                         concentrations (left) and urinary HEAA levels (right) from a 6-hour,
                         50-ppm inhalation exposure.
                                                                                                          B-17
                                      DRAFT - DO NOT CITE OR QUOTE

-------

1,4-Dioxane In human blood from a 6-hour, 50 ppm
100.0 -
c
o
o 10.0-
Blood 1,4-Dioxane Cor
(mg/L)
P T-^
-^ O
exposure

,n-n- l a
*i v_
/n *
1
'/ \
! \

	 Young et al. (1977) empirical model
	 Sweeney et al. (2008) PC - UCL
	 Sweeney et al. (2008) PC - Central
	 Sweeney et al. (2008) PC - LCL
Young et al. (1977) observation
data
ta
\




] 5 10 15
Time (hrs)






Cumulative HEAA in human urine from a 6-hour, 50 ppm
exposure
700-
600 -
£•3500-
Cumulative Ur
HEAA Amount
O O O O O

D n
D
D
D

jyr — — Young et al. (1977) empirical me
j%r 	 Sweeney et al. (2008) PC - UC
ff/
jf* 	 Sweeney et al. (2008) PC - LCL
^/^* n Young et al. (1977) observation



del
tral
data
3 5 10 15 20 25
Time (hrs)
            Source: Reprinted with permission of Oxford Journals, Sweeney et al. (2008): Used with permission of Taylor & Francis,
     Young et al. (1977).
            Figure B-14 Comparisons of the range of PBPK model predictions from upper and
                       lower boundaries on partition coefficients with empirical model
                       predictions and experimental observations for blood 1,4-dioxane
                       concentrations (left) and urinary HEAA levels (right) from a 6-hour,
                       50-ppm inhalation exposure.
     Table B-3   PBPK metabolic and elimination parameter values resulting from recalibration of the
                human model using biologically plausible values for physiological flow rates3 and
                selected upper and lower boundary values for tissue:air partition coefficients
Source of partition
coefficients
Maximum rate for 1,4-dioxane
metabolism (Vmaxc)b
Metabolic dissociation constant
(Km)c
HEAA urinary elimination rate
constant (kme)d
aCardiac output = 17.0 L/hour/kg BWU'4'
bmg/hour/kg BW075
cmg/L
dhour-1
Leung and Paustenbach (1990)
For maximal Vd
14.95
5.97
0.18
alveolar ventilation =
For minimal Vd
18.24
0.0001
0.17
17.7 L/hour/kg BWU'4
Sweeney et al. (2008)
For maximal Vd For minimal Vd
17.37 21.75
4.88 0.0001
0.26 0.19

     B.5.4    Alternative Model Parameterization

1           Since the PBPK model does not predict the experimental observations of Young et al. (1977)
2    when parameterized by biologically plausible values, an exercise was performed to explore alternative
3    parameters and values capable of producing an adequate fit of the data. Since the metabolism of
4    1,4-dioxane appears to be linear in humans for a 50-ppm exposure (Young et al., 1977). the parameters
5    VmaxC and Km were replaced by a zero-order, non-saturable metabolism rate constant, kLC. This rate
6    constant was fitted to the experimental blood 1,4-dioxane data using partition coefficient values of
                                   DRAFT - DO NOT CITE OR QUOTE
                                                                                                   B-18

-------
1    Sweeney et al. (2008) to minimize the Vd (i.e., maximize the blood 1,4-dioxane levels). The resulting
2    model predictions are shown in Figure B-15. As before, the maximum blood 1,4-dioxane levels were
3    approximately sevenfold lower than the observed values.
16.1
c 14 1
O
1 12.1
c
u
e 10.1
o
o 3-
g o> 81
I"
*-" 4 1
•o
O
a 2.1
01
1
1 ,4-Dioxane in human blood from a 6-hour, 50 ppm
exposure: ktc(3.0) fitted to all observations
a
>**"*«
^ »
to •
* P

-------
                                     1,4-Dioxane in human blood from a 6-hour, 50 ppm
                                     exposure: kLC(0.1) fitted 1 to 6-hour observations
                                                        	Young et al. (1977) empirical
                                                            model
                                                   6      8     10
                                                    Time (hrs)
                                                                    12
                                                                         14
             Figure B-16   Predictions of blood 1,4-dioxane concentration following calibration
                         of a zero-order metabolism rate constant, kLC, to only the exposure
                         phase of the experimental data.
 1           Finally, the model was re-calibrated by simultaneously fitting kLC and the slowly perfused
 2    tissue:air partition coefficient to the experimental data with no bounds on possible values (except that
 3    they be non-zero). The fitted slowly perfused tissue:air partition coefficient was an extremely low (and
 4    biologically unlikely) value of 0.0001.  The resulting model predictions, however, were closer to the
 5    observations than even the empirical model predictions (Figure B-17). These exercises show that better
 6    fits to the observed blood 1,4-dioxane kinetics are achieved only when parameter values are adjusted in a
 7    way that corresponds to a substantial decrease in apparent Vd of 1,4-dioxane in the human, relative to the
 8    rat (e.g., decreasing the slowly perfused tissue:air partition coefficient to extremely low values, relative to
 9    observations). Downward adjustment of the elimination parameters (e.g., decreasing kLC) increases the
10    predicted blood concentrations of 1,4-dioxane, achieving better agreement with observations during the
11    exposure phase of the experiment; however, it results in unacceptably slow elimination kinetics, relative
12    to observations following cessation of exposure. These observations suggest that some other process not
13    captured in the present PBPK model structure is responsible for the species differences in 1,4-dioxane Vd
14    and the inability to reproduce the human experimental inhalation data with biologically plausible
15    parameter value s.
                                      DRAFT - DO NOT CITE OR QUOTE
                                                                                                         B-20

-------
                                      1,4-Dioxane in human blood from a 6-hour,
                                                50 ppm exposure
                                                     rj Young etal. (1977)
                                                        observation data
                                                           10
                                                                 12
                                                                      14
                                                  Time (hrs)
             Figure B-17 Predictions of blood 1,4-dioxane concentration following simultaneous
                        calibration of a zero-order metabolism rate constant, kLC, and slowly
                        perfused tissue:air partition coefficient to the experimental data.
      B.6  Conclusions

 1           The rat and human empirical models of Young et al. (1978b: 1978a: 1977) were successfully
 2    implemented in acslXtreme and perform identically to the models reported in the published papers
 3    (Figures 3-3 through 3-6), with the exception of the lower predicted HEAA concentrations and early
 4    appearance of the peak HEAA levels in rat urine. The early appearance of peak HEAA levels cannot
 5    presently be explained, but may result from manipulations of kme or other parameters by Young et al.
 6    (1978b;  1978a) that were not reported. The lower predictions of HEAA levels are likely due to reliance on
 7    a standard urine volume production rate in the absence  of measured (but unreported) urine volumes.
 8    While the human urinary HEAA predictions were lower than observations, this is due to parameter fitting
 9    of Young et al. (1977). No model output was published in Young et al. (1977) for comparison. The
10    empirical models were modified to allow for user-defined inhalation exposure levels. However, no
11    modifications were made to model oral exposures because adequate data to parameterize such
12    modifications do not exist for rats or humans. The inhalation Young et aL (1977) model failed to provide
13    adequate fits to the subchronic exposure plasma levels of 1.4-dioxane in rats using the data from the Kasai
14    et al. (2008) study.

15           Several procedures were applied to the human PBPK model to determine if an adequate fit of the
16    model to the empirical model output or experimental observations could be attained using biologically
17    plausible values for the model parameters. The re-calibrated model predictions for blood 1,4-dioxane
18    levels do not come within 10-fold of the experimental values using measured tissue: air partition
19    coefficients from Leung and Paustenbach (1990) or Sweeney et al. (2008) (Figure B-9 and Figure B-10).
20    Use of a slowly perfused tissue: air partition coefficient 10-fold lower than measured values produces
                                                                                                     B-21
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1    exposure-phase predictions that are much closer to observations, but does not replicate the elimination
 2    kinetics (Figure B-l 1). Re-calibration of the model with upper bounds on the tissue:air partition
 3    coefficients results in predictions that are still six- to sevenfold lower than empirical model prediction or
 4    observations (Figure B-l 3 and Figure B-l 4). Exploration of the model space using an assumption of
 5    first-order metabolism (valid for the 50-ppm inhalation exposure) showed that an adequate fit to the
 6    exposure and elimination data can be achieved only when unrealistically low values are assumed for the
 7    slowly perfused tissue:air partition coefficient (Figure B-17). Artificially low values for the other
 8    tissue:air partition coefficients are not expected to improve the model fit, because the sensitivity analysis
 9    to exert less influence on blood 1,4-dioxane than VmaxC and Km. This suggests that the model structure is
10    insufficient to capture the apparent 10-fold species difference in the blood 1,4-dioxane Vd between rats
11    and humans. In the absence of actual measurements for the human slowly perfused tissue:air partition
12    coefficient, high uncertainty exists for this model parameter value. Differences in the ability of rat and
13    human blood to bind  1,4-dioxane may contribute to the difference in Vd. However, this is expected to be
14    evident in very different values for rat and human blood:air partition coefficients, which is not the case
15    (Table B-l). Therefore, some other, as yet unknown, modification to model structure may be necessary.
16    Sweeney et al. (2008) PBPK model provided an overall improvement on previous models: however, the
17    Sweeney et al. (2008) inhalation model  predictions of animal and human data were problematic.
      B.7  aclsXtreme  Code for the Young et al. Empricial  Model for
            1,4-Dioxane in  Rats

18    PROGRAM: Young (1978b) rat.csl
19    !	
20    ! Created by Michael Lumpkin, Syracuse Research Corporation, 08/06
21    ! This program implements the 1-compartment empirical model for 1,4-dioxane
22    ! in rats, developed by Young et al. (1978a: 1978bX Program was modified to run
23    ! in ACSL Xtreme and to include user-defined i.v. and inhalation concentrations
24    !(MLumpkin, 08/06)
9S    I
Z,,J    ! ————————————————————————————————————————————————————————
26
27    INITIAL
28
29    !*****Timing and Integration Commands*****
30    ALGORITHM IALG=2  IGear integration algorithm for stiff systems
31    IMERROR %%%%=0.01 IRelative error for lead in plasma
32    NSTEPS NSTP=1000    INumber of integration steps per communication interval
3 3    CINTERVAL CINT=0.1 ! Communication interval
34    CONSTANT TSTART=0.!Start of simulation (hr)
35    CONSTANT TSTOP=70. !End of simulation (hr)
36
37    !*****MODEL PARAMETERS*****
38    CONSTANT BW=0.215  Body weight (kg)
39    CONSTANT MINVOL=0.238 Irespiratory minute volume (L/min) estimated from Young et al. (1978)
40    CONSTANT IVDOSE = 0. !IV dose (mg/kg)!
41    CONSTANT CONC = 0. ! inhalation concentration (ppm)
42
43    CONSTANT MOLWT=88.105 !mol weight of 1,4-dioxane
44    CONSTANT TCHNG=6.0        Exposure pulse 1 width (hr)
45    CONSTANT TDUR=24.0 Exposure duration (hr)

                                                                                                    B-22
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1   CONSTANT TCHNG2=120.0 Exposure pulse 2 width (hr)
 2   CONSTANT TDUR2=168.0 Exposure duration 2 (hr)
 3
 4   CONSTANT Vmax=4.008 !(mcg/mL/hr)
 5   CONSTANT Km=6.308 !(mcg/mL)
 6   CONSTANT Kinh=0.43   Ipulmonary absorption constant (/hr)
 7   CONSTANT Ke=0.0149 !(/hr)
 8   CONSTANT Kme=0.2593 !(/hr)
 9   CONSTANT Vd=0.3014 !(L)
10
11   rV=IVDOSE*BW
12   AmDIOXi=IV
13
14   END                 ! Of Initial Section
15
16   DYNAMIC
17   DERIVATIVE
18
19   ! * * * Dioxane inhalation concentration * * *
20   CIZONE=PULSE(0.0, TDUR, TCHNG) * PULSE(0.0, TDUR2, TCHNG2)
21           IFirst pulse is hours/day, second pulse is hours/week
22   CI=CONC*CIZONE*MOLWT/24450.    !Convert to mg/L
23
24   I * * * Dioxane metabolism/1 st order elimination * * *
25   dAmDIOX=(Kinh*CI*(MINVOL*60))-((Vmax*(AmDIOX))/(Km+(AmDIOX)))-(Ke*(AmDIOX))
26   AmDIOX=INTEG(dAmDIOX,AmDIOXi)
27   ConcDIOX=AmDIOX/Vd Iplasma dioxane concentration (mcg/mL)
28   AUCDIOX=INTEG(ConcDIOX,0) Iplasma dioxane AUC
29
30   ! * * * HE AA production and 1 st order metabolism * * *
31   dAmHEAA=((Vmax*(AmDIOX))/(Km+(AmDIOX)))-(Kme*(AmHEAA))
32   AmHEAA=INTEG(dAmHEAA,0.)
33   ConcHEAA=AmHEAA/Vd Iplasma HEAA concentration
34
35   ! * * * 1 st order dioxane elimination to urine * * *
36   dAmDIOXu=(Ke*(AmDIOX))*0.35
37   AmDIOXu=INTEG(dAmDIOXu,0.)
38   ConcDIOXu=Ke*AmDIOX*0.35/1.45e-3  lurine production approx 1.45e-3 L/hr in SD rats
39
40   ! * * * 1 st order dioxane exhaled * * *
41   dAmDIOXex=(Ke*(AmDIOX))*0.65
42   AmDIOXex=INTEG(dAmDIOXex,0.)
43
44   ! * * * 1 st order HEAA elimination to urine * * *
45   dAniHEAAu=(Kme*(AmHEAA))
46   AmHEAAu=INTEG(dAmHEAAu,0.)
47   ConcHEAAu=Kme*AmHEAA/1.45e-3 lurine production approx 1.45e-3 L/hr in SD rats
48
49   END lof Derivative Section
50
51   DISCRETE
52
53   END    lof Discrete Section
54
55   TERMT (T .GT. TSTOP)
56
57   END lof Dynamic Section
58
59   TERMINAL

                                                                                                 B-23
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2   END   ! of Terminal Section
 3
 4   END   ! of Program
      B.8   aclsXtreme Code for the Young etal. Empricial  Model for
            1,4-Dioxane in Humans

 5    PROGRAM: Young (1977) human.cs!
 6    !	
 7    ! Created by Michael Lumpkin, Syracuse Research Corporation, 01/06
 8    ! This program implements the 1-compartment model for 1,4-dioxane in humans,
 9    ! developed by Young et al., (1977). Program was modified to run
10    ! in acslXtreme (MLumpkin, 08/06)
11    !	
12
13    INITIAL
14
15    !*****Timing and Integration Commands*****
16    ALGORITHM IALG=2  ! Gear integration algorithm for stiff systems
17    IMERROR %%%%=0.01 IRelative error for lead in plasma
18    NSTEPS NSTP=1000    INumber of integration steps per communication interval
19    CINTERVALCINT=0.1 !Communication interval
20    CONSTANT TSTART=0.!Start of simulation (hr)
21    CONSTANT TSTOP= 120.       !End of simulation (hr)
22
23    |*****MODEL PARAMETERS*****
24    ! CONST ANT D AT A=l  lOptimization dataset
25    CONSTANT MOLWT=88.105 !mol weight for 1,4-dioxane
26    CONSTANT DOSE=0.   IDose (mg/kg
27    CONSTANT CONC=0.  llnhalation concentration (ppm)
28    CONSTANT BW=84.1   Body weight (kg)
29    CONSTANT MINVOL=7.0      Ipulmonary minute volume (L/min)
30    CONSTANT F=1.0             IFraction of dose absorbed
31    CONST ANT kinh= 1.06  IRate constant for inhalation (mg/hr); optimized by MHL
32    CONSTANT ke=0.0033  IRate constant for dioxane elim to urine (hr-1)
33    CONSTANT km=0.7096 IRate constant for metab of dioxane to HEAA (hr-1)
34    CONSTANT kme=0.2593 IRate constant for transfer from rapid to blood (hr-1)
35    CONSTANT VdDkg=0.104      I Volume of distribution for dioxane (L/kg BW)
36
37    CONSTANT VdMkg=0.480      I Volume of distribution for HEAA (L/kg BW)
38    CONSTANT OStart=0.   I Time of first oral dose (hr)
39    CONSTANT OPeriod= 120.       I Oral Dose pulse period (hr)
40    CONSTANT OWidth= 1. I Width (gavage/drink time) of oral dose (hr)
41
42    CONSTANT IStart=0.    ITime of inhalation onset (hr)
43    CONSTANT IPeriod=120.       llnhalation pulse period (hr)
44    CONSTANT IWidth=6.  I Width (duration) of inhalation exposure (hr)
45
46    END                 I Of Initial Section
47
48    DYNAMIC
49
50    DERIVATIVE
51    !****VARIABLES and DEFINED VALUES*****
52    VdD=BW*VdDkg      I Volume of distribution for dioxane

                                                                                                B-24
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1   VdM=BW*VdMkg      ! Volume of distribution for HEAA
 2
 3   InhalePulse=PULSE(IStart,IPeriod,IWidth)
 4   Inhale=CONC*InhalePulse*MOLWT/24450.      !Convert to mg/L
 5
 6   !*****DIFFERENTIAL EQUATIONS FOR COMPARTMENTS****
 7
 8   ! *** Dioxane in the body (plasma) ***
 9   dAMTbD=(Kinh*Inhale*(MINVOL*60))-(AMTbD*km)-(AMTbD*ke)
10   AMTbD=INTEG(dAMTbD,0.)
11   CbD=AMTbD/VdD
12   AUCbD=INTEG(CbD,0)
13
14   i * * * HEAA in the body (plasma) * * *
15   dAMTbM=AMTbD*km-AMTbM*kme
16   AMTbM=INTEG(dAMTbM,0.)
17   CbM=AMTbM/VdM
18
19   ! * * * Cumulative Dioxane in the urine * * *
20   dAMTuD=(AMTbD*ke)
21   AMTuD=INTEG(dAMTuD,0.)
22
23   ! *** Cumulative HEAA in the urine ***
24   dAMTuM=(AMTbM*kme)
25   AMTuM=INTEG(dAMTuM,0.)
26
27   END   ! Of Derivative Section
28
29   DISCRETE
30
31   END                 !of Discrete Section
32
3 3   TERMT (T . GT. TSTOP)
34
35   END   ! Of Dynamic Section
36
37   TERMINAL
38
39   END                 ! of Terminal Section
40
41   END                 ! of Program
     B.9   aclsXtreme Code for the Reitz et al. PBPK Model  For 1,4-Dioxane

42   (Reitz etal. 1990)
43   PROGRAM: DIOXANE.CSL (Used in Risk Estimation Procedures)
44    ! Added a venous blood compartment and 1 st order elim of metab.'
45    Mass Balance Checked OK for Inhal, IV, Oral, and Water RHR'
46    IDefined Dose Surrogates for Risk Assessment 01/04/89'
47    Modified the Inhal Route to use PULSE for exposure conditions'
48    Modifications by GLDiamond, Aug2004, marked as !**
49    !
50    Metabolism of dioxane modified by MLumpkin, Oct2006, to include 1st order
51    lor saturable kinetics. For 1st order, set VmaxC=0; for M-Menten, set K1C=0.
52    !
53   INITIAL


                                                                                               B-25
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    INTEGER I
 3    1=1
 4    ! ARRAY TDATA(20) ! CONSTANT TDATA=999, 19*1.0E-6 !**
 5    CONSTANT BW = 0.40 !'Body weight (kg)'
 6    CONSTANT QPC =15.! 'Alveolar ventilation rate (1/hr)'
 7    CONSTANT QCC =15. ! 'Cardiac output (1/hr)'
 8
 9    IFlows to Tissue Compartments'
10    CONSTANT QLC = 0.25 !'Fractional blood flow to liver'
11    CONSTANT QFC = 0.05 I'Fractional blood flow to fat'
12    CONSTANT QSC = 0.18 I'Fractional blood flow to slow'
13    QRC= 1.0-(QFC + QSC + QLC)
14    CONSTANT SPDC =1.0! diffusion constant for slowly perfused tissues
15
16    ! Volumes of Tissue/Blood Compartments'
17    CONSTANT VLC = 0.04 !'Fraction liver tissue'
18    CONSTANT VFC = 0.07 I'Fraction fat tissue'
19    CONSTANT VRC = 0.05 I'Fraction Rapidly Perf tissue'
20    CONSTANT VBC = 0.05 I'Fraction as Blood'
21    VSC = 0.91 - (VLC + VFC + VRC + VBC)
22
23    IPartition Coefficients'
24    CONSTANT PLA = 1557.  I'Liver/air partition coefficient'
25    CONSTANT PFA =851.1 'Fat/air partition coefficient'
26    CONSTANT PSA = 2065.  I'Muscle/air (Slow Perf) partition'
27    CONSTANT PRA = 1557.  I'Richly perfused tissue/air partition'
28    CONSTANT PB = 1850. I'Blood/air partition coefficient'
29
30    I Other Compound Specific Parameters'
31    CONSTANT MW = 88.1 I'Molecular weight (g/mol)'
32    CONSTANT KLC = 12.0 I temp zero-order metab constant
33    CONSTANT VMAXC = 13.8 I'Maximum Velocity of Metabol.'
34    CONSTANT KM = 29.4 I'Michaelis Menten Constant'
35    CONSTANT ORAL = 0.0 I'Oral Bolus Dose (mg/kg)'
36    CONSTANT KA = 5.0 I'Oral uptake rate (/hr)'
37    CONSTANT WATER = 0.0 I 'Cone in Water (mg/liter, ppm)'
38    CONSTANT WDOSE=0.0       I Water dose (mg/kg-day) **
39    CONSTANT IV = 0.0 1'IV dose (mg/kg)'
40    CONSTANT CONC = 0.0 I'Inhaled concentration (ppm)'
41    CONSTANT KME = 0.276 I'Urinary Elim constant for met (hr-1)'
42
43    ITiming commands'
44    CONSTANT TSTOP = 50 I 'Length of experiment (hrs)'
45    CONSTANT TCHNG = 6 I 'Length of inhalation exposure (hrs)'
46    CINTERVALCINT=0.1
47    CONSTANT WIDD=24. !**
48    CONSTANT PERD=24. !**
49    CONSTANT PERW=168.       !**
50    CONSTANT WIDW=168.       !**
51    CONSTANT DAT=0.017!**
52
53    ! Scaled parameters calculated in this  section of Program'
54    QC=QCC*BW**0.74
55          QP=QPC*BW**0.74
56    QL=QLC*QC
57          QF=QFC*QC
58          QS=QSC*QC
59          QR=QRC*QC
                                                                                                  B-26
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    VL=VLC*BW
 2          VF=VFC*BW
 3          VS=VSC*BW
 4          VR=VRC*BW
 5          VB=VBC*BW
 6    PL=PLA/PB
 7          PR=PRA/PB
 8          PS=PSA/PB
 9          PF=PFA/PB
10          KL = KLC*bw**0.7 ! Zero-order metab constant
11          VMAX = VMAXC*BW**0.7
12    DOSE = ORAL*B W ! 'Initial Amount in Stomach'
13    ABO = IV*B W  ! 'Initial Amount in Blood'
14    IDRINK = 0.102*BW**0.7*WATER/24 I'lnput from water (mg/hr)' ! **
15    !DRINKA = 0.102*BW**0.7*WATER/DAT I'Inputfrom water (mg/hr)' !**
16          DRTNKA=WDOSE*BW/DAT
17    CV = ABO/VB ! 'Initialize CV
18
19   END !'End of INITIAL'
20
21   DYNAMIC
22
23          ALGORITHM IALG = 2 ! 'Gear method for stiff systems'
24          TERMT(T .GE. TSTOP )
25          CR = AR/VR
26          CS = AS/VS
27          CF = AF/VF
28          BODY = AL + AR + AS + AF + AB+ TUMMY
29          BURDEN = AM + BODY
30          TMASS = BURDEN + AX + AMEX
31
3 2   ! Calculate the Interval Excretion Data here:'
33    !      DAX = AMEX-AMEX2
34    !      IF(DOSE .LE. 0.0 .AND. IV .LE. 0.0 ) GO TO SKIP1
35    !      PCTAX = 100*(AX - AX2)/(DOSE + IV*BW)
36    !      PCTMX = 100*(AMEX - AMEX2)/(DOSE + IV*BW)
37   !      SKIP!.. CONTINUE
38    !      IF(T .LT. TDATA(I) .OR. I .GE. 20 ) GO TO SKIP
39    !      AX2=AX
40   !      AMEX2=AMEX
41   !      1=1+1
42   !      SKIP.. CONTINUE
43
44   IDISCRETE EXPOSE
45   ! CIZONE = 1.0 ! CALL LOGD(.TRUE.) Turns on inhalation exposure?
46   !END
47   IDISCRETE CLEAR
48   ! CIZONE = 0.0 ! CALL LOGD(.TRUE.)
49   !END
50
51   DERIVATIVE
52
53   !Use Zero-Crossing Form of DISCRETE Function Here'
54   ! SCHEDULE command must be in DERIVATIVE section'
55    ! DAILY = PULSE (0.0, PERI, TCHNG)
56    ! WEEKLY = PULSE (0.0, PER2, LEN2 )
57    ! SWITCHY = DAILY * WEEKLY
58   !SCHEDULE EXPOSE .XP. SWITCHY - 0.995
59   !SCHEDULE CLEAR .XN. SWITCHY - 0.005
                                                                                             B-27
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2   DAILY=PULSE(0.0,PERD,WIDD)
 3   WEEKLY=PULSE(0.0,PERW,WIDW)
 4   SWITCHY = DAILY * WEEKLY
 5
 /T   1^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^:^IVTOfllfiGfi 1~TGT"G fOT" ^VOTlff^
 7          CI = CONC * MW / 24451.0 * SWITCHY!**
 8
 9    ! CA = Concentration in arterial blood (mg/1)'
10    CA = (QC*CV+QP*CI)/(QC+(QP/PB))
11    CX=CA/PB
12
13          DRINK=DRINKA* SWITCHY   !**
14
15    ! TUMMY = Amount in stomach'
16    RTUMMY = -KA*TUMMY
17    TUMMY = INTEG(RTUMMY,DOSE)
18    ! RAX = Rate of Elimination in Exhaled air'
19    RAX = QP*CX
20    AX = INTEG(RAX, 0.0)
21
22    ! AS = Amount in slowly perfused tissues (mg)'
23    RAS = SPDC*(CA-CVS) !now governed by diffusion-limited constant, SPDC, instead of QS
24    AS = INTEG(RAS,0.)
25    CVS = AS/(VS*PS)
26
27    ! AR = Amount in rapidly perfused tissues (mg)'
28    RAR = QR*(CA-CVR)
29    AR = INTEG(RAR,0.)
30    CVR = AR/(VR*PR)
31
32    ! AF = Amount in fat tissue (mg)'
33    RAF = QF*(CA-CVF)
34    AF = INTEG(RAF,0.)
35    CVF = AF/(VF*PF)
36
37    ! AL = Amount in liver tissue (mg)'
38    RAL = QL*(CA-CVL) - KL*CVL - VMAX*CVL/(KM+CVL) + KA*TUMMY + DRINK
39          AL = INTEG(RAL,0.)
40    CVL = AL/(VL*PL)
41
42   Metabolism comments updated by EDM on 2/1/10
43    ! AM = Amount metabolized (mg)'
44    RMEX = (KL*CVL)+(VMAX*CVL/(KM+CVL)) IRate of 1,4-dioxane metabolism
45    RAM = (KL*CVL)+(VMAX*CVL)/(KM+CVL) - KME* AM IRate of change of metabolite in body
46
47          AM = INTEG(RAM, 0.0) ! 'Amt Metabolite in body
48    CAM = AM/B W ! 'Cone Metabolite in body'
49    AMEX = INTEG(KME*AM, 0.0) I'Amt Metabolite Excreted via urine'
50
51    ! AB = Amount in Venous Blood'
52    RAB = QF*CVF + QL*CVL + QS*CVS + QR*CVR - QC*CV
53    AB = INTEG(RAB, ABO)
54    CV = AB/VB
55    AUCV = INTEG(CV, 0.0)
56
57   IPossible Dose Surrogates for Risk Assessment Defined Here'
58
59    CEX = 0.667*CX + 0.333*CI I'Conc inExhal Air'

                                                                                                 B-28
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    AVECON = PLA * (CEX+CI)/2 I'Ave Cone in Nose Tissue'
 2    AUCCON = INTEG(AVECON, 0.0) I'Area under Curve (Nose)'
 3
 4    AUCMET = INTEG(CAM, 0.0) I'Area under Curve (Metab)'
 5
 6    CL = AL/VL I'Cone Liver Tissue'
 7    AUCL = INTEG(CL, 0.0) I'Area under Curve (Liver)'
 8          AAUCL=AUCL/TIME
 9
10    I Dose Surrogates are Average Area under Time/Cone Curve per 24 hrs'
11    IF (T .GT. 0) TIME=T
12    dayS = TIME/24.0
13    NOSE = AUCCON/D AYS I 'Nasal Turbinates'
14    LIVER = AUCL/DAYS I 'Liver Tissues'
15    METAB = AUCMET/DAYS I'Stable Metabolite'
16
17    END I 'End of dynamic'
18
19    END I End of TERMINAL
20
21    END I 'End of PROGRAM
                                                                                                B-29
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     APPENDIX  C.     DETAILS   OF  BMD  ANALYSIS  FOR
         ORAL  RFD  FOR   1,4-DIOXANE
     C.1   Cortical Tubule  Degeneration

 1          All available dichotomous models in the Benchmark Dose Software (version 2.1.1) were fit to the
 2   incidence data shown in Table  C-l, for cortical tubule degeneration in male and female Osborne-Mendel
 3   rats exposed to 1,4-dioxane in the drinking water (NCI. 1978). Doses associated with a BMR of a 10%
 4   extra risk were calculated.
     Table C-1   Incidence of cortical tubule degeneration in Osborne-Mendel rats exposed to
                 1,4-dioxane in drinking water for 2 years
Males (mg/kg-day)
0 240
0/3 1a 20/3 1b
(65%)
Females (mg/kg-day)
530
27/33b
(82%)
0
0/3 1a
350
0/34
640
10/32b
(31%)
     Statistically significant trend for increased incidence by Cochran-Armitage test (p < 0.05) performed for this review.
     blncidence significantly elevated compared to control by Fisher's exact test (p < 0.05) performed for this review.
     Source: NCI (1978).

 6           As assessed by the %2 goodness-of-fit test, several models in the software provided adequate fits
 7   to the data for the incidence of cortical tubule degeneration in male and female rats (%2 p > 0.1)
 8   (Table C-2). Comparing across models, a better fit is indicated by a lower AIC value (U.S. EPA. 2000a).
 9   As assessed by Akaike's  Information Criterion (AIC), the log-probit model provided the best fit to the
10   cortical  tubule degeneration incidence data for male rats (Table C-2, Figure C-l) and could be used to
11   derive a POD of 38.5 mg/kg-day for this endpoint. The Weibull model provided the best fit to the data for
12   female rats (Table C-2, Figure C-5) and could be used to derive a POD of 452.4 mg/kg-day for this
13   endpoint. For those models that exhibit adequate fit, models with the lower AIC values are preferred.
14   Differences in AIC values of less than 1 are generally not considered important. BMDS modeling results
15   for all dichotomous models are shown in Table C-2.
                                                                                                     C-l
                                    DRAFT - DO NOT CITE OR QUOTE

-------
Table C-2   Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
            data for cortical tubule degeneration in male and female Osborne-Mendel rats (NCI.
            1978) exposed to 1,4-dioxane in drinking water
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(mg/kg-day)
BMDLio
(mg/kg-day)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
74.458
89.0147
75.6174
74.168
74.458
88.782
74.458
74.458
0.6514
0.0011
1
0.7532
0.6514
0.0011
0.6514
0.6514
0
-1.902
0
0
0
-1.784
0
0
28.80
88.48
20.85
51.41
28.80
87.10
28.80
28.80
22.27
65.84
8.59
38.53
22.27
66.32
22.27
22.27
Female
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
41.9712
43.7495
41.7501
43.7495
48.1969
43.7495
41.75
52.3035
0.945
0.9996
0.9999
0.9997
0.1443
0.9997
0.9999
0.03
0.064
0
0
0
-1.693
0
0
-2.086
524.73
617.44
591.82
584.22
399.29
596.02
596.45
306.21
437.08
471.92
447.21
436.19
297.86
456.42
452.36
189.49
a p-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
"Power restricted to > 1.
°Slope restricted to > 1.
dBetas restricted to >0.

Source:  NCI (1978).
                                 DRAFT - DO NOT CITE OR QUOTE
                                                                                                      C-2

-------
                                    LogProbit Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
      •
      I
      C
      o
      13
      ro
               0.8
               0.6
         0.4
               0.2
                                 100
                                       200
300
400
500
                                                    dose
        14:4902/01 2010

            Source: NCI (1978).
            Figure C-l  BMD Log-probit model of cortical tubule degeneration incidence data
                       for male rats exposed to 1,4-dioxane in drinking water for 2 years to
Probit Model.  (Version: 3.1;  Date:  05/16/2008)
Input Data File: C:\14DBMDS\lnp_nci_mrat_cortdeg_Lnp-BMR10-restrict.(d)
Gnuplot Plotting File: C:\14DBMDS\lnp_nci_mrat_cortdeg_Lnp-BMR10-restrict.plt
                                               Mon  Feb  01  14:49:17  2010

 BMDS Model Run

The form of the probability function is:

 P[response] = Background +  (1-Background)  *  CumNorm(Intercept+Slope*Log(Dose)),

 where CumNorm(.) is the cumulative normal  distribution function

 Dependent variable = Effect
 Independent variable = Dose
 Slope parameter is restricted  as  slope  >=  1

 Total number of observations = 3
 Total number of records with missing values  = 0
 Maximum number of iterations = 250
 Relative Function Convergence  has  been  set to:  le-008
 Parameter Convergence has been set to:  le-008
 User has chosen the log transformed model
                                                                                                C-3
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Default Initial  (and Specified) Parameter Values
 2    background = 0
 3    intercept = -5.14038
 4    slope = 1
 5
 6
 7   Asymptotic Correlation Matrix of Parameter Estimates
 8    (*** The model parameter(s) -background -slope have been estimated at a boundary
 9   point, or have been specified by the user, and do not appear in the correlation
10   matrix)
11
12    intercept
13    intercept 1
14
15
16    Parameter Estimates
17
18    95.0% Wald Confidence Interval
19   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
20   background 0 NA
21    intercept -5.22131 0.172682 -5.55976 -4.88286
22    slope 1 NA
23
24   NA - Indicates that this parameter has hit a bound implied by some ineguality
25   constraint and thus has no standard error.
26
27
28
29    Analysis of Deviance Table
30
31    Model Log(likelihood) # Param's Deviance Test d.f. P-value
32    Full model -35.8087 3
33    Fitted model -36.084 1 0.550629 2 0.7593
34    Reduced model -65.8437 1 60.07 2 <.0001
35
36    AIC: 74.168
37
38
39    Goodness of Fit
40    Scaled
41    Dose Est._Prob. Expected Observed Size Residual
42    	
43    0.0000 0.0000 0.000 0.000 31 0.000
44    240.0000 0.6023 18.672 20.000 31 0.487
45    530.0000 0.8535 28.166 27.000 33 -0.574
46
47    ChiA2 = 0.57 d.f. = 2 P-value = 0.7532
48
49
50    Benchmark Dose Computation
51   Specified effect =0.1
52   Risk Type = Extra risk
53   Confidence level = 0.95
54    BMD = 51.4062
55    BMDL = 38.5284
                                                                                              C-4
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                       Weibull Model with 0.95 Confidence Level
        T3
        =1

 Total number of  observations = 3
 Total number of  records with missing values = 0
 Maximum number of iterations = 250
 Relative Function Convergence has been set  to: le-008
 Parameter Convergence has been set  to: le-008
 Default Initial  (and  Specified)  Parameter Values
                                                                                                C-5
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Background = 0.015625
 2    Slope = 1.55776e-010
 3    Power = 3.33993
 4
 5
 6    Asymptotic Correlation Matrix  of  Parameter  Estimates
 7    (*** The model parameter(s)  -Background  -Power  have  been estimated at a boundary
 8   point, or have been specified by the user, and  do  not  appear in the correlation
 9   matrix)
10
11    Slope
12    Slope -1.$
13
14    Parameter Estimates
15    95.0% Wald Confidence Interval
16   Variable Estimate Std. Err.  Lower  Conf.  Limit Upper  Conf.  Limit
17   Background 0 NA
18    Slope 1.15454e-051 1.#QNAN  1.#QNAN 1.#QNAN
19    Power 18 NA
20
21   NA - Indicates that this parameter has hit a bound implied by some ineguality
22   constraint and thus has no standard error.
23
24    Analysis of Deviance Table
25
26    Model Log(likelihood) # Param's Deviance Test  d.f.  P-value
27    Full model -19.8748  3
28    Fitted model -19.875 1 0.000487728 2  0.9998
29    Reduced model -32.1871 1 24.6247  2 <.0001
30
31    AIC: 41.75
32
33
34    Goodness of Fit
35    Scaled
36    Dose Est._Prob. Expected Observed Size  Residual
37    	
38    0.0000 0.0000 0.000  0.000 31 0.000
39    350.0000 0.0000 0.000 0.000 34 -0.016
40    640.0000 0.3125 9.999 10.000 32 0.000
41
42    ChiA2 = 0.00 d.f. =  2 P-value  = 0.9999
43
44
45    Benchmark Dose Computation
46   Specified effect =0.1
47   Risk Type = Extra risk
48   Confidence level =0.95
49    BMD = 596.445
50    BMDL = 452.359
     C.2  Liver hyperplasia

51          All available dichotomous models in the Benchmark Dose Software (version 2.1.1) were fit to the
52   incidence data shown in Table C-3, for liver hyperplasia in male and female F344/DuCrj rats exposed to
53   1,4-dioxane in the drinking water (Kano et al., 2009; JBRC. 1998). Benchmark doses associated with a
54   BMR of a 10% extra risk were calculated.
                                                                                               C-6
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      Table C-3   Incidence of liver hyperplasia in F344/DuCrj rats exposed to 1,4-dioxane in drinking
                  water3
     	Males (mg/kg-day)	Females (mg/kg-day)	
           0	11	55	274	0	18	83	429
          3/40	2/45	9/35a	12/22°	2/38b	2/37	9/38	24/24°
      aDose information from Kano et al. (2009) and incidence data from sacrificed animals from JBRC (1998).
      "Incidence significantly elevated compared
      ""Incidence significantly elevated compared 1
      Sources: Kano et al. (2009): JBRC (1998).
"Incidence significantly elevated compared to control by x2 test (p < 0.05).
Incidence significantly elevated compared to control by x2 test (p < 0.01).
 1           For incidence of liver hyperplasia in F344 male rats, the logistic, probit, and dichotomous-Hill
 2    models all exhibited a statistically significant lack of fit (i.e., %2 p-value < 0.1; see Table C-4), and thus
 3    should not be considered further for identification of a POD. All of the remaining models exhibited
 4    adequate fit, but the AIC values for the gamma, multistage, quantal-linear, and Weibull models were
 5    lower than the AIC values for the log-logistic and log-probit models. Finally, the AIC values for gamma,
 6    multistage, quantal-linear, and Weibull models in Table C-4 are equivalent and, in this case, essentially
 7    represent the same model. Therefore, consistent with the external review draft Benchmark Dose
 8    Technical Guidance (U.S. EPA. 2000aX any of them with equal AIC values (gamma, multistage,
 9    quantal-linear, or Weibull) could be used to identify a POD for this endpoint of 23.8 mg/kg-day.

10           For liver hyperplasias in F344 female rats exposed to 1,4-dioxane, the quantal-linear and
11    dichotomous-Hill models did not result in a good fit (i.e., %2/rvalue < 0.1; See Table C-4). The
12    multistage (3-degree) model had the lowest AIC value and was selected as the best-fitting model.
13    Therefore, consistent with the BMD technical guidance document (U.S. EPA. 2000a). the BMDL from
14    the multistage (3-degree) model was selected to yield a POD for this endpoint of 27.1 mg/kg-day.
                                                                                                           C-7
                                      DRAFT - DO NOT CITE OR QUOTE

-------
Table C-4   Benchmark dose modeling results based on the incidence of liver hyperplasias in
             male and female F344 rats exposed to 1,4-dioxane in drinking water for 2 years
Model
AIC
Scaled „.._.
p-va,ue- Reside, (mg™^y)
BMDLio
(mg/kg-day)
Male
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage0
(2 degree)
Probit
Weibullb
Quantal-Linear
Dichotomous-Hill
114.172
117.047
115.772
115.57
114.172
116.668
114.172
114.172
117.185
0.3421
0.0706
0.1848
0.1431
0.3421
0.0859
0.3421
0.3421
NCe
0.886
1.869
0.681
1.472
0.886
1.804
0.886
0.886
-0.2398
35.90
83.56
33.39
54.91
35.90
76.69
35.90
35.90
32.01
23.81
63.29
16.96
37.05
23.81
58.57
23.81
23.81
14.84
Female
Gammab
Logistic
Log-logistic0
Log-probitc
Multistage0
(2 degree)
Multistage0
(3 degree)
Probit
Weibullb
Quantal-Linear
Dichotomous-Hill
78.8357
77.0274
78.8357
78.8357
76.9718
76.8351
77.0308
78.8349
87.3833
2972.99
0.9783
0.9174
0.9781
0.9781
0.9563
0.9999
0.9095
0.9995
0.0245
NCe
0
-0.016
0
0
-0.107
0
0.017
0
-1.116
0
70.78
54.66
77.72
74.64
56.06
65.28
52.53
66.47
21.52
NCe
40.51
41.11
51.21
50.97
31.17
27.08
38.44
36.14
15.61
NCe
ap-Value from the x^ goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to > 1.
°Slope restricted to 5 1.
dBetas restricted to >0.
eNC=Not calculated.

Sources: Kano et al. (2009): JBRC (1998).
                                  DRAFT - DO NOT CITE OR QUOTE
                                                                                                          C-8

-------
                                   Gamma Multi-Hit Model with 0.95 Confidence Level
 1
 2
 o
 J
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
        •5
        I
        C
        o
        13
        ro
                                              100
                                                    150
200
250
                                                     dose
          14:35 12/042009
            Figure C-3  BMD gamma model of liver hyperplasia incidence data for F344 male
                       rats exposed to 1,4-dioxane in drinking water for 2 years to support
                       results Table C-4.
Gamma Model.  (Version: 2.13; Date:  05/16/2008)
Input Data File:
Z:\14Dioxane\BMDS\gam_jbrcl998_mrat_liver_hyper_Gam-BMR10-Restrict.(d)
Gnuplot Plotting File:
Z:\14Dioxane\BMDS\gam_jbrcl998_mrat_liver_hyper_Gam-BMR10-Restrict.plt
                                         Fri  Dec  04  14:35:02  2009

 BMDS Model Run

 The form of the probability function is:

 P[response]= background+(1-background)*CumGamma[siope*dose,power],
 where CumGamma(.)  is the cummulative Gamma  distribution function

 Dependent variable = Effect
 Independent variable = Dose
 Power parameter is restricted as power >=1

 Total number of observations = 4
 Total number of records with missing values = 0
 Maximum number of iterations = 250
 Relative Function Convergence has  been set  to:  le-008
 Parameter Convergence has been set to:  le-008
 Default Initial  (and Specified) Parameter Values
                                                                                               C-9
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Background = 0.0853659
 2    Slope = 0.00479329
 3    Power = 1.3
 4
 5
 6    Asymptotic Correlation Matrix of Parameter Estimates
 7    (*** The model parameter(s) -Power have been estimated at a boundary point, or have
 8   been specified by the user, and do not appear in the correlation matrix )
 9
10    Background Slope
11   Background 1 -0.36
12    Slope -0.36 1
13
14    Parameter Estimates
15    95.0% Wald Confidence Interval
16   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17   Background 0.0569658 0.0278487 0.00238329 0.111548
18    Slope 0.00293446 0.000814441 0.00133818 0.00453073
19    Power 1 NA
20
21   NA - Indicates that this parameter has hit a bound implied by some ineguality
22   constraint and thus has no standard error.
23
24    Analysis of Deviance Table
25
26    Model Log(likelihood) # Param's Deviance Test d.f. P-value
27    Full model -53.9471 4
28    Fitted model -55.0858 2 2.27725 2 0.3203
29    Reduced model -67.6005 1 27.3066 3 <.0001
30
31    AIC: 114.172
32
33
34    Goodness of Fit
35    Scaled
36    Dose Est._Prob. Expected Observed Size Residual
37    	
38    0.0000 0.0570 2.279 3.000 40 0.492
39    11.0000 0.0869 3.911 2.000 45 -1.011
40    55.0000 0.1975 6.913 9.000 35 0.886
41    274.0000 0.5780 12.715 12.000 22 -0.309
42
43    ChiA2 = 2.15 d.f. = 2 P-value = 0.3421
44
45
46    Benchmark Dose Computation
47   Specified effect =0.1
48   Risk Type = Extra risk
49   Confidence level = 0.95
50    BMD = 35.9046
51    BMDL = 23.8065
                                                                                             C-10
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                               Multistage Model with 0.95 Confidence Level
I
C
o
         0.1
                                       100
150
200
250
                                               dose
  14:35 12/042009
    Figure C-4  BMD multistage (2 degree) model of liver hyperplasia incidence data
                for F344 male rats exposed to 1,4-dioxane in drinking water for 2 years
                to support results
                                                                                          C-ll
                            DRAFT - DO NOT CITE OR QUOTE

-------
            Table C-4.

 1   ====================================================================
 2   Multistage Model.  (Version: 3.0; Date: 05/16/2008)
 3   Input Data File:
 4   Z:\14Dioxane\BMDS\mst_jbrcl998_mrat_liver_hyper_Mst-BMR10-restrict.(d)
 5   Gnuplot Plotting File:
 6   Z:\14Dioxane\BMDS\mst_jbrcl998_mrat_liver_hyper_Mst-BMR10-Restrict.plt
 7                                           Fri Dec 04 14:35:06 2009
 8   ====================================================================
 9    BMDS Model Run
10   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11    The form of the probability function is:
12
13    P [response] = background +  (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2) ]
14
15    The parameter betas are restricted to be positive
16
17    Dependent variable = Effect
18    Independent variable = Dose
19
20    Total number of observations = 4
21    Total number of records with missing values = 0
22    Total number of parameters in model = 3
23    Total number of specified parameters = 0
24    Degree of polynomial = 2
25
26
27    Maximum number of iterations = 250
28    Relative Function Convergence has been set to: le-008
29    Parameter Convergence has been set to: le-008
30
31
32
33    Default Initial Parameter Values
34    Background = 0.0750872
35    Beta(l) = 0.00263797
36    Beta(2) = 0
37
38
39    Asymptotic Correlation Matrix of Parameter Estimates
40   (*** The model parameter(s) -Beta(2) have been estimated at a boundary point, or have
41   been specified by the user, and do not appear in the correlation matrix)
42
43    Background Beta(l)
44   Background 1 -0.49
45    Beta(l) -0.49 1
46
47
48    Parameter Estimates
49    95.0% Wald Confidence Interval
50   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
51   Background 0.0569658 * * *
52    Beta(l) 0.00293446 * * *
53    Beta(2) 0 * * *
54
55   * - Indicates that this value is not calculated.
56
57
58
59    Analysis of Deviance Table
60
61    Model Log(likelihood) # Param's Deviance Test d.f. P-value
62    Full model -53.9471 4
63    Fitted model -55.0858 2 2.27725 2 0.3203
64    Reduced model -67.6005 1 27.3066 3 <.0001


                                                                                              C-12
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
 AIC: 114.172
 Goodness of Fit
 Scaled
 Dose Est._Prob. Expected  Observed Size Residual
 0.0000 0.0570 2.279 3.000  40  0.492
 11.0000 0.0869 3.911 2.000  45  -1.011
 55.0000 0.1975 6.913 9.000  35  0.886
 274.0000 0.5780 12.715  12.000  22  -0.309

 ChiA2 = 2.15 d.f. = 2 P-value  = 0.3421
 Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
 BMD = 35.9046
 BMDL = 23.8065
 BMDU = 82.1206

Taken together,  (23.8065,  82.1206)  is  a 90% two-sided confidence interval  for  the  BMD
                                       Weibull Model with 0.95 Confidence Level
        o
        13
           0.8


           0.7


           0.6


           0.5


           0.4


           0.3


           0.2


           0.1
                             Weibull
                        BMDL
                           BMD
                                   50
                                         100
150
200
250
                                                      dose
          14:35 12/042009
            Figure C-5 BMD Weibull model of liver hyperplasia incidence data for F344 male
                       rats exposed to 1,4-dioxane in drinking water for 2 years to support the
                       results in
                                                                                               C-13
                                   DRAFT - DO NOT CITE OR QUOTE

-------
            Table C-4.

 1   ====================================================================
 2   Weibull Model using Weibull Model  (Version: 2.12; Date: 05/16/2008)
 3   Input Data  File:
 4   Z:\14Dioxane\BMDS\wei_jbrcl998_mrat_liver_hyper_Wei-BMR10-Restrict.(d)
 5   Gnuplot Plotting File:
 6   Z:\14Dioxane\BMDS\wei_jbrcl998_mrat_liver_hyper_Wei-BMR10-Restrict.plt
 7                                           Fri Dec 04 14:35:08 2009
 8   ====================================================================
 9    BMDS Model Run
10   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11    The form of the probability function is:
12
13    P[response] = background +  (1-background)*[1-EXP(-slope*dose/xpower)]
14
15    Dependent  variable = Effect
16    Independent variable = Dose
17    Power parameter is restricted as power >=1
18
19    Total number of observations = 4
20    Total number of records with missing values = 0
21    Maximum number of iterations = 250
22    Relative Function Convergence has been set to: le-008
23    Parameter  Convergence has been set to: le-008
24
25
26
27    Default Initial  (and Specified) Parameter Values
28    Background = 0.0853659
29    Slope = 0.00253609
30    Power = 1
31
32
33    Asymptotic Correlation Matrix of Parameter Estimates
34   (** The model parameter(s) -Power have been estimated at a boundary point, or have
35   been specified by the user, and do not appear in the correlation matrix  )
36
37    Background Slope
38   Background  1 -0.36
39    Slope -0.36 1
40
41
42    Parameter  Estimates
43    95.0% Wald Confidence Interval
44   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
45   Background  0.0569661 0.0278498 0.00238155 0.111551
46    Slope 0.00293445 0.000814445 0.00133816  0.00453073
47    Power 1 NA
48
49   NA - Indicates that this parameter has hit a bound implied by some ineguality
50   constraint  and thus has no standard error.
51
52
53    Analysis of Deviance Table
54
55    Model Log(likelihood) # Param's Deviance Test d.f. P-value
56    Full model -53.9471 4
57    Fitted model -55.0858 2 2.27725 2 0.3203
58    Reduced model -67.6005 1 27.3066 3 <.0001
59
60    AIC: 114.172
61
62
63    Goodness of Fit
64    Scaled


                                                                                              C-14
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
 Dose Est._Prob. Expected Observed Size Residual
 0.0000 0.0570 2.279  3.000  40 0.492
 11.0000 0.0869 3.911  2.000 45 -1.011
 55.0000 0.1975 6.913  9.000 35 0.886
 274.0000 0.5780  12.715  12.000 22 -0.309

 ChiA2 = 2.15 d.f.  =  2  P-value = 0.3421
 Benchmark Dose Computation
Specified effect =0.1
Risk Type = Extra  risk
Confidence level = 0.95
 BMD = 35.9047
 BMDL = 23.8065
                                    Quantal Linear Model with 0.95 Confidence Level
        o
        13
           0.8


           0.7


           0.6


           0.5


           0.4


           0.3


           0.2


           0.1
                                    Quantal Linear
                        BMDL
                           BMD
                                   50
                                         100
150
200
250
                                                      dose
          14:35 12/042009
            Figure C-6 BMD quantal-linear model of liver hyperplasia incidence data for F344
                       male rats exposed to 1,4-dioxane in drinking water for 2 years to
                       support the results in
                                                                                                C-15
                                   DRAFT - DO NOT CITE OR QUOTE

-------
            Table C-4.

 1   ===================================
 2   Quantal Linear Model using Weibull Model  (Version: 2.12; Date: 05/16/2008)
 3   Input Data  File: Z:\14Dioxane\BMDS\qln_jbrcl998_mrat_liver_hyper_Qln-BMR10.(d)
 4   Gnuplot Plotting File: Z:\14Dioxane\BMDS\gln_jbrcl998_mrat_liver_hyper_Qln-BMR10.plt
 5                                           Fri Dec 04 14:35:09 2009
 6   ====================================================================
 7    BMDS Model Run
 8   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 9    The form of the probability function is:
10
11    P[response] = background +  (1-background)*[1-EXP(-slope*dose)]
12
13
14    Dependent  variable = Effect
15    Independent variable = Dose
16
17    Total number of observations = 4
18    Total number of records with missing values = 0
19    Maximum number of iterations = 250
20    Relative Function Convergence has been set to: le-008
21    Parameter  Convergence has been set to: le-008
22
23    Default Initial (and Specified) Parameter Values
24    Background = 0.0853659
25    Slope = 0.00253609
26    Power = 1  Specified
27    Asymptotic Correlation Matrix of Parameter Estimates
28   (*** The model parameter(s) -Power have been estimated at a boundary point, or have
29   been specified by the user, and do not appear in the correlation matrix)
30
31    Background Slope
32   Background  1 -0.36
33    Slope -0.36 1
34
35
36
37    Parameter  Estimates
38    95.0% Wald Confidence Interval
39   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40   Background  0.0569665 0.02785 0.00238157 0.111551
41    Slope 0.00293447 0.000814452 0.00133818  0.00453077
42
43
44
45    Analysis of Deviance Table
46
47    Model Log(likelihood) # Param's Deviance Test d.f. P-value
48    Full model -53.9471 4
49    Fitted model -55.0858 2 2.27725 2 0.3203
50    Reduced model -67.6005 1 27.3066 3 <.0001
51
52    AIC: 114.172
53
54
55    Goodness of Fit
56    Scaled
57    Dose Est._Prob. Expected Observed Size Residual
58    	~	
59    0.0000 0.0570 2.279 3.000 40 0.492
60    11.0000 0.0869 3.911 2.000 45 -1.011
61    55.0000 0.1975 6.913 9.000 35 0.886
62    274.0000 0.5780 12.716 12.000 22 -0.309
63
64    Chi^2 = 2.15 d.f.  = 2 P-value = 0.3421


                                                                                              C-16
                                  DRAFT - DO NOT CITE OR QUOTE

-------
1
2
3
4
5
6
7
 Benchmark  Dose Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence  level = 0.95
 BMD = 35.9044
 BMDL = 23.8065
                                      Multistage Model with 0.95 Confidence Level
        O

       I
        c
        O
                0.8
                0.6
                0.4
                0.2  -
         10:3005/21 2010
            Source: JBRC (1998).
            Figure C-7  BMD Multistage model (third (3°)of liver hyperplasia incidence data for
                       F344 female rats exposed to 1,4-dioxane in drinking water for 2 years to
                       support the results in
                                                                                                 C-17
                                   DRAFT - DO NOT CITE OR QUOTE

-------
            Table C-4.

 1   ====================================================================
 2   Multistage Model.  (Version: 3.0; Date: 05/16/2008)
 3   Input Data File:
 4   H:\14Dioxane\BMDS\mst_jbrcl998_frat_liver_hyper_Mst-BMR10-Restrict-3deg.(d)
 5   Gnuplot Plotting File:
 6   H:\14Dioxane\BMDS\mst_jbrcl998_frat_liver_hyper_Mst-BMR10-Restrict-3deg.plt
 7                                                  Fri May 21 10:30:14 2010
 8   ====================================================================
 9    BMDS Model Run
10   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
11   The form of the probability function is:
12
13   P[response] = background +
14   (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
15
16   The parameter betas are restricted to be positive
17
18    Dependent variable = Effect
19    Independent variable = Dose
20
21    Total number of observations = 4
22    Total number of records with missing values = 0
23    Total number of parameters in model = 4
24    Total number of specified parameters = 0
25    Degree of polynomial = 3
26
27    Maximum number of iterations = 250
28    Relative Function Convergence has been set to: le-008
29    Parameter Convergence has been set to: le-008
30
31    Default Initial Parameter Values
32    Background = 0
33    Beta(l) = 0
34    Beta(2) = 0
35    Beta(3) = 1.2696e+012
36
37    Asymptotic Correlation Matrix of Parameter Estimates
38
39   (*** The model parameter(s) -Beta(l), -Beta(2) have been estimated at  a boundary
40   point, or have been specified by the user, and do not appear  in the correlation
41   matrix)
42
43    Background Beta(3)
44   Background 1 -0.55
45    Beta(3) -0.55 1
46
47
48    Parameter Estimates
49
50    95.0% Wald Confidence Interval
51    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
52   Background 0.0523101 * * *
53    Beta(l) 0 * * *
54    Beta(2) 0 * * *
55    Beta(3) 3.78712e-007 * * *
56
57   * - Indicates that this value is not calculated.
58
59
60    Analysis of Deviance Table
61
62    Model Log(likelihood) # Param's Deviance Test d.f. P-value
63    Full model -36.4175 4
64    Fitted model -36.4175 2 0.00016582 2 0.9999


                                                                                              C-18
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Reduced model -79.9164 1 86.9979 3 <.0001
 2
 3    AIC: 76.8351
 4
 5    Goodness of Fit
 6    Scaled
 7    Dose Est._Prob. Expected Observed Size Residual
 9    0.0000 0.0523 1.988 2.000 38 0.009
10    18.0000 0.0544 2.013 2.000 37 -0.009
11    83.0000 0.2368 8.999 9.000 38 0.000
12    429.0000 1.0000 24.000 24.000 24 0.000
13
14    ChiA2 = 0.00 d.f. = 2 P-value = 0.9999
15
16    Benchmark Dose Computation
17   Specified effect =0.1
18   Risk Type = Extra risk
19   Confidence level = 0.95
20    BMD = 65.2814
21    BMDL = 27.0766
22    BMDU = 91.3457
23
24   Taken together, (27.0766, 91.3457)  is a 90% two-sided confidence interval for the BMD
                                                                                             C-19
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      APPENDIX  D.     DETAILS  OF   BMD  ANALYSIS   FOR
         ORAL  CSF   FOR   1,4-DIOXANE


 1           Dichotomous models available in the Benchmark Dose Software (BMDS) (version 2.1.1) were fit
 2    to the incidence data for hepatocellular carcinoma and/or adenoma for mice and rats, as well as nasal
 3    cavity tumors, peritoneal mesotheliomas, and mammary gland adenomas in rats exposed to 1,4-dioxane in
 4    the drinking water. Doses associated with a benchmark response (BMR) of a 10% extra risk were
 5    calculated. BMD10 and BMDL10 values from the best fitting model, determined by adequate global- fit (%2
 6    p > 0.1) and AIC values, are reported for each endpoint (U.S. EPA. 2000a). If the multistage cancer
 7    model is not the best fitting model for a particular endpoint, the best-fitting multistage cancer model for
 8    that endpoint is also presented as a point of comparison.

 9           A summary of the model predictions for the Kano et al. (2009) study are shown in Table D-l. The
10    data and BMD modeling results are presented separately for each dataset as follows:

11                     •  Hepatic adenomas and carcinomas in female F344 rats (Table D-2 and
12                        Table D-3; Figure D-l)

13                     •  Hepatic adenomas and carcinomas in male F344 rats (Table D-4 and Table D-5;
14                        Figure D-2 and Figure D-3)

15                     •   Significant tumor incidence data at sites other than the liver (i.e., nasal cavity,
16                        mammary gland, and peritoneal) in male and female F344 rats (Table D-6)

17                             o  Nasal cavity tumors in female F344 rats (Table D-7; Figure D-4)

18                             o  Nasal cavity tumors in male F344 rats (Table D-8; Figure D-5)

19                             o  Mammary  gland adenomas in female F344 rats (Table D-9; Figure D-6
20                                and Figure D-7)

21                             o  Peritoneal mesotheliomas in male F344 rats (Table D-10; Figure D-8 and
22                                Figure D-9)

23                     •  Hepatic adenomas and carcinomas in female BDF1 mice (Table D-l 1,
24                        Table D-l2, and Table D-l3; Figure D-10, Figure D-l 1, Figure D-l2, and
25                        Figure D-l3)

26                     •  Hepatic adenomas and carcinomas in male BDF1 mice (Table D-l4 and
27                        Table D-15; Figure D-14 and Figure D-15)

28           Data and BMD modeling results from the additional chronic bioassays (NCI. 1978; Kociba et al..
29    1974) were evaluated for comparison with the data from Kano et al. (2009). These results are presented as
30    follows:

31                     •   Summary of BMDS dose-response modeling estimates associated with liver and
32                        nasal tumor incidence data resulting from chronic oral exposure to 1,4-dioxane in
33                        rats and mice (Table D-l6)
                                                                                                    D-l
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1                     •  Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in
 2                        male and female Sherman rats (combined) (Kociba et al.. 1974) treated with
 3                        1,4-dioxane in the drinking water for 2 years (Table D-17)

 4                            o  BMDS dose-response modeling results for incidence of hepatocellular
 5                               carcinoma in male and female Sherman rats (combined) (Kociba et al..
 6                               1974) exposed to 1,4-dioxane in drinking water for 2 years (Table D-18;
 7                               Figure D-16 and Figure D-17)

 8                            o  BMDS dose-response modeling results for incidence of nasal squamous
 9                               cell carcinoma in male and female Sherman rats (combined) (Kociba et
10                               al.. 1974) exposed to 1,4-dioxane in the drinking water for 2 years
11                               (Table D-19; Figure D-18)

12                     •  Incidence of nasal cavity squamous cell carcinoma and hepatocellular adenoma
13                        in Osborne-Mendel rats (NCI, 1978) exposed to 1,4-dioxane in the drinking
14                        water (Table D-20)

15                            o  BMDS dose-response modeling results for incidence of hepatocellular
16                               adenoma in female Osborne-Mendel rats (NCI. 1978) exposed to
17                               1,4-dioxane in the drinking water for 2 years (Table D-21; Figure D-19
18                               and Figure D-20)

19                            o  BMDS dose-response modeling results for incidence of nasal cavity
20                               squamous cell carcinoma in female Osborne-Mendel rats  (NCI. 1978)
21                               exposed to 1,4-dioxane in the drinking water for 2 years (Table D-22;
22                               Figure D-21 and Figure D-22)

23                            o  BMDS dose-response modeling results for incidence of nasal cavity
24                               squamous cell carcinoma in male Osborne-Mendel rats (NCI. 1978)
25                               exposed to 1,4-dioxane in the drinking water for 2 years (Table D-23;
26                               Figure D-23 and Figure D-24)

27                     •  Incidence of hepatocellular adenoma or carcinoma in male and female B6C3F]
28                        mice (NCI. 1978) exposed to 1,4-dioxane in drinking water (Table D-24)

29                            o  BMDS dose-response modeling results for the combined incidence of
30                               hepatocellular adenoma or carcinoma in female B6C3Fi mice (NCI.
31                               1978) exposed to 1,4-dioxane in the drinking water for 2 years
32                               (Table D-25; Figure D-25)

33                            o  BMDS dose-response modeling results for incidence of combined
34                               hepatocellular adenoma or carcinoma in male B6C3Fi mice (NCI. 1978)
35                               exposed to 1,4-dioxane in the drinking water for 2 years (Table D-26;
36                               Figure D-26 and Figure D-27).
      D.1   General Issues  and Approaches  to BMDS  Modeling
      D.1.1    Combining Data on Adenomas  and Carcinomas

37          The incidence of adenomas and the incidence of carcinomas within a dose group at a site or tissue
38    in rodents are sometimes combined. This practice is based upon the hypothesis that adenomas may

                                                                                                  D-2
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1    develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a; McConnell et al..
 2    1986). The incidence at high doses of both tumors in rat and mouse liver is high in the key study (Kano et
 3    al., 2009). The incidence of hepatic adenomas and carcinomas was summed without double-counting
 4    them so as to calculate the combined incidence of either a hepatic carcinoma or a hepatic adenoma in
 5    rodents.

 6           The variable N is used to denote the total number of animals tested in the dose group. The
 7    variable Y is used here to denote the number of rodents within a dose group that have characteristic X,
 8    and the notation Y(X) is used to identify the number with a specific characteristic X. Modeling was
 9    performed on the adenomas and carcinomas separately and the following combinations of tumor types:

10                      •   Y(adenomas) = number of animals with adenomas, whether or not carcinomas
11                         are present;
12                      •   Y(carcinomas) = number of animals with carcinomas, whether or not adenomas
13                         are also present;
14                      •   Y(either adenomas or carcinomas) = number of animals with adenomas or
15                         carcinomas, not both = Y(adenomas) + Y(carcinomas) - Y(both adenomas and
16                         carcinomas);
17                      •   Y(neither adenomas nor carcinomas) = number of animals with no adenomas and
18                         no carcinomas = N - Y(either adenomas or carcinomas).
      D.1.2    Model Selection Criteria

19           Multiple models were fit to each dataset. The model selection criteria used in the BMD technical
20    guidance document (U.S. EPA. 2000a) were applied as follows:

21                     •   p-value for goodness-of-fit > 0.10
22                     •   AIC smaller than other acceptable models
23                     •   %2 residuals  as small as possible
24                     •   No systematic patterns of deviation of model from data
25           Additional criteria were applied to eliminate implausible dose-response functions:

26                     •   Monotonic dose-response functions, e.g. no negative coefficients of polynomials
27                         in MS models
28                     •   No infinitely steep dose-response functions near 0 (control dose), achieved by
29                         requiring the estimated parameters "power" in the Weibull and Gamma models
30                         and "slope" in the log-logistic model to have values >  1.
31           Because no single set of criteria covers all contingencies, an extended list of preferred models are
32    presented below in Table D-l.
                                                                                                    D-3
                                    DRAFT - DO NOT CITE OR QUOTE

-------
    D.1.3   Summary

1         The BMDS models recommended to calculate rodent BMD and BMDL values and corresponding

2   human BMDnED and BMDLHED values are summarized in Table D-l.
    Table D-1  Recommended models for rodents exposed to 1,4-dioxane in drinking water (Kano et
              al.. 2009)
Endpoint
Model
selection
criterion
Model Type
AIC
p-value
BMDa
mg/kg-day
BMDLa BMDHEDa BMDLHEDa
mg/kg-day mg/kg-day mg/kg-day
Female F344 Rat
Hepatic
Tumors
Mammary
Gland
Tumors
Nasal
Cavity
Tumors
Lowest AIC
Lowest AIC
Lowest AIC
Multistage
(2 degree)
LogLogistic
Multistage
(3 degree)
91.
194
42.
5898
.151
6063
0.4516
0.8874
0.9966
79.83
161.01
381.65
58
81
282
.09
.91
.61
19
40
94
.84
.01
.84
14.43
20.35
70.23
Male F344 Rat
Hepatic
Tumors
Peritoneal
Meso-thel
ioma
Nasal
Cavity
Tumors
Female BDF1
Hepatic
Tumors
Lowest AIC
Lowest AIC
Lowest AIC
Mouse
Lowest AIC
BMR 50%
Probit
Probit
Multistage
(3 degree)

LogLogistic
LogLogistic
147
138
24

176
176
.787
.869
.747

.214
.214
0.9867
0.9148
0.9989

0.1421
0.1421
62.20
93.06
328.11

5.54
49.88"
51
76
245

3
32.
.12
.32
.63

.66
93°
17
26
91

0
7.
.43
.09
.97

.83
51b
14.33
21.39
68.85

0.55
4.95°
Male BDF1 Mouse
Hepatic
Tumors
Lowest AIC
Log-Logistic
248
.839
0.3461
34.78
16
.60
5
.63
2.68
    "Values for BMR 10% unless otherwise noted.
    bBMR 50%.
    D.2  Female F344 Rats: Hepatic  Carcinomas and Adenomas

3         The incidence data for hepatic carcinomas and adenomas in female F344 rats (Kano et al.. 2009)
4   are shown in Table D-2.
                                                                                         D-4
                               DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-2   Data for hepatic adenomas and carcinomas in female F344 rats (Kano et al.. 2009)


Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
0
3
0
3
47
50
18
1
0
1
49
50
83
6
0
6
44
50
429
48
10
48
2
50
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009)
1          Note that the incidence of rats with adenomas, with carcinomas, and with either adenomas or
2   carcinomas are monotone non-decreasing functions of dose except for 3 female rats in the control group.
3   These data therefore appear to be appropriate for dose-response modeling using BMDS.

4          The results of the BMDS modeling forthe entire suite of models are presented in Table D-3.
                                                                                                   D-5
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-3   BMDS dose-response modeling results for the combined incidence of hepatic
                 adenomas and carcinomas in female F344 rats (Kano et al.. 2009)

Model
Gamma
Logistic
LogLogistic
LogProbit"
AIC
93.1067
91.7017
93.102
93.0762
p-value
0.3024
0.4459
0.3028
0.3074
BMDio
mg/kg-day
89.46
93.02
88.34
87.57
BMDLio
mg/kg-day
62.09
71.60
65.52
66.19
x2a
0.027
0.077
0.016
0.001
BMDio HED
mg/kg-day
22.23
23.12
21.95
21.76
BMDLio HED
mg/kg-day
15.43
17.79
16.28
16.45
                    114094
                                   °-0001
            25.58
           19.92
        -1.827
                                                                         6.36
             4.95
     Multistage-Cancer
     (2 degree)	
                                          79.83
                       58.09    -0.408
                              19.84
                                14.43
gSefanC6r     93-2682    °-2747
                                               92.81
                                                      59.31
                                0.077
                              23.06
                                14.74
     Probit
                    91.8786    0.3839
            85.46
           67.84    -0.116
                   21.24
                                                                                      16.86
     Weibull
                    93.2255    0.2825
            92.67
           59.89
         0.088
                                                                        23.03
             14.88
     Quantal-Linear
                    114.094
0.0001
25.58
19.92
                                                              -1.827
6.36
4.95
     Dichotomous-Hill
                    4458.37
                                  0
     aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
     "Slope restricted > 1.
     ""Best-fitting model.
     dValue unable to be calculated (NC: not calculated) by BMDS.
                                    Multistage Cancer Model with 0.95 Confidence Level
        I
        c
        o
        t3
        (O
                 0.8
                 0.6
            0.4
                 0.2
                                              Multistage Cancer  	
                                             Linear extrapolation  	
                             BMDL
                                     BMD
                                50
                                       100
                                               150
                                                      200     250
                                                        dose
                                                                      300
                                                                              350
                                                                                     400
                                                                                             450
          07:20 10/262009

            Source: Used with permission of Elsevier, Ltd., Kano et al. (2009).

            Figure D-l Multistage BMD model (2 degree) for the combined incidence of hepatic
                        adenomas and carcinomas in female F344 rats.
1
2
3
4
Multistage Cancer Model.   (Version: 1.7;  Date:  05/16/2008)
Input  Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_hepato_adcar_Msc-BMR10-2poly.(d)
                                     DRAFT - DO NOT CITE OR QUOTE
                                                                                                        D-6

-------
 1   Gnuplot Plotting File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_hepato_adcar_Msc-BMR10-2poly.plt
 3   Mon Oct 26 08:20:52 2009
 4   ====================================================================
 5    BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7
 8   The form of the probability function is:
 9   P [response] = background +  (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]
10
11   The parameter betas are restricted to be positive
12
13   Dependent variable = Effect
14   Independent variable = Dose
15
16   Total number of observations = 4
17   Total number of records with missing values = 0
18   Total number of parameters in model = 3
19   Total number of specified parameters = 0
20   Degree of polynomial = 2
21
22   Maximum number of iterations =250
23   Relative Function Convergence has been set to: le-008
24   Parameter Convergence has been set to: le-008
25
26   Default Initial Parameter Values
27   Background = 0.0281572
28   Beta(l) = 0
29   Beta(2) = 1.73306e-005
30
31   Asymptotic Correlation Matrix of Parameter Estimates  (*** The model parameter(s)
32   -Beta(l)have been estimated at a boundary point, or have been specified by  the user,
33   and do not appear in the correlation matrix )
34
35    Background Beta(2)
36   Background 1 -0.2
37   Beta(2) -0.2 1
38
39                                     Parameter Estimates
40    95.0% Wald Confidence Interval
41   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
42   Background 0.0362773 * * *
43   Beta(l) 0 * * *
44   Beta(2) 1.65328e-005 * * *
45
46   * - Indicates that this value is not calculated.
47
48
49    Analysis of Deviance Table
50
51    Model Log(likelihood) # Param's Deviance Test d.f. P-value
52    Full model -42.9938 4
53    Fitted model -43.7949 2 1.60218 2 0.4488
54    Reduced model -120.43 1 154.873 3 <.0001
55
56    AIC: 91.5898
57
58    Goodness of Fit
59    Scaled
60    Dose Est._Prob. Expected Observed Size Residual
61    	
62    0.0000 0.0363 1.814 3.000 50 0.897
63    18.0000 0.0414 2.071 1.000 50 -0.760
64    83.0000 0.1400 7.001 6.000 50 -0.408
65    429.0000 0.9540 47.701 48.000 50 0.202
66
67   ChiA2 = 1.59 d.f.  = 2 P-value = 0.4516

                                                                                              D-7
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
Benchmark Dose Computation

Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
 BMD = 79.8299
 BMDL = 58.085
 BMDU = 94.0205

Taken together,  (58.085 , 94.0205) is a 90% two-sided  confidence  interval  for the BMD

Multistage Cancer Slope Factor =  0.00172161
15

16

17
18
D.3  Male  F344 Rats:  Hepatic  Carcinomas and Adenomas

             The data for hepatic adenomas and carcinomas in male F344 rats (Kano et al.. 2009) are
             shown in Table D-4.
Table D-4
Data for hepatic adenomas and carcinomas in male F344 rats (Kano et al., 2009)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
0
3
0
3
47
50
11
4
0
4
46
50
55
7
0
7
43
50
274
32
14
39
11
50
Source: Used with permission from Elservier, Ltd., Kano et al. (2009).
19

20          Note that the incidence of rats with hepatic adenomas, carcinomas, and with either adenomas or
21   carcinomas are monotone non-decreasing functions of dose. These data therefore appear to be appropriate
22   for dose-response modeling using BMDS.

23          The results of the BMDS modeling for the entire suite of models tested using the data for hepatic
24   adenomas and carcinomas for male F344 rats are presented in Table D-5.
                                                                                               D-8
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-5   BMDS dose-response modeling results for the combined incidence of adenomas and
                 carcinomas in livers of male F344 rats (Kano et al.. 2009)

Model
Gamma
Logistic
LogLogistic
LogProbit"
AIC
149.884
147.813
149.886
149.913
p-value
0.7257
0.9749
0.7235
0.6972
BMDio
mg/kg-day
62.41
68.74
62.10
61.70
BMDLio
mg/kg-day
30.79
55.39
34.61
37.49
x2a
-0.03
0.097
-0.021
-0.003
BMDio HED
mg/kg-day
17.49
19.27
17.41
17.29
BMDLio HED
mg/kg-day
8.63
15.53
9.70
10.51
     Multistage-Cancer
     (1 degree)	
                     152.836   0.0978
23.82
18.34
-0.186
6.68
5.14
     Multistage-Cancer
     (2 degree)	
                     149.814   0.8161
61.68
28.26
-0.063
17.29
7.92
     Multistage-Cancer
     (3 degree)	
                     149.772   0.9171
63.62
27.49
-0.024
17.83
7.71
     Probitc
                     147.787   0.9867
62.20
51.12
 -0.05
17.43
14.33
     Weibull
                     149.856   0.7576
62.63
30.11
-0.039
17.56
8.44
     Quantal-Linear
                     152.836   0.0978
23.82
18.34
-0.186
6.68
5.14
     Dichotomous-Hill
                     4441.71
 NCT
 NCT
  0
  0
  0
     "Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
     "Slope restricted > 1.
     ""Best-fitting model.
     dValue unable to be calculated (NC: not calculated) by BMDS.
                                          Probit Model with 0.95 Confidence Level
      I
      C
      o
      '•
               0.8
               0.6
           0.4
               0.2
                                                 100
                                                              150
                                                                           200
                                                                                         250
                                                          dose
        07:32 10/26 2009

            Source: Used with permission from Elservier, Ltd., Kano et al. (2009).

            Figure D-2 Probit BMD model for the combined incidence of hepatic adenomas and
                        carcinomas in male F344 rats.
1
2
o
J
4
Probit Model.  (Version:  3.1;  Date: 05/16/2008)
Input  Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-BMR10.(d)
                                     DRAFT - DO NOT CITE OR QUOTE
                                                                                                         D-9

-------
 1   Gnuplot Plotting File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_hepato_adcar_Prb-BMR10.plt
 3   Mon Oct 26 08:32:08 2009
 4   ====================================================================
 5   BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7
 8   The form of the probability function is:
 9   P[response] = CumNorm(Intercept+Slope*Dose),
10   where CumNorm(.)  is the cumulative normal distribution function
11
12   Dependent variable = Effect
13   Independent variable = Dose
14   Slope parameter is not restricted
15
16   Total number of observations = 4
17   Total number of records with missing values = 0
18   Maximum number of iterations =250
19   Relative Function Convergence has been set to: le-008
20   Parameter Convergence has been set to: le-008
21
22
23   Default Initial (and Specified) Parameter Values
24   background = 0 Specified
25   intercept = -1.51718
26   slope = 0.00831843
27
28   Asymptotic Correlation Matrix of Parameter Estimates
29   (*** Thg model parameter(s) -background have been estimated at a boundary point, or
30   have been specified by the user, and do not appear in the correlation matrix )
31
32    intercept slope
33   intercept 1 -0.69
34   slope -0.69 1
35
36
37                                     Parameter Estimates
38    95.0% Wald Confidence Interval
39   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40   intercept 1.53138 0.160195 -1.84535 -1.2174
41   slope 0.00840347 0.000976752 0.00648907 0.0103179
42
43
44    Analysis of Deviance Table
45
46    Model Log(likelihood) # Param's Deviance Test d.f. P-value
47    Full model -71.8804 4
48    Fitted model -71.8937 2 0.0265818 2 0.9868
49    Reduced model -115.644 1 87.528 3 <.0001
50
51    AIC: 147.787
52
53
54    Goodness of Fit
55    Scaled
56    Dose Est._Prob.  Expected Observed Size Residual
57    	
58    0.0000 0.0628 3.142 3.000 50 -0.083
59    11.0000 0.0751 3.754 4.000 50 0.132
60    55.0000 0.1425 7.125 7.000 50 -0.050
61    274.0000 0.7797 38.985 39.000 50 0.005
62
63    ChiA2 = 0.03 d.f. = 2 P-value = 0.9867
64
65    Benchmark Dose Computation
66
67   Specified effect =0.1

                                                                                             D-10
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
Risk Type = Extra  risk
Confidence level = 0.95
 BMD = 62.1952
 BMDL = 51.1158
                                  Multistage Cancer Model with 0.95 Confidence Level
                                             Multistage Cancer 	
                                           Linear extrapolation 	
       T3
       
-------
 1
 2   Maximum number of iterations =250
 3   Relative Function Convergence has been set to: le-008
 4   Parameter Convergence has been set to: le-008
 5
 6   Default Initial Parameter Values
 7   Background = 0.0623822
 8   Beta(l) = 0.00142752
 9   Beta(2) = 0
10   Beta(3) = 5.14597e-008
11   Asymptotic Correlation Matrix of Parameter Estimates
12   (*** The model parameter(s) -Beta(2)have been estimated at a boundary point, or have
13   been specified by the user, and do not appear in the correlation matrix  )
14
15    Background Beta(l) Beta(3)
16   Background 1 -0.67 0.58
17   Beta(l) -0.67 1 -0.95
18   Beta(3) 0.58 -0.95 1
19
20
21    Parameter Estimates
22
23    95.0% Wald Confidence Interval
24   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
25   Background 0.0619918 * * *
26   Beta(l) 0.001449 * * *
27   Beta(2) 0 * * *
28   Beta(3) 5.11829e-008 * * *
29
30   * - Indicates that this value is not calculated.
31
32
33
34    Analysis of Deviance Table
35
36    Model Log(likelihood) # Param's Deviance Test d.f. P-value
37    Full model -71.8804 4
38    Fitted model -71.8858 3 0.0107754 1 0.9173
39    Reduced model -115.644 1 87.528 3 <.0001
40
41    AIC: 149.772
42
43
44    Goodness of Fit
45    Scaled
46    Dose Est._Prob. Expected Observed Size Residual
47    	~	
48    0.0000 0.0620 3.100 3.000 50 -0.058
49    11.0000 0.0769 3.844 4.000 50 0.083
50    55.0000 0.1412 7.059 7.000 50 -0.024
51    274.0000 0.7799 38.997 39.000 50 0.001
52
53    Chi^2 = 0.01 d.f. = 1 P-value = 0.9171
54
55
56    Benchmark Dose Computation
57
58   Specified effect =0.1
59   Risk Type = Extra risk
60   Confidence level = 0.95
61    BMD = 63.6179
62    BMDL = 27.4913
63    BMDU = 123.443
64
65   Taken together,  (27.4913, 123.443) is a 90% two-sided confidence interval for the BMD
66
67   Multistage Cancer Slope Factor = 0.00363752

                                                                                             D-12
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     D.4  F344 Rats:  Tumors  at Other Sites

1           The data for tumors at sites other than the liver in male and female F344 rats (Kano et al.. 2009)
2    are shown in Table D-6. Note that the incidence of rats with these endpoints are monotone non-decreasing
3    functions (except female peritoneal mesotheliomas). These data therefore appear to be appropriate for
4    dose-response modeling using BMDS.

            Table D-6   Data for significant tumors at other sites in male and female F344 rats
                       (Kano et al.. 2009)
Dose (mg/kg-day)
Tumor site and type
Nasal cavity squamous cell carcinoma
Peritoneal mesothelioma
Mammary gland adenoma
Total number per group
Female
0
0
1
6
50
18
0
0
7
50
83
0
0
10
50
429
7
0
16
50
0
0
2
0
50
Male
11
0
2
1
50
55
0
5
2
50
274
3
28
2
50
     Source: Used with permission from Elsevier, Ltd., Kano et al., (2009).

5           The results of the BMDS modeling for the entire suite of models are presented in Table D-7
6    through Table D-10 for tumors in the nasal cavity, mammary gland, and peritoneal cavity.
                                                                                                   D-13
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-7   BMDS dose-response modeling results for the incidence of nasal cavity tumors in
                 female F344 rats3 (Kano et al.. 2009)

Model
Gamma
Logistic
LogLogistic
LogProbitc
(7d^TCancer
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)d
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
44.4964
44.4963
44.4963
44.4963
45.6604
43.0753
42.6063
44.4963
44.4963
45.6604
46.4963
p-value
1
1
1
1
0.6184
0.9607
0.9966
1
1
0.6184
0.9997
mg/kg-day
403.82
421.54
413.69
400.06
375.81
366.07
381.65
414.11
414.86
375.81
413.96
BMDLio
mg/kg-day
269.03
351.74
268.85
260.38
213.84
274.63
282.61
333.31
273.73
213.84
372.57
x2b
0
0
0
0
0.595
0.109
0.021
0
0
0.595
1.64x10'B
mg/kg-day
100.35
104.75
102.80
99.42
93.39
90.97
94.84
102.91
103.09
93.39
102.87
BMDLio HED
mg/kg-day
66.85
87.41
66.81
64.71
53.14
68.24
70.23
82.83
68.02
53.14
92.58
     "Nasal cavity tumors in female F344 rats include squamous cell carcinoma and esthesioneuro-epithelioma.
     ""Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
     °Slope restricted > 1.
     dBest-fitting model.
                                    Multistage uancer Model with u.yt> uontidence Level
                 0.3
                0.25
                 0.2
                0.15
                 0.1
                0.05
                                                Multistage Cancer
                                               Linear extrapolation
                                                                 BMDL
                                                                                       BMD
                                 50
                                        100
                                                150
                                                        200     250
                                                          dose
                                                                        300
                                                                                350
                                                                                        400
                                                                                                 450
        07:28 10/26 2009

            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-4  Multistage BMD model (3 degree) for nasal cavity tumors in female
                        F344 rats.
1    =========================================================
2    Multistage Cancer Model .  (Version:  1.7;  Date:  05/16/2008)
3    Input Data File:
     L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-BMR10-3poly. (d)
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                      D-14

-------
 1   Gnuplot Plotting File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_nasal_car_Msc-BMR10-3poly.plt
 3   Mon Oct 26 08:28:58 2009
 4   ====================================================================
 5    BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7   The form of the probability function is: P[response] = background +
 8   (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
 9
10   The parameter betas are restricted to be positive
11
12   Dependent variable = Effect
13   Independent variable = Dose
14   Total number of observations = 4
15   Total number of records with missing values = 0
16   Total number of parameters in model = 4
17   Total number of specified parameters = 0
18   Degree of polynomial = 3
19
20   Maximum number of iterations =250
21   Relative Function Convergence has been set to: le-008
22   Parameter Convergence has been set to: le-008
23
24   Default Initial Parameter Values
25   Background = 0
26   Beta(l) = 0
27   Beta(2) = 0
28   Beta(3) = 1.91485e-009
29   Asymptotic Correlation Matrix of Parameter Estimates
30   (*** The model parameter(s) -Background -Beta(l) -Beta(2)
31   have been estimated at a boundary point, or have been specified by the user,
32   and do not appear in the correlation matrix )
33
34    Beta(3)
35    Beta (3) 1
36
37    Parameter Estimates
38
39    95.0% Wald Confidence Interval
40   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
41   Background 0 * * *
42   Beta(l) 0 * * *
43   Beta(2) 0 * * *
44   Beta(3) 1.89531e-009 * * *
45
46   * - Indicates that this value is not calculated.
47
48
49    Analysis of Deviance Table
50
51    Model Log(likelihood) # Param's Deviance Test d.f. P-value
52    Full model -20.2482 4
53    Fitted model -20.3031 1 0.109908 3 0.9906
54    Reduced model -30.3429 1 20.1894 3 0.0001551
55
56    AIC: 42.6063
57
58
59    Goodness of Fit
60    Scaled
61    Dose Est._Prob. Expected Observed Size Residual
62    	
63    0.0000 0.0000 0.000 0.000 50 0.000
64    18.0000 0.0000 0.001 0.000 50 -0.024
65    83.0000 0.0011 0.054 0.000 50 -0.233
66    429.0000 0.1390 6.949 7.000 50 0.021
67

                                                                                              D-15
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1     ChiA2 = 0.06 d.f.  = 3 P-value = 0.9966
 2
 3
 4     Benchmark Dose Computation
 5
 6    Specified effect = 0.1
 7    Risk Type = Extra risk
 8    Confidence level = 0.95
 9     BMD = 381.651
10     BMDL = 282.609
11     BMDU = 500.178
12
13    Taken together, (282.609,  500.178)  is a 90%  two-sided confidence  interval  for  the BMD
14
15    Multistage Cancer Slope Factor = 0.000353846
                                                                                             D-16
                                 DRAFT - DO NOT CITE OR QUOTE

-------
Table D-8   BMDS dose-response modeling results for the incidence of nasal cavity tumors in
             male F344 rats3 (Kano et al.. 2009)

Model
Gamma
Logistic
LogLogistic
LogProbitc
AIC
26.6968
26.6968
26.6968
26.6968
p-value
1
1
1
1
BMDio
mg/kg-day
299.29
281.06
288.31
303.06
BMDLio
mg/kg-day
244.10
261.29
245.29
238.86
x2b
0
0
0
0
BMDio HED
mg/kg-day
83.89
78.78
80.81
84.94
BMDLio HED
mg/kg-day
68.42
73.24
68.75
66.95
Multistage-Cancer
(1 degree)	
26.0279    0.8621
           582.49
             256.43
            0.384
           163.28
              71.88
Multistage-Cancer
(2 degree)	
24.9506
 0.988
365.19
242.30
0.073
102.37
67.92
Multistage-Cancer
(3 degree)d	
 24.747
0.9989
328.11
245.63
0.015
91.97
68.85
Probit
26.6968
   1
287.96
257.01
  0
80.72
72.04
Weibull
26.6968
   1
288.00
246.36
  0
80.73
69.06
Quantal-Linear
26.0279    0.8621
           582.49
             256.43
            0.384
           163.28
              71.88
Dichotomous-Hill
28.6968    0.9994
           290.52
             261.47
          6.25x10"
           81.44
              73.29
"Nasal cavity tumors in male F344 rats include squamous cell carcinoma, Sarcoma: NOS, rhabdomyosarcoma, and
esthesioneuro-epithelioma.
bMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
°Slope restricted > 1.
dBest-fitting model.
                                   Multistage cancer ivioaei wnn u.ao (^onnaence i_evei
  o
 'o
  CO
            0.15
              0.1
            0.05
                                                Multistage Cancer
                                              Linear extrapolation
                                                                         BMDL
                                                                                                  BMO
                                 50
                                            100
                                                        150        200
                                                          dose
                                                                               250
                                                                                          300
   07:34 10/26 2009

        Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

        Figure D-5 Multistage BMD model (3 degree) for nasal cavity tumors in male F344
                    rats.
Multistage Cancer Model.   (Version:  1.7; Date: 05/16/2008)
                                                                                                      D-17
                                 DRAFT - DO NOT CITE OR QUOTE

-------
 1   Input Data File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_nasal_car_Msc-BMR10-3poly.(d)
 3   Gnuplot Plotting File:
 4   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_nasal_car_Msc-BMR10-3poly.plt
 5   Mon Oct 26 08:34:20 2009
 6   ====================================================================
 7   BMDS Model Run
 8   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 9   The form of the probability function is: P[response] = background +
10   (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
11
12   The parameter betas are restricted to be positive
13
14   Dependent variable = Effect
15   Independent variable = Dose
16   Total number of observations = 4
17   Total number of records with missing values = 0
18   Total number of parameters in model = 4
19   Total number of specified parameters = 0
20   Degree of polynomial = 3
21
22   Maximum number of iterations =250
23   Relative Function Convergence has been set to: le-008
24   Parameter Convergence has been set to: le-008
25
26   Default Initial Parameter Values
27   Background = 0
28   Beta(l) = 0
29   Beta(2) = 0
30   Beta(3) = 3.01594e-009
31
32
33   Asymptotic Correlation Matrix of Parameter Estimates
34
35   (*** The model parameter(s) -Background -Beta(l) -Beta (2)
36   have been estimated at a boundary point, or have been specified by the user,
37   and do not appear in the correlation matrix )
38
39    Beta(3)
40    Beta (3) 1
41
42
43    Parameter Estimates
44
45    95.0% Wald Confidence Interval
46   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
47   Background 0 * * *
48   Beta(l) 0 * * *
49   Beta(2) 0 * * *
50   Beta(3) 2.98283e-009 * * *
51
52   * - Indicates that this value is not calculated.
53
54
55
56    Analysis of Deviance Table
57
58    Model Log(likelihood) # Param's Deviance Test d.f. P-value
59    Full model -11.3484 4
60    Fitted model -11.3735 1 0.0502337 3 0.9971
61    Reduced model -15.5765 1 8.45625 3 0.03747
62
63    AIC: 24.747
64
65
66    Goodness of Fit
67    Scaled

                                                                                              D-18
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Dose Est._Prob. Expected Observed Size Residual
 2	
 3    0.0000 0.0000 0.000 0.000 50 0.000
 4    11.0000 0.0000 0.000 0.000 50 -0.014
 5    55.0000 0.0005 0.025 0.000 50 -0.158
 6    274.0000 0.0595 2.976 3.000 50 0.015
 7
 8    ChiA2 = 0.03 d.f. = 3 P-value = 0.9989
 9
10
11    Benchmark Dose Computation
12
13   Specified effect =0.1
14   Risk Type = Extra risk
15   Confidence level = 0.95
16    BMD = 328.108
17    BMDL = 245.634
18    BMDU = 1268.48
19
20   Taken together, (245.634, 1268.48)  is a 90% two-sided confidence interval for the BMD
21
22   Multistage Cancer Slope Factor = 0.00040711
                                                                                             D-19
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Table D-9   BMDS dose-response modeling results for the incidence of mammary gland
            adenomas in female F344 rats (Kano et al.. 2009)

Model
Gamma
Logistic
Log Logistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
194.222
194.475
194.151
195.028
194.222
194.222
194.222
194.441
194.222
194.222
197.916
p-value
0.8559
0.7526
0.8874
0.5659
0.8559
0.8559
0.8559
0.7656
0.8559
0.8559
NC°
BMDio
mg/kg-day
176.66
230.35
161.01
270.74
176.66
176.66
176.66
223.04
176.65
176.65
94.06
BMDLio
mg/kg-day
99.13
159.73
81.91
174.66
99.13
99.13
99.13
151.60
99.13
99.13
14.02
x2a
0.465
0.612
0.406
-0.075
0.465
0.465
0.465
0.596
0.465
0.465
3.49x10'b
BMDio HED
mg/kg-day
43.90
57.24
40.01
67.28
43.90
43.90
43.90
55.43
43.90
43.90
23.37
BMDLio HED
mg/kg-day
24.63
39.69
20.35
43.41
24.63
24.63
24.63
37.67
24.63
24.63
3.48
aMaximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Best-fitting model.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
                                  Log-Logistic Model with 0.95 Confidence Level
 o
 OS
           0.5
           0.4
           0.3
           0.2
           0.1
                                   100     150      200     250      300     350     400     45(
   11:31 02/01 2010

        Source: Use with permission from Elsevier, Ltd., Kano et al. (2009).

        Figure D-6 LogLogistic BMD model for mammary gland adenomas in female F344
                   rats.


Logistic Model.  (Version:  2.12;  Date: 05/16/2008)


                                DRAFT - DO NOT CITE OR QUOTE
                                                                                                  D-20

-------
 1   Input Data File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.(d)
 2   Gnuplot Plotting File: C:\14DBMDS\lnl_kano2009_frat_mamm_ad_Lnl-BMR10-Restrict.plt
 3                                                  Mon Feb 01 11:31:31 2010
 4   ====================================================================
 5    BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7    The form of the probability function is:
 8
 9    P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
10
11    Dependent variable = Effect
12    Independent variable = Dose
13    Slope parameter is restricted as slope >= 1
14
15    Total number of observations = 4
16    Total number of records with missing values = 0
17    Maximum number of iterations = 250
18    Relative Function Convergence has been set to: le-008
19    Parameter Convergence has been set to: le-008
20
21    User has chosen the log transformed model
22
23    Default Initial Parameter Values
24    background = 0.12
25    intercept = -7.06982
26    slope = 1
27   Asymptotic Correlation Matrix of Parameter Estimates
28
29   (*** Thg model parameter(s) -slope have been estimated at a boundary point,  or have
30   been specified by the user, and do not appear in the correlation matrix )
31
32    background intercept
33   background 1 -0.53
34    intercept -0.53 1
35
36    Parameter Estimates
37
38    95.0% Wald Confidence Interval
39    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
40   background 0.130936 * * *
41    intercept -7.2787 * * *
42    slope 1 * * *
43
44   * - Indicates that this value is not calculated.
45
46
47
48    Analysis of Deviance Table
49
50    Model Log(likelihood) # Param's Deviance Test d.f. P-value
51    Full model -94.958 4
52    Fitted model -95.0757 2 0.235347 2 0.889
53    Reduced model -98.6785 1 7.4409 3 0.0591
54
55    AIC: 194.151
56
57
58    Goodness of Fit
59    Scaled
60    Dose Est._Prob. Expected Observed Size Residual
61    	
62    0.0000 0.1309 6.547 6.000 50 -0.229
63    18.0000 0.1416 7.080 7.000 50 -0.032
64    83.0000 0.1780 8.901 10.000 50 0.406
65    429.0000 0.3294 16.472 16.000 50 -0.142
66
67    ChiA2 = 0.24 d.f. = 2 P-value = 0.8874

                                                                                             D-21
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 Benchmark Dose  Computation
Specified effect =0.1
Risk Type = Extra risk
Confidence level = 0.95
 BMD = 161.012
 BMDL = 81.9107
                                  Multistage Cancer Model with 0.95 Confidence Level
                0.5
                0.4
                0.3
                0.2
                0.1
                                            Multistage Cancer 	
                                           Linear extrapolation
                              50
                                     100
                                             150
                                                    200    250
                                                      dose
                                                                   300
                                                                          350
                                                                                  400
                                                                                         450
          07:27 10/26 2009

            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-7 Multistage BMD model (1 degree) for mammary gland adenomas in
                       female F344 rats.
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Multistage Cancer Model.  (Version: 1.7; Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_mamm_ad_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_frat_mamm_ad_Msc-BMR10-lpoly.plt
Mon Oct 26 08:27:02  2009

 BMDS Model Run

The form of the probability function is:

P[response] = background  + (1-background)*[1-EXP(-betal*dose/xl)]

The parameter betas  are restricted to be positive
Dependent variable  = Effect
Independent variable = Dose

Total number  of  observations = 4
Total number  of  records with missing values =  0
Total number  of  parameters in model = 2
Total number  of  specified parameters = 0
Degree of polynomial = 1
                                                                                                D-22
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2   Maximum number of iterations =250
 3   Relative Function Convergence has been set to: le-008
 4   Parameter Convergence has been set to: le-008
 5
 6   Default Initial Parameter Values
 7   Background = 0.136033
 8   Beta(l) = 0.000570906
 9   Asymptotic Correlation Matrix of Parameter Estimates
10
11    Background Beta(l)
12   Background 1 -0.58
13   Beta(l) -0.58 1
14
15
16   Parameter Estimates
17
18    95.0% Wald Confidence Interval Variable Estimate Std. Err. Lower Conf. Limit Upper
19   Conf. Limit
20   Background .133161 * * *
21   Beta(l) 0.000596394 * * *
22
23   * - Indicates that this value is not calculated.
24
25
26
27    Analysis of Deviance Table
28
29    Model Log(likelihood) # Param's Deviance Test d.f. P-value
30    Full model -94.958 4
31    Fitted model -95.111 2 0.305898 2 0.8582
32    Reduced model -98.6785 1 7.4409 3 0.0591
33
34    AIC: 194.222
35
36
37    Goodness of Fit
38    Scaled
39    Dose Est._Prob. Expected Observed Size Residual
40    	
41    0.0000 0.1332 6.658 6.000 50 -0.274
42    18.0000 0.1424 7.121 7.000 50 -0.049
43    83.0000 0.1750 8.751 10.000 50 0.465
44    429.0000 0.3288 16.442 16.000 50 -0.133
45
46    ChiA2 = 0.31 d.f. = 2 P-value = 0.8559
47
48
49    Benchmark Dose Computation
50
51   Specified effect =0.1
52   Risk Type = Extra risk
53   Confidence level = 0.95
54    BMD =176.663
55    BMDL = 99.1337
56    BMDU = 501.523
57
58   Taken together, (99.1337, 501.523) is a 90% two-sided confidence interval for the BMD
59
60   Multistage Cancer Slope Factor = 0.00100874
                                                                                             D-23
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-10  BMDS dose-response modeling results for the incidence of peritoneal mesotheliomas
                 in male F344 rats (Kano et al.. 2009)

Model
Gamma
Logistic
LogLogistic
LogProbit"
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
140.701
139.016
140.699
140.69
140.826
140.747
140.747
138.869
140.709
140.826
2992
p-value
0.9189
0.8484
0.9242
0.9852
0.3617
0.8135
0.8135
0.9148
0.8915
0.3617
NCa
BMDio
mg/kg-day
73.52
103.52
72.56
70.29
41.04
77.73
77.73
93.06
74.77
41.04
NCa
BMDLio
mg/kg-day
35.62
84.35
36.37
52.59
30.51
35.43
35.43
76.32
35.59
30.51
NC°
x2a
0.018
0.446
0.014
0.001
-1.066
0.067
0.067
0.315
0.027
-1.066
0
BMDlOHED
mg/kg-day
20.61
29.02
20.34
19.70
11.50
21.79
21.79
26.09
20.96
11.50
0
BMDLio HED
mg/kg-day
9.98
23.65
10.19
14.74
8.55
9.93
9.93
21.39
9.97
8.55
0
     "Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
     "Slope restricted > 1.
     ""Best-fitting model.
     dValue unable to be calculated (NC: not calculated) by BMDS.
                                          Probit Model with 0.95 Confidence Level
1
2
3
4
 o

 £5
                0.7
                0.6
                0.5
                0.4
                0.3
                0.2
                0.1
                 0   -
                                                  100
                                                               150
                                                                             200
                                                                                          250
                                                           dose
        07:41 10/262009

            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-8  Probit BMD model for peritoneal mesotheliomas in male F344 rats.
Probit Model.  (Version:  3.1;  Date:  05/16/2008)
Input  Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_peri_meso_Prb-BMR10.(d)
                                                                                                      D-24
                                    DRAFT - DO NOT CITE OR QUOTE

-------
 1   Gnuplot Plotting File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kano2009_mrat_peri_meso_Prb-BMR10.plt
 3   Mon Oct 26 08:41:29 2009
 4   ====================================================================
 5   BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7
 8   The form of the probability function is: P[response] = CumNorm(Intercept+Slope*Dose) ,
 9   where CumNorm(.)  is the cumulative normal distribution function
10
11   Dependent variable = Effect
12   Independent variable = Dose
13   Slope parameter is not restricted
14
15   Total number of observations = 4
16   Total number of records with missing values = 0
17   Maximum number of iterations =250
18   Relative Function Convergence has been set to: le-008
19   Parameter Convergence has been set to: le-008
20
21   Default Initial (and Specified) Parameter Values
22   background = 0 Specified
23   intercept = -1.73485
24   slope = 0.00692801
25
26   Asymptotic Correlation Matrix of Parameter Estimates
27   (*** The model parameter(s) -background have been estimated at a boundary point, or
28   have been specified by the user, and do not appear in the correlation matrix )
29
30    intercept slope
31    intercept 1 -0.75
32   slope -0.75 1
33
34                                     Parameter Estimates
35    95.0% Wald Confidence Interval
36   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
37   intercept -1.73734 0.18348 -2.09695 -1.37772
38   slope 0.00691646 0.000974372 0.00500672 0.00882619
39
40    Analysis of Deviance Table
41    Model Log(likelihood) # Param's Deviance Test d.f. P-value
42    Full model -67.3451 4
43    Fitted model -67.4344 2 0.178619 2 0.9146
44    Reduced model -95.7782 1 56.8663 3 <.0001
45    AIC: 138.869
46
47    Goodness of Fit
48    Scaled
49    Dose Est._Prob.  Expected Observed Size Residual
50    	
51    0.0000 0.0412 2.058 2.000 50 -0.041
52    11.0000 0.0483 2.417 2.000 50 -0.275
53    55.0000 0.0874 4.370 5.000 50 0.315
54    274.0000 0.5627 28.134 28.000 50 -0.038
55
56    ChiA2 = 0.18 d.f. = 2 P-value = 0.9148
57    Benchmark Dose Computation
58   Specified effect =0.1
59   Risk Type = Extra risk
60   Confidence level = 0.95
61    BMD = 93.0615
62    BMDL = 76.3242
                                                                                             D-25
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                  Multistage Cancer Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
  o
  t3
                0.7


                0.6


                0.5


                0.4


                0.3


                0.2


                0.1
                                             Multistage Cancer
                                           Linear extrapolation
                           BMDL
                                          BMD
                                   50
                                              100
                                                          150
                                                                      200
                                                                                  250
                                                      dose
         07:41 10/26 2009
            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-9 Multistage BMD (2 degree) model for peritoneal mesotheliomas in male
                       F344 rats.
Multistage Cancer Model.  (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_peri_meso_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mrat_peri_meso_Msc-BMR10-2poly.plt
Mon Oct 26 08:41:28  2009

BMDS Model Run
The form of the probability function is:

P [response] = background + (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]

The parameter betas  are  restricted to be positive
Dependent variable  =  Effect
Independent variable  = Dose

Total number  of  observations = 4
Total number  of  records with missing values = 0
Total number  of  parameters in model = 3
Total number  of  specified parameters = 0
 Degree of polynomial = 2

Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial  Parameter Values
Background =  0.0358706
                                                                                                D-26
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1   Beta(l) = 0.000816174
 2   Beta(2) = 7.47062e-006
 3
 4
 5   Asymptotic Correlation Matrix of Parameter Estimates
 6
 7    Background Beta(l)  Beta(2)
 8   Background 1 -0.67 0.59
 9   Beta(l) -0.67 1 -0.98
10   Beta(2) 0.59 -0.98 1
11
12                                     Parameter Estimates
13    95.0% Wald Confidence Interval
14   Variable Estimate Std. Err.  Lower Conf.  Limit Upper Conf.  Limit
15   Background 0.0366063 * * *
16   Beta(l) 0.000757836 * * *
17   Beta(2) 7.6893e-006 * * *
18
19   * - Indicates that this value is not calculated.
20
21    Analysis of Deviance Table
22
23    Model Log(likelihood) # Param's Deviance Test d.f.  P-value
24    Full model -67.3451 4
25    Fitted model -67.3733 3 0.056567 1 0.812
26    Reduced model -95.7782 1  56.8663 3 <.0001
27
28    AIC: 140.747
29
30
31    Goodness of Fit
32    Scaled
33    Dose Est._Prob. Expected  Observed Size  Residual
34    	~	
35    0.0000 0.0366 1.830 2.000 50 0.128
36    11.0000 0.0455 2.275 2.000 50 -0.186
37    55.0000 0.0972 4.859 5.000 50 0.067
38    274.0000 0.5605 28.027 28.000 50 -0.008
39
40    Chi^2 = 0.06 d.f. = 1 P-value = 0.8135
41
42
43    Benchmark Dose Computation
44
45   Specified effect =0.1
46   Risk Type = Extra risk
47   Confidence level =0.95
48    BMD = 77.7277
49    BMDL = 35.4296
50    BMDU = 118.349
51
52   Taken together, (35.4296,  118.349)  is a  90% two-sided  confidence  interval  for  the BMD
53
54   Multistage Cancer Slope Factor = 0.0028225
     D.5  Female BDF1 Mice: Hepatic Carcinomas  and Adenomas

55          Data for female BDF1 mouse hepatic carcinomas and adenomas are shown in Table D-l 1. Note
56   that the incidence of carcinomas and the incidence of either adenomas or carcinomas are monotone
57   non-decreasing functions of dose. These data therefore appear to be appropriate for dose-response
58   modeling using BMDS. However, the incidence of adenomas clearly reaches a peak value at

                                                                                             D-27
                                 DRAFT - DO NOT CITE OR QUOTE

-------
 1   66 mg/kg-day and then decreases sharply with increasing dose. This cannot be modeled by a multistage
 2   model using only non-negative coefficients. To some extent the incidence of "either adenomas or
 3   carcinomas" retains some of the inverted-U shaped dose-response of the adenomas, which dominate
 4   based on their high incidence  at the lowest dose groups (66 and 278 mg/kg-day), thus is not well
 5   characterized by any multistage model.

     Table D-11  Data for hepatic adenomas and carcinomas in female BDF1 mice (Kano et al.. 2009)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
0
5
0
5
45
50
66
31
6
35
15
50
278
20
30
41
9
50
964
3
45
46
4
50
Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).
 6           The results of the BMDS modeling for the entire suite of models for hepatic adenomas and
 7   carcinomas in female BDF1 mice are presented in Table D-12. The multistage models did not provide
 8   reasonable fits to the incidence data for hepatocellular adenoma or carcinoma in female BDF1 mice. The
 9   log-logistic model provided the best-fit to the data as indicated by the AIC and/>-value as was chosen as
10   the best-fitting model to carry forward in the analysis; however, this model resulted in a BMDLio much
11   lower than the response level at the lowest dose in the study (Kano et al.. 2009). Thus, the log-logistic
12   model was run for BMRs of 30 and 50%. The output from these models are shown in Figures D-11 and
13   D-12. A summary of the  BMD results for BMRs of 10, 30, and 50% are shown in Table D-13. Using a
14   higher BMR resulted in BMDLs closer to the lowest observed response data, and a BMR of 50% was
15   chosen to carry forward in the analysis.

16           The graphical output from fitting these models suggested that a simpler model obtained by
17   dropping the data point for the highest dose (964 mg/kg-day) might also be adequate. This was tested and
18   the results did not affect the choice of the model, nor significantly affect the resulting BMDs and BMDLs.
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                    D-28

-------
Table D-12  BMDS dose-response modeling results for the combined incidence of hepatic
            adenomas and carcinomas in female BDF1 mice (Kano et al.. 2009)
Model
Gamma
Logistic
Log Logistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
203.331
214.951
176.214
198.354
203.331
203.331
203.331
217.671
203.331
203.331
7300.48
p-value
0
0
0.1421
0
0
0
0
0
0
0
NCa
BMDio
mg/kg-day
26.43
58.05
5.54
26.37
26.43
26.43
26.43
69.89
26.43
26.43
NCa
BMDLio
mg/kg-day
19.50
44.44
3.66
19.57
19.50
19.50
19.50
56.22
19.50
19.50
NCa
x23
-2.654
3.201
-0.121
-1.166
-2.654
-2.654
-2.654
3.114
-2.654
-2.654
0
BMDlOHED
mg/kg-day
3.98
8.74
0.83
3.97
3.98
3.98
3.98
10.5
3.98
3.98
0
BMDLio HED
mg/kg-day
2.94
6.69
0.55
2.95
2.94
2.94
2.94
8.46
2.94
2.94
0
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
b Best-fitting model, lowest AIC value.
°Slope restricted > 1.
dValue unable to be calculated (NC: not calculated) by BMDS.
Table D-13  BMDS LogLogistic dose-response modeling results using BMRs of 10, 30, and 50%
            for the combined incidence of hepatic adenomas and carcinomas in female
            BDF1 mice (Kano et al.. 2009).
BMR
10%
30%
50%
AIC
176.214
176.214
176.214
p-value
0.1421
0.1421
0.1421
BMD
mg/kg-day
5.54
21.38
49.88
BMDL
mg/kg-day
3.66
14.11
32.93
x2a
-0.121
-0.121
0
BMDHED
mg/kg-day
0.83
3.22
7.51
BMDLHED
mg/kg-day
0.55
2.12
4.95
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
                                                                                                D-29
                               DRAFT - DO NOT CITE OR QUOTE

-------
                                   Log-Logistic Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
      •
      I
      C
      o
      13
      ro
               0.8
               0.6
               0.4
               0.2
                                 Log-Logistic
                 Bli/IDLJBMD
                                  200
                                          400
600
800
1000
                                                    dose
        11:2605/122010
            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-10   LogLogistic BMD model for the combined incidence of hepatic
                       adenomas and carcinomas in female BDF1 mice with a BMR of 10%.
Logistic Model.  (Version:  2.12;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR10-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR10-Restrict.p
It
                                                Wed May 12 11:26:35 2010

 BMDS Model Run

 The form of the probability function is:
 P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]

 Dependent variable = Effect
 Independent variable =  Dose
 Slope parameter is restricted as slope >= 1

 Total number of observations = 4
 Total number of records with missing values = 0
 Maximum number of iterations = 250
 Relative Function Convergence has been set to: le-008
 Parameter Convergence has been set to: le-008

 User has chosen the log transformed model
                                                                                                D-30
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    Default Initial Parameter Values
 3    background = 0.1
 4    intercept = -4.33618
 5    slope = 1
 6
 7    Asymptotic Correlation Matrix of Parameter Estimates
 8    (*** The model parameter(s) -slope have been estimated at a boundary point, or have
 9   been specified by the user, and do not appear in the correlation matrix )
10
11    background intercept
12   background 1 -0.32
13    intercept -0.32 1
14
15    Parameter Estimates
16
17    95.0% Wald Confidence Interval
18    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
19   background 0.105265 * * *
20    intercept -3.90961 * * *
21    slope 1 * * *
22
23   * - Indicates that this value is not calculated.
24
25    Analysis of Deviance Table
26
27    Model Log(likelihood) # Param's Deviance Test d.f. P-value
28    Full model -84.3055 4
29    Fitted model -86.107 2 3.6029 2 0.1651
30    Reduced model -131.248 1 93.8853 3 <.0001
31
32    AIC: 176.214
33
34
35    Goodness of Fit
36    Scaled
37    Dose Est._Prob. Expected Observed Size Residual
38    	~	
39    0.0000 0.1053 5.263 5.000 50 -0.121
40    66.0000 0.6149 30.743 35.000 50 1.237
41    278.0000 0.8639 43.194 41.000 50 -0.905
42    964.0000 0.9560 47.799 46.000 50 -1.240
43
44    Chi^2 = 3.90 d.f. = 2 P-value = 0.1421
45
46
47    Benchmark Dose Computation
48   Specified effect =0.1
49   Risk Type = Extra risk
50   Confidence level =0.95
51    BMD = 5.54218
52    BMDL = 3.65848
53
                                                                                             D-31
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                   Log-Logistic Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
      •
      I
      C
      o
      13
      ro
               0.8
               0.6
               0.4
               0.2
                                 Log-Logistic
                  EMDLBI
                  BMD
                                  200
                                          400
600
800
1000
                                                    dose
        11:2605/122010
            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-ll    LogLogistic BMD model for the combined incidence of hepatic
                       adenomas and carcinomas in female BDF1 mice with a BMR of 30%.
Logistic Model.  (Version:  2.12;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR30-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR30-Restrict.p
It
                                                Wed May 12 11:26:36 2010

 BMDS Model Run

 The form of the probability function is:
 P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]

 Dependent variable = Effect
 Independent variable =  Dose
 Slope parameter is restricted as slope >= 1

 Total number of observations = 4
 Total number of records with missing values = 0
 Maximum number of iterations = 250
 Relative Function Convergence has been set to: le-008
 Parameter Convergence has been set to: le-008
 User has chosen the log transformed model
                                                                                               D-32
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    Default Initial Parameter Values
 2    background = 0.1
 3    intercept = -4.33618
 4    slope = 1
 5
 6    Asymptotic Correlation Matrix of Parameter Estimates
 7    (*** The model parameter(s) -slope have been estimated at a boundary point, or have
 8   been specified by the user, and do not appear in the correlation matrix)
 9
10    background intercept
11   background 1 -0.32
12    intercept -0.32 1
13
14    Parameter Estimates
15
16    95.0% Wald Confidence Interval
17    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
18   background 0.105265 * * *
19    intercept -3.90961 * * *
20    slope 1 * * *
21
22   * - Indicates that this value is not calculated.
23
24
25    Analysis of Deviance Table
26
27    Model Log(likelihood) # Param's Deviance Test d.f. P-value
28    Full model -84.3055 4
29    Fitted model -86.107 2 3.6029 2 0.1651
30    Reduced model -131.248 1 93.8853 3 <.0001
31
32    AIC: 176.214
33
34
35    Goodness of Fit
36    Scaled
37    Dose Est._Prob. Expected Observed Size Residual
38    	~	
39    0.0000 0.1053 5.263 5.000 50 -0.121
40    66.0000 0.6149 30.743 35.000 50 1.237
41    278.0000 0.8639 43.194 41.000 50 -0.905
42    964.0000 0.9560 47.799 46.000 50 -1.240
43
44    Chi^2 = 3.90 d.f. = 2 P-value = 0.1421
45
46
47    Benchmark Dose Computation
48   Specified effect =0.3
49   Risk Type = Extra risk
50   Confidence level =0.95
51    BMD = 21.377
52    BMDL = 14.1113
                                                                                             D-33
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                   Log-Logistic Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
      •
      I
      C
      o
      13
      ro
               0.8
               0.6
               0.4
               0.2
                                 Log-Logistic
                   BMDL
                    BMD
                                  200
                                          400
600
800
1000
                                                    dose
        11:2605/122010
            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-12   LogLogistic BMD model for the combined incidence of hepatic
                       adenomas and carcinomas in female BDF1 mice with a BMR of 50%.
Logistic Model.  (Version:  2.12;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR50-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_fmouse_hepato_adcar_Lnl-BMR50-Restrict.p
It
                                                Wed May 12 11:26:36 2010

 BMDS Model Run

 The form of the probability function is:
 P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]

 Dependent variable = Effect
 Independent variable =  Dose
 Slope parameter is restricted as  slope >= 1

 Total number of observations = 4
 Total number of records with missing values = 0
 Maximum number of iterations = 250
 Relative Function Convergence has been set to: le-008
 Parameter Convergence has been set to: le-008

 User has chosen the log transformed model
                                                                                               D-34
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    Default Initial Parameter Values
 3    background = 0.1
 4    intercept = -4.33618
 5    slope = 1
 6
 7    Asymptotic Correlation Matrix of Parameter Estimates
 8    (*** The model parameter(s) -slope have been estimated at a boundary point, or have
 9   been specified by the user, and do not appear in the correlation matrix)
10
11    background intercept
12   background 1 -0.32
13    intercept -0.32 1
14
15    Parameter Estimates
16
17    95.0% Wald Confidence Interval
18    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
19   background 0.105265 * * *
20    intercept -3.90961 * * *
21    slope 1 * * *
22
23   * - Indicates that this value is not calculated.
24
25    Analysis of Deviance Table
26
27    Model Log(likelihood) # Param's Deviance Test d.f. P-value
28    Full model -84.3055 4
29    Fitted model -86.107 2 3.6029 2 0.1651
30    Reduced model -131.248 1 93.8853 3 <.0001
31
32    AIC: 176.214
33
34    Goodness of Fit
35    Scaled
36    Dose Est._Prob. Expected Observed Size Residual
37    	
38    0.0000 0.1053 5.263 5.000 50 -0.121
39    66.0000 0.6149 30.743 35.000 50 1.237
40    278.0000 0.8639 43.194 41.000 50 -0.905
41    964.0000 0.9560 47.799 46.000 50 -1.240
42
43    ChiA2 = 3.90 d.f. = 2 P-value = 0.1421
44
45
46    Benchmark Dose Computation
47   Specified effect =0.5
48   Risk Type = Extra risk
49   Confidence level = 0.95
50    BMD = 49.8797
51    BMDL = 32.9263
                                                                                             D-35
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                 Multistage Cancer Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
      •
      I
      C
      o
      13
      ro
               0.8
               0.6
               0.4
               0.2
                                           Multistage Cancer
                                          Linear extrapolation
                   liMDLB
                   BMD
                                  200
                                          400
600
800
1000
                                                    dose
   11:2605/122010

       Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

       Figure D-13    Multistage BMD model (1 degree) for the combined incidence of
                  hepatic adenomas and carcinomas in female BDF1 mice.


Multistage Cancer Model.  (Version:  1.7; Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_fmouse_hepato_adcar_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_fmouse_hepato_adcar_Msc-BMR10-lpoly.plt
                                                Wed May 12 11:26:31 2010

 BMDS Model  Run

 The form of the probability function is:
 P[response] = background +  (1-background)*[1-EXP(-betal*dose/xl)]

 The parameter betas are  restricted to be positive

 Dependent variable  = Effect
 Independent variable = Dose

 Total number of observations = 4
 Total number of records  with missing values = 0
 Total number of parameters  in model = 2
 Total number of specified parameters = 0
 Degree of polynomial = 1

 Maximum number  of  iterations = 250
 Relative Function  Convergence  has  been set to: le-008
                                                                                                D-36
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    Parameter Convergence has been set to: le-008
 2
 3    Default Initial Parameter Values
 4    Background = 0.51713
 5    Beta(l) = 0.00201669
 6
 7    Asymptotic Correlation Matrix of Parameter Estimates
 8
 9    Background Beta(l)
10   Background 1 -0.65
11    Beta(l) -0.65 1
12
13    Parameter Estimates
14
15    95.0% Wald Confidence Interval
16    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17   Background 0.265826 * * *
18    Beta(l) 0.00398627 * * *
19
20   * - Indicates that this value is not calculated.
21
22    Analysis of Deviance Table
23
24    Model Log(likelihood) # Param's Deviance Test d.f. P-value
25    Full model -84.3055 4
26    Fitted model -99.6653 2 30.7195 2 2.1346928e-007
27    Reduced model -131.248 1 93.8853 3 <.0001
28
29    AIC: 203.331
30
31    Goodness of Fit
32    Scaled
33    Dose Est._Prob. Expected Observed Size Residual
34    	~	
35    0.0000 0.2658 13.291 5.000 50 -2.654
36    66.0000 0.4357 21.783 35.000 50 3.770
37    278.0000 0.7576 37.880 41.000 50 1.030
38    964.0000 0.9843 49.213 46.000 50 -3.651
39
40    Chi^2 = 35.65 d.f. = 2 P-value = 0.0000
41
42
43    Benchmark Dose Computation
44   Specified effect = 0.1
45
46   Risk Type = Extra risk
47   Confidence level =0.95
48    BMD = 26.4309
49    BMDL = 19.5045
50    BMDU = 37.5583
51
52   Taken together, (19.5045, 37.5583) is  a 90% two-sided confidence interval for the BMD
53
54   Multistage Cancer Slope Factor = 0.00512702
     D.6  Male  BDF1 Mice: Hepatic Carcinomas and Adenomas

55          Data for hepatic carcinomas and adenomas in male BDF1 mice (Kano et al., 2009) are shown in
56   Table D-14. Note that the incidence of carcinomas and the incidence of either adenomas or carcinomas
57   are monotone non-decreasing functions of dose. These data therefore appear to be appropriate for
58   dose-response modeling using BMDS. However, the incidence of adenomas clearly reaches a peak value

                                                                                             D-37
                                  DRAFT - DO NOT CITE OR QUOTE

-------
1    at 191 mg/kg-day and then decreases sharply with increasing dose. This cannot be modeled by a
2    multistage model using only non-negative coefficients. To some extent the incidence of "either adenomas
3    or carcinomas or both" retains some of the inverted-U shaped dose-response of the adenomas, which
4    dominate based on their high incidence at the lowest dose groups (49 and 191 mg/kg-day), thus is not
5    well characterized by any multistage model.

     Table D-14  Data for hepatic adenomas and carcinomas in male BDF1 mice (Kano et al.. 2009)
Tumor type
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenomas or carcinomas
Neither adenomas nor carcinomas
Total number per group
Dose (mg/kg-day)
0
9
15
23
27
50
49
17
20
31
19
50
191
23
23
37
13
50
677
11
36
40
10
50
     Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

6           The results of the BMDS modeling for the entire suite of models for hepatic adenomas and
7    carcinomas in male BDF1 mice are presented in Table D-15.
     Table D-15  BMDS dose-response modeling results for the combined incidence of hepatic
                 adenomas and carcinomas in male BDF1 mice (Kano et al.. 2009)

Model
Gamma
Logistic
Log Logistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
250.551
251.187
248.839
252.244
250.551
250.551
250.551
251.326
250.551
250.551
250.747
p-value
0.1527
0.112
0.3461
0.0655
0.1527
0.1527
0.1527
0.1048
0.1527
0.1527
NC°
BMDio
mg/kg-da
y
70.99
91.89
34.78
133.53
70.99
70.99
70.99
97.01
70.99
70.99
11.60
BMDLio
mg/kg-da
y
44.00
61.98
16.60
78.18
44.00
44.00
44.00
67.36
44.00
44.00
1.63
x2a
0.605
0.529
0.656
0.016
0.605
0.605
0.605
0.518
0.605
0.605
-1.25x10'b
BMDlOHED
mg/kg-day
11.48
14.86
5.63
21.60
11.48
11.48
11.48
15.69
11.48
11.48
1.88
BMDLio HED
mg/kg-day
7.12
10.02
2.68
12.64
7.12
7.12
7.12
10.90
7.12
7.12
0.26
     "Maximum absolute x2 residual deviation between observed and predicted count.
     "Best-fitting model.
     °Slope restricted > 1.
     Value unable to be calculated (NC: not calculated) by BMDS.
Values much larger than 1 are undesirable.
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                     D-38

-------
        "0

        I
        d
        O
        t3
        cc
                0.9
                0.8
                0.7
           0.6
                0.5
                0.4
                0.3
                                     Log-Logistic Model with 0.95 Confidence Level
                                   Log-Logistic
                     3MDL  BMD
                                100
                                    200
300
400
500
600
700
                                                      dose
         07:30 10/262009

            Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

            Figure D-14   LogLogistic BMD model for the combined incidence of hepatic
                       adenomas and carcinomas in male BDF1 mice.
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Logistic Model.  (Version:  2.12;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_mmouse_hepato_adcar_Lnl-BMR10-Restrict.(
d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_kano2009_mmouse_hepato_adcar_Lnl-BMR10-Restrict.p
It
Thu Nov 12 09:09:36  2009

 BMDS Model Run

The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]

Dependent variable = Effect
Independent variable =  Dose
Slope parameter is restricted as slope >= 1

Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has  been set to: le-008

User has chosen the  log transformed model

Default Initial Parameter  Values
 background = 0.46
 intercept = -5.58909
 slope = 1
 Asymptotic Correlation Matrix of Parameter Estimates
                                                                                               D-39
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    (*** The model parameter(s) -slope have been estimated at a boundary point, or have
 3   been specified by the user, and do not appear in the correlation matrix )
 4
 5    background intercept
 6   background 1 -0.69
 7    intercept -0.69 1
 8
 9
10    Parameter Estimates
11
12    95.0% Wald Confidence Interval
13    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
14   background 0.507468 * * *
15    intercept -5.74623 * * *
16    slope 1 * * *
17
18   * - Indicates that this value is not calculated.
19
20
21    Analysis of Deviance Table
22
23    Model Log(likelihood) # Param's Deviance Test d.f. P-value
24    Full model -121.373 4
25    Fitted model -122.419 2 2.09225 2 0.3513
26    Reduced model -128.859 1 14.9718 3 0.001841
27
28    AIC: 248.839
29
30
31    Goodness of Fit
32    Scaled
33    Dose Est._Prob. Expected Observed Size Residual
34    	~	
35    0.0000 0.5075 25.373 23.000 50 -0.671
36    49.0000 0.5741 28.707 31.000 50 0.656
37    191.0000 0.6941 34.706 37.000 50 0.704
38    677.0000 0.8443 42.214 40.000 50 -0.863
39
40    Chi^2 = 2.12 d.f. = 2 P-value = 0.3461
41
42
43    Benchmark Dose Computation
44   Specified effect = 0.1
45   Risk Type = Extra risk
46   Confidence level =0.95
47    BMD = 34.7787
48    BMDL = 16.5976
                                                                                             D-40
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                  Multistage Cancer Model with O.9b Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
        o
        I
                0.9
                0.8
                0.7
                0.6
                0.5
                0.4
                0.3
                                            Multistage Cancer
                                           Linear extrapolation
                       BMDL
                               BMD
                                100
                                         200
                                                   300
                                                            400
                                                                     500
                                                                               600
                                                                                        700
                                                     dose
     07:30 10/26 2009

       Source: Used with permission from Elsevier, Ltd., Kano et al. (2009).

       Figure D-15    Multistage BMD model (1 degree) for the combined incidence of
                  hepatic adenomas and carcinomas in male BDF1 mice.


Multistage Cancer Model.  (Version:  1.7; Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mmouse_hepato_adcar_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kano2009_mmouse_hepato_adcar_Msc-BMR10-lpoly.plt
Mon Oct 26 08:30:50  2009

 BMDS Model  Run
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-betal*dose/xl)]

The parameter betas  are  restricted to be positive

Dependent variable = Effect
Independent variable = Dose

Total number of observations = 4
Total number of records  with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008

Default Initial Parameter Values
Background = 0.573756
Beta(l) = 0.00123152

Asymptotic Correlation Matrix of Parameter Estimates
                                                                                                D-41
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    Background Beta(l)
 2   Background 1 -0.58
 3   Beta(l) -0.58 1
 4
 5
 6   Parameter Estimates
 7
 8    95.0% Wald Confidence Interval
 9   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
10   Background 0.545889 * * *
11   Beta(l) 0.00148414 * * *
12
13   * - Indicates that this value is not calculated.
14
15
16
17    Analysis of Deviance Table
18
19    Model Log(likelihood) # Param's Deviance Test d.f. P-value
20    Full model -121.373 4
21    Fitted model -123.275 2 3.80413 2 0.1493
22    Reduced model -128.859 1 14.9718 3 0.001841
23
24    AIC: 250.551
25
26
27    Goodness of Fit
28    Scaled
29    Dose Est._Prob. Expected Observed Size Residual
30    	~	
31    0.0000 0.5459 27.294 23.000 50 -1.220
32    49.0000 0.5777 28.887 31.000 50 0.605
33    191.0000 0.6580 32.899 37.000 50 1.223
34    677.0000 0.8337 41.687 40.000 50 -0.641
35
36    Chi^2 = 3.76 d.f. = 2 P-value = 0.1527
37
38
39    Benchmark Dose Computation
40
41   Specified effect =0.1
42   Risk Type = Extra risk
43   Confidence level =0.95
44    BMD = 70.9911
45    BMDL = 44.0047
46    BMDU = 150.117
47
48   Taken together, (44.0047, 150.117) is a 90% two-sided confidence interval for the BMD
49
50   Multistage Cancer Slope Factor = 0.00227248
     D.7  BMD Modeling Results from Additional Chronic Bioassays

51          Data and BMDS modeling results for the additional chronic bioassays (NCI. 1978; Kociba et al.,
52   1974) were evaluated for comparison with the Kano et al. (2009) study. These results are presented in the
53   following sections.

54          The BMDS dose-response modeling estimates and HEDs that resulted are presented in detail in
55   the following sections and a summary is provided in Table D-16.
                                                                                             D-42
                                  DRAFT - DO NOT CITE OR QUOTE

-------
    Table D-16  Summary of BMDS dose-response modeling estimates associated with liver and nasal
                tumor incidence data resulting from chronic oral exposure to 1,4-dioxane in rats and
                mice
Endpoint
Model
selection
criterion
Model Type
AIC
p-value
Kociba et al., (1974) Male and Female (combined) Sherman
Hepatic
Tumors3
Nasal
Cavity
Tumors'3
NCI, (1978)
Hepatic
Turners0
Nasal
Cavity
Tumorsb
NCI. (1978)
Nasal
Cavity
Tumorsb
NCI, (1978)
Hepatic
Tumorsd
NCI, (1978)
Hepatic
Tumorsd
Lowest AIC
Lowest AIC
Probit
Multistage
(3 degree)
84.3126
26.4156
0.606
0.9999
BMDio
mg/kg-da
y
Rats
1113
1717

.94
.16
BMDLio
mg/kg-day

920.62
1306.29
BMDlOHED
mg/kg-day

290.
448.

,78
24
BMDLio HED
mg/kg-day

240.
340.

,31
,99
Female Osborne-Mendel Rats
Lowest AIC
Lowest AIC
LogLogistic
LogLogistic
84.2821
84.2235
0.7333
0.2486
111.
155.
46
32
72.41
100.08
28.
40.
,75
,07
18.
25.
68
82
Male Osborne-Mendel Rats
Lowest AIC
Female B6C3Fi
Lowest AIC,
Multistage
model
LogLogistic
Mice
Multistage
(2 degree)
92.7669

85.3511
0.7809

1
56.

160.
26

68
37.26

67.76
16.

23.
,10

,12
10.

9.
66

,75
Male B6C3Fi Mice
Lowest AIC
Gamma
177.539
0.7571
601.
69
243.92
87.
98
35.
,67
    "Incidence of hepatocellular carcinoma.
    blncidence of nasal squamous cell carcinoma.
    Incidence of hepatocellular adenoma.
    dlncidence of hepatocellular adenoma or carcinoma.
     D.7.1    Hepatocellular Carcinoma and Nasal Squamous  Cell Carcinoma
             (Kociba etal.. 1974)

1          The incidence data for hepatocellular carcinoma and nasal squamous cell carcinoma are presented
2    in Table D-17. The predicted BMD10HED and BMDL10HED values are also presented in Table D-18 and
3    Table D-19 for hepatocellular carcinomas and nasal squamous cell carcinomas, respectively.
                                                                                                 D-43
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Table D-17  Incidence of hepatocellular carcinoma and nasal squamous cell carcinoma in male
            and female Sherman rats (combined) (Kociba et al.. 1974) treated with 1,4-dioxane in
            the drinking water for 2 years
          Animal Dose (mg/kg-day)             Incidence of hepatocellular      Incidence of nasal
       .         *   i    j *    i  j    i                    .a                squamous cell
       (average of male and female dose)                 carcinoma                     .      a
       v     a                        '                                           carcinoma
0
14
121
1,307
1/1 06b
0/110
1/106
10/66d
0/1 06C
0/110
0/106
3/66d
aRats surviving until 12 months on study.
bp < 0.001; positive dose-related trend (Cochran-Armitage test).
°p < 0.01; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.

Source: Used with permission from Elsevier, Ltd., Kociba et al. (1974).
Table D-18  BMDS dose-response modeling results for the incidence of hepatocellular carcinoma
            in male and female Sherman rats (combined) (Kociba et al.. 1974) exposed to
            1,4-dioxane in the drinking water for 2 years
Model
Gamma
Logistic
LogLogistic
LogProbit"
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)
Probitc
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
86.2403
84.3292
86.2422
84.4246
85.1187
86.2868
86.2868
84.3126
86.2443
85.1187
1503.63
p-value
0.3105
0.6086
0.3103
0.5977
0.3838
0.3109
0.3109
0.606
0.3104
0.3838
NCa
BMDio
mg/kg-day
985.13
1148.65
985.62
1036.97
940.12
1041.72
1041.72
1113.94
998.33
940.12
NCa
BMDLio
mg/kg-day
628.48
980.95
611.14
760.29
583.58
628.56
628.56
920.62
629.93
583.58
NCa
x23
-0.005
-0.004
-0.005
-0.011
0.279
-0.006
-0.006
-0.005
-0.005
0.279
0
BMDio HED
mg/kg-day
257.15
299.84
257.28
270.68
245.40
271.92
271.92
290.78
260.60
245.40
0
BMDLio HED
mg/kg-day
164.05
256.06
159.53
198.46
152.33
164.07
164.08
240.31
164.43
152.33
0
"Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
"Slope restricted > 1.
""Best-fitting model.
dValue unable to be calculated (NC: not calculated) by BMDS.
                                                                                                    D-44
                                DRAFT - DO NOT CITE OR QUOTE

-------
                                        Probit Model with 0.95 Confidence Level
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
        I
        c
        o
        t3
        ns
                 0.25
                  0.2
                 0.15
                  0.1
                 0.05
                               Probit
                                                                BMDL
                                                                               BMD
                                  200
                                           400
                                                     600
                                                       dose
                                                              800
                                                                       1000
                                                                                 1200
          11:54 10/27 2009
            Source: Used with permission from Elsevier, Ltd., Kociba et al. (1974).

            Figure D-16   Probit BMD model for the incidence of hepatocellular carcinoma in
                       male and female Sherman rats exposed to 1,4-dioxane in drinking
                       water.
Probit Model.  (Version:  3.1;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\pro_kociba_mf_rat_hepato_car_Prb-BMR10.plt
Tue Oct 27 12:54:14 2009

 BMDS Model Run
                                                              is the cumulative  normal
The form of the probability function is:
P[response] = CumNorm(Intercept+Slope*Dose),where CumNorm(
distribution function

Dependent variable  = Effect
Independent variable =  Dose
Slope parameter is  not  restricted

Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has  been set to: le-008

Initial  (and Specified) Parameter Values
background = 0 Specified
intercept = -2.62034
slope = 0.0012323
Asymptotic Correlation  Matrix of  Parameter Estimates
(*** The model parameter(s)  -background have been estimated at a boundary  point,  or
have been specified by  the user,  and do not appear in the correlation matrix  )
                                                                                               D-45
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    intercept slope
 3   intercept 1 -0.82
 4   slope -0.82 1
 5
 6
 7   Parameter Estimates
 8
 9    95.0% Wald Confidence Interval
10   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
11   intercept -2.55961 0.261184 -3.07152 -2.0477
12   slope 0.00117105 0.000249508 0.000682022 0.00166008
13
14
15    Analysis of Deviance Table
16
17    Model Log(likelihood) # Param's Deviance Test d.f. P-value
18    Full model -39.3891 4
19    Fitted model -40.1563 2 1.53445 2 0.4643
20    Reduced model -53.5257 1 28.2732 3 <.0001
21
22    AIC: 84.3126
23
24
25    Goodness of Fit
26    Scaled
27    Dose Est._Prob. Expected Observed Size Residual
28    	
29    0.0000 0.0052 0.555 1.000 106 0.598
30    14.0000 0.0055 0.604 0.000 110 -0.779
31    121.0000 0.0078 0.827 1.000 106 0.191
32    1307.0000 0.1517 10.014 10.000 66 -0.005
33
34    ChiA2 = 1.00 d.f. = 2 P-value = 0.6060
35
36
37    Benchmark Dose Computation
38
39   Specified effect =0.1
40   Risk Type = Extra risk
41   Confidence level = 0.95
42    BMD = 1,113.94
43    BMDL = 920.616
                                                                                             D-46
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                                   Multistage Cancer Model with 0.95 Confidence Level
 1
 2
 o
 6
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
 T3

 •§

 <
 CZ
 CD

 CO
                0.25
                 0.2
                0.15
                 0.1
                0.05
                                              Multistage Cancer
                                             Linear extrapolation
                                                BMDL
                                                                       BMD
                        0
                                 200
                                           400
                                                     600
                                                       dose
                                                               800
                                                                        1000
                                                                                  1200
   11:54 10/272009

       Source: Used with permission from Elsevier, Ltd., Kociba et al. (1974).

       Figure D-17   Multistage BMD model (1 degree) for the incidence of hepatocellular
                  carcinoma in male and female Sherman rats exposed to 1,4-dioxane in
                  drinking water.


Multistage Cancer Model.  (Version: 1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_hepato_car_Msc-BMR10-lpoly.(d)
Gnuplot Plotting  File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_hepato_car_Msc-BMR10-lpoly.plt
Tue Oct 27 12:54:10  2009

BMDS Model Run
The form of the probability function is:

P[response] = background + (1-background)*[1-EXP(-betal*doseAl)]

The parameter betas  are  restricted to be positive

Dependent variable = Effect
Independent variable = Dose

Total number of observations = 4
total number of records  with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1

Maximum number of iterations =250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.000925988
                                                                                                D-47
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1   Beta(l) = 0.000124518
 2
 3
 4   Asymptotic Correlation Matrix of Parameter Estimates
 5    Background Beta(l)
 6   Background 1 -0.44
 7   Beta(l) -0.44 1
 8
 9
10   Parameter Estimates
11
12    95.0% Wald Confidence Interval
13   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
14   Background 0.0038683 * * *
15   Beta(l) 0.000112071 * * *
16
17   * - Indicates that this value is not calculated.
18
19
20    Analysis of Deviance Table
21
22    Model Log(likelihood) # Param's Deviance Test d.f. P-value
23    Full model -39.3891 4
24    Fitted model -40.5594 2 2.34056 2 0.3103
25    Reduced model -53.5257 1 28.2732 3 <.0001
26
27    AIC: 85.1187
28
29
30    Goodness of Fit
31    Scaled
32    Dose Est._Prob. Expected Observed Size Residual
33    	
34    0.0000 0.0039 0.410 1.000 106 0.923
35    14.0000 0.0054 0.597 0.000 110 -0.775
36    121.0000 0.0173 1.832 1.000 106 -0.620
37    1307.0000 0.1396 9.213 10.000 66 0.279
38
39    Chi^2 = 1.92 d.f. = 2 P-value = 0.3838
40
41
42    Benchmark Dose Computation
43
44   Specified effect = 0.1
45   Risk Type = Extra risk
46   Confidence level =0.95
47    BMD = 940.124
48    BMDL = 583.576
49    BMDU = 1,685.88
50
51   Taken together, (583.576, 1685.88) is a 90% two-sided confidence interval for the BMD
52
53   Multistage Cancer Slope Factor = 0.000171357
                                                                                             D-48
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Table D-19  BMDS dose-response modeling results for the incidence of nasal squamous cell
            carcinoma in male and female Sherman rats (combined) (Kocibaet al.. 1974) exposed
            to 1,4-dioxane in the drinking water for 2 years
Model
Gamma
Logistic
LogLogistic
LogProbit"
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Multistage-Cancer
(3 degree)0
Probit
Weibull
Quantal-Linear
Dichotomous-Hill
AIC
28.4078
28.4078
28.4078
28.4078
27.3521
26.4929
26.4156
28.4078
28.4078
27.3521
30.4078
p-value
1
1
1
1
0.9163
0.9977
0.9999
1
1
0.9163
0.9997
BMDio
mg/kg-day
1,572.09
1,363.46
1,464.77
1,644.38
3,464.76
1,980.96
1,717.16
1,419.14
1,461.48
3,464.76
1,465.77
BMDLio
mg/kg-day
1,305.86
1,306.67
1,306.06
1,305.49
1,525.36
1,314.37
1,306.29
1,306.44
1,306.11
1,525.35
1319.19
x2a
0
0
0
0
0.272
0.025
0.002
0
0
0.272
5.53x10"'
BMDio HED
mg/kg-day
410.37
355.91
382.35
429.24
904.42
517.10
448.24
370.44
381.50
904.42
382.62
BMDLio HED
mg/kg-day
340.87
341.09
340.93
340.78
398.17
343.10
340.99
341.03
340.94
398.17
344.35
"Maximum absolute x^
"Slope restricted > 1.
""Best-fitting model.
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
                               Multistage Cancer Model with 0.95 Confidence Level
 f
 o
 OS
           0.14
           0.12
            0.1
           0.08
           0.06
           0.04
           0.02
                           Multistage Cancer
                         Linear extrapolation
                                                                                        BMID
   06:25 10/272009
                    0      200    400    600    800     1000   1200   1400    1600    1800
                                                    dose
       Figure D-18   Multistage BMD model (3 degree) for the incidence of nasal
                  squamous cell carcinoma in male and female Sherman rats exposed to
                  1,4-dioxane in drinking water.
                                                                                              D-49
                               DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2   ====================================================================
 3   Multistage Cancer Model.  (Version: 1.7; Date: 05/16/2008)
 4   Input Data File:
 5   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_nasal_car_Msc-BMR10-3poly.(d)
 6   Gnuplot Plotting File:
 7   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_kociba_mf_rat_nasal_car_Msc-BMR10-3poly.plt
 8   Tue Oct 27 07:25:02 2009
 9   ====================================================================
10    BMDS Model Run
11   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
12
13   The form of the probability function is:
14
15   P[response] = background +
16   (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2-beta3*dose/x3) ]
17
18   The parameter betas are restricted to be positive
19
20   Dependent variable = Effect
21   Independent variable = Dose
22
23   Total number of observations = 4
24   Total number of records with missing values = 0
25   Total number of parameters in model = 4
26   Total number of specified parameters = 0
27   Degree of polynomial = 3
28
29   Maximum number of iterations =250
30   Relative Function Convergence has been set to: le-008
31   Parameter Convergence has been set to: le-008
32   Default Initial Parameter Values
33   Background = 0
34   Beta(l) = 0
35   Beta(2) = 0
36   Beta(3) = 2.08414e-011
37
38
39   Asymptotic Correlation Matrix of Parameter Estimates
40
41    (*** The model parameter(s) -Background -Beta(l) -Beta(2)
42   have been estimated at a boundary point, or have been specified by the user,
43   and do not appear in the correlation matrix )
44
45    Beta(3)
46    Beta (3) 1
47
48
49                                     Parameter Estimates
50
51    95.0% Wald Confidence Interval
52   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
53   Background 0 * * *
54   Beta(l) 0 * * *
55   Beta(2) 0 * * *
56   Beta(3) 2.08088e-011 * * *
57
58   * - Indicates that this value is not calculated.
59
60
61
62    Analysis of Deviance Table
63
64    Model Log(likelihood) # Param's Deviance Test d.f. P-value
65    Full model -12.2039 4
66    Fitted model -12.2078 1 0.00783284 3 0.9998
67    Reduced model -17.5756 1 10.7433 3 0.0132

                                                                                              D-50
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    AIC: 26.4156
 3
 4
 5    Goodness of Fit
 6    Scaled
 7    Dose Est._Prob. Expected Observed  Size  Residual
 9    0.0000 0.0000  0.000  0.000  106  0.000
10    14.0000 0.0000  0.000  0.000  110  -0.003
11    121.0000 0.0000  0.004  0.000  106 -0.063
12    1307.0000 0.0454 2.996  3.000 66 0.002
13
14    ChiA2 = 0.00 d.f. =  3  P-value  = 0.9999
15
16
17    Benchmark Dose  Computation
18
19   Specified effect  =0.1
20   Risk Type = Extra risk
21   Confidence level  =0.95
22    BMD = 1,717.16
23    BMDL = 1,306.29
24    BMDU = 8,354.46
25
26   Taken together,  (1306.29, 8354.46)  is a  90%  two-sided confidence interval for the BMD
27
28   Multistage Cancer Slope  Factor  = 7.65529e-005
     D.7.2    Nasal Cavity Squamous Cell Carcinoma and Liver Hepatocellular
              Adenoma  in Osborne-Mendel  Rats (NCI. 1978)

29          The incidence data for hepatocellular adenoma (female rats) and nasal squamous cell carcinoma
30   (male and female rats) are presented in Table D-20. The log-logistic model adequately fit both the male
31   and female rat nasal squamous cell carcinoma data, as well as female hepatocellular adenoma incidence
32   data. For all endpoints and genders evaluated in this section, compared to the multistage models, the
33   log-logistic model had a higher p-value, as well as both a lower AIC and lower BMDL. The results of the
34   BMDS modeling for the entire suite  of models are presented in Table D-21 through Table D-23.
                                                                                              D-51
                                  DRAFT - DO NOT CITE OR QUOTE

-------
Table D-20  Incidence of nasal cavity squamous cell carcinoma and hepatocellular adenoma in
            Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in the drinking water
Male rat Animal Dose (mg/kg-day)a
0
Nasal cavity squamous cell carcinoma 0/33C
240°
12/26°
530
16/33°
Female rat Animal Dose (mg/kg-day)a
0
Nasal cavity squamous cell carcinoma 0/34C
Hepatocellular adenoma 0/31C
350
10/30"
10/30°
640
8/29°
11/29°
"Tumor incidence values were adjusted for mortality (NCI. 1978).
bGroup not included in statistical analysis by NCI (1978) because the dose group was started a year earlier without
appropriate controls.
°p S 0.001; positive dose-related trend (Cochran-Armitage test).
dp < 0.001; Fisher's Exact test.

Source: NCI (1978).
                                                                                                     D-52
                                 DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-21  BMDS dose-response modeling results for the incidence of hepatocellular adenoma
                in female Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in the drinking
                water for 2 years
Model
Gamma
Logistic
Log Logistic"
LogProbit
AIC
84.6972
92.477
84.2821
85.957
p-value
0.5908
0.02
0.7333
0.3076
BMDio
mg/kg-day
132.36
284.09
111.46
209.47
BMDLio
mg/kg-day
94.06
220.46
72.41
160.66
x2a
0
1.727
0
1.133
BMDlOHED
mg/kg-day
34.144
73.29
28.75
54.04
BMDLio HED
mg/kg-day
24.26
56.87
18.68
41.45
     Multistage-Cancer
     (1 degree)	
                  84.6972
0.5908
132.36
 94.06
        34.14
            24.26
     Multistage-Cancer
     (2 degree)	
                  84.6972
0.5908
132.36
 94.06
        34.14
            24.26
     Probit
                  91.7318
0.0251
267.02
207.18
1.7
68.88
53.44
     Weibull
                  84.6972
0.5908
132.36
 94.06
        34.14
            24.26
     Quantal-Linear
                  84.6972
0.5908
132.36
 94.06
 0
34.14
24.26
"Maximum absolute x2
bBest-fitting model.
                     residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
                                     Log-Logistic Model with O.9b Confidence Level
      T3
      
-------
 1   ===============
 2    BMDS Model Run
 4   The form of the probability function is:
 5   P [response] = background+ (1-background) / [1+EXP (-intercept-slope*Log (dose) ) ]
 6
 7   Dependent variable = Effect
 8   Independent variable = Dose
 9   Slope parameter is restricted as slope >= 1
10
11   Total number of observations = 3
12   Total number of records with missing values = 0
13   Maximum number of iterations =250
14   Relative Function Convergence has been set to: le-008
15   Parameter Convergence has been set to: le-008
16
17   User has chosen the log transformed model
18
19   Default Initial Parameter Values
20   background = 0
21   intercept = -6.62889
22   slope = 1
23
24   Asymptotic Correlation Matrix of Parameter Estimates
25
26   (*** The model parameter (s) -background -slope have been estimated at a boundary
27   point, or have been specified by the user, and do not appear in the correlation
28   matrix)
29
30    intercept
31    intercept 1
32
33                                     Parameter Estimates
34
35    95.0% Wald Confidence Interval
36   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
37   background 0 * * *
38   intercept -6.91086 * * *
39   slope 1 * * *
40
41   * - Indicates that this value is not calculated.
42
43
44    Analysis of Deviance Table
45
46    Model Log (likelihood) # Param' s Deviance Test d.f. P-value
47    Full model -40.8343 3
48    Fitted model -41.141 1 0.613564 2 0.7358
49    Reduced model -50.4308 1 19.1932 2 <.0001
50
51    AIC: 84.2821
52
53
54    Goodness of Fit
55    Scaled
56    Dose Est._Prob. Expected Observed Size Residual
57    ------------------------------------------------------------------------
58    0.0000 0.0000 0.000 0.000 31 0.000
59    350.0000 0.2587 8.536 10.000 33 0.582
60    640.0000 0.3895 12.464 11.000 32 -0.531
61
62    ChiA2 = 0.62 d.f. = 2 P-value = 0.7333
63
64
65    Benchmark Dose Computation
66
67   Specified effect =0.1
                                                                                             D-54
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1   Risk Type  =  Extra risk
 2   Confidence level  = 0.95
 3    BMD =  111.457
 4    BMDL = 72.4092
                                  Multistage Cancer Model with 0.95 Confidence Level
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
                0.5
                0.4
                0.3
                0.2
                0.1
                                            Multistage Cancer
                                           Linear extrapolation
                            BMDL
                                     BMD
                                100
                                          200
         06:32 10/27 2009

            Source: NCI (1978).
                                                    300
                                                      dose
                                                              400
                                                                         500
                                                                                   600
            Figure D-20   Multistage BMD model (1 degree) for the incidence of hepatocellular
                       adenoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
                       drinking water.
Multistage Cancer Model.  (Version:  1.7;  Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_hepato_ad_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_hepato_ad_Msc-BMR10-lpoly.plt
Tue Oct 27 07:32:16  2009

BMDS Model Run
The form of the probability function is:

P[response] = background +  (1-background)*[1-EXP(-betal*dose/xl)]

The parameter betas  are  restricted to be positive

Dependent variable = Effect
Independent variable = Dose

Total number of observations = 3
Total number of records  with missing values = 0
Total number of parameters  in model = 2
Total number of specified parameters = 0



                              DRAFT - DO NOT CITE OR QUOTE
                                                                                               D-55

-------
 1   Degree of polynomial = 1
 2
 3   Maximum number of iterations =250
 4   Relative Function Convergence has been set to: le-008
 5   Parameter Convergence has been set to: le-008
 6
 7
 8   Default Initial Parameter Values
 9   Background = 0.0385912
10   Beta(l) = 0.000670869
11   Asymptotic Correlation Matrix of Parameter Estimates
12
13   (*** The model parameter(s) -Background have been estimated at a boundary point, or
14   have been specified by the user, and do not appear in the correlation matrix)
15
16    Beta(l)
17    Beta(l) 1
18
19
20
21                                     Parameter Estimates
22
23    95.0% Wald Confidence Interval
24   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
25   Background 0 * * *
26   Beta(l) 0.00079602 * * *
27
28   * - Indicates that this value is not calculated.
29
30
31
32    Analysis of Deviance Table
33
34    Model Log(likelihood) # Param's Deviance Test d.f. P-value
35    Full model -40.8343 3
36    Fitted model -41.3486 1 1.02868 2 0.5979
37    Reduced model -50.4308 1 19.1932 2 <.0001
38
39    AIC: 84.6972
40
41
42    Goodness of Fit
43    Scaled
44    Dose Est._Prob. Expected Observed Size Residual
45    	
46    0.0000 0.0000 0.000 0.000 31 0.000
47    350.0000 0.2432 8.024 10.000 33 0.802
48    640.0000 0.3992 12.774 11.000 32 -0.640
49
50    ChiA2 = 1.05 d.f. = 2 P-value = 0.5908
51
52
53    Benchmark Dose Computation
54
55   Specified effect =0.1
56   Risk Type = Extra risk
57   Confidence level = 0.95
58    BMD = 132.359
59    BMDL = 94.0591
60    BMDU = 194.33
61
62   Taken together,  (94.0591, 194.33 ) is a 90% two-sided confidence interval for the BMD
63
64   Multistage Cancer Slope Factor = 0.00106316
                                                                                             D-56
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-22  BMDS dose-response modeling results for the incidence of nasal cavity squamous
                cell carcinoma in female Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in
the drinking water for 2 years
Model
Gamma
Logistic
Log Logistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
AIC
84.7996
92.569
84.2235
87.3162
84.7996
84.7996
91.9909
84.7996
84.7996
p-value
0.1795
0.0056
0.2486
0.0473
0.1795
0.1795
0.0064
0.1795
0.1795
BMDio
mg/kg-day
176.28
351.51
155.32
254.73
176.28
176.28
328.46
176.28
176.28
BMDLio
mg/kg-day
122.27
268.75
100.08
195.76
122.27
122.27
251.31
122.27
122.27
x2a
1.466
2.148
0
1.871
1.466
1.466
2.136
1.466
1.466
BMDio HED
mg/kg-day
45.47
90.68
40.07
65.71
45.47
45.47
84.73
45.47
45.47
BMDLio HED
mg/kg-day
31.54
69.33
25.82
50.50
31.54
31.54
64.83
31.54
31.54
"Maximum absolute x2
bBest-fitting model.
°Slope restricted > 1 .
                    residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
                                     Log-Logistic Model with 0.95 Confidence Level
1
2
3
4
               0.5
               0.4
               0.3
               0.2
               0.1
                                   Log-Logistic
                             BMDL      BMD
        06:30 10/27 2009

            Source: NCI (1978).
                                 100       200        300       400
                                                       dose
                                                                           500
                                                                                     600
       Figure D-21   LogLogistic BMD model for the incidence of nasal cavity squamous
                  cell carcinoma in female Osborne-Mendel rats exposed to 1,4-dioxane in
                  drinking water.


Logistic Model.  (Version:  2.12;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA HPAG\14Dioxane\BMDS\lnl nci  frat nasal car  Lnl-BMRlO-Restrict.(d)
                                                                                                  D-57
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1   Gnuplot Plotting File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_frat_nasal_car_Lnl-BMR10-Restrict.plt
 3   Tue Oct 27 07:30:09 2009
 4   ====================================================================
 5    BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7
 8   The form of the probability function is:
 9
10   P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
11
12
13   Dependent variable = Effect
14   Independent variable = Dose
15   Slope parameter is restricted as slope >= 1
16
17   Total number of observations = 3
18   Total number of records with missing values = 0
19   Maximum number of iterations =250
20   Relative Function Convergence has been set to: le-008
21   Parameter Convergence has been set to: le-008
22
23
24   User has chosen the log transformed model
25
26
27   Default Initial Parameter Values
28   background = 0
29   intercept = -6.64005
30   slope = 1
31
32
33   Asymptotic Correlation Matrix of Parameter Estimates
34   (*** The model parameter(s) -background -slope have been estimated at a boundary
35   point, or have been specified by the user, and do not appear in the correlation
36   matrix)
37
38    intercept
39    intercept 1
40
41
42                                     Parameter Estimates
43
44    95.0% Wald Confidence Interval
45   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
46   background 0 *
47   intercept -7.24274
48   slope 1 * * *
49
50   * - Indicates that this value is not calculated.
51
52    Analysis of Deviance Table
53
54    Model Log(likelihood) # Param's Deviance Test d.f. P-value
55    Full model -39.7535 3
56    Fitted model -41.1117 1 2.71651 2 0.2571
57    Reduced model -47.9161 1 16.3252 2 0.0002851
58
59    AIC: 84.2235
60
61    Goodness of Fit
62    Scaled
63    Dose Est._Prob. Expected Observed Size Residual
64    	~	
65    0.0000 0.0000 0.000 0.000 34 0.000
66    350.0000 0.2002 7.008 10.000 35 1.264
67    640.0000 0.3140 10.992 8.000 35 -1.090

                                                                                             D-58
                                  DRAFT - DO NOT CITE OR QUOTE
* *

    * * *

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
       = 2.78 d.f.  =  2  P-value = 0.2486
 Benchmark Dose Computation

Specified effect  =  0.1
Risk Type = Extra risk
Confidence level  =  0.95
 BMD = 155.324
 BMDL = 100.081
                                   Multistage Cancer Model with 0.95 Confidence Level
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
        "8
        "o
        I
                0.5
                0.4
                0.3
                0.2
                0.1
                                             Multistage Cancer
                                           Linear extrapolation
                                                                                   600
          06:30 10/27 2009
            Source: NCI (1978).

            Figure D-22   Multistage BMD model (1 degree) for the incidence of nasal cavity
                       squamous cell carcinoma in female Osborne-Mendel rats exposed to
                       1,4-dioxane in drinking water.
Multistage Cancer Model.  (Version:  1.7; Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_nasal_car_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_frat_nasal_car_Msc-BMR10-lpoly.plt
Tue Oct 27 07:30:12  2009

 BMDS Model Run

The form of the probability function is:
P[response] = background  +  (1-background)*[1-EXP(-betal*doseAl) ]

The parameter betas  are restricted to be positive

Dependent variable = Effect
Independent variable = Dose

Total number of observations = 3
                                                                                               D-59
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1   Total number of records with missing values = 0
 2   Total number of parameters in model = 2
 3   Total number of specified parameters = 0
 4   Degree of polynomial = 1
 5
 6   Maximum number of iterations =250
 7   Relative Function Convergence has been set to: le-008
 8   Parameter Convergence has been set to: le-008
 9
10   Default Initial Parameter Values
11   Background = 0.0569154
12   Beta(l) = 0.00042443
13
14   Asymptotic Correlation Matrix of Parameter Estimates
15   (*** The model parameter(s) -Background have been estimated at a boundary point, or
16   have been specified by the user, and do not appear in the correlation matrix)
17
18    Beta(l)
19    Beta(l) 1
20
21                                     Parameter Estimates
22
23    95.0% Wald Confidence Interval
24   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
25   Background 0 * * *
26   Beta(l) 0.000597685 * * *
27
28   * - Indicates that this value is not calculated.
29
30    Analysis of Deviance Table
31
32    Model Log(likelihood) # Param's Deviance Test d.f. P-value
33    Full model -39.7535 3
34    Fitted model -41.3998 1 3.29259 2 0.1928
35    Reduced model -47.9161 1 16.3252 2 0.0002851
36
37    AIC: 84.7996
38
39    Goodness of Fit
40    Scaled
41    Dose Est._Prob. Expected Observed Size Residual
42    	
43    0.0000 0.0000 0.000 0.000 34 0.000
44    350.0000 0.1888 6.607 10.000 35 1.466
45    640.0000 0.3179 11.125 8.000 35 -1.134
46
47    ChiA2 = 3.44 d.f. = 2 P-value = 0.1795
48
49    Benchmark Dose Computation
50   Specified effect =0.1
51   Risk Type = Extra risk
52   Confidence level = 0.95
53    BMD = 176.281
54    BMDL = 122.274
55    BMDU = 271.474
56
57   Taken together, (122.274, 271.474) is a 90% two-sided confidence interval for the BMD
58
59   Multistage Cancer Slope Factor = 0.000817837
                                                                                             D-60
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-23  BMDS dose-response modeling results for the incidence of nasal cavity squamous
                cell carcinoma in male Osborne-Mendel rats (NCI. 1978) exposed to 1,4-dioxane in the
drinking water for 2 years
Model
Gamma
Logistic
Log Logistic"
LogProbitc
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)
Probit
Weibull
Quantal-Linear
AIC
93.6005
103.928
92.7669
95.0436
93.6005
93.6005
103.061
93.6005
93.6005
p-value
0.5063
0.0061
0.7809
0.2373
0.5063
0.5063
0.0078
0.5063
0.5063
BMDio
mg/kg-day
73.94
179.05
56.26
123.87
73.94
73.94
168.03
73.94
73.94
BMDLio
mg/kg-day
54.724
139.26
37.26
95.82
54.72
54.72
131.61
54.72
54.72
x2a
0
2.024
0
1.246
0
0
2.024
0
0
BMDlOHED
mg/kg-day
21.17
51.25
16.10
35.46
21.16
21.16
48.10
21.17
21.17
BMDLio HED
mg/kg-day
15.66
39.86
10.66
27.43
15.66
15.66
37.67
15.66
15.66
"Maximum absolute x2
bBest-fitting model.
°Slope restricted > 1 .
                    residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
                                     Log-Logistic Model with 0.95 Confidence Level
1
2
o
J
4
5
6
  o
  I
  d
  o
  CO
               0.7
               0.6
               0.5
               0.4
               0.3
               0.2
               0.1
                                  100
                                               200
                                                           300
                                                                        400
                                                                                     500
                                                      dose
    06:27 10/27 2009
       Source: NCI (1978).

       Figure D-23   LogLogistic BMD model for the incidence of nasal cavity squamous
                  cell carcinoma in male Osborne-Mendel rats exposed to 1,4-dioxane in
                  drinking water.


Logistic Model.  (Version: 2.12;  Date: 05/16/2008)
Input Data  File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_mrat_nasal_car_Lnl-BMR10-Restrict.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\lnl_nci_mrat_nasal_car_Lnl-BMR10-Restrict.plt
                                   DRAFT - DO NOT CITE OR QUOTE
                                                                                                 D-61

-------
 1   Tue Oct 27 07:27:57 2009
 2   ====================================================================
 3   BMDS Model Run
 4   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 5
 6   The form of the probability function is:
 7   P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
 8
 9   Dependent variable = Effect
10   Independent variable = Dose
11   Slope parameter is restricted as slope >= 1
12
13   Total number of observations = 3
14   Total number of records with missing values = 0
15   Maximum number of iterations =250
16   Relative Function Convergence has been set to: le-008
17   Parameter Convergence has been set to: le-008
18
19   User has chosen the log transformed model
20
21   Default Initial Parameter Values
22   background = 0
23   intercept = -6.08408
24   slope = 1
25
26   Asymptotic Correlation Matrix of Parameter Estimates
27   (*** The model parameter(s) -background -slope have been estimated at a boundary
28   point, or have been specified by the user, and do not appear in the correlation
29   matrix)
30
31    intercept
32    intercept 1
33
34                                     Parameter Estimates
35
36    95.0% Wald Confidence Interval
37   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
38   background 0 * * *
39   intercept -6.2272 * * *
40   slope 1 * * *
41
42   * - Indicates that this value is not calculated.
43
44    Analysis of Deviance Table
45
46    Model Log(likelihood) # Param's Deviance Test d.f. P-value
47    Full model -45.139 3
48    Fitted model -45.3835 1 0.488858 2 0.7832
49    Reduced model -59.2953 1 28.3126 2 <.0001
50
51    AIC: 92.7669
52
53                                       Goodness of Fit
54    Scaled
55    Dose Est._Prob. Expected Observed Size Residual
56    	~	
57    0.0000 0.0000 0.000 0.000 33 0.000
58    240.0000 0.3216 10.612 12.000 33 0.517
59    530.0000 0.5114 17.388 16.000 34 -0.476
60
61    Chi^2 = 0.49 d.f. = 2 P-value = 0.7809
                                                                                             D-62
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 Benchmark Dose Computation

Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
 BMD = 56.2596
 BMDL = 37.256
                                   Multistage Cancer Model with 0.95 Confidence Level
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
       o
      't5
       CO
               0.7
               0.6
               0.5
               0.4
               0.3
               0.2
               0.1
                                              Multistage Cancer
                                            Linear extrapolation
                    ;
                         BMDL  BMD
                       0

        06:28 10/27 2009

            Source: NCI (1978).
                              100
200
300
400
500
                                                   dose
       Figure D-24   Multistage BMD model (1 degree) for the incidence of nasal cavity
                  squamous cell carcinoma in male Osborne-Mendel ratsexposed to
                  1,4-dioxane in drinking water.


Multistage Cancer Model.  (Version:  1.7;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mrat_nasal_car_Msc-BMR10-lpoly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mrat_nasal_car_Msc-BMR10-lpoly.plt
                                               Tue Oct  27  07:28:00  2009

 BMDS Model  Run

The form of  the probability function is:
P[response]  = background  +  (1-background)*[1-EXP(-betal*dose/xl)]

The parameter betas are restricted  to be positive

Dependent variable = Effect
Independent  variable = Dose

Total number of observations  =  3
Total number of records with  missing values =  0
Total number of parameters  in model = 2
Total number of specified parameters = 0
                                                                                               D-63
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1   Degree of polynomial = 1
 2
 3   Maximum number of iterations =250
 4   Relative Function Convergence has been set to: le-008
 5   Parameter Convergence has been set to: le-008
 6   Default Initial Parameter Values
 7   Background = 0.0578996
 8   Beta(l) = 0.00118058
 9
10   Asymptotic Correlation Matrix of Parameter Estimates
11   (*** The model parameter(s) -Background have been estimated at a boundary point, or
12   have been specified by the user, and do not appear in the correlation matrix)
13
14    Beta(l)
15    Beta(l) 1
16
17                                     Parameter Estimates
18
19    95.0% Wald Confidence Interval
20   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
21   Background 0 * * *
22   Beta(l) 0.00142499 * * *
23
24   * - Indicates that this value is not calculated.
25
26    Analysis of Deviance Table
27
28    Model Log(likelihood) # Param's Deviance Test d.f. P-value
29    Full model -45.139 3
30    Fitted model -45.8002 1 1.32238 2 0.5162
31    Reduced model -59.2953 1 28.3126 2 <.0001
32
33    AIC: 93.6005
34
35    Goodness of Fit
36    Scaled
37    Dose Est._Prob. Expected Observed Size Residual
38    	~	
39    0.0000 0.0000 0.000 0.000 33 -0.000
40    240.0000 0.2896 9.558 12.000 33 0.937
41    530.0000 0.5301 18.024 16.000 34 -0.695
42
43    ChiA2 = 1.36 d.f. = 2 P-value = 0.5063
44
45    Benchmark Dose Computation
46   Specified effect =0.1
47   Risk Type = Extra risk
48   Confidence level =0.95
49    BMD = 73.9379
50    BMDL = 54.7238
51    BMDU = 103.07
52
53   Taken together, (54.7238, 103.07 )  is a 90% two-sided confidence interval for the BMD
54
55   Multistage Cancer Slope Factor = 0.00182736
     D.7.3    Hepatocellular Adenoma or Carcinoma in B6C3Fi Mice (NCI. 1978)

56          The incidence data for hepatocellular adenoma or carcinoma in male and female mice are
57   presented in Table D-24. The 2-degree polynomial model (betas restricted > 0) was the lowest degree
58   polynomial that provided an adequate fit to the female mouse data (Figure D-25), while the gamma model


                                                                                             D-64
                                  DRAFT - DO NOT CITE OR QUOTE

-------
1    provided the best fit to the male mouse data (Figure D-26). The results of the BMDS modeling for the
2    entire suite of models are presented in Table D-25 and Table D-26 for the female and male data,
3    respectively.
                                                                                                   D-65
                                   DRAFT - DO NOT CITE OR QUOTE

-------
Table D-24  Incidence of hepatocellular adenoma or carcinoma in male and female B6C3F1 mice
            (NCI. 1978) exposed to 1,4-dioxane in drinking water
Male mouse Animal Dose (mg/kg-day)a
0
8/49°
720
19/50°
830
28/47°
Female mouse Animal Dose (mg/kg-day)a
0
0/50°
380
21/48C
860
35/37c
aTumor incidence values were not adjusted for mortality.
bp < 0.001, positive dose-related trend (Cochran-Armitage test).
°p < 0.001 by Fisher's Exact test pair-wise comparison with controls.
dp = 0.014.

Source: NCI (1978).
Table D-25  BMDS dose-response modeling results for the combined incidence of hepatocellular
            adenoma or carcinoma in female B6C3FT mice (NCI. 1978) exposed to 1,4-dioxane in
            the drinking water for 2 years
Model
Gamma
Logistic
Log Logistic
LogProbit"
Multistage-Cancer
(1 degree)
Multistage-Cancer
(2 degree)0
Probit
Weibull
Quantal-Linear
AIC
85.3511
89.1965
85.3511
85.3511
89.986
85.3511
88.718
85.3511
89.986
p-value
1
0.0935
1
1
0.0548
1
0.1165
1
0.0548
BMDio
mg/kg-day
195.69
199.63
228.08
225.8
49.10
160.68
188.24
161.77
49.10
BMDLio
mg/kg-day
105.54
151.35
151.16
150.91
38.80
67.76
141.49
89.27
38.80
x2a
0
0.675
0
0
0
0
-1.031
0
0
BMDlOHED
mg/kg-day
28.16
28.72
32.82
32.49
7.06
23.12
27.08
23.28
7.065
BMDLio HED
mg/kg-day
15.19
21.78
21.75
21.71
5.58
9.75
20.36
12.84
5.58
"Maximum absolute x^
"Slope restricted > 1.
""Best-fitting model.
residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
                                                                                                  D-66
                                DRAFT - DO NOT CITE OR QUOTE

-------
                                 Multistage Cancer Model with 0.95 Confidence Level
 1
 1
 2
 o
 J
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
      I
               0.8
               0.6
               0.4
               0.2
                                           Multistage Cancer  	
                                          Linear extrapolation
                       BMDL
                                   BMD
                             100
                                    200
                                            300
                                                    400     500
                                                     dose
                                                                   600
                                                                          700
                                                                                  800
                                                                                         900
   06:36 10/27 2009

       Source: NCI (1978).

       Figure D-25   Multistage BMD model (2 degree) for the incidence of hepatocellular
                  adenoma or carcinoma in female B6C3F! mice exposed to 1,4-dioxane in
                  drinking water.

Multistage Cancer Model.  (Version:  1.7;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_fmouse_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_fmouse_hepato_adcar_Msc-BMR10-2poly.plt
Tue Oct 27 07:36:26 2009

 BMDS Model  Run
The form of the probability  function is:
P [response] = background  +  (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2 ) ]

The parameter betas are restricted to be  positive

Dependent variable = Effect
Independent variable = Dose

Total number of observations =  3
Total number of records with missing values  = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations  =250
Relative Function Convergence has been set to:  le-008
Parameter Convergence has been set  to:  le-008

Default Initial Parameter Values
Background = 0
Beta(l) = 2.68591e-005
                                                                                               D-67
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1   Beta(2) = 3.91383e-006
 2
 3
 4   Asymptotic Correlation Matrix of Parameter Estimates
 5    (*** The model parameter(s) -Background have been estimated at a boundary point, or
 6   have been specified by the user, and do not appear in the correlation matrix)
 7
 8    Beta(l) Beta(2)
 9    Beta(l) 1 -0.92
10    Beta(2) -0.92 1
11
12
13                                     Parameter Estimates
14
15    95.0% Wald Confidence Interval
16   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
17   Background 0 * * *
18   Beta(l) 2.686e-005 * * *
19   Beta(2) 3.91382e-006 * * *
20
21   * - Indicates that this value is not calculated.
22
23
24    Analysis of Deviance Table
25
26    Model Log(likelihood) # Param's Deviance Test d.f. P-value
27    Full model -40.6756 3
28    Fitted model -40.6756 2 3.20014e-010 1 1
29    Reduced model -91.606 1 101.861 2 <.0001
30
31    AIC: 85.3511
32
33    Goodness of Fit
34    Scaled
35    Dose Est._Prob. Expected Observed Size Residual
36    	
37    0.0000 0.0000 0.000 0.000 50 0.000
38    380.0000 0.4375 21.000 21.000 48 0.000
39    860.0000 0.9459 35.000 35.000 37 0.000
40
41    Chi^2 = 0.00 d.f. = 1 P-value = 1.0000
42
43
44    Benchmark Dose Computation
45   Specified effect =0.1
46   Risk Type = Extra risk
47   Confidence level =0.95
48    BMD = 160.678
49    BMDL = 67.7635
50    BMDU = 186.587
51
52   Taken together,  (67.7635, 186.587) is a 90% two-sided confidence interval for the BMD
53

54                Multistage  Cancer  Slope Factor =  0.00147572
                                                                                             D-68
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table D-26  BMDS dose-response modeling results for the combined incidence of hepatocellular
                 adenoma or carcinoma in male B6C3F1 mice (NCI. 1978) exposed to 1,4-dioxane in
drinking water
Model
Gammab
Logistic
LogLogistic
LogProbit"
AIC
177.539
179.9
179.443
179.443
p-value
0.7571
0.1189
NCC
NCC
BMDio
mg/kg-day
601.69
252.66
622.39
631.51
BMDLio
mg/kg-day
243.92
207.15
283.04
305.44
x2a
-0.233
0.214
0
0
BMDlOHED
mg/kg-day
87.98
36.94
91.01
92.34
BMDLio HED
mg/kg-day
35.67
30.29
41.39
44.66
     Multistage-Cancer
     (1 degree)	
                    180.618
0.0762
164.29
117.37
0.079
24.02
17.16
     Multistage-Cancer
     (2 degree)	
                    179.483
0.1554
354.41
126.24
0.124
51.82
18.46
     Probit
                    179.984
0.1128
239.93
196.90
0.191
35.08
28.79
     Weibull
                    179.443
 NCC
608.81
249.71
  0
89.02
36.51
     Quantal-Linear
                    180.618
0.0762
164.29
117.37
0.079
24.02
17.16
     "Maximum absolute x2 residual deviation between observed and predicted count. Values much larger than 1 are undesirable.
     "Best-fitting model.
     "Value unable to be calculated (NC: not calculated) by BMDS.
     dSlope restricted > 1.
                                     Gamma Multi-Hit Model with 0.95 Confidence Level
        o
       '•o
                 0.7
                 0.6
                 0.5
                 0.4
                 0.3
                 0.2
                 0.1
                                      Gamma Multi-Hit
                                        BMDL
                                                                           BMD
         06:34 10/27 2009

            Source: NCI (1978).
                                100     200     300     400     500     600     700
                                                         dose
                                                                                          800
            Figure D-26    Gamma BMD model for the incidence of hepatocellular adenoma or
                        carcinoma in male B6C3Fi mice exposed to 1,4-dioxane in drinking
                        water.
1
2
3
4
Gamma Model.  (Version: 2.13;  Date: 05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-BMR10-Restrict.(d)
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                     D-69

-------
 1   Gnuplot Plotting File:
 2   L:\Priv\NCEA_HPAG\14Dioxane\BMDS\gam_nci_mmouse_hepato_adcar_Gam-BMR10-Restrict.plt
 3   Tue Oct 27 07:34:35 2009
 4   ====================================================================
 5   BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7
 8   The form of the probability function is:
 9   P[response]= background+(1-background)*CumGamma[slope*dose,power],
10   where CumGamma(.)  is the cummulative Gamma distribution function
11
12   Dependent variable = Effect
13   Independent variable = Dose
14   Power parameter is restricted as power >=1
15
16   Total number of observations = 3
17   Total number of records with missing values = 0
18   Maximum number of iterations =250
19   Relative Function Convergence has been set to: le-008
20   Parameter Convergence has been set to: le-008
21
22   Default Initial (and Specified) Parameter Values
23   Background =0.17
24   Slope = 0.000671886
25   Power =1.3
26
27   Asymptotic Correlation Matrix of Parameter Estimates
28   (*** The model parameter(s) -Power have been estimated at a boundary point, or have
29   been specified by the user, and do not appear in the correlation matrix)
30
31    Background Slope
32   Background 1 -0.52
33    Slope -0.52 1
34
35                                     Parameter Estimates
36    95.0% Wald Confidence Interval
37   Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
38   Background 0.160326 0.0510618 0.060247 0.260405
39   Slope 0.0213093 0.000971596 0.019405 0.0232136
40   Power 18 NA
41
42   NA - Indicates that this parameter has hit a bound implied by some ineguality
43   constraint and thus has no standard error.
44
45    Analysis of Deviance Table
46
47    Model Log(likelihood) # Param's Deviance Test d.f. P-value
48    Full model -86.7213 3
49    Fitted model -86.7693 2 0.096042 1 0.7566
50    Reduced model -96.715 1 19.9875 2 <.0001
51
52    AIC: 177.539
53
54    Goodness of Fit
55    Scaled
56    Dose Est._Prob. Expected Observed Size Residual
57    	
58    0.0000 0.1603 7.856 8.000 49 0.056
59    720.0000 0.3961 19.806 19.000 50 -0.233
60    830.0000 0.5817 27.339 28.000 47 0.196
61
62    ChiA2 = 0.10 d.f. = 1 P-value = 0.7571
63   Benchmark Dose Computation
64   Specified effect = 0.1
65   Risk Type = Extra risk
66   Confidence level = 0.95
67    BMD = 601.692

                                                                                             D-70
                                  DRAFT - DO NOT CITE OR QUOTE

-------
                    BMDL = 243.917
                                  Multistage Cancer Model with 0.95 Confidence Level
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
 T3

 •§



  C=
  O

  CO
               0.7
               0.6
               0.5
               0.4
               0.3
               0.2
               0.1
                                            Multistage Cancer
                                           Linear extrapolation 	
                            BMDL
                                                   BMD
                       0

         06:34 10/272009

            Source: NCI (1978).
                              100
                                      200
                                              300
                                                 400
                                                 dose
                                                              500
                                                                      600
                                                                              700
                                                                                      800
            Figure D-27   Multistage BMD model (2 degree) for the incidence of hepatocellular
                       adenoma or carcinoma in male B6C3F! mice exposed to 1,4-dioxane in
                       drinking water.
Multistage Cancer Model.  (Version:  1.7;  Date:  05/16/2008)
Input Data File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_mmouse_hepato_adcar_Msc-BMR10-2poly.(d)
Gnuplot Plotting File:
L:\Priv\NCEA_HPAG\14Dioxane\BMDS\msc_nci_iranouse_hepato_adcar_Msc-BMR10-2poly.plt
Tue Oct 27 07:34:42  2009

BMDS Model Run
The form of the probability  function is:  P[response]  = background +
(1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2) ]

The parameter betas  are  restricted to be  positive

Dependent variable = Effect
Independent variable = Dose

Total number of observations =  3
Total number of records  with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations =250
Relative Function Convergence has  been set to:  le-008
Parameter Convergence has been  set to:  le-008
Default Initial Parameter Values
                                                                                               D-71
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
Background = 0.131156
Beta(l) = 0
Beta(2) = 9.44437e-007

Asymptotic Correlation Matrix of Parameter Estimates
(*** The model parameter(s) -Beta(l) have been estimated at a boundary point, or have
been specified by the user, and do not appear in the correlation matrix)

 Background Beta(2)
Background 1 -0.72
 Beta(2) -0.72 1
                                  Parameter Estimates

 95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
Background 0.1568 * * *
Beta(l) 0 * * *
Beta(2) 8.38821e-007 * * *

    Indicates that this value is not calculated.
 Analysis of Deviance Table

 Model Log(likelihood) # Param's Deviance Test d.f. P-value
 Full model -86.7213 3
 Fitted model -87.7413 2 2.04001 1 0.1532
 Reduced model -96.715 1 19.9875 2 <.0001

 AIC: 179.483
 Goodness of Fit
 Scaled
 Dose Est._Prob. Expected Observed Size Residual

 0.0000 0.1568 7.683 8.000 49 0.124
 720.0000 0.4541 22.707 19.000 50 -1.053
 830.0000 0.5269 24.764 28.000 47 0.946

       = 2.02 d.f. = 1 P-value = 0.1554
 Benchmark Dose Computation

Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
 BMD = 354.409
 BMDL = 126.241
 BMDU = 447.476

Taken together, (126.241, 447.476) is a 90% two-sided confidence interval for the BMD

Multistage Cancer Slope Factor = 0.000792138
                                                                                             D-72
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      APPENDIX   E.     COMPARISON  OF  SEVERAL   DATA
         REPORTS  FOR  THE  JBRC   2-YEAR  1,4-DIOXANE
         DR INKING   WATE  R   STUDY

 1           As described in detail in Section 4.2.1.2.6 of this Toxicological Review ofl, 4-Dioxane, the JBRC
 2    conducted a 2-year drinking water study on the effects of 1,4-dioxane in both sexes of rats and mice. The
 3    results from this study have been reported three times, once as conference proceedings (Yamazaki et al..
 4    1994). once as a detailed laboratory report (JBRC. 1998). and once as a published manuscript (Kano et
 5    al.. 2009). After the External Peer Review  draft of the Toxicological Review of 1,4-Dioxane (U.S. EPA.
 6    2009b) had been released, the Kano et al. (2009) manuscript was published; thus, minor changes to the
 7    Toxicological Review of 1,4-Dioxane occurred.

 8           The purpose of this appendix is to  provide a clear and transparent comparison of the reporting of
 9    this 2-year 1,4-dioxane drinking water study. The variations included: (1) the level of detail on dose
10    information reported; (2) categories for incidence data reported (e.g., all animals or sacrificed animals);
11    and (3) analysis of non- and neoplastic  lesions. Even though the data contained in the reports varied, the
12    differences were minor and did not did not significantly affect the qualitative or quantitative cancer
13    assessment.

14           Tables  contained within this appendix provide a comparison of the variations in the reported data
15    (Kano et al.. 2009; JBRC. 1998; Yamazaki et al.. 1994). Table E-l and Table E-2 show the histological
16    nonneoplastic findings provided for male and female F344 rats, respectively. Table E-3 and Table E-4
17    show the histological nonneoplastic findings provided for male and female F344 rats, respectively.
18    Table E-3 and Table E-4 show the histological neoplastic findings provided for male and female F344
19    rats, respectively. Table E-5 and Table  E-6 show the histological nonneoplastic findings provided for
20    male and female F344 rats, respectively. Table E-7 and Table E-8 show the histological neoplastic
21    findings provided for male and female Crj:BDFl mice, respectively.
                                                                                                      E-l
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table E-1    Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
               JBRC drinking water study in male F344 rats





Nasal respiratory
epithelium; nuclear
enlargement
Nasal respiratory
epithelium; squamous cell
metaplasia
Nasal respiratory
epithelium; squamous cell
hyperplasia
Nasal gland; proliferation
Nasal olfactory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
respiratory metaplasia
Nasal olfactory epithelium;
atrophy
Lamina propria; hydropic
change
Lamina propria; sclerosis
Nasal cavity; adhesion
Nasal cavity; inflammation
Hyperplasia; liver
Spongiosis hepatis; liver
Clear cell foci; liver
Acidophilic cell foci; liver
Basophilic cell foci; liver
Mixed-cell foci; liver
Nuclear enlargement;
kidney proximal tubule
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Yamazaki et al. (1994)a

JBRC 11998}"
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000 0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-da}
Not reported
Not reported
Not reported
0/50 0/50 0/50 31/50
Not reported
0/50 0/50 0/50 2/50
Not reported
0/50 0/50 0/50 5/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
3/50 2/10 10/50 24/50
Not reported
12/50 20/50 25/50 40/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
r«ntmi 8- 41' 209-
Co£V 24 121 586
W [16] (81) (398)
0/50 0/50 0/50 26/50
0/40 0/45 0/35 12/226
0/50 0/50 0/50 31/50
0/40 0/45 0/35 15/226
0/50 0/50 0/50 2/50
0/40 0/45 0/35 1/22
Not reported
Not reported
0/50 0/50 5/50 38/50
0/40 0/45 4/35 20/226
12/50 11/50 20/50 43/50
10/40 11/45 17/35 22/22e
0/50 0/50 0/50 36/50
0/40 0/45 0/35 17/226
0/50 0/50 0/50 46/50
0/40 0/45 0/35 20/226
0/50 0/50 1/50 44/50
0/40 0/45 1/35 20/226
0/50 0/50 0/50 48/50
0/40 0/45 0/35 21/226
0/50 0/50 0/50 13/50
0/40 0/45 0/35 7/22e
3/50 2/50 10/50 24/50
3/40 2/45 9/35f 12/226
12/50 20/50 25/50 40/50
12/40 20/45 21/35f 21/226
3/50 3/50 9/50 8/50
3/40 3/45 9/35f 7/22e
Not reported
Not reported
7/50 11/50 6/50 16/50
7/40 11/45 6/35 8/22f
2/50 8/50 14/50 13/50
2/40 8/45 14/356 22/22e
0/50 0/50 0/50 50/50
0/40 0/45 0/35 22/22e
0 200 1,000 5,000
fir
n 11+ 55+ 274+
u 1 3 18
0/50 0/50 0/50 26/50"
Not reported
0/50 0/50 0/50 31/50"
Not reported
0/50 0/50 0/50 2/50
Not reported
Not reported
Not reported
0/50 0/50 5/50 38/50"
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
3/50 3/50 9/50 8/50
Not reported
12/50 8/50 7/50 5/50
Not reported
7/50 11/50 8/50 16/50'
Not reported
2/50 8/50 14/506 13/50"
Not reported
Not reported
Not reported
 Dose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups.
    Statistical test results were not reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in
    the external peer review draft of this document (U.S. EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
 JBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
                                                                                                                          E-2
                                       DRAFT - DO NOT CITE OR QUOTE

-------
Table E-2   Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
           JBRC drinking water study in female F344 rats
                          Yamazaki et al. (1994)a
Kano et al. (2009)
                                              Drinking water concentration (ppm)
0 200 1,000 5,000 0
Calculated
Not reported
Nasal respiratory
epithelium; nuclear
enlargement
Nasal respiratory
epithelium;
squamous cell
metaplasia
Nasal respiratory
epithelium;
squamous cell
hyperplasia
Nasal gland;
proliferation
Nasal olfactory
epithelium; nuclear
enlargement
Nasal olfactory
epithelium;
respiratory
metaplasia
Nasal olfactory
epithelium; atrophy
Lamina propria;
hydropic change
Lamina propria;
slerosis
Nasal cavity;
adhesion
Nasal cavity;
inflammation
Liver; hyperplasia
Liver; spongiosis
hepatis
Liver; cyst formation
Liver; clear cell foci
Liver; acidophilic cell
foci
Liver; basophilic cell
foci
Liver; mixed-cell foci
Kidney proximal
tubule; nuclear
enlargement
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Not reported
Not reported
0/50 0/50 0/50 35/50
Not reported
0/50 0/50 0/50 5/50
Not reported
0/50 0/50 0/50 11/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
3/50 2/50 11/50 47/50
Not reported
0/50 0/50 1/50 20/50
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control
(0)
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
2/50
1/38
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
0/50
0/38
3/50
2/38
0/50
0/38
0/50
0/38
200 1,000
5,000 0 200 1,000
5,000
Dose (Intake [mg/kg-day])"'1'
12- 56-
29 149
(21) (103)
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 28/50
0/37 24/38e
0/50 2/50
0/37 1/38
0/50 1/50
0/37 1/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 0/50
0/37 0/38
0/50 1/50
0/37 1/38
2/50 11/50
2/37 9/38
0/50 1/50
0/37 1/38
1/50 1/50
1/37 0/38
307-
720
(514)
13/50
7/24e
35/50
18/246
5/50
4/24f
11/50
8/24e
39/50
22/24e
42/50
24/24e
40/50
22/24e
46/50
23/24e
48/50
23/24e
46/50
24/24e
15/50
7/24e
47/50
24/24e
20/50
14/246
8/50
5/24f
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
1/50
1/38
0/50
0/38
1/50 3/50
1/37 3/38
0/50 6/50
0/37 6/38
11/50
7/24f
39/50
22/24e
n 18+ 83+
0 3 14
0/50 0/50 0/50
429+
69
13/50"
Not reported
0/50 0/50 0/50
35/50"
Not reported
0/50 0/50 0/50
5/50
Not reported
Not reported
Not reported
0/50 0/50 28/50"
39/50"
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
1/50 1/50 5/50
4/50
Not reported
1/50 1/50 1/50
1/50
Not reported
23/50 27/50 31/50
8/50"
Not reported
1/50 1/50 3/50
11/50'
Not reported
Not reported
Not reported
                              DRAFT - DO NOT CITE OR QUOTE
                                                                                              E-3

-------
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups.
    Statistical test results were not reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in
    the external peer review draft of this document (U.S. EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation.  The mean shown in this table was used in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
dJBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
Table E-3    Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
                drinking water study in male F344 rats

Yamazaki et al. (1994)a


0 200 1,000 5,000
JBRC(1998)P
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day

Not reported
r«ntmi 8- 41' 209-
Co,ftrol 24 121 586
1°' (16) (81) (398)
0 200 1,000 5,000
)"""
n 11+ 55+ 274+
0 1 3 18
Nasal cavity
Squamous cell carcinoma
Sarcoma NOS
Rabdomyosarcoma
Esthesioneuroepithelioma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50
0/50 0/50
3/50
Not reported
0/50
0/50 0/50
2/50
Not reported
0/50
0/50 0/50
1/50
Not reported
0/50
0/50 0/50
1/50
Not reported
0/50
0/50 0/50
3/50"
Not reported
0/50
0/50 0/50
2/50
Not reported
0/50
0/50 0/50
1/50
Not reported
0/50
0/50 0/50
1/50
Not reported
0/50

0/50

0/50

0/50

0/50
Not
0/50
Not
0/50
Not
0/50
Not
0/50
reported
0/50
reported
0/50
reported
0/50
reported
3/50"

2/50

1/50

1/50

Liver
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma
or carcinoma
Tumors at other sites
Peritoneum
mesothelioma
Subcutis fibroma
Mammary gland
fibroadenoma
Mammary gland
adenoma
Mammary gland
fibroadenoma
or adenoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals

All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50 2/50 4/50 24/50
Not reported
0/50 0/50 0/50 14/50
Not reported
Not reported
Not reported

2/50 2/50 5/50 28/50
Not reported
5/50 3/50 5/50 12/50
Not reported
1/50 1/50 0/50 4/50
Not reported
0/50 0/50 0/50 0/50
Not reported
Not reported
Not reported
0/50 2/50 4/49 24/50"'"
Not reported
0/50 0/50 0/49 14/50"'"
Not reported
0/50 2/50 4/49 33/50"'"
Not reported

2/50 2/50 5/50 28/50"'°
Not reported
5/50 3/50 5/50 12/50"
Not reported
1/50 1/50 0/50 4/50"
Not reported
Not reported
Not reported
Not reported
Not reported
3/50 4/50 7/50 32/50"'"
Not reported
0/50 0/50 0/50 14/50"'"
Not reported
3/50 4/50 7/50 39/50"'"
Not reported

2/50 2/50 5/50 28/50"'°
Not reported
5/50 3/50 5/50 12/50"
Not reported
1/50 1/50 0/50 4/50"
Not reported
0/50 1/50 2/50 2/50
Not reported
1/50 2/50 2/50 6/50°
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
bJBRC(1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used i
the external peer review draft of this document (U.S. EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation.  The mean shown in this table was used in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
dp<0.01 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01.
                                          DRAFT - DO NOT CITE OR QUOTE
                                                                                                                                  E-4

-------
Table E-4    Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
               drinking water study in female F344 rats

Yamazaki et al. (1994)a


0 200 1,000 5,000
JBRC(1998)P
Kano et al. (2009)
Drinking water concentration (ppm)
0 200 1,000 5,000 | 0 200 1,000 5,000
Calculated Dose (Intake [mg/kg-day]f "

Not Reported
r«ntmi 12' 56' 307-
Co,n,V 29 149 720
W (21) (103) (514)
n 18+ 83+ 429+
u 3 14 69
Nasal cavity
Squamous cell
carcinoma
Sarcoma NOS
Rabdomyosarcoma
Esthesioneuroepithelio
ma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50

0/50

0/50

0/50

0/50 0/50
Not
reported
0/50 0/50
Not
reported
0/50 0/50
Not
reported
0/50 0/50
Not
reported
7/50

0/50

0/50

1/50

0/50 0/50
Not
Not
Not
Not
Not
0/50 0/50
Not
0/50 7/50""
reported
reported
reported
reported
reported
0/50 1/50
reported
0/bO

0/bO

0/bO

0/bO

0/50
Not
0/50
Not
0/50
Not
0/50
Not
0/50
reported
0/50
reported
0/50
reported
0/50
reported
7/50""

0/50

0/50

1/50

Liver
Hepatocellular
adenoma
Hepatocellular
carcinoma
Hepatocellular
adenoma or
carcinoma
Tumors at other sites
Peritoneum
mesothelioma
Subcutis fibroma
Mammary gland
fibroadenoma
Mammary gland
adenoma
Mammary gland
fibroadenoma
or adenoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals

All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
1/50 0/50 5/50 38/50
Not reported
0/50 0/50 0/50 10/50
Not reported
Not reported
Not reported

1/50 0/50 0/50 0/50
Not reported
0/50 2/50 1/50 0/50
Not reported
3/50 2/50 1/50 3/50
Not reported
6/50 7/50 10/50 16/50
Not reported
Not reported
Not reported
1/50 0/50 5/50 38/50e'f
Not reported
1/50 0/50 0/50 10/50e'f
Not reported
1/50 0/50 5/50 40/50e'f
Not reported

Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
6/50 7/50 10/50 16/50d'f
Not reported
Not reported
Not reported
3/50 1/50 6/50 48/5(r
Not reported
0/50 0/50 0/50 10/5(r
Not reported
3/50 1/50 6/50 48/5(r
Not reported

1/50 0/50 0/50 0/50
Not reported
0/50 2/50 1/50 0/50
Not reported
3/50 2/50 1/50 3/50
Not reported
6/50 7/50 10/50 16/50"'
Not reported
8/50 8/50 11/50 18/50"'
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups.
    Statistical test results were not reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in
    the external peer review draft of this document (U.S.  EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used  in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
"p< 0.05 by Fisher's Exact test.
ep<0.01 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01.
                                                                                                                           E-5
                                       DRAFT - DO NOT CITE OR QUOTE

-------
Table E-5    Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
               JBRC drinking water study in male Crj:BDF1 mice

Yamazaki et al.
(1994f
JBRC(1998)M
Kano et al. (2009)
Drinking water concentration (ppm)
0 500 2,000 8,000 1 0 500
2,000 8,000 | 0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-day]p

Nasal respiratory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
nuclear enlargement
Nasal olfactory epithelium;
atrophy
Nasal cavity; inflammation
Tracheal epithelium; atrophy
Tracheal epithelium; nuclear
enlargement
Bronhcial epithelium; nuclear
enlargement
Bronchial epithelium; atrophy
Lung/bronchial; accumlation of
foamy cells
Liver; angiectasis
Kidney proximal tubule; nuclear
enlargement
Testis; mineralization
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
All
animals
Sacrificed
animals
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
r«ntr«i 37- 144- 451-
ColF01 94 358 1086
u (66) (251) (768)
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
1/50 2/50
1/31 1/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
0/50 0/50
0/31 0/33
1/50 0/50
1/31 0/33
2/50 3/50
2/31 2/33
0/50 0/50
0/31 0/33
40/50 42/50
28/31 30/33
0/50 31/50
0/25 19/266
9/50 49/50
7/25e 26/26e
1/50 48/50
0/25 26/26e
1/50 25/50
1/25 15/266
0/50 42/50
0/25 24/26e
0/50 17/50
0/25 12/266
0/50 41/50
0/25 24/26e
0/50 43/50
0/25 26/26e
0/50 27/50
0/25 22/26e
4/50 16/50
3/25 8/26f
0/50 39/50
0/25 22/26e
38/50 34/50
24/25f 21/26f
n 49+ 191+ 677+
0 5 21 74
0/50 0/50 0/50 31/506
Not reported
0/50 0/50 9/50e 49/506
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose groups.
    Statistical test results were not reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in the
    external peer review draft of this document (U.S. EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical Review
    of 1,4-Dioxane (U.S. EPA, 2010).
dJBRC did not report statistical significance for the "All animals" comparison.
ep<0.01 by x2 test.
fp < 0.05 by x2 test.
                                                                                                                          E-6
                                       DRAFT - DO NOT CITE OR QUOTE

-------
Table E-6    Nonneoplastic lesions: Comparison of histological findings reported for the 2-year
               JBRC drinking water study in female Crj:BDF1 mice

Yamazaki et
(1994f
al.
JBRC (1998)"
Kano et al.
(2009)

Drinking water concentration (ppm)

0 500 2,000
8,000
0
500
2,000
8,000 | 0
500
2,000
8,000
Calculated Dose (Intake

Nasal respiratory
epithelium; Nuclear
enlargement
Nasal olfactory
epithelium; Nuclear
enlargement
Nasal respiratory
epithelium; Atrophy
Nasal olfactory
epithelium; Atrophy
Nasal cavity;
Inflammation
Tracheal epithelium;
Atrophy
Bronhcial epithelium:
Nuclear enlargement
Bronchial epithelium;
Atrophy
Lung/bronchial;
Accumlation of foamy
cells
Kidney proximal
tubule; Nuclear
enlargement
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Control
0
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
2/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
45-
109
(77)
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
0/50
0/29
1/50
1/29
0/50
0/29
1/50
1/29
0/50
0/29
192-
454
(323)
0/50
0/17
41/50
17/176
0/50
0/17
1/50
0/17
7/50
5/1 7e
2/50
1/17
22/50
13/176
7/50
3/17
4/50
3/17
0/50
0/17
[mg/kg-day])"'1'
759-
1374
(1066)
41/50
5/5e
33/50
1/5
26/50
1/5
42/50
5/5e
42/50
5/5e
49/50
5/5e
48/50
5/5e
50/50
5/5e
45/50
5/5e
8/50
0/5
0 ^ 278 ± 40 964 ± 88
0/50 0/50 0/50 41/50"
Not reported
0/50 0/50 41/50" 33/50"
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
aDose rates mg/kg-day]) were not provided in Yamazaki et al. (1994). Drinking water concentrations (ppm) of 1,4-dioxane were used to identify the dose
    groups. Statistical test results were not reported.
'Statistical analysis was not performed for data on 'All animals' in the JBRC (1998) report.
CJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in
    the external peer review draft of this document (U.S. EPA, 2009b).
dKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
ep<0.01 by chi-square test.
                                                                                                                          E-7
                                       DRAFT - DO NOT CITE OR QUOTE

-------
Table E-7    Neoplastic  lesions: Comparison of histological findings reported for the 2-year JBRC
               drinking water study in male Crj:BDF1  mice

Yamazaki et al. (1994)a| JBRC (1998)p
Kano et al. (2009)
Drinking water concentration (ppm)

0 500 2,000 8,000
0 500 2,000 8,000
Calculated Dose (Intake [mg/kg-da

Not reported
r«ntml 37' 144' 451'
Control 94 358 1086
0 (66) (251) (768)
0 500 2,000 8,000
yir
n 49+ 191+ 677+
u 5 21 74
Nasal cavity
Esthesioneuroepithelioma
Adenocarcinoma
All Animals
Sacrificed
animals
All Animals
Sacrificed
animals
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 0/50
Not reported
0/50 0/50 0/50 1/50
Not reported
Not reported
Not reported
0/50 0/50 0/50 1/50
Not reported
0/50 0/50 0/50 0/50
Not reported
Liver
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenoma
or carcinoma
All Animals
Sacrificed
animals
All Animals
Sacrificed
animals
All Animals
Sacrificed
animals
7/50 16/50 22/50 8/50
Not reported
15/50 20/50 23/50 36/50
Not reported
Not reported
Not reported
7/50 16/50 22/50" 8/50
Not reported
15/50 20/50 23/50 36/50"'"
Not reported
21/50 31/50 37/50 39/50"'"
Not reported
9/50 17/50 23/50" 11/50
Not reported
15/50 20/50 23/50 36/50""
Not reported
23/50 31/50 37/50" 40/50""
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations of 1,4-dioxane were used to identify the dose
    groups. Statistical test results were not reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in
    the external peer review draft of this document (U.S. EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
dp£ 0.05 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01.
fp£0.01 by Fisher's Exact test.
                                                                                                                           E-S
                                       DRAFT - DO NOT CITE OR QUOTE

-------
Table E-8    Neoplastic lesions: Comparison of histological findings reported for the 2-year JBRC
               drinking water study in female Crj:BDF1 mice

Yamazaki et al.
(1994f
JBRC(1998)b
Kano et al. (2009)
Drinking water concentration ppm)

0 500 2,000 8,000
0 500 2,000 8,000
0
Calculated Dose (Intake [mg/kg-day]P

Not reported
r«ntr«i 46- 192- 759'
Control 1og 454 1374
u (77) (323) (1066)
0

500 2,000 8,000

66 ± 278 ± 964 ±
10 40 88
Nasal Cavity
Esthesioneruoepithelioma
Adenocarcinoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
0/50
0/50 0/50
0/50
Not reported
0/50
0/50 0/50
1/50
Not reported
Not
Not
reported
reported
0/50 0/50 0/50 1/50
Not
reported
0/50

0/50

0/50
Not
0/50
Not
0/50
reported
0/50
reported
0/50

1/50

Liver
Hepatocellular adenomas
Hepatocellular carcinomas
Either adenoma
or carcinoma
All animals
Sacrificed
animals
All animals
Sacrificed
animals
All animals
Sacrificed
animals
4/50 30/50 20/50 2/50
Not reported
0/50 6/50 30/50 45/50
Not reported
Not reported
Not reported
4/50
30/50" 20/50"
2/50"
Not reported
0/50
6/50' 30/50"
45/50"'"
Not reported
4/50
34/50" 41/50"
46/50"'"
Not reported
5/50 31/50" 20/50" 3/50
Not reported
0/50 6/50' 30/50" 45/50"'"
Not reported
5/50 35/50" 41/50" 46/50"'"
Not reported
aDose rates (mg/kg-day) were not provided in Yamazaki et al. (1994). Drinking water concentrations (ppm) of 1,4-dioxane were used to identify the dose
groups. Statistical test results were not reported.
bJBRC (1998) reported an estimated chemical intake range (of doses) for the animals; and the midpoint of the range (shown in parentheses) was used in
    the external peer review draft of this document (U.S.  EPA, 2009b).
cKano et al. (2009) reported a mean intake dose for each  group ± standard deviation. The mean shown in this table was used in the 2010 Toxicolgical
    Review of 1,4-Dioxane (U.S. EPA, 2010).
dp<0.01 by Fisher's Exact test.
'Significantly decreased by Cochran-Armitage test for trend p < 0.05
f p < 0.05 by Fisher's Exact test.
'Significantly increased by Peto test for trend p < 0.01
                                                                                                                            E-9
                                        DRAFT - DO NOT CITE OR QUOTE

-------
     APPENDIX  F.     DETAILS   OF  BMD  ANALYSIS  FOR
         INHALATION   RFC  FOR  1,4-DIOXANE

     F.1    Centrilobular Necrosis of the  Liver
 1           All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
 2    incidence data shown in Table F-l. for centrilobular necrosis of the liver in male F344/DuCrj rats exposed
 3    to 1.4-dioxane vapors for 2 years (Kasai et al., 2009). Doses associated with a BMR of a 10% extra risk
 4    were calculated.
     Table F-1   Incidence of centrilobular necrosis of the liver in F344/DuCrj rats exposed to
                1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
1/50

3/50
250
6/50
(12%)
1,250
12/50a
(24%)
     ap < 0.01 by Fisher's exact test.
     Source: Kasai et al. (2009).

 5          As assessed by the y2 goodness-of-fit test, several models in the software provided adequate fits
 6   to the incidence data of centrilobular necrosis of the liver in male rats (y2 p > 0.1) (Table F-2). Comparing
 7   across adequately Fitting models, the BMDL estimates were not within threefold difference of each
 8   other. Therefore, in accordance with EPA BMD technical guidance (U.S. EPA. 2000aX the adequately
 9   Fitting model that resulted in the lowest BMDL was selected as appropriate for deriving a POD which was
10   the Dichotomous-Hill model. BMDS modeling results for all dichotomous models are shown in Table F-2
11   and the model plot (Figure F-l) and output for the selected Dichotomous-Hill model are included
12   immediately after the table.
                                                                                                 F-l
                                   DRAFT - DO NOT CITE OR QUOTE

-------
Table F-2    Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
             data for centrilobular necrosis of the liver in male F344/DuCrj rats exposed to
             1,4-dioxane vapors (Kasai et al.. 2009)

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
Dichotomous-Hill0'
e
129.692
131.043
129.465
132.067
129.692
130.889
129.692
129.692
130.404
0.5099
0.2794
0.568
0.1645
0.5099
0.2992
0.5099
0.5099
0.7459
0.786
-0.142
0.676
-0.175
0.786
-0.167
0.786
0.786
-0.179
502.444
794.87
453.169
801.17
502.445
756.192
502.461
502.461
219.51
308.113
609.269
258.687
539.489
308.112
567.169
308.113
308.113
59.5598
a p-Value from the x goodness-of-fit test for the selected model. Values <0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
"Power restricted to > 1.
cSlope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit model based on lowest BMDL.

Source:  Kasai et al. (2009).
 .Q
 0.4


0.35


 0.3


0.25


 0.2


0.15


 0.1


0.05


   0
   16:18 01/12 2011
                             Dichotomous-Hill Model with 0.95 Confidence Level
                                 Dichotomous-Hill
                   BMDL
                               BMD
                     0
                             200
                                      400
                                               600
                                               dose
                                                        800
                                                                 1000
                                                                          1200
        Figure F-l  BMD Dichotomous Hill model of centrilobular necrosis incidence data
                    for male rats exposed to 1,4-dioxane vapors for 2 years to support the
                    results in Table F-2.
Dichotomous Hill  Model.  (Version:  1.2; Date:  12/11/2009)
                                                                                                       F-2
                                 DRAFT - DO NOT CITE OR QUOTE

-------
 1   Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
 2   files/dhl Centr necrosis liver Dhl-BMRlO-Restrict.(d)
 3           Gnuplot Plotting File:  C:/Documents and Settings/pgillesp/Desktop/BMDS
 4   files/dhl Centr necrosis liver Dhl-BMRlO-Restrict.pit
 5                                                  Wed Jan I2_ 16:34:41 2011
 6   ====================================================================
 7    BMDS Model Run
 8   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 9    The form of the probability function is:
10
11    P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
12    where: 0 <= g < 1, 0 < v <= 1
13    v is the maximum probability of response predicted by the model,
14    and v*g is the background estimate of that probability.
15
16    Dependent variable = Effect
17    Independent variable = Dose
18    Slope parameter is restricted as slope  >= _!
19
20    Total number of observations = £
21    Total number of records with missing values = 0_
22    Maximum number of iterations = 250
23    Relative Function Convergence has been  set to: le-008
24    Parameter Convergence has been set to:  le-008
25
26    Default Initial Parameter Values
27    v = -9999~
28    g = -9999
29    intercept = -8.08245
30    slope = I
31
32
33    Asymptotic Correlation Matrix of Parameter Estimates
34   (*** The model parameter(s) -slope have  been estimated at a boundary point, or have
35   been specified by the user, and do not appear in  the correlation matrix)
36
37    v g intercept
38    v I -0.25 -0789
39    g -0.25 I 0.016
40    intercept -0.89 0.016 I
41
42
43    Parameter Estimates
44
45    95.0% Wald Confidence Interval
46    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
47    v 0.311077 0.156196 0 . 0oT9~3876 0/617216
48    g 0.0709966 0.0662298 -0.0588115 0.200805
49    intercept -6.06188 1.34538 -8.69878 -3.42498
50    slope I NA
51
52   NA 2. Indicates that this parameter has hit a. bound implied by some ineguality
53   constraint and thus has no standard error.
54
55
56    Analysis of Deviance Table
57
58    Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
59    Full model -62.1506 4
60    Fitted model -62.2022 3 0.103279 I 0.7479
61    Reduced model -69.3031 I 14.305 3 0.002518
62
63    AIC: 130.404
64
65    Goodness of Fit
66    Scaled
67    Dose Est. Prob. Expected Observed Size  Residual

                                                                                               F-3
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2     0.0000 0.0221  1.104  1.000 5£ -0.100
 3     50.0000 0.0522  2.612 3.000 5£ 0.247
 4     250.0000 0.1285  6.423 6.000 5j3 -0.179
 5     1250.0000 0.2372  11.861 12.000 5_0  0.046
 6
 7     Chi"2 = 0.10 d.f.  =  I P-value =  0.7459
 8
 9
10     Benchmark Dose  Computation
11    Specified effect  = 0.1
12    Risk Type = Extra  risk
13    Confidence level  = 0.95
14     BMP = 219.51
15     BMDL = 59.5598
      F.2    Spongiosis Hepatis

16           All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
17    incidence data shown in Table F-3i for spongiosis hepatis of the liver in male F344/DuCrj rats exposed to
18    1.4-dioxane vapors for 2 years (Kasai et al.. 2009). Doses associated with a BMR of a 10% extra risk
19    were calculated.
      Table F-3   Incidence of spongiosis hepatis of the liver in F344/DuCrj rats exposed to 1,4-dioxane
                 via inhalation for 2 years
1 ,4-dioxane vapor concentration (ppm)

7/50
(14%)

6/50
250
13/50
(26%)
1,250
19/50a
(38%)
      ap < 0.01 by Fisher's exact test.
      Source: Kasai et al. (2009).


20          As assessed by. the £ goodness-of-fit test, several models in the software provided adequate fits
21    to the incidence data of spongiosis of the liver in male rats (y2 p > 0.1) (Table F-4). BMDL estimates for
22    all adequately Fitting models were not within threefold difference of each other (U.S. EPA. 2000a).
23    Therefore, in accordance with EPA BMP technical guidance (U.S. EPA. 2000a). the adequately fitting
24    model that resulted m the lowest BMDL was selected as appropriate for deriving a POD which was the
25    dichotomous-Hill model. However, the dichotomous-Hill model, warned that the BMDL estimate was
26    "imprecise at best" (see Figure F-2 and subsequent textual model output). Comparing across aU models
27    (excluding the dichotomous-hill model), a better fit is indicated by a lower AIC value since the BMDL
28    estimates for all appropriately Fitting models were within threefold difference of each other (U.S.
29    EPA. 2000a). As assessed by the AIC. the log-logistic model provided the best fit to the spongiosis
30    incidence data for male rats (Table F-4. Figure F-3 and subsequent textual model output) and could be
31    used to derive a POD for this endpoint.
                                                                                                    F-4
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-4   Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
                 data for spongiosis hepatis of the liver in male F344/DuCrj rats (NCI. 1978) exposed to
                 1,4-dioxane vapors
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-log isticc'T
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
Dichotomous-Hillc'
e
206.472
207.141
206.229
208.147
206.472
207.06
206.472
206.472
206.364
0.4482
0.3159
0.5102
0.1825
0.4482
0.3292
0.4482
0.4482
0.4671
1.031
1.242
0.912
1.536
1.031
1.223
1.031
1.031
1.031
369.422
537.295
314.34
633.557
369.422
515.483
369.422
369.422
289.919
224.993
392.318
172.092
414.718
224.993
371.644
224.993
224.993
59.69
     a p-Value from the x goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
     significant lack of fit, and thus a different model should be chosen.
     "Power restricted to > 1.
     cSlope restricted to > 1.
     dBetas restricted to > 0.
     eModel output warned that the BMDL estimate was "imprecise at best".
     'Bold indicates best-fit model based on lowest AIC.

     Source: Kasai et al. (2009).
                                       Dichotomous-Hill Model with 0.95 Confidence Level
                     0.5
                     0.4
                     0.3
                     0.2
                     0.1
                                                                                   1200
               16:5201/122011
            Figure F-2  BMD Dichotomous-Hill model of spongiosis hepatis incidence data for
                        male rats exposed to 1,4-dioxane vapors for 2 years to support the
                        results in Table F-4.
2    Dichotomous Hill Model.  (Version:  1.2;  Date:  12/11/2009)
3    Input Data File: C:/Documents and  Settings/pgillesp/Desktop/BMDS
4    files/dhl  spong hepa liver  Dhl-BMRlO-Restrict.(d)
5            Gnuplot Plotting File:  C:/Documents and  Settings/pgillesp/Desktop/BMDS
6    files/dhl  spong hepa liver  Dhl-BMRlO-Restrict.pit
                                                                                                        F-5
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1                                                  Wed Jan 12_ 16:52:46 2011
 2   ====================================================================
 3    BMDS Model Run
 4   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 5    The form of the probability function is:
 6
 7    P[response] = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
 8    where: 0 <= g < 1, 0 < v <= 1
 9    v is the maximum probability of response predicted by the model,
10    and v*g is the background estimate of that probability.
11
12    Dependent variable = Effect
13    Independent variable = Dose
14    Slope parameter is restricted as slope  >= _!
15
16    Total number of observations = £
17    Total number of records with missing values =  0_
18    Maximum number of iterations = 250
19    Relative Function Convergence has been  set to: le-008
20    Parameter Convergence has been set to:  le-008
21
22    Default Initial Parameter Values
23    v = -9999~
24    g = -9999
25    intercept = -8.74962
26    slope = 1.13892
27
28    Asymptotic Correlation Matrix of Parameter Estimates
29   (*** The model parameter(s) -v -slope have been estimated at  a. boundary point,  or have
30   been specified by the user, and do not appear in the correlation matrix j_
31
32    g intercept
33    g I -0.53
34    intercept -0.53 I
35
36    Parameter Estimates
37
38    95.0% Wald Confidence Interval
39    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf. Limit
40    v I NA
41    g 0.125 0.0332679 0.0597961 0.190204
42    intercept -7.86683 0.396424 -8.6438 -7.08985
43    slope I NA
44
45   NA ^ Indicates that this parameter has hit a. bound implied by some ineguality
46   constraint and thus has no standard error.
47
48    Analysis of Deviance Table
49
50    Model Log(likelihood) #_ Param's Deviance Test  d.f. P-value
51    Full model -100.45 4
52    Fitted model -101.182 2 1.46273 2 0.4813
53    Reduced model -106.633 I 12.3646 3 0.006233
54
55    AIC: 206.364
56
57    Goodness of Fit
58    Scaled
59    Dose Est. Prob. Expected Observed Size  Residual
60
61    0.0000 0.1250 6.250 7.000 _50_ 0.321
62    50.0000 0.1415 7.073 6.000 5_0 -0.435
63    250.0000 0.2015 10.075 13.000 _50_ 1.031
64    1,250.0000 0.4084 20.420 19.000 _50_ -0.409
65
66    Chi"2 = 1.52 d.f. = 2 P-value = 0.4671
67

                                                                                               F-6
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10

11
 Benchmark Dose Computation
Specified effect  =  0.1
Risk Type = Extra risk
Confidence level  =  0.95
 BMP = 289.919

 Warning: BMDL computation is at best imprecise for  these  data
 BMDL = 59.69
                                     Log-Logistic Model with 0.95 Confidence Level
            I
                   0.5
                   0.4
              0.3
                   0.2
                   0.1
                                   Log-Logistic
                             BMDL
                                        BMD
                                  200
                                           400
             16:5201/122011
                                                    600
                                                    dose
                                                             800
                                                                     1000
                                                                              1200
            Figure F-3  BMD Log-Logistic model of spongiosis hepatis incidence data for male
                       rats exposed to 1,4-dioxane vapors for 2 years to support the results in
                       Table F-4.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
Logistic Model.  (Version:  2.13;  Date: 10/28/2009)
Input Data File:  C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl spong hepa  liver Lnl-BMRlO-Restrict.(d)
        Gnuplot Plotting File: C:/Documents and Settings/pgillesp/Desktop/BMDS
files/lnl spong hepa  liver Lnl-BMRlO-Restrict.pit
                                                Wed Jan I2_ 16:52:44  2011

 BMDS Model Run

The form of the probability function is:

 P[response] = backgrounds- (1-background)/[1+EXP(-intercept-siope*Log(dose))]

 Dependent variable = Effect
 Independent variable = Dose
 Slope parameter  is restricted as slope >= _!

 Total number of  observations =  £
 Total number of  records with missing values = 0_
 Maximum number of  iterations =  250
                                                                                                 F-7
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    Relative Function Convergence has been set to: le-008
 2    Parameter Convergence has been set to: le-008
 3
 4    User has chosen the log transformed model
 5
 6    Default Initial Parameter Values
 7    background = 0.14
 8    intercept = -8.74962
 9    slope = 1.13892
10
11   Asymptotic Correlation Matrix of Parameter Estimates
12    (*** The model parameter(s) -slope have been estimated at a boundary point,  or have
13   been specified by the user, and do not appear in the correlation matrix)
14
15    background intercept
16   background I -0.54
17    intercept -0.54 I
18
19    Parameter Estimates
20    95.0% Wald Confidence Interval
21    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
22   background 0.1376? 1 1 1~~
23    intercept -7.9477 111
24    slope 1111
25
26   1 2. Indicates that this value is not calculated.
27
28
29    Analysis of Deviance Table
30
31    Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
32    Full model -100.45 £
33    Fitted model -101.115 ,2 1.3283 2_ 0.5147
34    Reduced model -106.633 I 12.3646 3_ 0.006233
35
36    AIC: 206.229
37
38
39    Goodness of Fit
40    Scaled
41    Dose Est. Prob. Expected Observed Size Residual
42
43    0.0000 0.1377 6.885 7.000 _50_ 0.047
44    50.0000 0.1527 7.633 6.000 5J) -0.642
45    250.0000 0.2077 10.385 13.000 _50_ 0.912
46    1250.0000 0.4019 20.097 19.000 5_0 -0.316
47
48    Chi"2 = 1.35 d.f. = £ P-value = 0.5102
49
50
51    Benchmark Dose Computation
52   Specified effect =0.1
53   Risk Type = Extra risk
54   Confidence level = 0.95
55    BMP = 314.34
56    BMDL = 172.092
     F.3   Squamous Cell Metaplasia
57          All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
58   incidence data shown in Table F-5. for squamous cell metaplasia of the respiratory epithelium in male
                                                                                               F-8
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1   F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
 2   a 10% extra risk were calculated.
      Table F-5   Incidence of squamous cell metaplasia of the respiratory epithelium in F344/DuCrj
                 rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50
0/50
250
7/50b
(14%)
1,250
44/50a
     ap < 0.01 by Fisher's exact test.
     bp < 0.05 by Fisher's exact test.
     Source: Kasai et al. (2009).

 3           For incidence of squamous cell metaplasia in F344/DuCrj male rats, the logistic and probit
 4   models all exhibited a statistically significant lack of fit (i.e.. y2 p-va\ue < 0.1; see Table F-6). and thus
 5   should not be considered further for identification of a POD. All of the remaining models exhibited
 6   adequate fit. The BMDL estimates for all appropriately Fitting models were within threefold
 7   difference of each other, indicating that BMDL selection should be made based on model fit (U.S.
 8   EPA. 2000a). As assessed by the AIC. the Log-probit model provided the best fit to the squamous cell
 9   metaplasia data for male rats (Table F-6, Figure F-4^ and could be used to derive a POD for this
10   endpoint.
                                                                                                       F-9
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-6   Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
                 data for squamous cell metaplasia of the respiratory epithelium in male F344/DuCrj
                 rats exposed to 1,4-dioxane vapors (Kasai et al.. 2009)

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc' e
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-Hillc
81.687
89.4148
81.5252
81.23
82.6875
87.9361
82.1236
92.9215
83.1888
0.8682
0.0464
0.9142
0.9894
0.6188
0.0779
0.7679
0.0198
0.9995

1.806
0.131
0.032
0.605
1.681

-1.76

218.38
370.443
218.218
217.79
231.294
337.732
218.435
87.682
240.867
150.329
288.535
158.293
159.619
141.025
268.424
145.383
68.8015
161.945
     1 p-Value from the x2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
     significant lack of fit, and thus a different model should be chosen.
     bPower restricted to > 1.
     cSlope restricted to > 1.
     dBetas restricted to > 0.
     eBold indicates best-fit model based on lowest AIC.

     Source: Kasai et al. (2009).
                                        LogProbit Model with 0.95 Confidence Level
            "8
            •

            .
            o
                    0.8
                    0.6
                    0.4
                    0.2
                                                                                    1200
             13:11 01/132011
            Figure F-4  BMD Log-probit model of squamous cell metaplasia of the respiratory
                        epithelium incidence data for male rats exposed to 1,4-dioxane vapors
                        for 2 years to support the results in Table F-6.
1    Probit Model.  (Version:  3.2; Date:  10/28/2009)
2    Input Data File:  C:/Documents and  Settings/pgillesp/Desktop/BMDS
3    files/lnp squ  cell meta  re Lnp-BMRlO-Restrict.(d)
                                                                                                       F-10
                                     DRAFT - DO NOT CITE OR QUOTE

-------
 1           Gnuplot Plotting File: C:/Documents and Settings/pgi11esp/Desktop/BMDS
 2   files/lnp sgu cell meta re Lnp-BMRlO-Restrict.pit
 3                                                  Thu Jan 13 13:11:09 2011
 4   ====================================================================
 5    BMDS Model Run
 f)   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 7    The form of the probability  function is:
 8
 9    P[response] = Background +  (1-Background) ^_ CumNorm(Intercept+Slope*Log(Dose)),
10           where CumNorm(.)  is the cumulative normal distribution function
11
12    Dependent variable = Effect
13    Independent variable = Dose
14    Slope parameter is restricted as slope  >= _!
15
16    Total number of observations = £
17    Total number of records with missing values = 0_
18    Maximum number of iterations = 250
19    Relative Function Convergence has been  set to: le-008
20    Parameter Convergence has been set to:  le-008
21
22    User has chosen the log transformed model
23
24    Default Initial  (and Specified) Parameter Values
25    background = 0_
26    intercept = -6.76507
27    slope = 1.09006
28
29    Asymptotic Correlation Matrix of Parameter Estimates
30   (*** The model parameter(s) -background  have been estimated at a. boundary point,  or
31   have been specified by the user, and do  not appear in the correlation matrix)
32
33    intercept slope
34    intercept I -0.99
35    slope -0.99 I
36
37    Parameter Estimates
38
39    95.0% Wald Confidence Interval
40    Variable Estimate Std. Err.  Lower Conf. Limit Upper Conf. Limit
41   background 0_ NA
42    intercept -8.86173 1.2226 -11.258 -6.46548
43    slope 1.40803 0.193057 1.02965 1.78642
44
45   NA 2. Indicates that this parameter has hit a. bound implied by some ineguality
46   constraint and thus has no standard error.
47
48    Analysis of Deviance Table
49
50    Model Log(likelihood)  #_ Param's Deviance Test d.f. P-value
51    Full model -38.5944 4
52    Fitted model -38.615 2_ 0.041197 £ 0.9796
53    Reduced model -113.552 I 149.916 3 <.OOQ1
54
55    AIC: 81.23
56
57    Goodness of Fit
58    Scaled
59    Dose Est. Prob. Expected Observed Size  Residual
60
61    0.0000 0.0000 0.000 0.000 5£ 0.000
62    50.0000 0.0004 0.020 0.000 5_0 -0.141
63    250.0000 0.1384 6.922 7.000 _50_ 0.032
64    1250.0000 0.8808 44.038 44.000 5J) -0.017
65
66    Chi"2 = 0.02 d.f. = _2 P-value = 0.9894
67

                                                                                              F-ll
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    Benchmark Dose Computation
 3   Specified effect =  0.1
 4   Risk  Type = Extra risk
 5   Confidence level =  0.95
 6    BMP  = 217.79
 7    BMDL = 159.619
     F.4   Squamous Cell Hyperplasia

 8          All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
 9   incidence data shown in Table F^ for squamous cell hyperplasia of the respiratory epithelium in male
10   F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
11   a 10% extra risk were calculated.
     Table F-7   Incidence of squamous cell hyperplasia of the respiratory epithelium in F344/DuCrj
                 rats exposed to 1,4-dioxane via inhalation for 2 years
     	1,4-dioxane vapor concentration (ppm)	
     	50	250	1,250	
              0/50                  0/50                  1/50                      10/50a
     	(2%)	(20%)	
     ap < 0.01 by Fisher's exact test.
     Source: Kasai et al. (2009).


12          For incidence of squamous cell hyperplasia in F344/DuCrj male rats, the logistic, probit. and

13   quantal-linear models all exhibited a statistically significant lack of fit (i.e.. y2 p-value < 0.1; see
14   Table F-8). and thus should not be considered further for identification of a POD. All of the remaining

15   models exhibited adequate fit. The BMDL estimates for all appropriately Fitting models were within

16   threefold difference of each other, indicating that BMDL selection should be made based on model

17   fit (U.S. EPA. 2000a). As assessed by the AIC. the Log-probit model provided the best fit to the
18   squamous cell hyperplasia data for male rats (Table F-8i Figure F-5 and subsequent textual model output).
19   and could be used to derive a POD for this endpoint.
                                                                                                   F-12
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-8    Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
                 data for squamous cell hyperplasia of the respiratory epithelium in male F344/DuCrj
                 rats exposed to 1,4-dioxane vapors (Kasai et al.. 2009)

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc' e
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-Hillc
81.687
89.4148
81.5252
81.23
82.6875
87.9361
82.1236
92.9215
83.1888
0.8682
0.0464
0.9142
0.9894
0.6188
0.0779
0.7679
0.0198
0.9995

1.806
0.131
0.032
0.605
1.681

-1.76

218.38
370.443
218.218
217.79
231.294
337.732
218.435
87.682
240.867
150.329
288.535
158.293
159.619
141.025
268.424
145.383
68.8015
161.945
     1 p-Value from the x2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
     significant lack of fit, and thus a different model should be chosen.
     bPower restricted to > 1.
     cSlope restricted to > 1.
     dBetas restricted to > 0.
     eBold indicates best-fit model based on lowest AIC.
     Source: Kasai et al. (2009).
                                         LogProbit Model with 0.95 Confidence Level
                    0.35
                     0.3
                    0.25
                     0.2
                    0.15
                     0.1
                    0.05
                                                                                   1200
             13:2501/132011
            Figure F-5  BMD Log-probit model of squamous cell hyperplasia of the respiratory
                        epithelium incidence data for male rats exposed to 1,4-dioxane vapors
                        for 2 years to support the results in Table F-8.
1    Probit Model.   (Version: 3.2;  Date: 10/28/2009)
2    Input Data File: C:/Documents and Settings/pgillesp/Desktop/BMDS
3    files/lnp  squ cell  hyper re Lnp-BMRlO-Restrict.(d)
4            Gnuplot Plotting File: C:/Documents  and Settings/pgillesp/Desktop/BMDS
5    files/lnp  sgu cell  hyper re Lnp-BMRlO-Restrict.pit
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                      F-13

-------
 1                                                  Thu Jan 13_ 13:25:05 2011
 2   ====================================================================
 3    BMDS Model Run
 4   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 5    The form of the probability function is:
 6
 7    P[response] = Background +  (1-Background) ^_ CumNorm(Intercept+Slope*Log(Dose)),
 8           where CumNorm(.)  is the cumulative normal distribution function
 9
10    Dependent variable = Effect
11    Independent variable = Dose
12    Slope parameter is restricted as slope  >= _!
13
14    Total number of observations = £
15    Total number of records with missing values = 0_
16    Maximum number of iterations = 250
17    Relative Function Convergence has been  set to: le-008
18    Parameter Convergence has been set to:  le-008
19
20    User has chosen the log transformed model
21
22    Default Initial  (and Specified) Parameter Values
23    background = £
24    intercept = -7.75604
25    slope = I
26
27    Asymptotic Correlation Matrix of Parameter Estimates
28   (*** The model parameter(s) -background  -slope have been estimated at  a boundary
29   point, or have been specified by the user, and do not appear  in  the correlation
30   matrix)
31
32    intercept
33    intercept _!
34
35    Parameter Estimates
36
37    95.0% Wald Confidence Interval
38    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
39   background C) NA
40    intercept -7.90911 0.186242 -8.27414 -7.54408
41    slope I NA
42
43   NA 2. Indicates that this parameter has hit a. bound implied by some ineguality
44   constraint and thus has no standard error.
45
46    Analysis of Deviance Table
47
48    Model Log (likelihood)  #_ Param' s Deviance Test d. f. P-value
49    Full model -29.9221 £
50    Fitted model -30.2589 I 0.673572 3 0.8794
51    Reduced model -42.5964 I 25.3487 3_ <.OOQ1
52
53    AIC: 62.5177
54
55    Goodness of Fit
56    Scaled
57    Dose Est. Prob. Expected Observed Size  Residual
58
59    0.0000 0.0000 0.000 0.000 _50_ 0.000
60    50.0000 0.0000 0.002 0.000 5J) -0.040
61    250.0000 0.0085 0.424 1.000 50 0.
62    1250.0000 0.2182 10.911 10.000 5_0 -0.312
63
64    Chi"2 = 0.89 d.f. = 3 P-value = 0.8282
65
66
67    Benchmark Dose Computation

                                                                                              F-14
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Specified  effect = 0.1
 2    Risk Type  =  Extra risk
 3    Confidence level = 0.95
 4     BMP = 755.635
 5     BMDL = 560.86
      F.5    Respiratory Metaplasia

 6           All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
 7    incidence data shown in Table F^ for respiratory metaplasia of the olfactory epithelium in male
 8    F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (NCI. 1978). Doses associated with a BMR of
 9    a 10% extra risk were calculated.
      Table F-9   Incidence of respiratory metaplasia of the olfactory epithelium in F344/DuCrj rats
                 exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)

11/50
(22%)

34/50
250
49/50 a
(98%)
1,250
48/50a
(96%)
      ap < 0.01 by Fisher's exact test.
      Source: Kasai et al. (2009).

10           As assessed by the y2 goodness-of-fit test, no models in the software provided adequate fits to the
11    data for the incidence of respiratory metaplasia of the olfactory epithelium in male rats (y1 p > 0.1)
12    (Table F-10). However, given that first non-control dose had a response level substantially above the
13    desired BMR (i.e. 10%). the use of BMP methods included substantial model uncertainty. The model
14    uncertainty associated with this dataset is related to low-dose extrapolation and consistent with BMP
15    technical guidance document (U.S. EPA. 2000a). all available dichotomous models in the Benchmark
16    Dose Software (version 2.1.2) were  fit to the incidence data shown in Table F-9 with the highest dose
17    group omitted. As assessed by the y2 goodness-of-fit test, the logistic, log-logistic, log-probit and probit
18    models all exhibited a statistically significant lack of fit (i.e.. y2 p-va\ue < 0.1:See Table F-l 1). and thus
19    should not be considered further for identification of a POD. The BMDL estimates for all appropriately
20    Fitting models were within threefold difference of each other, indicating that BMDL selection should
21    be made based on model fit (U.S. EPA. 2000a). The AIC values for gamma, multistage, quantal-linear.
22    and Weibull models in Table F-l 1 are equivalent and the lowest and, in this case, essentially represent the
23    same model. Therefore, consistent with the external review draft Benchmark Dose Technical Guidance
24    (U.S. EPA. 2000a). any of them with equal AIC values (gamma, multistage, quantal-linear. or Weibull)
25    could be used to identify a POD for this  endpoint. The model plot for the gamma model (Figure F-6)
26    and output are included immediately after the table.
                                                                                                     F-15
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table F-10  Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
             data for respiratory metaplasia of olfactory epithelium in male F344/DuCrj rats (Kasai
             et al.. 2009) exposed to 1,4-dioxane vapors

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-Hillc
179.68
191.339
152.72
161.267
179.68
198.785
179.68
179.68
150.466
0
0
0.0285
0
0
0
0
0
NA
-2.07
1.788
0.039
-0.39
-2.07
1.479
-2.07
-2.07

17.4082
34.2946
4.05465
14.3669
17.4082
61.4378
17.4082
17.4082
38.8552
12.3829
24.5917
1.90233
10.3023
12.3829
45.9091
12.3829
12.3829
31.4727
ap-Value from the x2 good ness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to > 1.
cSlope restricted to > 1.
dBetas restricted to >0.

Source: Kasai et al. (2009).
Table F-11   Good ness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
             data for respiratory metaplasia of olfactory epithelium with high dose group dropped
             in male F344/DuCrj rats (Kasai et al.. 2009) exposed to 1,4-dioxane vapors

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma0' e
Logistic
Log-logistic0
Log-probitc
Multistage
(2degree)d'e
Probit
Weibull"
Quantal-Linear6
129.463
133.583
131.182
131.182
129.463
136.121
129.463
129.463
0.5815
0.0119
NA
NA
0.5815
0.0066
0.5815
0.5815
-0.106
-1.031


-0.106
-1.511
-0.106
-0.106
6.46848
12.5197
14.2075
12.2114
6.46847
15.2883
6.46847
6.46847
4.73742
9.34421
3.77044
7.80131
4.73742
11.6855
4.73742
4.73742
1 p-Value from the x2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
significant lack of fit, and thus a different model should be chosen.
bPower restricted to > 1.
cSlope restricted to > 1.
dBetas restricted to >0.
eBold indicates best-fit models based on lowest AIC.

Source:  Kasai et al. (2009).
                                  DRAFT - DO NOT CITE OR QUOTE
                                                                                                        F-16

-------
                                   Gamma Multi-Hit Model with 0.95 Confidence Level
            I
                   0.8
              0.6
                   0.4
                   0.2
                                                                                250
             16:2401/132011
            Figure F-6  BMD Gamma model of respiratory metaplasia of olfactory epithelium
                       incidence data for male rats exposed to 1,4-dioxane vapors for 2 years
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Gamma Model.  (Version: 2.15;  Date:  10/28/2009)
Input Data File: C:/Documents  and  Settings/pgillesp/Desktop/BMDS
files/gam resp meta no high dose Gam-BMRlO-Restrict.(d)
        Gnuplot Plotting File: C:/Documents and  Settings/pgillesp/Desktop/BMDS
files/gam resp meta no high dose Gam-BMRlO-Restrict.pit
                                               Thu  Jan~13  16:24:15  2011

 BMDS Model Run

 The form of the probability  function  is:

 P[response]= background+(1-background)*CumGamma[slope*dose,power],
        where CumGamma(.)  is the cummulative  Gamma  distribution function

 Dependent variable = Effect
 Independent variable = Dose
 Power parameter is restricted as  power >=1

 Total number of observations  = 3_
 Total number of records with  missing  values  = 0_
 Maximum number of iterations  = 250
 Relative Function Convergence has been set  to:  le-008
 Parameter Convergence has been set  to:  le-008

 Default Initial  (and Specified) Parameter Values
 Background = 0.230769
 Slope = 0.022439
 Power = 1.3

 Asymptotic Correlation Matrix  of  Parameter Estimates
(*** The model parameter(s)  -Power have  been estimated at a boundary point, or have
been specified by the user,  and do not appear in the correlation matrix)

 Background Slope
Background _! -0 . 33
 Slope -0.33 1
                                                                                               F-17
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    Parameter  Estimates
 3
 4    95.0% Wald Confidence  Interval
 5    Variable Estimate  Std.  Err.  Lower Conf.  Limit Upper Conf. Limit
 6   Background  0.226249 Q.05'88535 0.110898 0.3416
 7    Slope 0.0162883  0.00320976 0.00999729 0.0225793
 8    Power I NA
 9
10   NA 2.  Indicates  that this parameter has hit a. bound implied by some inequality
11   constraint  and  thus has  no standard error.
12
13    Analysis of Deviance Table
14
15    Model Log (likelihood)  #_ Param' s Deviance Test d. f.  P-value
16    Full model -62.5908 3
17    Fitted model -62.7313  ,2 0.280907 I 0.5961
18    Reduced model  -99.1059  I 73.0301 2_ <.OOQ1
19
20    AIC: 129.463
21
22    Goodness of Fit
23    Scaled
24    Dose Est.  Prob.  Expected Observed Size Residual
25
26    0.0000 0.2262  11.312 11.000  _50_ -0.106
27    50.0000 0.6573 32.865  34.000 5J) 0.338
28    250.0000 0.9868  49.341  49.000 _50_ -0.422
29
30    Chi^2 = 0.30 d.f.  = I  P-value = 0.5815
31
32    Benchmark  Dose Computation
33   Specified effect  =  0.1
34   Risk  Type = Extra risk
35   Confidence  level  =  0.95
36    BMP  = 6.46848
37    BMDL = 4.73742
     F.6   Atrophy

38                 All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were
39                 fit to the incidence data shown in Table F-12. for atrophy of the olfactory epithelium in
40                 male F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (Kasai etal.. 2009).
41                 Doses associated with a BMR of a 10% extra risk were calculated.
                                                                                               F-18
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-12  Incidence of respiratory metaplasia of the olfactory epithelium in F344/DuCrj rats
                 exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)
0
0/50
50
40/50 a
(80%)
250
47/50 a
(94%)
1,250
48/50a
(96%)
     ap < 0.01 by Fisher's exact test.
     Source: Kasai et al. (2009).

1           As assessed by the y2 goodness-of-fit test, the gamma, logistic, log-probit. multistage, probit.
2    Weibull. and quantal-linear models all exhibited a statistically significant lack of fit (i.e.. y2/>-value < 0.1;
3    see Table F-13). and thus should not be considered further for identification of a POD. The BMDL
4    estimates for all appropriately Fitting models were within threefold difference of each other.
5    indicating that BMDL selection should be made based on model fit (U.S. EPA. 2000a). As assessed by
6    the AIC. the Log-logistic model provided the best fit to the atrophy data for male rats (Table F-13.
7    Figure F-?X and could be used to derive a POD for this endpoint. However, given that first non-control
8    dose had a response level substantially above the desired BMR (i.e. 10%). the use of BMP methods
9    included substantial model uncertainty.
                                                                                                      F-19
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-13  Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
                 data for atrophy of olfactory epithelium in male F344/DuCrj rats (Kasai et al.. 2009)
                 exposed to 1,4-dioxane vapors
Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-log isticc'e
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-Hillc
159.444
190.692
93.9074
117.337
159.444
200.626
159.444
159.444
95.5314
0
0
0.3023
0
0
0
0
0
1

4.342


0
3.943



9.93187
33.9373
1.67195
9.42745
9.9319
61.9146
9.9319
9.9319
2.93951
8.14152
25.4454
1.01633
7.20318
8.14152
47.107
8.14152
8.14152
0.544697
     1 p-Value from the x2 goodness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
     significant lack of fit, and thus a different model should be chosen.
     bPower restricted to > 1.
     cSlope restricted to > 1.
     dBetas restricted to >0.
     eBold indicates best-fit model based on lowest AIC.
     Source: Kasai et al. (2009).
                                       Log-Logistic Model with 0.95 Confidence Level
                    0.8
                    0.6
                    0.4
                    0.2
                     0
                                     Log-Logistic
                                                                                      1
                      BMDLBMD
                                   200
                                             400
             09:5301/142011
                                                       600
                                                       dose
                                                                800
                                                                          1000
                                                                                   1200
            Figure F-7  BMD Log-Logistic model of atrophy of olfactory epithelium incidence
                        data for male rats exposed to 1,4-dioxane vapors for 2 years to support
                        the results in Table F-13.
1    Logistic Model.  (Version:  2.13; Date:  10/28/2009)
2    Input Data File:  C:/Documents and  Settings/pgillesp/Desktop/BMDS
3    files/lnl atrophy Lnl-BMRlO-Restrict.(d)
4             Gnuplot Plotting File:  C:/Documents and  Settings/pgillesp/Desktop/BMDS
5    files/lnl atrophy Lnl-BMRlO-Restrict.pit
                                     DRAFT - DO NOT CITE OR QUOTE
                                                                                                       F-20

-------
 1                                                  Fri Jan \A_ 09:53:22 2011
 2   ====================================================================
 3    BMDS Model Run
 4   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 5    The form of the probability function is:
 6    P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]
 7
 8    Dependent variable = Effect
 9    Independent variable = Dose
10    Slope parameter is restricted as slope  >= _!
11
12    Total number of observations = £
13    Total number of records with missing values = 0_
14    Maximum number of iterations = 250
15    Relative Function Convergence has been  set to: le-008
16    Parameter Convergence has been set to:  le-008
17
18    User has chosen the log transformed model
19
20    Default Initial Parameter Values
21    background = 0_
22    intercept = -3.48908
23    slope = I
24
25    Asymptotic Correlation Matrix of Parameter Estimates
26    (*** The model parameter(s) -background  -slope have been estimated at  a  boundary
27   point, or have been specified by the user, and do not appear in  the correlation
28   matrix)
29
30    intercept
31    intercept _!
32
33    Parameter Estimates
34    95.0% Wald Confidence Interval
35    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
36   background 0_ 1 1 1
37    intercept -2.71122 111
38    slope 1111
39
40   1 2. Indicates that this value is not calculated.
41
42    Analysis of Deviance Table
43
44    Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
45    Full model -44.7657 £
46    Fitted model -45.9537 I 2.37596 3 0.4981
47    Reduced model -126.116 I 162.701 3_ <.OOQ1
48
49    AIC: 93.9074
50
51    Goodness of Fit
52    Scaled
53    Dose Est. Prob. Expected Observed Size  Residual
54
55    0.0000 0.0000 0.000 0.000 _50_ 0.000
56    50.0000 0.7687 38.433 40.000 5J) 0.525
57    250.0000 0.9432 47.161 47.000 _50_ -0.099
58    1250.0000 0.9881 49.405 48.000 5_0 -1.833
59
60    Chi"2 = 3.65 d.f. = 3 P-value = 0.3023
61
62    Benchmark Dose Computation
63   Specified effect = 0.1
64   Risk Type = Extra risk
65   Confidence level = 0.95
66    BMP = 1.67195
67    BMDL = 1.01633

                                                                                              F-21
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      F.7    Hydropic Change

 1           All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
 2    incidence data shown in Table F-14. for hydropic change of the lamina propria in the nasal cavity of male
 3    F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (Kasai et al.. 2009). Doses associated with a
 4    BMR of a 10% extra risk were calculated.
     Table F-14  Incidence of hydropic change of the lamina propria in the nasal cavity of F344/DuCrj
                 rats exposed to 1,4-dioxane via inhalation for 2 years
1,4-dioxane vapor concentration (ppm)

0/50

2/50
(4%)
250
36/50 a
(72%)
1,250
49/50a
     ap < 0.01 by Fisher's exact test.
     Source: Kasai et al., (2009).

 5          For incidence of hydropic change of the lamina propria in F344/DuCrj male rats, the gamma.
 6   logistic, multistage, probit. Weibull. and quantal-linear models all exhibited a statistically significant lack
 7   of fit (i.e.. y2/>-value < 0.1; see Table F-16). and thus should not be considered further for identification
 8   of a POD. The BMDL estimates for all appropriately Fitting models were within threefold difference
 9   of each other, indicating that BMDL selection should be made based on model fit (U.S.  EPA. 2000a).
10   As assessed by the AIC. the Log-logistic model provided the best fit to the hydropic change of the lamina
11   propria data for male rats (Table F-15i Figure F-8 and subsequent text output), and could be used to
12   derive a POD of for this endpoint.
                                                                                                    F-22
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-15  Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
                 data for hydropic change of the lamina propria in the nasal cavity of male F344/DuCrj
                 rats exposed to 1,4-dioxane vapors (Kasai et al.. 2009)

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-log isticc'e
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
ichotomous-Hillc
98.3441
117.957
90.5388
91.5881
99.3482
136.585
100.225
99.3482
91.8937
0.0002
0
0.6819
0.3458
0.0256
0
0.0033
0.0256
1
-1.321
-1.143
-0.333
-0.538
-2.411
-2.099
-1.899
-2.411

51.979
89.2909
68.5266
63.0852
28.7899
92.6118
39.1371
28.7899
73.1032
28.7632
70.6131
46.7808
44.5657
22.6831
74.3784
23.9762
22.6831
49.2687
     ap-Value from the x2 good ness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
     significant lack of fit, and thus a different model should be chosen.
     bPower restricted to > 1.
     cSlope restricted to > 1.
     dBetas restricted to >0.
     eBold indicates best-fit model based on lowest AIC.
     Source: Kasai et al. (2009).
                                       Log-Logistic Model with 0.95 Confidence Level
                   0.8
                   0.6
                   0.4
                   0.2
                                     Log-Logistic
                        BMDL
                              BMD
                                   200
                                             400
             10:3001/142011
                                                      600
                                                       dose
                                                                800
                                                                         1000
                                                                                   1200
            Figure F-8  BMD Log-logistic model of hydropic change of lamina propria (nasal
                        cavity) incidence data for male rats exposed to 1,4-dioxane vapors for 2
                        years to support the results in Table F-16.
1    Logistic  Model.  (Version: 2.13;  Date:  10/28/2009)
2    Input Data File:  C:/Documents  and Settings/pgillesp/Desktop/BMDS
3    files/lnl hydrpic Lnl-BMRlO-Restrict.(d)
4            Gnuplot Plotting File: C:/Documents  and Settings/pgillesp/Desktop/BMDS
5    files/lnl hydrpic Lnl-BMRlO-Restrict.pit
                                    DRAFT - DO NOT CITE OR QUOTE
                                                                                                      F-23

-------
 1   Fri Jan \A_ 10:30:47 2011
 2   =========================
 3    BMDS Model Run
 4   ~~~~~~~~~~~~~~~~
 5    The form of the probability function  is:
 6    P[response] = background+(1-background)/[1+EXP(-intercept-siope*Log(dose))]
 7
 8    Dependent variable = Effect
 9    Independent variable = Dose
10    Slope parameter is restricted  as  slope  >= _!
11
12    Total number of observations = £
13    Total number of records with missing  values  = 0_
14    Maximum number of iterations = 250
15    Relative Function Convergence  has been  set to: le-008
16    Parameter Convergence has been set  to:  le-008
17
18    User has chosen the log transformed model
19
20    Default Initial Parameter Values
21    background = 0_
22    intercep^=~-Il. 5745
23    slope = 2.19638
24
25    Asymptotic Correlation Matrix  of  Parameter Estimates
26    (*** The model parameter(s) -background  have  been  estimated  at  a. boundary point,  or
27   have been specified by the user, and do  not appear in the  correlation matrix)
28
29    intercept slope
30    intercept I -0.99
31    slope -0.99 I
32
33    Parameter Estimates
34    95.0% Wald Confidence Interval
35    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf.  Limit
36   background 0_ ^_ ^_ ^_
37    intercept -12.1316 111
38    slope 2.3501 111"
39
40   1 2. Indicates that this value is not calculated.
41
42    Analysis of Deviance Table
43
44    Model Log (likelihood) #_ Param' s Deviance Test d. f. P-value
45    Full model -42.9468 £
46    Fitted model -43.2694 ,2 0. 645129  ,2  0.7243
47    Reduced model -136.935 I 187.976  3_  <.OOQ1
48
49    AIC: 90.5388
50
51    Goodness of Fit
52    Scaled
53    Dose Est. Prob. Expected Observed Size  Residual
54
55    0.0000 0.0000 0.000 0.000 _50_ 0.000
56    50.0000 0.0503 2.515 2.000 5J)  -0.333
57    250.0000 0.6994 34.969 36.000  _50_  0.318
58    1250.0000 0.9903 49.515 49.000 5_0 -0.744
59
60    Chi"2 = 0.77 d.f. = 2 P-value  = 0.6819
61
62    Benchmark Dose Computation
63   Specified effect = 0.1
64   Risk Type = Extra risk
65   Confidence level = 0.95
66    BMP = 68.5266
67    BMDL = 46.7808

                                                                                              F-24
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      F.8    Sclerosis

 1           All available dichotomous models in the Benchmark Dose Software (version 2.1.2) were fit to the
 2    incidence data shown in Table F-16. for sclerosis of the lamina propria in the nasal cavity of male
 3    F344/DuCrj rats exposed to 1.4-dioxane vapors for 2 years (Kasai et al., 2009). Doses associated with a
 4    BMR of a 10% extra risk were calculated.
      Table F-16  Incidence of sclerosis of the lamina propria in the nasal cavity of F344/DuCrj rats
                 exposed to 1,4-dioxane via inhalation for 2 years
     	1,4-dioxane vapor concentration (ppm)	
     	50	250	1,250	
             0/50                0/50                    22/50a                       40/50a
     	(44%)	(80%)	
     ap < 0.01 by Fisher's exact test.
     Source: Kasai et al. (2009).


 5           As assessed by the y2 goodness-of-fit test, all models with the exception of the dichotomous-hill
 6   model, exhibited a statistically significant lack of fit (i.e.. y2 p-value < 0. l:See Table F-17). and thus
 7   should not be considered further for identification of a POD. Since the dichotomous-hill model provided
 8   the only fit to the sclerosis of the lamina propria data for male rats as assessed by the y.2 goodness-of-fit
 9   test (Table F-17. Figure F-9 and subsequent text output), it could be considered to derive a POD for this
10   endpoint; however, the model output warned that the BMDL estimate was "imprecise at best".
                                                                                                      F-25
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     Table F-17  Goodness-of-fit statistics and BMD10 and BMDL10 values from models fit to incidence
                 data for sclerosis of the lamina propria in the nasal cavity of male F344/DuCrj rats
                 exposed to 1,4-dioxane vapors (Kasai et al.. 2009)

Model
AIC
p-valuea
Scaled Residual
of Interest
BMDio
(ppm)
BMDLio
(ppm)
Male
Gamma"
Logistic
Log-logistic0
Log-probitc
Multistage
(2 degree)d
Probit
Weibull"
Quantal-Linear
Dichotomous-Hillc'
e
134.416
161.562
130.24
127.784
132.436
159.896
132.436
132.436
124.633
0.0123
0
0.0683
0.0829
0.0356
0
0.0356
0.0356
0.9994
-1.89
4.542
-1.579
-0.995
-1.949
4.619
-1.949
-1.949
0
75.4489
244.217
86.3863
109.558
71.9719
231.856
71.9719
71.9719
206.74
57.6938
196.446
52.4762
88.1232
57.6471
191.419
57.6471
57.6471
167.46
     ap-Value from the x good ness-of-fit test for the selected model. Values < 0.1 indicate that the model exhibited a statistically
     significant lack of fit, and thus a different model should be chosen.
     "Power restricted to > 1.
     cSlope restricted to > 1.
     dBetas restricted to >0.
     eModel output warned that the BMDL estimate was "imprecise at best".

     Source: Kasai et al. (2009).
 1
 2
 o
 J
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Dichotomous  Hill Model.   (Version:  1.2;  Date: 12/11/2009)
Input Data File: C:/Documents and  Settings/pgillesp/Desktop/BMDS
files/dhl sclerosis  Dhl-BMRlO-Restrict.(d)
        Gnuplot  Plotting File: C:/Documents  and Settings/pgillesp/Desktop/BMDS
files/dhl sclerosis  Dhl-BMRlO-Restrict.pit
                                                 Fri Jan \A_ 10:53:28  2011

 BMDS Model  Run

 The form of the probability  function is:
 P[response]  = v*g +(v-v*g)/[1+EXP(-intercept-slope*Log(dose))]
 where: 0 <= g < 1,  0 < v <=  1
 v is the maximum probability of response predicted by the  model,
 and v*g is  the background estimate  of  that probability.
 Dependent  variable = Effect
 Independent  variable = Dose
 Slope parameter is restricted  as  slope >= _!

 Total number of observations = £
 Total number of records with missing values = 0_
 Maximum number of iterations = 250
 Relative Function Convergence  has been set to: le-008
 Parameter  Convergence has been set  to: le-008

 Default Initial Parameter Values
 v = -9999
 g = -9999
 intercept  =  -11.4511
 slope = ~.?6444

 Asymptotic Correlation Matrix  of  Parameter Estimates
(*** The model parameter(s) -g  have  been estimated at a  boundary point, or have  been
specified by  the user, and do not  appear in the correlation matrix)
                                                                                                  F-26
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1    v intercept slope
 2    v 1 0.00074 -0.00078
 3    intercept 0.00074 I ^1
 4    slope -0.00078 ^\_ I
 5
 6    Parameter Estimates
 7
 8    95.0% Wald Confidence Interval
 9    Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
10    v 0.8 0.0565686 0.689128 0.'910872
11    g £ NA
12    intercept -62.1804 4133.38 -8163.46 8039.1
13    slope 11.2979 748.603 -1455.94 1478.53
14
15   NA 2. Indicates that this parameter has hit a. bound  implied by  some  inequality
16   constraint and thus has no standard error.
17
18
19    Analysis of Deviance Table
20
21    Model Log(likelihood) #_ Param's Deviance Test d.f. P-value
22    Full model -59.3166 4
23    Fitted model -59.3166 3 1.23973e-006 I 0.9991
24    Reduced model -123.82 I 129.007 3 <.OOQ1
25
26    AIC: 124.633
27
28    Goodness of Fit
29    Scaled
30    Dose Est. Prob. Expected Observed Size Residual
31
32    0.0000 0.0000 0.000 0.000 _50_  0.000
33    50.0000 0.0000 0.000 0.000 5_0 -0.001
34    250.0000 0.4400 22.000 22.000 _50_ 0.000
35    1250.0000 0.8000 40.000 40.000 5J) -0.000
36
37    Chi^2 = 0.00 d.f. = I P-value = 0.9994
38
39    Benchmark Dose Computation
40   Specified effect = 0.1
41   Risk Type = Extra risk
42   Confidence level = 0.95
43    BMP = 206.74
44
45    Warning: BMDL computation is  at best imprecise for these data
46    BMDL = 167.46
                                                                                              F-27
                                  DRAFT - DO NOT CITE OR QUOTE

-------
I
                          Dichotomous-Hill Model with 0.95 Confidence Level
        0.8
        0.6
        0.4
        0.2
                             Dichotomous-Hill
                     BMDL  BMD
                          200
                                   400
  10:5301/142011
                                            600
                                            dose
                                                     800
                                                             1000
                                                                      1200
 Figure F-9 BMD Log-logistic model of sclerosis of lamina propria (nasal cavity)
            incidence data for male rats exposed to 1,4-dioxane vapors for 2 years
            to support the results in Table F-18.
                                                                                            F-28
                         DRAFT - DO NOT CITE OR QUOTE

-------
      APPENDIX  G.     RFC  DERIVATION:   ALTERNATIVE
         APPROACH  IN   THE   APPLICATION  OF  THE
         DOSIMETRIC   ADJUSTMENT   FACTOR

 1           For the derivation of a RfC based upon an animal study, the selected POD must be adjusted to
 2    reflect the human equivalent concentration (HEC). and uncertainty factors (UFs) must be applied to
 3    account for recognized uncertainties in the use of the available data. The HEC is calculated by the
 4    application of the appropriate dosimetric adjustment factor (DAF). in accordance with the U.S. EPA RfC
 5    methodology (U.S. EPA. 1994). DAFs are ratios of animal and human physiological parameters, and are
 6    dependent on the nature of the contaminant (particle or gas) and the target site (e.g.. respiratory tract or
 7    systemic) (U.S. EPA. 1994). UFs are used as appropriate and are an order of magnitude (10) or a reduced
 8    order of magnitude (3 or 1). For the derivation of the RfC. the composite UFs are applied to the HEC.

 9           1.4-Dioxane is miscible with water and has a high blood:air partition coefficient. Typically.
10    highly water-soluble and directly reactive chemicals (i.e. Category 1 gases) partition greatly into the
11    upper respiratory tract, induce portal-of-entry effects, and do not accumulate significantly in the blood.
12    1.4-Dioxane induces effects throughout the respiratory tract, liver, and kidneys; and has been measured in
13    the blood after inhalation exposure (Kasai et al.. 2008). The observations of systemic (nonrespiratory)
14    effects and measured blood levels resulting from 1.4-dioxane exposure clearly indicate that this
15    compound is absorbed into the bloodstream and distributed throughout the body. Furthermore, the lack of
16    an anterior to posterior gradient for the nasal effects induced by 1.4-dioxane is not typical of chemicals
17    which are predominantly directly reactive. Thus. 1.4-dioxane might be best described as a water-soluble
18    and non-directly reactive gas. Gases such as these are readily taken up into respiratory tract tissues and
19    can also diffuse into the blood capillaries (Medinsky and Bond. 2001). The effects in the olfactory
20    epithelium may be the result of the metabolism of 1.4-dioxane to an acid metabolite; however, for the
21    reasons stated above it is unclear whether or not these effects are solely the result of portal-of-entry or
22    systemic delivery. A similar pattern of effects were observed after oral exposure to 1.4-dioxane (JBRC.
23    1998: Kanoetal.. 2009) .

24           In  consideration of all the evidence, the human equivalent concentration (HEC) for 1.4-dioxane
25    was calculated in this assessment by application of the appropriate dosimetric adjustment factor (DAF)
26    for systemic acting gases (i.e. Category 3 gases) to the POD for the co-critical effects (olfactory
27    epithelium atrophy and respiratory metaplasia), and adjusted for exposure duration (POp_ADj ^
28    32.2 mg/m3). However, since 1.4-dioxane is miscible with water and may induce portal-of-entry effects.
29    an alternative calculation of the HEC for 1.4-dioxane. based upon the application of a DAF for
30    portal-of-entry acting gases (i.e.. Category 1) was derived and is provided below in Section G.I.

31           Uncertainly factors applied in this assessment included factors of 10 for LOAEL-to-NOAEL
32    extrapolation. 10 for human interindividual variability. 3 for animal-to-human extrapolation, and 3 for
33    database deficiencies (See Section 5.2.4. for details).
                                                                                                       G-l
                                     DRAFT - DO NOT CITE OR QUOTE

-------
     G.1  Application of DAF  for Category 1 Gases

 1          In accordance with the guidance for deriving inhalation RfCs (U.S. EPA. 1994). a DAF based on
 2   the regional gas dose ratio (RGDR) for a gas with portal-of-entry respiratory effects (i.e., extrathoracic:
 3   nasal region to the larynx) was derived by using: 1) a calculated ventilation rate (VE) of 0.254 L/minute
 4   ,based on the average body weight of the male F344 rats reported in the principal study (Kasai et al..
 5   2009); 2) a default VE value of 13.8 L/minute for humans; and 3) default extrathoracic region surface area
 6   (SA) values of 15.0 cm2 for the rat and 200 cm2 for humans. The resulting equation is as follows:

                                                     VE(rat)/SA(rat)
                                        RGDR =      EV   "
                                                 VE(human)/SA (human)
                                                   _ 0.254/15
                                                   = 13.8/200
 7                                                   = 0.25

 8          Applying the RGDR of 0.25 to the POD for the co-critical effects, adjusted for exposure duration:
 9   (PODAnr. 32.2 mg/m3) yields a HEC (TOD^) of 8.1 mg/m3 :

10                       PODggr (mg/m3) = PODAm (mg/m3) x RGDR
11                              = 32.2 mg/m3 x Q.25
12                              =8.1 mg/m3
     G.2  Application of Uncertainty Factors
13          A composite UF of 1.000 was determined for the derivation of the RfC. As stated above, the
14   composite UF of 1.000 includes factors of 10 for LOAEL-to-NOAEL extrapolation. 10 for human
15   interindividual variability. 3 for animal-to-human extrapolation, and 3 for database deficiencies.

16          Applying the composite UF of 1.000 to the HEC (POD^) of 8.1 mg/m3 yields an RfC of 0.008
17   or 8 x 1(T3 mg/m3.

18                                       RfC = PODHEc/UF
19                                          =8.1 mg/m3/I.OOP
20                                          = 0.008 or 8 x 1Q~3 mg/m3
                                                                                               G-2
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     APPENDIX  H.     DETAILS  OF   BMD  ANALYSIS  FOR
         INHALATION   UNIT  RISK  FOR   1,4-DIOXANE

 1          Multistage cancer models available in the Benchmark Dose Software (BMDS) (version 2.2beta)
 2   were fit to the incidence data for hepatocellular carcinoma and/or adenoma, nasal cavity squamous cell
 3   carcinoma, renal cell carcinoma, peritoneal mesothelioma. and mammary gland fibroadenoma. Zymbal
 4   gland adenoma, and subcutis Fibroma in rats exposed to 1.4-dioxane vapors for 2 years (Kasai et al..
 5   2009). Concentrations associated with a benchmark response (BMR) of a 10% extra risk were calculated.
 6   BMCio and BMCLjo values from the best fitting model, determined by adequate global- fit (y1 p > 0.1)
 7   and AIC values, are  reported for each endpoint (U.S. EPA. 2000a). Given the multiplicity of tumor sites.
 8   basing the IUR on one tumor site will underestimate the carcinogenic potential of 1.4-dioxane. A
 9   Bayesian analysis was performed using WinBUGS (Spiegelhalter et al.. 2003). freeware developed by the
10   MRC Biostatistical Unit. Cambridge. United Kingdom (available at
11   http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml) and reported in detail in Section H.3. In
12   addition, the combined tumor analysis was also performed using the beta version of the BMDS
13   MSCombo model (BMDS Version 2.2beta) and is included in Section H.4. The results of both analyses
14   were comparable.

15          A summary  of the BMDS model predictions for the Kasai et al. (2009) study are shown in
16   Table H-l.
     H.1  General Issues and Approaches to  BMDS and  Multitumor
           Modeling
     H.1.1    Combining Data tumor types

17          The incidence of adenomas and the incidence of carcinomas within a dose group at a site or tissue
18   in rodents are sometimes combined. This practice is based upon the hypothesis that adenomas may
19   develop into carcinomas if exposure at the same dose was continued (U.S. EPA. 2005a; McConnell et al..
20   1986). In the same manner and was done for the oral cancer assessment (Appendix D). the incidence of
21   hepatic adenomas and carcinomas was summed without double-counting them so as to calculate the
22   combined incidence of either a hepatic carcinoma or a hepatic adenoma in rodents.

23                 The remaining of the tumor types were assumed to occur independently.
     H.I.2    Summary

24          The BMDS models recommended to calculate rodent BMCio and BMCLjo values for individual
25   tumor types and combined tumor analysis are summarized in Table H-l. The first order multistage models

                                                                                                H-l
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
for most tumor types were selected because they resulted in the lowest AIC values; however, for renal cell
carcinoma and Zymbal aland adenoma, the lowest AIC
model was not the first order model. In
the third order model resulted in the lowest AIC (first (1°)-, second (2°)-, and third (3°)-degree
BMDS.
models
were evaluated); however, using the MCMC approach in WinBUGS, the third order (3°) multistage
model did not converge while the second order(2°) model did converge
and Zymbal gland adenoma, the second order
(WinBugs) approach and the BMDS (Version
. Thus, for renal cell carcinoma
(2°) multistage model was used in both the MCMC
2.2 beta)
MSCombo approach for direct comparison of
results. These results are shown below in Table H-l.
Table H-1 Summary of BMC10 and BMCL10 model
combined tumor analysis for male rats
2009)
Multistage
Endpoint Model
Degree
Nasal squamous cell carcinoma First (1°)
Hepatocellular First (1°)
adenoma/carcinoma 1 — _
Renal cell carcinoma Third (3°)
Peritoneal mesothelioma First (1°)
Mammary gland fibroadenoma First (1°)
Zymbal gland adenoma Third (3°)
Subcutis fibroma3 First (1 °)
WinBUGS multitumor analysis"
BMDS Version 2.2beta MSCombo
AIC
49.03
127.9
29.99
155.4
86.29
29.99
89.2


results for individual turr
exposed to 1,4-dioxane \
D-value x2 Residual
of Interest
0.9607
0.6928
0.9984
0.8509
0.7904
0.9984
0.5245


0.176
-0.763
0.017
-0.204
-0.149
0.017
0.537


ior types and
rapors (Kasai et al.,


BMCIO(ppm) B,MCL1°
(ppm)
1107.04
252.80
1355.16
82.21
1635.46
1355.16
141.762
39.2
40.4
aHigh-dose dropped. See Section H.2.6 for details.
bln MCMC approach, the simulations for the four-parameter third order(3°) multistage model did not converge for renal
and Zymbal gland adenomas. Second order (2°) multistage model was used instead.
629.95
182.26
16.15
64.38
703.03
16.15
81.9117
31.4
30.3
cell carcinomas
     H.2  BMDS  Model Output for Multistage Cancer Models for Inidividual
           Tumor Types

 9         For tumor incidence data reported in the Kasai et al. (2009) 2-year inhalation bioassay. multistage
10   cancer models of first (1°)-. second (2°)-. and third (3°)degrees were implemented BMDS (Version
11   2.2Beta). Incidence data used for BMP analysis are shown in Table H-2^ Tumor incidence for mammary
12   gland adenoma was excluded from this analysis since only 1 tumor of this type was found across all
13   doses.
                                                                                           H-2
                                 DRAFT - DO NOT CITE OR QUOTE

-------
     Table H-2   Incidence of tumors in male F344/DuCrj rats exposed to 1,4-dioxane vapor by
                 whole-body inhalation for 2 years
Effect
Nasal squamous cell carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Hepatocellular adenoma or carcinoma
Renal cell carcinoma
Peritoneal mesothelioma
Mammary gland fibroadenoma
Zymbal gland adenoma
Subcutis fibroma
1,4-dioxane vapor concentration (ppm)
0 (clean air)
0/50
1/50
0/50
1/50
0/50
2/50
1/50
0/50
1/50
50
0/50
2/50
0/50
2/50
0/50
4/50
2/50
0/50
4/50

1/50
3/50
1/50
4/50
0/50
14/50a
3/50
0/50
9/50a
1,250
6/50°'c
21/50a'c
2/50
22/50a'c
4/50c
41/50a'c
5/50a
4/50c

     ap < 0.01 by Fisher's exact test.
     bp < 0.05 by Fisher's exact test.
     cp < 0.01 by Peto's test for dose-related trend.
     dp < 0.05 by Peto's test for dose-related trend.
     eProvided via personal communication from Dr. Tatsuya Kasai (2008) to Dr. Reeder Sams on 12/23/2008. Statistics
      were not reported for these data by study authors, so statistical analyses were conducted by EPA.

     Source: Kasai et al. (2009) and Kasai personal communication (2008)
     H.2.1    Nasal Squamous Cell  Carcinoma

1           The incidence data for nasal squamous cell carcinoma were monotonic non-decreasing functions
2    of dose: therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
3    BMDS modeling for the multistage cancer model for first (1°)-. second (2°)-. and third (3°)-degree
4    polynomials are shown in Table H-3i The first (l°)-degree polynomial was the best fitting model based on
5    AIC. The plot (Figure H-l) and model output for the first (l°)-degree model are shown below.
                                                                                                         H-3
                                     DRAFT - DO NOT CITE OR QUOTE

-------
Table H-3   BMDS Multistage cancer dose-response modeling results for the incidence of nasal
           squamous cell carcinomas in male rats exposed to 1,4-dioxane vapors for 2-years
           (Kasai et al.. 2009)

Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
49.0308
50.8278
50.8278
p-value
0.9607
0.9087
0.9087
\2 Residual of
Interest
0.176
-0.021
-0.021
BMC-io
(ppm)
1,107.04
1,086.94
1,086.94
BMCLio
(ppm)
629.95
642.43
642.43
Best-fitting model based on AIC.
                            Multistage Cancer Model with 0.95 Confidence Level
 I
 C
 o
          0.25
           0.2
          0.15
           0.1
          0.05
                                       Multistage Cancer
                                      Linear extrapolation
                                             BMDL
                                                                               BMD
                           200
                                          400
600
 dose
800
1000
1200
   10:26 11/172010
       Figure H-l  Multistage model (First (l°)-degree) for male rat nasal squamous cell
                  carcinomas.
 1
 2
 3
 4
 5
 6
 7
 8
 Q    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~,
10     The form of the probability function is:
11
MS COMBO.  (Version:  1.4;  Date:  10/20/2010)
Input Data File: C:\Documents  and
Settings\emclanah\Desktop\BMD  14D Cancer\Data\New.(d)
        Gnuplot Plotting File: C:\Documents  and
Settings\emclanah\Desktop\BMD  14D Cancer\Data\New.plt
                                                Wed Nov r? 10:57:55 2010

 BMDS Model Run
                                                                                           H-4
                             DRAFT - DO NOT CITE OR QUOTE

-------
 1    P[response] = background +  (1-background)*[1-EXP(-betal*doseAl)]
 2
 3    The parameter betas are restricted  to be  positive
 4
 5    Dependent variable = EFFECT
 6    Independent variable = DOSE
 7
 8    Total number of observations = £
 9    Total number of records with missing values  =  0_
10    Total number of parameters in model = 2_
11    Total number of specified parameters = 0_
12    Degree of polynomial = \_
13
14    Maximum number of iterations = 250
15    Relative Function Convergence has been set to: le-008
16    Parameter Convergence has been set  to: le-008
17
18
19    Default Initial Parameter Values
20    Background = 0_
21    Beta(l) = 0.000104666
22
23    Asymptotic Correlation Matrix of Parameter Estimates
24    (***The model parameter(s) -Background have been estimated at  a  boundary point,  or
25   have been specified by the user, and do not appear  in the  correlation matrix _)_
26
27    Beta(l)
28    Beta(l) I
29
30    Parameter Estimates
31    95.0% Wald Confidence Interval
32    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf.  Limit
33   Background 0_ ^_ ^_ ^_
34    Beta(l) 9.51733e-005 ^ ^ ^
35
36   i z Indicates that this value is not calculated.
37
38    Analysis of Deviance Table
39
40    Model Log (likelihood) #_ Param' s Deviance  Test  d. f. P-value
41    Full model -23.2482 £
42    Fitted model -23.5154 I 0.534383 3  0.9113
43    Reduced model -30.3429 I 14.1894 3_  0.002658
44
45    AIC: 49.0308
46
47    Log-likelihood Constant 20.493267595834471
48
49
50                 Goodness  of  Fit
51    Scaled
52    Dose Est. Prob. Expected Observed Size Residual
53
54    0.0000 0.0000 0.000 C> _50_ 0.000
55    50.0000 0.0047 0.237 0 50 -0.
56    250.0000 0.0235 1.176  I _50_ -0.164
57    1,250.0000 0.1122 5.608 _6 _50_  0.176
58
59    Chi^2 = 0.30 d.f. = 3^  P-value =  0.9607
60
61
                                                                                               H-5
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    Benchmark  Dose Computation
 2
 3   Specified effect = 0.1
 4   Risk Type = Extra risk
 5   Confidence  level = 0.95
 6    BMP =  1107.04
 7    BMDL = 629.948
 8    BMDU = 2215.11
 9
10   Taken together, (629.948, 2215.11)  ijs  a 90% two-sided confidence  interval for the BMP
     H.2.2    Hepatocellular Adenoma and Carcinoma

11          The incidence data for the occurrence of either hepatocellular adenoma or carcinoma were
12   combined for this analysis as explained in H. 1.1. The incidence data were monotonic non-decreasing
13   functions of dose: therefore, these data are appropriate for dose-response modeling using BMDS. The
14   results of the BMDS modeling for the multistage cancer model for first-, second-, and third-degree
15   polynomials are shown in Table H-4. The lst-degree polynomial was the best fitting model based on AIC.
16   The plot (Figure H-2) and model output for the lst-degree model are shown below.
                                                                                                  H-6
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     Table H-4  BMDS Multistage cancer dose-response modeling results for the incidence of either
                hepatocellular adenoma or carcinoma in male rats exposed to 1,4-dioxane vapors for
                2-years (Kasai et al.. 2009)

Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
127.86
129.157
129.131
p-value
0.6928
0.7636
0.8
X^ Residual of
Interest
-0.763
-0.094
-0.068
BMCio
(ppm)
252.80
377.16
397.426
BMCLio
(ppm)
182.26
190.28
190.609
     "Best-fitting model based on AIC.
      I
      c
      o
               0.6
               0.5
               0.4
0.3
               0.2
               0.1
                                 Multistage Cancer Model with 0.95 Confidence Level
                            Multistage Cancer
                           Linear extrapolation
                          BMDL
                     BMD
                               200
                           400
600
 dose
800
1000
1200
        10:24 11/172010
            Figure H-2 Multistage model (First-degree (1°)) for male rat hepatocellular
                       adenomas and carcinomas.
 1    MS COMBO.  (Version:  1.4;  Date: 10/20/2010)
 2            Input Data File:  C:\Documents and
 3    Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
 4            Gnuplot Plotting  File:  C:\Documents and
 5    Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
 6                                                    Wed Nov r? 10:57:55  2010
 7
 8     BMDS Model Run
 Q    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10     The form  of  the  probability function is:
11            P[response] = background  + (1-background)*[1-EXP(-betal*doseAl)]
                                                                                                 H-7
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    The parameter betas are restricted to be positive
 3
 4    Dependent variable = EFFECT
 5    Independent variable = DOSE
 6
 7    Total number of observations = £
 8    Total number of records with missing values =  0_
 9    Total number of parameters in model = 2_
10    Total number of specified parameters = £
11    Degree of polynomial = I_
12
13    Maximum number of iterations = 250
14    Relative Function Convergence has been set to: le-008
15    Parameter Convergence has been set to: le-008
16
17    Default Initial Parameter Values
18    Background = 0.00480969
19    Beta(l) = 0.0004548
20
21    Asymptotic Correlation Matrix of Parameter Estimates
22
23    Background Beta(1)
24   Background I -0.53
25    Beta(l) -0.53 I
26
27    Parameter Estimates
28
29    95.0% Wald Confidence Interval
30    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf.  Limit
31   Background 0.0176~678 * * *
32    Beta(l) 0.000416776 111
33
34   1 z Indicates that this value is not calculated.
35
36    Analysis of Deviance Table
37
38    Model Log(likelihood) #_ Param's Deviance Test  d.f. P-value
39    Full model -61.5341 4
40    Fitted model -61.9302 2 0.792109 2 0.673
41    Reduced model -82.7874 I 42.5066 3 <.OOQ1
42
43    AIC: 127.86
44
45    Log-likelihood Constant 55.486699676972215
46
47    Goodness of Fit
48    Scaled
49    Dose Est. Prob. Expected Observed Size Residual
50
51    0.0000 0.0171 0.853 I _50_ 0.160
52    50.0000 0.0373 1.867 2 5J) 0.099
53    250.0000 0.1143 5.716 £ _50_ -0.763
54    1,250.0000 0.4162 20.810 22_ _50_ 0.342
55    Chi^2 = 0.73 d.f. = 2_ P-value = 0. 6928
56
57
58    Benchmark Dose Computation
59
60   Specified effect = 0.1
61   Risk Type = Extra risk
62   Confidence level = 0.95
63    BMP = 252.799
64    BMDL = 182.256
65    BMDU = 371.457
66
67   Taken together,  (182.256, 371.457) ijs a 90% two-sided confidence  interval  for  the  BMP

                                                                                               H-
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     H.2.3   Renal Cell Carcinoma and Zymbal Gland Adenoma

1          The incidence data for renal cell carcinomas and Zymbal gland adenomas were the same. These
2    data were monotonic non-decreasing functions of dose: therefore, these data are appropriate for
3    dose-response modeling using BMDS. The results of the BMDS modeling for the multistage cancer
4    model for first (1°)-. second (2°)- and third-degree (3°) polynomials are shown in Table H-5^ The
5    third-degree (3°)polynomial was the best fitting model based on AIC; however, when conducting the
6    multitumor analysis. WinBUGS was unable to converge using the third-degree (3°) model. Thus, the
7    second degree (2°) model was used in the multitumor analyses. The plots (Figure H-3 and Figure H-4)
8    and model outputs for both the second (2°)- and third-degree (3°) models are shown below.
                                                                                                 H-9
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     Table H-5   BMDS Multistage cancer dose-response modeling results for the incidence of renal
                cell carcinomas and Zymbal gland adenomas in male rats exposed to 1,4-dioxane
                vapors for 2-years (Kasai et al.. 2009)

Polynomial Degree
(1°) First
(2°) Second
(3°) Third3
AIC
31.6629
30.2165
29.9439
p-value
0.8004
0.9817
0.9984
X^ Residual of
Interest
0.446
0.085
0.017
BMCio
(ppm)
1,974.78
1,435.28
1,355.16
BMCLio
(ppm)
957.63
999.44
1,016.15
Best-fitting model based on AIC.
                                 Multistage Cancer Model with 0.95 Confidence Level
                0.2
               0.15
      T3
      £
      O
      c
      O
      •*=
      O
      (0
                0.1
               0.05
                                            Multistage Cancer
                                           Linear extrapolation
                                                              BMDL
                                                                                BMD
                               200
                                  400
600      800
    dose
1000
1200
1400
        10:1702/102011
            Figure H-3  Multistage model (Second-degree (2°)) for male rat renal cell
                       carcinomas and Zymbal gland adenomas.
 4
 5
 6
 7
 8
 9
10
11
 1    Multistage Cancer Model.  (Version:  1.9;  Date:  05/26/2010)
 2    Input Data File: C:/Documents  and
 3
Settings/emclanah/Desktop/BMD 14D Cancer/Data/msc  Kasai2009 renal Msc2-BMR10.(d)
        Gnuplot Plotting File:  C:/Documents and
Settings/emclanah/Desktop/BMD 14D Cancer/Data/msc  Kasai2009 renal Msc2-BMR10.pit
                                               Thu Feb K>  10:17:39  2011

 BMDS Model Run

 The form of the probability function  is:
                                                                                              H-10
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1    P [response] = background +  (1-background) * [1-EXP (-betal*dose/xl-beta2*dose/x2) ]
 2
 3    The parameter betas are restricted  to  be  positive
 4
 5    Dependent variable = EFFECT
 6    Independent variable = DOSE
 7
 8    Total number of observations = £
 9    Total number of records with missing values  =  0_
10    Total number of parameters in model =  3^
11    Total number of specified parameters = 0_
12    Degree of polynomial = 2_
13
14    Maximum number of iterations = 250
15    Relative Function Convergence has been set to:  le-008
16    Parameter Convergence has been set  to: le-008
17
18    Default Initial Parameter Values
19    Background = Q_
20    Beta(l) = 0
21    Beta(2) = 5.40386e-008
22
23    Asymptotic Correlation Matrix of Parameter Estimates
24    (*** The model parameter(s) -Background -Beta (1) have been estimated at a boundary
25   point, or have been specified by the user, and  do not appear in the correlation
26   matrix)
27
28    Beta(2)
29    Beta(2) I
30
31    Parameter Estimates
32    95.0% Wald Confidence Interval
33    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf.  Limit
34   Background 0_ ^_ ^_ ^_
35    Beta(l) C> 1 1 1 ~
36    Beta(2) 5.11454e-008 111
37
38   1 z Indicates that this value is not calculated.
39
40    Analysis of Deviance Table
41
42    Model Log(likelihood) #_ Param's Deviance  Test  d.f. P-value
43    Full model -13.9385 4
44    Fitted model -14.1082 I 0.339554 3  0.9524
45    Reduced model -19.6078 I 11.3387 3  0.01003
46
47    AIC: 30.2165
48
49    Goodness of Fit
50    Scaled
51    Dose Est. Prob. Expected Observed Size Residual
52
53    0.0000 0.0000 0.000 0.000 _50_ 0.000
54    50.0000 0.0001 0.006 0.000 5_0 -0.080
55    250.0000 0.0032 0.160 0.000 _50_ -0.400
56    1250.0000 0.0768 3.840 4.000 5J) 0.085
57
58    Chi^2 = 0.17 d.f. = 3^ P-value = 0.9817
59
                                                                                              H-ll
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
 Benchmark Dose Computation
Specified effect =  0.1
Risk Type = Extra risk
Confidence level =  0.95
 BMP = 17135.28
 BMDL = 999.44

 BMDU = 3,666.87

Taken together,  (999.44  _,_ 3, 666.87)  i_s a. 90% two-sided confidence  interval  for the BMP

Multistage Cancer Slope  Factor = 0.000100056
      T3
       
-------
 1
 2    Dependent variable = EFFECT
 3    Independent variable = DOSE
 4
 5    Total number of observations = £
 6    Total number of records with missing values  =  0_
 7    Total number of parameters in model = _4
 8    Total number of specified parameters = 0_
 9    Degree of polynomial = 3_
10
11    Maximum number of iterations = 250
12    Relative Function Convergence has been set to:  le-008
13    Parameter Convergence has been set to: le-008
14
15    Default Initial Parameter Values
16    Background = 0_
17    Beta(l) = C>
18    Beta(2) = £
19    Beta(3) = 4.2804e-011
20
21
22    Asymptotic Correlation Matrix of Parameter Estimates
23    (*** The model parameter(s) -Background -Beta(1)  -Beta(2)  have been estimated at a.
24   boundary point, or have been specified by  the user,  and do not appear in the
25   correlation matrix)
26
27    Beta(3)
28    Beta(3) I
29
30    Parameter Estimates
31
32    95.0% Wald Confidence Interval
33    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf.  Limit
34   Background 0_ ^_ ^_ ^_
35    Beta(l) C> 1 1 1 ~
36    Beta(2) £ 1 1 1
37    Beta(3) 4.23353e-011 ^ ^ ^
38
39   ^_ 2_ Indicates that this value is not calculated.
40
41    Analysis of Deviance Table
42
43    Model Log(likelihood) #_ Param's Deviance  Test  d.f.  P-value
44    Full model -13.9385 4
45    Fitted model -13.9719 I 0.0669578 3 0.9955
46    Reduced model -19.6078 I 11.3387 3 0.01003
47
48    AIC: 29.9439
49
50    Log-likelihood Constant 12.347138085809094
51
52
53    Goodness of Fit
54    Scaled
55    Dose Est. Prob. Expected Observed Size Residual
56
57    0.0000 0.0000 0.000 0 _50_ 0.000
58    50.0000 0.0000 0.000 0_ 5_0 -0.016
59    250.0000 0.0007 0.033 £ _50_ -0.182
60    1250.0000 0.0794 3.968 4_ 5J) 0.017
61
62    Chi^2 = 0.03 d.f. = 3^ P-value = 0.9984
63
64
65    Benchmark Dose Computation
66   Specified effect = 0.1
67   Risk Type = Extra risk

                                                                                              H-13
                                  DRAFT - DO NOT CITE OR QUOTE

-------
 1   Confidence level = 0.95
 2    BMP  =  1,355.16
 3    BMDL = 1,016.15
 4    BMDU = 3,393.6
 5
 6   Taken together, (1016.15,  3393. 6 )_ is_ a 90% two-sided confidence interval  for the BMP
     H.2.4   Peritoneal Mesothelioma

 7          The incidence data for peritoneal mesotheliomas were monotonic non-decreasing functions of
 8   dose; therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
 9   BMDS modeling for the multistage cancer model for 1st. 2nd. and 3rd-degree polynomials are shown in
10   Table H-6^ The lst-degree polynomial was the best fitting model based on AIC. The plot (Figure H-5] and
11   model output for the lst-degree model are shown below.
                                                                                                  H-14
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     Table H-6  BMDS Multistage cancer dose-response modeling results for the incidence of
                peritoneal mesothelioma in male rats exposed to 1,4-dioxane vapors for 2-years
                (Kasai et al.. 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
155.433
157.168
157.168
p-value
0.8509
0.8053
0.8053
X^ Residual of
Interest
-0.204
-0.204
0
BMCio
(ppm)
82.21
96.23
96.23
BMCLio
(ppm)
64.38
65.15
65.15
     a Best-fitting model based on AIC.
                                 Multistage Cancer Model with 0.95 Confidence Level
      I
      C
      o
               0.8
               0.6
               0.4
               0.2
                                           Multistage Cancer
                                         Linear extrapolation
                    BMDL BMD
                               200
400
600
 dose
800
1000
1200
        10:31 11/172010
            Figure H-5 Multistage model (First-degree (1°)) for male rat peritoneal
                       mesotheliomas.
 1    MS COMBO.  (Version:  1.4;  Date: 10/20/2010)
 2            Input Data File:  C:\Documents and
 3    Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
 4            Gnuplot Plotting  File:  C:\Documents and
 5    Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
 6                                                    Wed Nov r?  10:57:55  2010
 7
 8     BMDS Model  Run
 Q    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10     The form  of the  probability function is:
11            P[response]  = background  + (1-background)*[1-EXP(-betal*doseAl)]
                                                                                                H-15
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    The parameter betas are restricted to be positive
 3
 4    Dependent variable = EFFECT
 5    Independent variable = DOSE
 6
 7    Total number of observations = £
 8    Total number of records with missing values  =  0_
 9    Total number of parameters in model = 2_
10    Total number of specified parameters = £
11    Degree of polynomial = I_
12    Maximum number of iterations = 250
13    Relative Function Convergence has been set to: le-008
14    Parameter Convergence has been set to: le-008
15
16    Default Initial Parameter Values
17    Background = 0.0172414
18    Beta(l) = 0.00135351
19
20    Asymptotic Correlation Matrix of Parameter Estimates
21
22    Background Beta(1)
23   Background I -0.45
24    Beta(l) -0.45 I
25
26    Parameter Estimates
27    95.0% Wald Confidence Interval
28    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf.  Limit
29   Background 0.033631 ^ ^ ^
30    Beta(l) 0.00128167 ^ ^ ^
31
32   i z Indicates that this value is not calculated.
33
34    Analysis of Deviance Table
35
36    Model Log(likelihood) #_ Param's Deviance Test  d.f. P-value
37    Full model -75.553 £
38    Fitted model -75.7165 2_ 0.326905 2_ 0.8492
39    Reduced model -123.008 I 94.9105 3 <.OOQ1
40
41    AIC: 155.433
42
43    Log-likelihood Constant 68.666413125908832
44
45    Goodness of Fit
46    Scaled
47    Dose Est. Prob. Expected Observed Size Residual
48
49    0.0000 0.0336 1.682 2 _50_ 0.250
50    50.0000 0.0936 4.681 £ 5_0 -0.331
51    250.0000 0.2986 14.928 I4_ _50_ -0.287
52    1,250.0000 0.8053 40.265 £1 _50_ 0.263
53
54    Chi^2 = 0.32 d.f. = 2_ P-value = 0.8509
55
56    Benchmark Dose Computation
57   Specified effect = 0.1
58   Risk Type = Extra risk
59   Confidence level = 0.95
60    BMP = 82.2057
61    BMDL = 64.3808
62    BMDU = 107.497
63
64   Taken together,  (64.3808, 107.497) ijs a 90% two-sided confidence  interval  for  the  BMP
                                                                                              H-16
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     H.2.5   Mammary Gland Fibroadenoma

1           The incidence data for mammary gland fibroadenomas were monotonic non-decreasing functions
2    of dose: therefore, these data are appropriate for dose-response modeling using BMDS. The results of the
3    BMDS modeling for the multistage cancer model for first (1°)-. second (2°). and third (3°)-degree
4    polynomials are shown in Table H-7. Since quadratic and cubic terms of the multistage models evaluated
5    resulted in the estimates on the boundary, i.e. equal to 0. the first (l°)-degree polynomial was selected
6    based on model parsimony. The plot (Figure H-6) and model output for the first d°)-degree model are
7    shown below.
                                                                                                  H-17
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     Table H-7  BMDS Multistage cancer dose-response modeling results for the incidence of
                mammary gland fibroadenoma in male rats exposed to 1,4-dioxane vapors for 2-years
                (Kasai et al.. 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
86.29
86.29
86.29
p-value
0.7904
0.7904
0.7904
X^ Residual of
Interest
-0.149
-0.149
-0.149
BMCio
(ppm)
1,635.46
1,635.46
1,635.46
BMCLio
(ppm)
703.03
703.03
703.03
     aAII model fits were equivalent based on AIC. Selected 1 -degree model based on parsimony.
                                  Multistage Cancer Model with 0.95 Confidence Level
      I
      C
      o
                 0.2
               0.15
                 0.1
               0.05
                                             Multistage Cancer
                                            Linear extrapolation
                                              BMDL
                                                 BMID
                              200
400
600
800
dose
1000    1200    1400    1600
        10:34 11/172010
            Figure H-6 Multistage model (First-degree (1°)) for male rat mammary gland
                       fibroadenoma.
 1    MS COMBO.  (Version:  1.4; Date: 10/20/2010)
 2            Input Data File:  C:\Documents and
 3    Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.(d)
 4            Gnuplot Plotting File:  C:\Documents and
 5    Settings\emclanah\Desktop\BMD 14D Cancer\Data\New.plt
 6                                                    Wed Nov r?  10:57:55 2010
 7
 8     BMDS Model  Run
 Q    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10     The form  of the  probability function is:
11            P[response]  =  background + (1-background)*[1-EXP(-betal*doseAl)]
                                                                                                H-18
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    The parameter betas are restricted to be positive
 3
 4    Dependent variable = EFFECT
 5    Independent variable = DOSE
 6
 7    Total number of observations = £
 8    Total number of records with missing values =  0_
 9    Total number of parameters in model = 2_
10    Total number of specified parameters = £
11    Degree of polynomial = I_
12
13    Maximum number of iterations = 250
14    Relative Function Convergence has been set to: le-008
15    Parameter Convergence has been set to: le-008
16
17    Default Initial Parameter Values
18    Background = 0.0335609
19    Beta(l) = 5.91694e-005
20
21    Asymptotic Correlation Matrix of Parameter Estimates
22
23    Background Beta(1)
24   Background I -0. 61
25    Beta(l) -0.61 I
26
27    Parameter Estimates
28
29    95.0% Wald Confidence Interval
30    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf.  Limit
31   Background Q.03158'36 * * *
32    Beta(l) 6.44224e-005 111
33
34   1 z Indicates that this value is not calculated.
35
36     Analysis of Deviance Table
37
38    Model Log(likelihood) #_ Param's Deviance Test  d.f. P-value
39    Full model -40.9017 4
40    Fitted model -41.145 2 0.486662 2 0.784
41    Reduced model -42.5964 I 3.3895 3 0.3354
42
43    AIC: 86.29
44
45    Log-likelihood Constant 35.472345543489602
46
47    Goodness of Fit
48    Scaled
49    Dose Est. Prob. Expected Observed Size Residual
50
51    0.0000 0.0316 1.579 I _50_ -0.468
52    50.0000 0.0347 1.735 2 5J) 0.205
53    250.0000 0.0471 2.353 3 _50_ 0.432
54    1,250.0000 0.1065 5.326 _5 _50_ -0.149
55
56    Chi"2 = 0.47 d.f. = 2 P-value = 0.7904
57
58    Benchmark Dose Computation
59   Specified effect = 0.1
60   Risk Type = Extra risk
61   Confidence level = 0.95
62    BMP = 1,635.46
63    BMDL = 703.034
64    BMDU = 1.9523e+009
65
66   Taken together,  (703.034, 1.9523e+009) is^ a 90% two-sided  confidence  interval  for  the
67   BMP

                                                                                              H-19
                                  DRAFT - DO NOT CITE OR QUOTE

-------
     H.2.6   Subcutis Fibroma

1           The incidence data for subcutis fibroma were monotonic non-decreasing functions of dose for the
2    control (0 ppm). low (50 ppm). and mid-dose (250 ppm); however, the incidence rate at the high dose
3    (1.250 ppm) was lower than observed at the mid-dose. No BMDS model had reasonable fit to the data
4    without dropping the high dose. The results of the BMDS modeling for the multistage cancer model for
5    first (1°)-. second (2°). and third (3°)-degree polynomials with the high dose dropped are shown in
6    Table H-8. Since quadratic and cubic terms of multistage models evaluated resulted in the estimates on
7    the boundary, i.e. equal to 0. , the first (l°)-degree polynomial was selected based on model parsimony.
8    The plot (Figure H-7) and model output for the first (l°)-degree model are shown below.
                                                                                                  H-20
                                   DRAFT - DO NOT CITE OR QUOTE

-------
     Table H-8   BMDS Multistage cancer dose-response modeling results for the incidence of
                 subcutis fibromas in male rats exposed to 1,4-dioxane vapors for 2-years (Kasai et al.
                 2009)
Polynomial Degree
(1°) First3
(2°) Second
(3°) Third
AIC
89.2094
89.2094
89.2094
p-value
0.5245
0.5245
0.5245
X^ Residual of
Interest
0.537
0.537
0.537
BMCio
(ppm)
141.76
141.76
141.76
BMCLio
(ppm)
81.92
81.92
81.92
     aAII model fits were equivalent based on AIC. Selected 1 -degree model based on parsimony.
      I
       c
       o
                0.35
                 0.3
                0.25
                 0.2
0.15
                 0.1
                0.05
        10:56 11/172010
                                  Multistage Cancer Model with 0.95 Confidence Level
                                             Multistage Cancer
                                            Linear extrapolation
                                        BMDL
                                             BMD
                                    50
                                 100         150
                                      dose
200
250
            Figure H-7 Multistage model (First-degree (1°)) for male rat subcutis fibroma (high
                       dose dropped).
 1   MS COMBO.  (Version: 1.4; Date:  10/20/2010)
 2           Input Data  File:  C:\Documents  and
 3   Settings\emclanah\Desktop\BMD  14D  Cancer\Data\New.(d)
 4           Gnuplot  Plotting File:  C:\Documents  and
 5   Settings\emclanah\Desktop\BMD  14D  Cancer\Data\New.plt
 6                                                    Wed Nov r? 10:57:55  2010
 7
 8    BMDS Model  Run
 Q   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
10    The form of the probability function is:
11           P[response]  = background +  (1-background)*[1-EXP(-betal*doseAl)]
                                                                                                 H-21
                                   DRAFT - DO NOT CITE OR QUOTE

-------
 1
 2    The parameter betas are restricted to be positive
 3
 4    Dependent variable = EFFECT
 5    Independent variable = DOSE
 6
 7    Total number of observations = 3_
 8    Total number of records with missing values =  0_
 9    Total number of parameters in model = 2_
10    Total number of specified parameters = £
11    Degree of polynomial = I_
12
13    Maximum number of iterations = 250
14    Relative Function Convergence has been set to: le-008
15    Parameter Convergence has been set to: le-008
16
17    Default Initial Parameter Values
18    Background = 0.0327631
19    Beta(l) = 0.000673665
20
21
22    Asymptotic Correlation Matrix of Parameter Estimates
23
24    Background Beta(1)
25   Background I -0 . 68
26    Beta(l) -0.68 I
27
28    Parameter Estimates
29
30    95.0% Wald Confidence Interval
31    Variable Estimate Std. Err. Lower Conf. Limit  Upper Conf. Limit
32   Background 0.0262054 111
33    Beta(l) 0.00074322 111
34
35   1 2. Indicates that this value is not calculated.
36
37    Analysis of Deviance Table
38
39    Model Log (likelihood) #_ Param' s Deviance Test  d. f. P-value
40    Full model -42.4101 3
41    Fitted model -42.6047 2 0.389155 I 0.5327
42    Reduced model -46.5274 I 8.23466 2_ 0.01629
43
44    AIC: 89.2094
45
46    Log-likelihood Constant 37.900888781466982
47
48    Goodness of Fit
49    Scaled
50    Dose Est. Prob. Expected Observed Size Residual
51
52    0.0000 0.0262 1.310 I _50_ -0.275
53    50.0000 0.0617 3.086 £ 5£ 0.537
54    250.0000 0.1913 9.566 9 _50_ -0.204
55    Chi^2 = 0.41 d.f. = I P-value = 0.5245
56
57
58    Benchmark Dose Computation
59   Specified effect = 0.1
60   Risk Type = Extra risk
61   Confidence level = 0.95
62    BMP = 141.762
63    BMDL = 81.9117
64    BMDU = 364.364
65
66   Taken together,  (81.9117, 364.364) i_£ a 90% two-sided confidence  interval  for  the  BMP


                                                                                              H-22
                                  DRAFT - DO NOT CITE OR QUOTE

-------
      H.2.7    Multitumor analysis  using Bayesian Methods

 1           Given the multiplicity of tumor sites, basing the IUR on one tumor site will likely underestimate
 2    the carcinogenic potential of 1.4-dioxane. Simply pooling the counts of animals with one or more tumors
 3    (i.e.. counts of tumor bearing animals) would tend to underestimate the overall risk when tumors are
 4    independent across sites and ignores potential differences in the dose-response relationships across the
 5    sites (NRC. 1994; Bogen. 1990). NRC (1994) also noted that the assumption of independence across
 6    tumor types is not likely to produce substantial error in the risk estimates unless tumors are known  to be
 7    biologically dependent.

 8           Kopylev et al. (2009) describe a Markov Chain Monte Caro (MCMC) computational approach to
 9    calculating the dose associated with a specified composite risk under assumption of independence of
10    tumors. The current Guidelines for Carcinogen Risk Assessment recommend calculation of an upper
11    bound to account for uncertainty in the estimate (U.S. EPA. 2005a). For uncertainty characterization.
12    MCMC methods have the advantage of providing information about the full distribution of risk and/or
13    benchmark dose, which can be used in generating a confidence bound. This MCMC approach building on
14    the re-sampling approach recommended by Bogen (1990). and also provides a distribution of the
15    combined potency across sites.

16           For individual tumor data modeled using the multistage model:

17                                   P(d \q) = 1 - exp[-(qn + qLd + q.jf + ... +

18    the model for the combined tumor risk is still multistage, with a functional form that has the sum of
19    stage-specific multistage coefficients as the corresponding multistage coefficient;

20
21           The resulting equation for fixed extra risk (BMR) is polynomial in dose (when logarithms of both
22    sides are taken) and can be straightforwardly solved for a combined BMC. Computation of the confidence
23    bound on combined risk BMC can be accomplished via likelihood methods (BMDS-MSCOMBO).
24    re-sampling (bootstrap) or Bayesian methods.

25           The MCMC computations were conducted using WinBUGS (Spiegelhalter et al., 2003)(freeware
26    developed by the MRC Biostatistical Unit. Cambridge. United Kingdom, available at
27    http://www.mrc-bsu.cam.ac.uk/bugs/winbugs/contents.shtml).

28           In a Bavesian analysis, the choice of the appropriate prior is important. In the examples
29    developed by Kopylev et al. (2009). a diffuse (i.e.. high variance or low tolerance) Gaussian prior
30    restricted to be nonnegative was used; such diffuse priors performed reasonably well.

31           The mean and the 5th percentile of the posterior distribution of combined BMC provide estimates
32    of the mean BMC and the lower bound  on the BMC (BMCL). respectively, for the combined tumor risk.

33    The values calculated using this method were: mean BMCm 39.2ppm. and BMCLin 31.4.
                                                                                                    H-23
                                    DRAFT - DO NOT CITE OR QUOTE

-------
     H.3  Multitumor Analysis Using BMDS MSCOMBO (BETA)

 1          The combined tumor analysis was also performed with beta version of the MSCombo model in
 2    BMDS (Version 2.2beta). The model resulted in similar results to the Bayesian method and model output
 3    is shown below for the combined calculation.

 4
 5    **** start  of combined BMD and BMDL Calculations.****
 6    Combined Log-Likelihood -277.79874987953076
 7    Combined Log-likelihood Constant 246.62591390071873
 8
 9
10    Benchmark Dose Computation
11   Specified effect = 0.1
12   Risk Type = Extra risk
13   Confidence level = 0.95
14    BMD = 40.4937
15    BMDL = 32.331
                                                                                           H-24
                                 DRAFT - DO NOT CITE OR QUOTE

-------