DRAFT - DO NOT CITE OR QUOTE
EPA/63 5/R-08/019A
www.epa.gov/iris
&EPA
TOXICOLOGICAL REVIEW
OF
ETHYL TERTIARY BUTYL ETHER
(CAS No. 637-92-3)
In Support of Summary Information on the
Integrated Risk Information System (IRIS)
July 2009
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
-------
DISCLAIMER
This document is a preliminary draft for review purposes only. This information is
distributed solely for the purpose of pre-dissemination peer review under applicable information
quality guidelines. It has not been formally disseminated by EPA. It does not represent and
should not be construed to represent any Agency determination or policy. Mention of trade
names or commercial products does not constitute endorsement or recommendation for use.
07/14/2009
11
DRAFT - DO NOT CITE OR QUOTE
-------
CONTENTS - TOXICOLOGICAL REVIEW OF ETHYL TERTIARY BUTYL ETHER
(CAS NO. 637-92-3)
LIST OF TABLES vi
LIST OF FIGURES xii
LIST 01 ABBREVIATIONS AM) ACRONYMS xv
FOREWORD xvii
AUTHORS, CONTRIBUTORS, AND REVIEWERS xviii
1. INTRODUCTION 1
2. CHEMICAL AM) PHYSICAL INFORMATION 3
3. TOXICOKINETICS 6
3.1. ABSORPTION 6
3.2. DISTRIBUTION 9
3.3. METABOLISM 10
3.3.1. Metabolism in Humans 11
3.3.1.1. Metabolism of ETBE in Humans In Vivo 11
3.3.1.2. In Vitro Metabolism of ETBE Using Human Enzyme Preparations 12
3.3.2. Metabolism in Animals 14
3.3.2.1. Metabolism of ETBE in Animals In Vivo 14
3.3.2.2. Metabolism of ETBE in Animal Tissues In Vitro 15
3.4. ELIMINATION 16
3.4.1. Elimination in Humans 16
3.4.2. Elimination in Animals 17
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS 22
4. HAZARD IDENTIFICATION 25
4.1. STUDIES IN HUMANS - EPIDEMIOLY, CASE REPORTS, CLINICAL
CONTROLS 25
4.1.1. Studies in Humans 25
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AM) INHALATION 27
4.2.1. Subchronic Studies—Oral 27
4.2.2. Chronic Studies—Oral 27
4.2.3. Sub chroni c Studi e s—Inhal ati on 31
4.2.3.1. Subchronic Inhalation Studies—Rats 31
4.2.3.2. Subchronic Inhalation Studies—Mice 39
4.2.4. Chronic Studies—Inhalation 39
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND
INHALATION 40
4.4. OTHER DURATION- OR ENDPOINT-SPECIFIC STUDIES 49
4.4.1. Acute Studies 49
4.4.1.1. Oral 49
07/14/2009 iii DRAFT - DO NOT CITE OR QUOTE
-------
4.4.1.2. Inhalation 50
4.4.2. Direct Administration Studies 50
4.4.2.1. Dermal Administration 50
4.4.2.2. Ocular Administration 52
4.4.3. Neurological Studies 52
4.5. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF
I I Ii: MODE 01 ACTION 55
4.5.1. Genotoxicity 55
4.5.1.1. In Vitro Bacterial Assays 55
4.5.1.2. In Vitro Mammalian Assays 55
4.5.1.3. In Vivo Assays 56
4.5.2. Studies with ETBE-Gasoline Mixtures 56
4.5.3. Structure-Activity Relationship Evaluations 59
4.6. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS 59
4.6.1. Oral 60
4.6.2. Inhalation 63
4.6.3. Mode of Action Information 68
4.7. EVALUATION OI CARCINOGENICITY 73
4.7.1. Summary of Overall Weight-of-Evidence 73
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence 75
4.7.3. Mode of Action Information 78
4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES 79
4.8.1. Possible Childhood Susceptibility 79
4.8.2. Possible Gender Differences 80
4.8.3. Self-reported Sensitive Individuals 81
4.8.4. Other—Aging; Gene Polymorphisms 82
5. DOSE RESPONSE ASSESSMENTS 85
5.1. ORAL REFERENCE DOSE (RID) 85
5.1.1. Choice of Principal Study and Critical Effect - with Rationale and
Justification 85
5.1.2. Methods of Analysis 91
5.1.3. RfD Derivation—Including Application of Uncertainty Factors (UFs) 93
5.1.4. Previous RfD 93
5.2. INHALATION REFERENCE CONCENTRATION (RfC) 93
5.2.1. Choice of Principal Study and Critical Effect - with Rationale and
Justification 93
5.2.2. Methods of Analysis 97
5.2.2.1. Adjustment to a Human Equivalent Exposure Concentration 101
5.2.3. RfC Derivation—Including Application of Uncertainty Factors (UFs) 102
5.2.4. Previous RfC 103
5.2.5. Reference Value Comparison Information 103
07/14/2009
iv
DRAFT - DO NOT CITE OR QUOTE
-------
5.3. UNCERTAINTIES IN THE INHALATION REFERENCE
CONCENTRATION (RfC) 105
5.4. CANCER ASSESSMENT 107
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF HAZARD
AND DOSE RESPONSE 109
6.1. HUMAN HAZARD POTENTIAL 109
6.2. DOSE RESPONSE 110
6.2.1. Noncancer/ Oral 110
6.2.2. Noncancer / Inhalation Ill
6.2.3. Cancer / Oral Ill
7. REFERENCES 113
APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
COMMENTS AND DISPOSITION A-l
APPENDIX B. BMD CALCULATIONS FOR THE ORAL MINIMAL DATA
VALUE AND RfC B-l
B. 1. NONCANCER DOSE-RESPONSE ASSESSMENT FOR ORAL
EXPOSURE TO ETBE: BMD MODELING RESULTS B-l
B.2. NONCANCER DOSE-RESPONSE ASSESSMENT FOR INHALATION
EXPOSURE TO ETBE: BMD MODELING RESULTS B-55
APPENDIX C. DERIVATION OF THE ORAL MINIMAL DATA VALUE C-l
07/14/2009
v
DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF TABLES
2-1. Physicochemical properties of ETBE 4
3-1. Doses received by humans and F344 rats following inhalation exposure to and
oral ingestion of fuel oxygenates 8
3-2. Blood:tissue partition coefficients for gasoline ether additives and TBA 10
3-3. Elimination of [14C]-ETBE-derived radioactivity from rats and mice within 96 hours
following a single 6-hour inhalation exposure 19
3-4. Radioactivity in blood and kidney of rats and blood and liver of mice, following 6
hours of [14C]-ETBE inhalation exposure 19
4-1. Tumor incidences resulting from 2-year gavage exposure of Sprague-Dawley rats
to ETBE 30
4-2. Summary of significant results from the 13-week subchronic ETBE inhalation
study in F344 rats 34
4-3. Incidence of lesions in kidney, seminiferous tubules, and bone marrow of F344
rats from 13-week subchronic ETBE inhalation study 37
4-4. Mean number of regenerative foci and hyaline droplet severity in kidneys from
male F344 rats exposed to ETBE in a 13-week subchronic inhalation study 38
4-5. Cell division or LI in proximal tubule cells from male and female F344 rats
exposed to ETBE in a 13-week subchronic inhalation study 38
4-6. Summary of significant results from the 13-week subchronic ETBE inhalation
study in CD-I mice 41
4-7. Effects of oral ETBE treatment on parental Sprague-Dawley rats in a two-
generation reproduction and fertility study 46
4-8. Oral toxicity studies for ETBE 63
4-9. Subchronic inhalation toxicity studies for ETBE 67
4-10. Frequencies of gene polymorphisms of human CYP2A6 84
5-1. Summary of BMD modeling results of ETBE oral toxicity studies for selection
of principal study in Sprague-Dawley rats 89
5-2. Summary of BMD modeling results of ETBE inhalation toxicity studies for
selection of principal study 98
B-l. Mean dam body weight change (and SD) in Sprague-Dawley female rats orally
exposed to ETBE on GDs 5 through 20 B-2
07/14/2009 vi DRAFT - DO NOT CITE OR QUOTE
-------
B-2. A summary of BMDS (version 1.4.1) modeling results based on mean dam body
weight change in Sprague-Dawley female rats orally exposed to ETBE on GDs 5
through 20 B-3
B-3. Mean net dam body weight change (and SD) in Sprague-Dawley female rats
orally exposed to ETBE on GDs days 5 through 20 B-8
B-4. A summary of BMDS (version 1.4.1) modeling results based on mean net dam
body weight change in Sprague-Dawley female rats orally exposed to ETBE on
GDs 5 through 20 B-9
B-5. Mean F0 body weight change (and SD) in Sprague-Dawley male rats orally
exposed to ETBE in a two-generation reproductive toxicity study B-14
B-6. A summary of BMDS (version 1.4.1) modeling results based on mean F0 body
weight change in Sprague-Dawley male rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-15
B-7. Mean absolute F1 liver weight (and SD) in Sprague-Dawley male rats orally
exposed to ETBE in a two-generation reproductive toxicity study B-20
B-8. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
F1 liver weight in Sprague-Dawley male rats orally exposed to ETBE in a two-
generation reproductive toxicity study B-21
B-9. Mean relative F1 liver weight (and SD) in Sprague-Dawley male rats orally
exposed to ETBE in a two-generation reproductive toxicity study B-26
B-10. A summary of BMDS (version 1.4.1) modeling results based on mean relative
F1 liver weight in Sprague-Dawley male rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-27
B-l 1. Mean absolute F0 kidney weight (and SD) in Sprague-Dawley male rats
orally exposed to ETBE in a two-generation reproductive toxicity study B-28
B-12. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
F0 kidney weight in Sprague-Dawley male rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-29
B-13. Mean relative F0 kidney weight (and SD) in Sprague-Dawley male rats
orally exposed to ETBE in a two-generation reproductive toxicity study B-34
B-14. A summary of BMDS (version 1.4.1) modeling results based on mean relative
F0 kidney weight in Sprague-Dawley male rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-3 5
B-15. Mean absolute F1 kidney weight (and SD) in Sprague-Dawley male rats orally
exposed to ETBE in a two-generation reproductive toxicity study B-40
07/14/2009
vii
DRAFT - DO NOT CITE OR QUOTE
-------
B-16. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
F1 kidney weight in Sprague-Dawley male rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-41
B-17. Mean absolute F1 kidney weight (and SD) in Sprague-Dawley female rats
orally exposed to ETBE in a two-generation reproductive toxicity study B-42
B-18. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
F1 kidney weight in Sprague-Dawley female rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-43
B-19. Mean relative F1 kidney weight (and SD) in Sprague-Dawley male rats orally
exposed to ETBE in a two-generation reproductive toxicity study B-48
B-20. A summary of BMDS (version 1.4.1) modeling results based on mean relative
F1 kidney weight in Sprague-Dawley male rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-49
B-21. Mean relative F1 kidney weight (and SD) in Sprague-Dawley female rats
orally exposed to ETBE in a two-generation reproductive toxicity study B-50
B-22. A summary of BMDS (version 1.4.1) modeling results based on mean relative
F1 kidney weight in Sprague-Dawley female rats orally exposed to ETBE in a
two-generation reproductive toxicity study B-51
B-23. Mean LI (and SD) from the livers of CD-I female mice exposed to four different
concentrations of ETBE via inhalation for 13 weeks B-56
B-24. A summary of BMDS (version 1.4.1) modeling results based on mean LI data
from the livers of CD-I female mice exposed to ETBE via inhalation for 13 weeks B-57
B-25. A summary of BMDS (version 1.4.1) modeling results based on incidence of
centrilobular hypertrophy in the livers of CD-I male mice exposed to ETBE via
inhalation for 13 weeks B-63
B-26. Incidence of centrilobular hypertrophy in the livers of CD-I female mice
exposed to four different concentrations of ETBE via inhalation for 13 weeks B-67
B-27. A summary of BMDS (version 1.4.1) modeling results based on incidence of
centrilobular hypertrophy in the livers of CD-I female mice exposed to ETBE
via inhalation for 13 weeks B-68
B-28. Mean absolute liver weight (and SD) in CD-I male mice exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-72
B-29. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
liver weight in CD-I male mice exposed to ETBE via inhalation for 13 weeks B-73
B-30. Mean absolute liver weight (and SD) in CD-I female mice exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-78
07/14/2009
viii
DRAFT - DO NOT CITE OR QUOTE
-------
B-31. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
liver weight in CD-I female mice exposed to ETBE via inhalation for 13 weeks B-79
B-32. Mean absolute liver weight (and SD) in F344 male rats exposed to four different
concentrations of ETBE via inhalation for 13 weeks B-84
B-33. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
liver weight in F344 male rats exposed to ETBE via inhalation for 13 weeks B-85
B-34. Mean absolute liver weight (and SD) in F344 female rats exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-90
B-35. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
liver weight in F344 female rats exposed to ETBE via inhalation for 13 weeks B-91
B-36. Mean absolute liver weight (and SD) in Sprague-Dawley male rats exposed to
four different concentrations of ETBE via inhalation for 13 weeks B-96
B-37. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
liver weight in Sprague-Dawley male rats exposed to ETBE via inhalation for
13 weeks B-97
B-38. Mean absolute liver weight (and SD) in Sprague-Dawley female rats exposed
to four different concentrations of ETBE via inhalation for 13 weeks B-102
B-39. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
liver weight in Sprague-Dawley female rats exposed to ETBE via inhalation for
13 weeks B-103
B-40. Mean relative liver weight (and SD) in Sprague-Dawley male rats exposed to
four different concentrations of ETBE via inhalation for 4 weeks B-108
B-41. A summary of BMDS (version 1.4.1) modeling results based on mean relative
liver weight in Sprague-Dawley male rats exposed to ETBE via inhalation for
4 weeks B-109
B-42. Mean relative liver weight (and SD) in Sprague-Dawley female rats exposed
to four different concentrations of ETBE via inhalation for 4 weeks B-l 14
B-43. A summary of BMDS (version 1.4.1) modeling results based on mean relative
liver weight in Sprague-Dawley female rats exposed to ETBE via inhalation for
4 weeks B-l 15
B-44. Mean absolute kidney weight (and SD) in F344 male rats exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-120
B-45. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
kidney weight in F344 male rats exposed to ETBE via inhalation for 13 weeks B-121
B-46. Mean absolute kidney weight (and SD) in F344 female rats exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-126
07/14/2009
IX
DRAFT - DO NOT CITE OR QUOTE
-------
B-47. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
kidney weight in F344 female rats exposed to ETBE via inhalation for 13 weeks B-127
B-48. Mean absolute kidney weight (and SD) in Sprague-Dawley male rats exposed
to four different concentrations of ETBE via inhalation for 4 weeks B-132
B-49. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
kidney weight in Sprague-Dawley male rats exposed to ETBE via inhalation for
4 weeks B-133
B-50. Mean LI (and SD) in the kidney of F344 male rats exposed to four different
concentrations of ETBE via inhalation for 13 weeks B-138
B-51. A summary of BMDS (version 1.4.1) modeling results based on mean LI in the
kidney of F344 male rats exposed to ETBE via inhalation for 13 weeks B-139
B-52. Mean regenerative foci (and SD) in the kidney of F344 male rats exposed to
four different concentrations of ETBE via inhalation for 13 weeks B-144
B-53. A summary of BMDS (version 1.4.1) modeling results based on mean regenerative
foci in the kidney of F344 male rats exposed to ETBE via inhalation for 13 weeks B-145
B-54. Incidence of regenerative foci in the kidneys of F344 male rats exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-150
B-55. A summary of BMDS (version 1.4.1) modeling results based on incidence of
regenerative foci in the kidneys of F344 male rats exposed to ETBE via inhalation
for 13 weeks B-151
B-56. Mean absolute adrenal gland weight (and SD) in F344 male rats exposed to
four different concentrations of ETBE via inhalation for 13 weeks B-155
B-57. A summary of BMDS (version 1.4.1) modeling results based on mean
absolute adrenal gland weight in F344 male rats exposed to ETBE via
inhalation for 13 weeks B-156
B-58. Mean absolute adrenal gland weight (and SD) in F344 female rats exposed to
four different concentrations of ETBE via inhalation for 13 weeks B-161
B-59. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
adrenal gland weight in F344 female rats exposed to ETBE via inhalation for
13 weeks B-162
B-60. Mean absolute adrenal gland weight (and SD) in Sprague-Dawley male rats
exposed to four different concentrations of ETBE via inhalation for 4 weeks B-167
B-61. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
adrenal gland weight in Sprague-Dawley male rats exposed to ETBE via
inhalation for 4 weeks B-168
07/14/2009
x
DRAFT - DO NOT CITE OR QUOTE
-------
B-62. Incidence of bone marrow congestion in F344 female rats exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-173
B-63. A summary of BMDS (version 1.4.1) modeling results based on incidence of
bone marrow congestion in F344 female rats exposed to ETBE via inhalation for
13 weeks B-174
B-64. Mean degenerated spermatocytes (and SD) in the testes of F344 male rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks B-178
B-65. A summary of BMDS (version 1.4.1) modeling results based on mean
degenerated spermatocytes in the testes of F344 male rats exposed to ETBE
via inhalation for 13 weeks B-179
B-66. Mean absolute heart weight (and SD) in F344 female rats exposed to four
different concentrations of ETBE via inhalation for 13 weeks B-184
B-67. A summary of BMDS (version 1.4.1) modeling results based on mean absolute
heart weight in F344 female rats exposed to ETBE via inhalation for 13 weeks B-185
07/14/2009
XI
DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF FIGURES
2-1. Chemical structure of ETBE 4
3-1. Proposed Metabolism of ETBE 11
3-2. Structure of the PBTK model for ETBE and TBA 23
5-1. RfV comparison array for alternative PODs for inhalation data 104
B-l. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial or
linear) based on mean dam body weight change in Sprague-Dawley female rats
orally exposed to ETBE on GDs 5 through 20 B-4
B-2. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial or
linear) based on mean dam body weight change in Sprague-Dawley female rats
orally exposed to ETBE on GDs 5 through 20 B-10
B-3. BMDS (version 1.4.1) model output for the best-fit model (i.e., power) based on
mean F0 body weight change in Sprague-Dawley male rats orally exposed to
ETBE in a two-generation reproductive toxicity study B-l6
B-4. BMDS (version 1.4.1) model output for the best-fit model (i.e., Hill) based on
mean absolute F1 liver weight in Sprague-Dawley male rats orally exposed to
ETBE in a two-generation reproductive toxicity study B-22
B-5. BMDS (version 1.4.1) model output for the best-fit model (i.e., Hill) based on
mean absolute F0 kidney weight in Sprague-Dawley male rats orally exposed to
ETBE in a two-generation reproductive toxicity study B-30
B-6. BMDS (version 1.4.1) model output for the best-fit model (i.e., Hill) based on
mean relative F0 kidney weight in Sprague-Dawley male rats orally exposed to
ETBE in a two-generation reproductive toxicity study B-36
B-7. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute F1 kidney weight in Sprague-Dawley female
rats orally exposed to ETBE in a two-generation reproductive toxicity study B-44
B-8. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean relative F1 kidney weight in Sprague-Dawley female
rats orally exposed to ETBE in a two-generation reproductive toxicity study B-52
B-9. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean LI data from the livers of CD-I female mice exposed
to ETBE via inhalation for 13 weeks B-58
B-10. BMDS (version 1.4.1) model output for the best-fit model (i.e., log-probit)
based on incidence of centrilobular hypertrophy in the livers of CD-I male
mice exposed to ETBE via inhalation for 13 weeks B-64
07/14/2009 xii DRAFT - DO NOT CITE OR QUOTE
-------
B-l 1. BMDS (version 1.4.1) model output for the best-fit model (i.e., logistic)
based on incidence of centrilobular hypertrophy in the livers of CD-I female
mice exposed to ETBE via inhalation for 13 weeks B-69
B-12. BMDS (version 1.4.1) model output for the best-fit model (i.e., 2° polynomial)
based on mean absolute liver weight in CD-I male mice exposed to ETBE via
inhalation for 13 weeks B-74
B-13. BMDS (version 1.4.1) model output for the best-fit model (i.e., 2° polynomial)
based on mean absolute liver weight in CD-I female mice exposed to ETBE via
inhalation for 13 weeks B-80
B-14. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute liver weight in F344 male rats exposed to
ETBE via inhalation for 13 weeks B-86
B-15. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute liver weight in F344 female rats exposed to
ETBE via inhalation for 13 weeks B-92
B-16. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute liver weight in SD male rats exposed to ETBE via
inhalation for 13 weeks B-98
B-17. BMDS (version 1.4.1) model output for the best-fit model (i.e., 2° polynomial)
based on mean absolute liver weight in Sprague-Dawley female rats exposed to
ETBE via inhalation for 13 weeks B-104
B-18. BMDS (version 1.4.1) model output for the best-fit model (i.e., 2° polynomial)
based on mean relative liver weight in Sprague-Dawley male rats exposed to
ETBE via inhalation for 4 weeks B-l 10
B-19. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean relative liver weight in Sprague-Dawley female rats
exposed to ETBE via inhalation for 4 weeks B-l 16
B-20. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute kidney weight in F344 male rats exposed to
ETBE via inhalation for 13 weeks B-122
B-21. BMDS (version 1.4.1) model output for the best-fit model (i.e., 2° polynomial)
based on mean absolute kidney weight in F344 female rats exposed to ETBE via
inhalation for 13 weeks B-128
B-22. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute kidney weight in Sprague-Dawley male rats
exposed to ETBE via inhalation for 4 weeks B-l34
07/14/2009
Xlll
DRAFT - DO NOT CITE OR QUOTE
-------
B-23. BMDS (version 1.4.1) model output for the best-fit model (i.e., 2° polynomial)
based on mean LI in the kidney of F344 male rats exposed to ETBE via inhalation
for 13 weeks B-140
B-24. BMDS (version 1.4.1) model output for the best-fit model (i.e., logistic) based
on incidence of regenerative foci in the livers of F344 male rats exposed to ETBE
via inhalation for 13 weeks B-152
B-25. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute adrenal gland weight in F344 male rats exposed
to ETBE via inhalation for 13 weeks B-157
B-26. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean absolute adrenal gland weight in F344 female rats
exposed to ETBE via inhalation for 13 weeks B-163
B-27. BMDS (version 1.4.1) model output for the best-fit model (i.e., power) based
on mean absolute adrenal gland weight in Sprague-Dawley male rats exposed to
ETBE via inhalation for 4 weeks B-169
B-28. BMDS (version 1.4.1) model output for the best-fit model (i.e., 3° multistage)
based on incidence of bone marrow congestion in F344 female rats exposed to
ETBE via inhalation for 13 weeks B-175
B-29. BMDS (version 1.4.1) model output for the best-fit model (i.e., 1° polynomial
or linear) based on mean regenerative spermatocytes in the testes of F344 male
rats exposed to ETBE via inhalation for 13 weeks B-180
B-30. BMDS (version 1.4.1) model output for the best-fit model (i.e., Hill) based on
mean absolute heart weight in F344 female rats exposed to ETBE via inhalation
for 13 weeks B-186
C-l. Potential RfV comparison array for alternative PODs for oral data C-4
07/14/2009
xiv
DRAFT - DO NOT CITE OR QUOTE
-------
LIST OF ABBREVIATIONS AND ACRONYMS
ACso
concentration required to anesthetize half of study group
AIC
Akaike's Information Criterion
ALDH
aldehyde dehydrogenases
AUC
area under the curve
BMCL
benchmark concentration lower; 95% confidence limit
BMD
benchmark dose
BMDL
benchmark dose, lower 95% confidence limit
BMDS
Benchmark Dose Software
BMR
benchmark response
BrdU
5-Bromo-2'-deoxyuridine
CASE
computer automated structure evaluator
CASRN
Chemical Abstracts Service Registry Number
CHO
Chinese hamster ovary
CNS
central nervous system
CYP
cytochrome P450
DIPE
diisopropyl ether
ETBE
ethyl tertiary butyl ether
FOB
functional observation battery
GABA
y-aminobutyric acid
GC-MC
gas chromatography-mass spectrometry
GD
gestational day
G/ETBE
gasoline/ETBE vapor condensate
HBA
2-hydroxybutyrate
HEC
human equivalent concentration
HGPRT
hypoxanthine-guanine phosphoribosyl transferase
IRIS
Integrated Risk Information System
i.p.
intraperitoneally
Km
Michaelis constant
Kow
octanol:water partition coefficient
KP
permeability coefficient
LC50
median lethal concentration
LD50
median lethal dose
LI
Labeling index
LOAEL
lowest-observed-adverse-effect level
MCHC
mean corpuscular hemoglobin concentration
MCV
mean corpuscular volume
MN
micronucleus
MPD
2-methyl-l,2-propane diol
MTBE
methyl tertiary butyl ether
MULTICASE
multiple computer automated structure evaluation
NOAEL
no-ob served-adverse-effect level
NOEL
no-ob served-effect level
NRC
National Research Council
NTP
National Toxicology Program
PBPK
physiologically based pharmacokinetic
PBTK
physiologically based toxicokinetic
07/14/2009
xv
DRAFT - DO NOT CITE OR QUOTE
-------
PEX
polyethylene
PND
postnatal day
PFC
plaque-forming cell
POD
point of departure
RfC
reference concentration
RfD
reference dose
RfY
reference values
SAR
structure-activity relationships
SCE
sister chromatid exchange
SRBC
sheep red blood cell
SD
standard deviation
TAA
t-amyl alcohol
TAME
tertiary amyl methyl ether
TBA
tertiary butanol
UF
uncertainty Factors
U.S. EPA
U.S. Environmental Protection Agency
Vmax
Maximum substrate turnover velocity
WHO
World Health Organization
07/14/2009
xvi
DRAFT - DO NOT CITE OR QUOTE
-------
FOREWORD
The purpose of this Toxicological Review is to provide scientific support and rationale
for the hazard and dose-response assessment in IRIS pertaining to chronic exposure to ethyl
tertiary butyl ether (ETBE). It is not intended to be a comprehensive treatise on the chemical or
toxicological nature of ETBE.
The intent of Section 6, Major Conclusions in the Characterization of Hazard and Dose
Response, is to present the major conclusions reached in the derivation of the reference dose,
reference concentration and cancer assessment, where applicable, and to characterize the overall
confidence in the quantitative and qualitative aspects of hazard and dose response by addressing
the quality of data and related uncertainties. The discussion is intended to convey the limitations
of the assessment and to aid and guide the risk assessor in the ensuing steps of the risk
assessment process.
For other general information about this assessment or other questions relating to IRIS,
the reader is referred to EPA's IRIS Hotline at 202-566-1676 (phone), (202) 566-1749 (fax), or
hotline.iris@epa.gov (email address).
07/14/2009
xvii
DRAFT - DO NOT CITE OR QUOTE
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
CHEMICAL MANAGER
Andrew A. Rooney, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
AUTHORS
Andrew A. Rooney, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Ted Berner, M.S.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Washington, DC 20460
Reeder Sams, Ph.D.
National Center for Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
CONTRACTOR SUPPORT
Lutz W. Weber, Ph.D., DABT
Oak Ridge Institute for Science and Education
Oak Ridge, TN 37830
C. Clifford Conaway, Ph.D., DABT
Consulting Toxicologist
Mahopac, NY 10541
Janusz Z. Byczkowski, Ph.D., DABT
Toxicology Consultant
Fairborn, OH 45324
George Holdsworth, Ph.D.
Oak Ridge Institute for Science and Education
Oak Ridge, TN 37830
07/14/2009 xviii DRAFT - DO NOT CITE OR QUOTE
-------
REVIEWERS
This document has been reviewed by EPA scientists and interagency reviewers from
other federal agencies.
INTERNAL EPA REVIEWERS
Jane Caldwell, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
J. Michael Davis, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
Anthony B. DeAngelo, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Lynn Flowers, Ph.D., DABT
National Center for Environmental Assessment
Office of Research and Development
Michael G. Narotsky, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Jamie B. Strong, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
D. Charles Thompson, Ph.D., DABT
National Center for Environmental Assessment
Office of Research and Development
Jeffrey E. Welch, Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
Doug C. Wolf, D.V.M., Ph.D.
National Health and Environmental Effects Research Laboratory
Office of Research and Development
07/14/2009
xix
DRAFT - DO NOT CITE OR QUOTE
-------
1. INTRODUCTION
This document presents background information and justification for the Integrated Risk
Information System (IRIS) Summary of the hazard and dose-response assessment of ethyl
tertiary butyl ether (ETBE). IRIS Summaries may include oral reference dose (RfD) and
inhalation reference concentration (RfC) values for chronic and other exposure durations, and a
carcinogenicity assessment.
The RfD and RfC, if derived, provide quantitative information for use in risk assessments
for health effects known or assumed to be produced through a nonlinear (presumed threshold)
mode of action. The RfD (expressed in units of mg/kg-day) is defined as an estimate (with
uncertainty spanning perhaps an order of magnitude) of a daily exposure to the human
population (including sensitive subgroups) that is likely to be without an appreciable risk of
deleterious effects during a lifetime. The inhalation RfC (expressed in units of mg/m3) is
analogous to the oral RfD, but provides a continuous inhalation exposure estimate. The
inhalation RfC considers toxic effects for both the respiratory system (portal-of-entry) and for
effects peripheral to the respiratory system (extrarespiratory or systemic effects). Reference
values are generally derived for chronic exposures (up to a lifetime), but may also be derived for
acute (<24 hours), short-term (>24 hours up to 30 days), and subchronic (>30 days up to 10% of
lifetime) exposure durations, all of which are derived based on an assumption of continuous
exposure throughout the duration specified. Unless specified otherwise, the RfD and RfC are
derived for chronic exposure duration.
The carcinogenicity assessment provides information on the carcinogenic hazard
potential of the substance in question and quantitative estimates of risk from oral and inhalation
exposure may be derived. The information includes a weight-of-evidence judgment of the
likelihood that the agent is a human carcinogen and the conditions under which the carcinogenic
effects may be expressed. Quantitative risk estimates may be derived from the application of a
low-dose extrapolation procedure. If derived, the oral slope factor is a plausible upper bound on
the estimate of risk per mg/kg-day of oral exposure. Similarly, an inhalation unit risk is a
plausible upper bound on the estimate of risk per |ig/m3 air breathed.
Development of these hazard identification and dose-response assessments for ETBE has
followed the general guidelines for risk assessment as set forth by the National Research Council
(1983). EPA Guidelines and Risk Assessment Forum Technical Panel Reports that may have
been used in the development of this assessment include the following: Guidelines for the Health
Risk Assessment of Chemical Mixtures (U.S. EPA, 1986a), Guidelines for Mutagenicity Risk
Assessment (U.S. EPA, 1986b), Recommendations for and Documentation of Biological Values
for Use in Risk Assessment {U.S. EPA, 1988), Guidelines for Developmental Toxicity Risk
Assessment (U. S. EPA, 1991 a), Interim Policy for Particle Size and Limit Concentration Issues
in Inhalation Toxicity Studies (U.S. EPA, 1994a), Methods for Derivation of Inhalation
07/14/2009
1
DRAFT - DO NOT CITE OR QUOTE
-------
Reference Concentrations and Application of Inhalation Dosimetry (U.S. EPA, 1994b), Use of
the Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995), Guidelines for
Reproductive Toxicity Risk Assessment (U.S. EPA, 1996), Guidelines for Neurotoxicity Risk
Assessment (U.S. EPA, 1998b), Science Policy Council Handbook. Risk Characterization (U.S.
EPA, 2000a), Benchmark Dose Technical Guidance Document (U.S. EPA, 2000b),
Supplementary Guidance for Conducting Health Risk Assessment of Chemical Mixtures (U.S.
EPA, 2000c), A Review of the Reference Dose and Reference Concentration Processes (U.S.
EPA, 2002), Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), Supplemental
Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens (U.S. EPA,
2005b), Science Policy Council Handbook: Peer Review (U. S. EPA, 2006a), and A Framework
for Assessing Health Risks of Environmental Exposures to Children (U.S. EPA, 2006b).
The literature search strategy employed for this compound was based on the Chemical
Abstracts Service Registry Number (CASRN) and at least one common name. Any pertinent
scientific information submitted by the public to the IRIS Submission Desk was also considered
in the development of this document. The relevant literature was reviewed through January
2009.
07/14/2009
2
DRAFT - DO NOT CITE OR QUOTE
-------
2. CHEMICAL AND PHYSICAL INFORMATION
In 1990, as part of comprehensive amendments to the Clean Air Act, automobile
emissions were regulated in an effort to reduce CO and O3 pollution. The 1990 amendments to
the Clean Air Act mandated that, in areas with excessive levels of CO and O3 air pollution,
automotive gasoline must contain additives that improve automobile exhaust quality. The most
common fuel oxygenates in 1999 were methyl tertiary butyl ether (MTBE), ethanol, ETBE,
tertiary amyl methyl ether (TAME), diisopropyl ether (DIPE), and tertiary butanol (TBA), with
MTBE accounting for 85% of the oxygenates used in the United States, or roughly 15 billion
L/year (Blue Ribbon Panel on Oxygenates in Gasoline, 1999). Exact production numbers for
ETBE are not available; however, it should be mentioned that in the United States, ETBE usage
ranked far behind MTBE and ethanol. It should also be noted that the reduction in the use of
MTBE makes the use of other oxygenates more likely for two reasons: (1) to replace the volume
that MTBE previously contributed to gasoline (up to 15%), and (2) to meet air pollution
reduction goals previously addressed by MTBE. While the production of oxygenates has
continued in the United States after 2006 (U.S. DOE 2007a), the production of reformulated
gasoline with ethers (e.g., MTBE or ETBE) as a fuel additive stopped in 2006. The amount of
gasoline with ether as a fuel additive was effectively eliminated in 2006, going from a net
production of 2-4 million barrels per month in 2005 to 0 barrels by the end of 2006 (and the
import of reformulated gasoline with ether as a fuel additive dropped from several hundred
thousand barrels per day in 2005 to 0 barrels also by the end of 2006) (U.S. DOE, 2007b).
ETBE has been proposed as an oxygenate substitute for the use of MTBE in gasoline and was
used more widely than MTBE in some European counties by the late 1990s; however, the use of
ETBE has been relatively low in the United States (CFDC, 2001). Although ethanol, rather than
ETBE, has replaced MTBE in gasoline in the United States, ETBE remains an alternative
oxygenate for gasoline (CFDC, 2007).
ETBE is a colorless liquid with a characteristic strong odor that has been described as
being reminiscent of ether, gasoline, or varnish, or as being sweet; its taste has been
characterized as highly objectionable (Vetrano, 1993, unpublished report). An additional effect
of adding ETBE to gasoline is that it may increase the emission of acetaldehyde and
1,3-butadiene into the atmosphere (Schuetzle, 1994). The chemical structure of ETBE is
presented in Figure 2-1, and its main physicochemical properties are given in Table 2-1.
07/14/2009
3
DRAFT - DO NOT CITE OR QUOTE
-------
ch3
ch3—ch2 o—c—ch3
ch3
Figure 2-1. Chemical structure of ETBE.
Table 2-1. Physicochemical properties of ETBE
Characteristic
Reference
CASRN
637-92-3
Drogos and Diaz, 2002
Chemical formula
CsHmO
Molecular weight
102.18
Systematic name
2 -ethoxy -2 -methy lpropane
2-methyl-2-ethoxypropane
National Library of Medicine/Special
Information Services3
Synonyms
ethyl tert-butyl ether
ethyl tert-butyl oxide
methyl-2-ethoxypropane
tert-butyl ethyl ether
ETBE
Melting point
-94°C
Drogos and Diaz, 2002
Boiling point
67-73 °C
Vapor pressure
130-152 mm Hg@25°C
Density
0.73-0.74 g/cm3 @ 25°C
Water solubility
7,650-26,000 mg/L
Oil/water partition coefficient
(log kow) @ 25°C
1.48
1.74
Montgomery, 1994
Drogos and Diaz, 2002
Henry's law constant
2.7 x 10~3 atm-m3/mol @ 25°C
Drogos and Diaz, 2002
Odor
Detection threshold
Recognition threshold
0.013 ppm (0.054 mg/m3)
0.024 ppm (0.1 mg/m3)
Vetrano, 1993
Taste detection threshold (in water)
0.047 ppm (47 ng/L)
Odor detection threshold (in water)
0.049 ppm (49 ng/L)
Odor detection threshold (in water)
0.005 ppm (5 ng/L)
Durand and Dietrich, 2007
Conversion factors
1 ppm = 4.18 mg/m3
1 mg/m3 = 0.24 ppm
1 mg/m3 = 102,180 mmol/L
ppm = mg/m3 x 24.45 m3/mole -r-
molecular weight in g/mol
mmol/L = mg/m3 ^ molecular weight
in mg/mmol -r- 1,000 L/m3
"Available online at http://chem.sis.nlm.nih.gov/chemidplus/jsp/common/ChemFull.jsp
The use of ETBE as a gasoline additive, at an amount of up to 17% by weight, indicates
that this chemical could be produced in very large amounts depending on how widespread the
use of ETBE becomes within the gasoline supply chain. Environmental concern surrounding
07/14/2009
4
DRAFT - DO NOT CITE OR QUOTE
-------
fuel oxygenates has arisen not only in connection with automotive emissions, but also with the
potential of inhalation and/or dermal exposure while refueling motor vehicles. An additional
concern associated with fuel additives is derived from the relatively high aqueous solubility of
these additives and the fact that they have been shown to easily reach groundwater following
leakage or spills (U.S. EPA, 1998b), with the potential of subsequent oral (drinking water) or
dermal (bathing or showering) exposure.
The concentration of oxygenates, particularly ETBE and MTBE, in groundwater and
surface water is likely to exceed the concentrations of other components of gasoline. This
potential for groundwater contamination at higher concentrations by these oxygenates is due to
the fact that when compared to other components of gasoline (e.g., benzene, toluene,
ethylbenzene, and total xylenes), ETBE and MTBE have greater solubilities in water, are less
likely to adhere to soil particles, and are more likely to resist biodegradation (Deeb et al., 2001;
Fayolle et al., 2001; U.S. EPA, 1992). Therefore, ETBE and other oxygenates are likely to travel
farther and faster in groundwater than other gasoline constituents. Although data specifically on
ETBE associated with leaking underground fuel storage tanks are not available, the issue of
water contamination with oxygenates in general has been raised due to the more than 400,000
reported and confirmed releases from underground fuel storage tanks in the United States since
1988 (Rothenstein, 2004), more than half of which potentially contained MTBE (U.S. EPA,
1998b). Data from over 7,200 monitoring wells on 868 leaking underground fuel storage tanks
in the Los Angeles, California area collected by Shih et al. (2004) detected oxygenates roughly
in proportion to their usage (e.g, MTBE at 82.7% and ETBE at 8.9% of the leaking sites). Apart
from potential health concerns, the presence of MTBE in drinking water is associated with an
unpleasant odor and taste that is unacceptable for many people even at relatively low
concentrations (U.S. EPA, 1998b), and ETBE has odor and taste thresholds that allow ETBE to
be detected at even lower concentrations in water or air (i.e., the detection thresholds of ETBE,
5-47 |ig/L in water and 0.13 mg/m3 in air, are 2.5-25 times lower than similar values for MTBE)
(Durand and Dietrich, 2007; Vetrano, 1993).
Plastic plumbing pipe, particularly silane cross-linked polyethylene (PEX), represents an
additional potential source of ETBE in drinking water (Durand and Dietrich, 2007). Durand and
Dietrich (2007) measured ETBE leaching from a PEX pipe using a utility quick test designed for
evaluating taste, odor, migration, and leaching of materials in water distribution systems. ETBE
was observed leaching into tap water with and without the addition of free chlorine or
monochloramine using solid phase microextration/gas chromatography-mass spectrometry
(GC-MS). Aqueous concentrations of ETBE in the leachate ranged from 23 to >140 [j,g/L and
decreased with increased flushing. A team of 10 panelists were recruited and trained for several
weeks in flavor profile analysis in a research protocol approved according to the standards of the
Virginia Tech Institutional Review Board for human subjects. Panelists were able to smell
ETBE at a concentration of 5 [j,g/L (Durand and Dietrich, 2007).
07/14/2009
5
DRAFT - DO NOT CITE OR QUOTE
-------
3. TOXICOKINETICS
3.1. ABSORPTION
Most of the available data on the uptake of ETBE were obtained from volunteers. Nihlen
et al. (1998a) exposed eight healthy male volunteers (average age: 29 years) to 5, 25, and 50 ppm
ETBE by inhalation for 2 hours. Each volunteer was exposed at each concentration in sequence
with 2-week intervals between exposures. The study was performed according to the
Declaration of Helsinki after approval by the Regional Ethical Committee of the institution
where the study was performed, and written informed consent by the volunteers. The volunteers
performed light physical exercise (50 watts) on a bicycle ergometer during exposure. Exhaled
air was collected before exposure, every 30 minutes during exposure, and 6 times after exposure.
The concentrations of ETBE and its primary metabolite, TBA, were determined in exhaled air
samples. Blood was drawn before exposure, approximately every 10 minutes during, and for
1 hour following exposure, approximately every 30 minutes from 1 to 4 hours after exposure,
and an additional 4 times up to 48 hours after exposure. Urine was collected prior to exposure, at
0 and 2 hours, and at approximately 4, 7, 11, 20, 22, and 46 hours after exposure. ETBE, TBA,
and acetone concentrations were determined in blood and urine. The blood profiles of parent
compound and metabolites were similar at all three exposure levels and reflected exposure
concentrations, as judged by linear increases in blood area-under-the-curve (AUC) values for the
concentration-time curve calculated (but only reported in a graphical form) by the authors.
Acetone levels appeared to reflect not only ETBE exposure, but also the physical activity,
and were highly variable. Nihlen et al. (1998a) calculated the ETBE doses to the volunteers to
be 0.58, 2.9, and 5.8 mmol for the 5, 25, and 50 ppm exposure levels, respectively. The
concentrations of ETBE in blood rose sharply during the first 30 minute of exposure and kept
rising at a lower rate until the end of exposure, reaching peak concentrations of about 10, 5.4,
and 1.1 |iM at 50, 25, and 5 ppm, respectively. By 6 hours, they had fallen to very low levels
(<1 |iM) even after 50 ppm exposure. Based on blood AUC values for ETBE, the authors
calculated two types of respiratory uptake: net respiratory uptake = (concentration in inhaled air
- concentration in exhaled air) multiplied by the pulmonary ventilation; and respiratory uptake =
net respiratory uptake + amount exhaled during the exposure. During the 2 hours of exposure,
the authors calculated that 32-34% of each dose were retained by the volunteers (respiratory
uptake), and the net respiratory uptake was calculated to be 26% of the dose at all three exposure
levels. Over 24 hours the respiratory excretion was calculated as 45—50% of the respiratory
uptake, and since the net respiratory uptake and excretion do not consider the amount of ETBE
cleared during exposure, the net respiratory excretion was lower, at 30-31% of the net
respiratory uptake.
Amberg et al. (2000) exposed six volunteers (three males and three females, average age
28 ± 2 years) to 4 and 40 ppm of ETBE (actual exposure concentrations were 4.5 and 40.6 ppm,
07/14/2009 6 DRAFT - DO NOT CITE OR QUOTE
-------
respectively). The exposures lasted 4 hours, and the two concentrations were administered to the
same volunteers 4 weeks apart. These volunteers were healthy nonsmokers and were asked to
refrain from alcohol and medication intake from 2 days before until the end of the experiment.
The study was performed according to the Declaration of Helsinki after approval by the Regional
Ethical Committee of the institution where the study was performed, and written informed
consent was obtained from the volunteers. Urine was collected at 6-hour intervals for 72 hours.
Blood was drawn immediately after exposure and thereafter every 6 hours for 48 hours. ETBE
and its primary metabolite, TBA, were determined in blood, and the same two substances, plus
additional metabolites of TBA, were assessed in urine. The authors estimated the received doses
to be 1,092 |imol following exposure to 40 ppm ETBE, and 121 |imol following 4 ppm
exposure, respectively. These estimates were derived using a resting human respiratory rate of
9 L/minute (13 rnVday) and a retention factor for ETBE of 0.3, which was based on data reported
by Nihlen et al. (1998a). Amberg et al. (2000) also exposed F344 NH rats, 5/sex/dose concurrent
with the volunteers in the same exposure chamber. Blood was taken from the tail vein at the end
of the exposure period and urine was collected for 72 hours at 6-hour intervals following
exposure. Immediately after the 4-hour exposure period, the authors reported that blood levels of
ETBE were lower in the rats than in humans, although exact values were not reported. The
authors estimated that the rats received doses of 20.5 and 2.3 |imol at the 40 and 4 ppm
exposures, respectively, using an alveolar ventilation rate of 0.169 L/minute and a retention
factor of 0.3 for rats.
No published oral dosing studies of the absorption of ETBE in rats or humans were
identified. Dekant et al. (2001a) published a review article that presented an overview of their
studies of the toxicokinetics of ETBE, MTBE, and TAME in both humans and rats following
inhalation exposure at 4 and 40 ppm, respectively (see also Amberg et al., 2000; Bernauer et al.,
1998). In addition, MTBE and TAME were administered to humans in aqueous solution at 5 and
15 mg, respectively. A synopsis of their findings is presented in Table 3-1. The data may
provide some insight relative to uptake of ETBE following ingestion. The authors assumed
100% absorption of MTBE and TAME following ingestion.
07/14/2009 7 DRAFT - DO NOT CITE OR QUOTE
-------
Table 3-1. Doses received by humans and F344 rats following inhalation
exposure to and oral ingestion of fuel oxygenates
Dose received (jimol)
Percent of dose excreted
Dose received (jimol)
Percent of dose excreted
Inhalation
4 ppm exposure level
40 ppm exposure level
ETBE
Human
121
41
1,092
43
Rat
2.3
50
21
53
MTBE
Human
161
35
1,387
69
Rat
3.8
42
33
39
TAME
Human
102
53
1,033
58
Rat
1.9
40
20
42
Ingestion
5 mg dose
15 mg dose
MTBE
Human
57
46
170
49
TAME
Human
49
9
147
14
Source: Dekant et al. (2001a).
A comparison of the percentage of oral dose excreted versus the percentage of inhalation
dose excreted suggests that the assumption of 100% absorption was correct for MTBE, but most
likely not for TAME. If ainblood partition coefficients (see Section 3.2 for details) were the
only determinants of inhalation uptake, one would expect the dose received for ETBE to be
lower than those for both MTBE and TAME, because the ainblood partition coefficient for
ETBE (11.7) is lower than that of MTBE (17.7) and TAME (17.9), which are almost identical
(Nihlen et al., 1995), and the uptake of ETBE is lower than that of MTBE based on the data from
this laboratory. If the log octanol: water partition coefficients (log K0W)were the only
determinants (approximately 1.1 for MTBE, 1.48-1.74 for ETBE, and 1.55 for TAME [Table 2-
1; Drogos and Diaz, 2002]), then values for ETBE and TAME should be similar. Data in Table
3-1 support the latter hypothesis, but there are limited data for the evaluation of either
hypothesis. Note that, on a body weight basis, doses were about 500 times higher in rats than in
humans, although exposures were delivered under entirely identical conditions in the two species
(e.g., Amberg et al., 2000).
No studies investigating dermal absorption of ETBE were identified. However, since
dermal absorption of homologous organic substances is thought to be a function of the
octanol:water partition coefficient, ETBE may be assumed to penetrate rat skin relatively well.
For humans, Potts and Guy (1992) have proposed an equation (3-1) to calculate the dermal
permeability coefficient, Kp:
07/14/2009
8
DRAFT - DO NOT CITE OR QUOTE
-------
log Kp (cm/sec) = -6.3 + 0.71 x log kow - 0.0061 x (molecular weight) (3-1)
Using the log kow (identified as Koct in Potts and Guy, 1992) values for ETBE (0.95-2.2)
and MTBE (0.55-1.91) from Drogos and Diaz (2002), and converting cm/second values to
cm/hour, yields a Kp value for ETBE of 0.0020-0.016 cm/hour and, for comparison, 0.0012-
0.012 cm/hour for MTBE. These calculations predict that the dermal absorption rate of ETBE in
humans would be between 1.3 and 1.7 times as high as that of MTBE. The Kp for MTBE (i.e.,
0.028 cm/hour) calculated by Prah et al. (2004) was approximately twice as high as the above Kp
derived using equation 3-1. However, the data from Prah et al. (2004) were derived from human
subjects exposed to a single concentration and the authors themselves highlight experimental
variables such as the importance of temperature as well as exposure concentration for dermal
absorption.
ETBE is moderately absorbed following inhalation exposure in rats and humans and
blood levels of ETBE approached, but did not reach steady-state concentrations within 2 hours.
Nihlen et al. (1998a) calculated the net respiratory uptake of ETBE in humans to be 26%
compared to 38% for MTBE, which, as the authors point out, parallels the lower blood:air
partition coefficient for ETBE (11.7) compared to MTBE (17.7). The AUC for the
concentration-time curve was linearly related to ETBE exposure level, suggesting linear kinetics
up to 50 ppm. Although comparison of log kow values with MTBE suggest that dermal
absorption rates for ETBE would be higher than MTBE, no data are available on dermal or oral
absorption of ETBE.
3.2. DISTRIBUTION
In vivo data on the tissue distribution of ETBE are not available. However, Nihlen et al.
(1995) conducted a series of in vitro experiments, using blood samples from 10 human donors
(5 males, 5 females), to assess the partitioning of ETBE, MTBE, TAME, and TBA between air
and blood. Kaneko et al. (2000) performed a similar series of in vitro studies to determine
partition coefficients of ETBE, MTBE, TAME, TBA, and t-amyl alcohol (TAA) using tissues
from five male Wistar rats. Both studies reported efficient uptake of these substances from air
into blood, with blood:air partition coefficients of 11.7 and 11.6 for ETBE, 17.7 and 14.2 for
MTBE, 17.9 and 15.5 for TAME, and 462 and 531 for TBA for humans and rats respectively.
Note the similarity between the values for humans and rats and that the blood:air partition
coefficient for ETBE, MTBE, and TAME are much lower than for TBA. Nihlen et al. (1995)
also estimated oil:water partition (log kow) coefficients and obtained values of -0.56 for TBA,
0.90 for MTBE, 1.36 for ETBE, and 1.45 for TAME. These values have a similar ranking, but
are not identical, to those listed in a report by Drogos and Diaz (2002) (namely, 0.35 for TBA,
0.94-1.30 for MTBE, 1.48-1.74 for ETBE, and 1.55 for TAME). Nihlen et al. (1995) also used
07/14/2009
9
DRAFT - DO NOT CITE OR QUOTE
-------
these coefficients, and air:oil partition coefficients, to calculate blood:tissue partition
coefficients. These values are listed in Table 3-2.
Table 3-2. Blood:tissue partition coefficients for gasoline ether additives and
TBA
Partition coefficient
TBA
MTBE
ETBE
TAME
Blood:brain
1.05
1.40
2.34
2.58
Blood:muscle
1.06
1.18
1.78
1.93
Blood:fat
0.646
4.98
11.6
13.3
Blood:lung
1.02
0.783
0.835
0.837
Blood:kidney
1.06
1.04
1.42
1.51
Blood:liver
1.05
1.04
1.44
1.54
Source: Nihlen etal. (1995).
Nihlen et al. (1998a) exposed eight healthy male volunteers (average age: 29 years) to 5,
25, and 50 ppm ETBE by inhalation for 2 hours. The volunteers performed light physical
exercise during exposure. Profiles of ETBE, TBA, and acetone were established for blood
throughout exposure and for up to 22 hours thereafter. The same laboratory conducted studies
with MTBE by using the same experimental protocol. Net uptake of MTBE was 38% of the
dose (compared to 26% net uptake for ETBE) and net exhalation was 28% of the net uptake for
MTBE (compared to 31% net exhalation for ETBE) (Nihlen et al., 1998c). The results may
reflect the difference in blood:air partition coefficients between MTBE and ETBE (18 and 12,
respectively) (Nihlen et al., 1995), suggesting that MTBE has a higher tendency to partition into
human blood and tissues and is less likely to be eliminated by exhalation compared to ETBE.
Therefore, the high volume of distribution for ETBE in humans, 6.4 L/kg, as compared to
3.9 L/kg for MTBE (Nihlen et al., 1998a) is indicative of the higher partition coefficients for
blood:tissue for ETBE relative to MTBE, particularly the over twofold greater blood:fat partition
coefficient (11.6 and 4.98 for ETBE and MTBE respectively).
3.3. METABOLISM
The metabolism of ETBE has been studied in rats and humans using both in vivo and in
vitro methods. A schematic of the proposed metabolism of ETBE is presented in Figure 3-1. On
the basis of structures of the metabolites elucidated, ETBE is initially metabolized by
cytochrome P450 (CYP) enzymes via oxidative deethylation by introducing a hydroxy group
into the ethyl or methyl moieties of the molecule (Bernauer et al., 1998). The resulting
hemiacetal is unstable and decomposes spontaneously into TBA and acetaldehyde. In human
liver microsome preparations, this step is catalyzed mainly by CYP2A6, with some contribution
from CYP3A4 and CYP2B6, and possible contribution of CYP2E1 (Le Gal et al., 2001; Hong et
al., 1999a). Using data from rat hepatic microsome preparations, Turini et al. (1998) suggest that
07/14/2009
10
DRAFT - DO NOT CITE OR QUOTE
-------
CYP2B1 may be the lead enzyme for this step in rats. Acetaldehyde is oxidized to acetic acid
and eventually to carbon dioxide (CO2). TB A can be sulfated, glucuronidated, and excreted into
urine, or it can undergo further oxidation to form 2-methyl-l,2-propane diol (MPD),
2-hydroxybutyrate (HBA), and acetone. It should be noted that these metabolites have been
identified in humans and rats for both ETBE and MTBE. However, all the enzymes that perform
these metabolic steps have not been fully described. Excretion studies indicate that final
metabolism to C02 plays only a minor role.
HIE CH3 ^ TBA CH3
| CYP2B6 | | , sulfate
c2h5—o—c—ch3 >¦ HO—c—gh3 —ch3—c—o
ch3 ^ ch3 ch3
acetaldehyde
\
gh2oh
I MPD
* -A HO—C—CH3
acetic acid ,
CH-,
co2
Source: Adapted from Dekant et al., 2001a.
Figure 3-1. Proposed Metabolism of ETBE.
\
acetone
glucuroni de
HBA
COOH
HO C — CH3
>' ch3
Zhang et al. (1997) used computer models to predict the metabolites of ETBE and their
toxic effects. The metabolism model correctly predicted cleavage into TBA and acetaldehyde
and that TBA would undergo glucuronidation and sulfation. However, for the further
metabolism of TBA, the computer model predicted reductive steps leading to metabolites that
have not been identified in vivo or in vitro. The software did not predict the formation of MPD
or HBA, which have been found in vivo.
3.3.1. Metabolism in Humans
3.3.1.1. Metabolism of ETBE in Humans In Vivo
Nihlen et al. (1998a) exposed eight healthy male volunteers (average age: 29 years) to 0,
5, 25, and 50 ppm ETBE by inhalation for 2 hours. Profiles of ETBE, TBA, and acetone were
established for blood throughout exposure and for up to 22 hours thereafter. The blood profiles
of parent compound and metabolites were similar at all three exposure levels, and reflected
exposure concentrations, as judged by linear increases in concentration-time AUC values
07/14/2009
11
DRAFT - DO NOT CITE OR QUOTE
-------
calculated (but only reported graphically) by the authors. Acetone levels were highly variable
before, during, and after the exposure period.
The concentration of ETBE in blood rose sharply during the first 30 minutes of exposure
and kept rising at a lower rate until the end of exposure to reach peak concentrations of about 10,
5, and 1 |iM at 50, 25, and 5 ppm, respectively. By 6 hours, ETBE concentrations had fallen to
low levels even after exposures to 50 ppm. The blood concentration of TBA continued to rise
for the full 2-hour exposure period, with peak values of about 13 and 7 |iM at 50 and 25 ppm,
respectively. Blood concentrations leveled off for 3-4 hours, and then began a slow decline to
less than one-half maximum levels by 24 hours (TBA levels could not be determined following
5 ppm exposure). Acetone blood levels began to increase after about 1 hour of exposure, and
continued to increase after the end of exposure (high dose), or leveled off (lower doses and
controls) for about IV2 hours after exposure. Blood acetone levels fell rapidly during the next
half hour, but remained slightly above normal for the exposed volunteers until 4 hours after
exposure, when measurements were terminated.
Amberg et al. (2000) exposed six volunteers (three males and three females, average age:
28 ± 2 years) to 4 and 40 ppm of ETBE, respectively (actual exposure concentrations were
4.5 and 40.6 ppm, respectively). The exposures lasted 4 hours, and the two concentrations were
administered to the same volunteers 4 weeks apart. Urine was collected at 6-hour intervals for
72 hours. Blood was drawn immediately and at 4 or 6 hours after exposure, and thereafter every
6 hours for 48 hours. Levels of parent ETBE and its primary metabolite, TBA, were determined
in blood and urine. In urine, two further metabolites of TBA, MPD and HBA, were also assayed.
At an exposure level of 40 ppm, the peak concentration of TBA in blood was 13.9 ±
2.2 and 1.8 ± 0.2 |iM at 4 ppm. At the and low high exposure concentrations, TBA disappeared
from blood with half-lives of 9.8 ± 1.4, and 8.2 ± 2.2 hours. The time courses of metabolite
appearance in urine after 40 and 4 ppm were similar, but relative urinary levels of metabolites
after 4 ppm differed from those after 40 ppm. Using parent ETBE as the reference, molar ratios
for total urinary excretion were 1:25:107:580 (ETBE:TBA:MPD:HBA) after 40 ppm and
1:17:45:435 after 4 ppm. Individual variations were large, but the authors did not report any
gender differences in the metabolism of ETBE based on data from only three subjects of each
sex.
3.3.1.2. In Vitro Metabolism of ETBE Using Human Enzyme Preparations
The metabolism of ETBE has been studied in vitro using both human liver microsomes
and genetically engineered cells expressing individual human CYP isozymes. Hong et al.
(1997a) coexpressed human CYP2A6 or CYP2E1 with human CYP reductase in insect SF9
cells. In this system, in the presence of 1 mM ETBE, TBA was formed at rates of
13.6 nmol/min-nmol CYP2A6 and 0.8 nmol/min-nmol CYP2E1. Corresponding activities with
1 mM MTBE as the substrate were 6.1 and 0.7 nmol/min-nmol, respectively.
07/14/2009 12 DRAFT - DO NOT CITE OR QUOTE
-------
Hong et al. (1999a) obtained 15 human liver microsome samples and used them to
compare metabolic activities with ETBE, MTBE, and TAME as the substrates. They found that
the metabolism of all three substrates was highly correlated with certain CYP isozymes. The
highest degree of correlation was found for CYP2A6, which also displayed the highest turnover
numbers. The 15 samples displayed very large interindividual variations in metabolic activities,
with turnover numbers for ETBE ranging from 179-3,130 pmol/minute-mg protein. Michaelis
constant (Km) values, estimated in three human liver microsomal samples using MTBE, ranged
from 28-89 |iM, with maximum substrate turnover velocity (Vmax) values ranging from 215—
783 pmol/minute-mg protein. The Vmax/Km ratios, however, varied only between 7.7 and 8.8.
As part of CYP inhibition studies in the same paper, human liver microsomes were co-incubated
with MTBE, ETBE, or TAME in the presence of chemicals or specific antibodies to inhibit
either CYP2A6 or CPY2E1. For chemical inhibition, coumarin was dissolved in 2 |iL of
methanol and added to the liver microsomes prior to initiation of the reaction. For antibody
inhibition, monoclonal antibodies against human CPY2A6 and CYP2E1 were preincubated with
liver microsomes prior to incubation with the rest of the reaction mixture. Methanol alone
caused approximately 20% inhibition of MTBE, ETBE, and TAME. Coumarin, a CYP2A6
substrate, caused a significant dose-dependent inhibition of all three oxidants with a maximal
inhibition of ETBE of 99% at 100 |iM coumarin. Antibodies against CYP2A6 inhibited
metabolism of MTBE, ETBE, and TAME by 75-95%. In contrast, there was no inhibition by
the antibody against CYP2E1. The same anti-CYP2El antibody inhibited over 90% of CPY2E1
activity assayed as N-nitrosodimethylamine in the liver microsomes. In the same paper, these
authors introduced several specific human CYPs into human P-lymphoblastoid cells and
measured metabolic activities with ETBE and MTBE as the substrates. They established a
correlation ranking for ETBE metabolism (to TBA) by 10 human CYP isozymes: 2A6 > 3A4 «
2B6 « 3A4/5 » 2C9 > 2E1 ~ 2C19 » 1A2 ~ 2D6 ~ 1A2. They characterized the correlation with
CYP2A6 as high, with 3A4, 3A5, and 2B6 as good, and with 2C9, 2E1, and 2C19 as poor, and
the remaining three CYP activities showed no correlation with ETBE metabolism. They also
reported direct enzyme activities toward ETBE as the substrate (in pmol TBA formed per minute
and pmol CYP): 2A6-1.61; 2E1-0.34; 2B6-0.18; and 1A2-0.13. CYPs 1B1, 2C8, 2C9, 2C19,
and 2D6 were not investigated. CYP1A2, which showed activity toward ETBE, did not
metabolize MTBE to TBA. CYP4A11 showed considerable activity toward MTBE, but very
low activity toward ETBE and TAME. CYP3 A4 and 1 Al did not metabolize ETBE or MTBE in
this system, but displayed considerable activity toward TAME. The authors conclude that
CYP2A6 is the major enzyme responsible for the oxidative metabolism of MTBE, ETBE, and
TAME in human livers. Furthermore, they conclude that the results of the correlation analysis
and antibody inhibition study strongly suggest that CYP2E1 is not a major enzyme responsible
for metabolism of MTBE, ETBE, or TAME.
07/14/2009
13
DRAFT - DO NOT CITE OR QUOTE
-------
Le Gal et al. (2001) used similar human cytochrome preparations as Hong et al. (1999a)
(i.e., from deceased human donors) or genetically modified human P-lymphoblastoid cells to
elucidate the metabolism of ETBE, MTBE, and TAME. They identified as primary metabolites
formaldehyde from MTBE and TAME, acetaldehyde from ETBE, TAA from TAME, and TBA
from ETBE and MTBE. The human microsomes showed higher catalytic activity towards
MTBE and TAME at 0.5 mM, compared to ETBE, but very similar activities at substrate
concentrations of 10 mM. Le Gal et al. (2001) confirmed the wide interindividual variation of
activities previously reported by Hong et al. (1999a, 1997b). Using MTBE as the substrate, they
found a highly significant correlation with CYP2A6 activities and a lesser, but still significant,
correlation with CYP3A4 activities. No correlations could be established for 1 Al, 1A2, or 2E1
activities. However, using substrate concentrations of 0.5 and 10 mM, they found that 2A6 and
3E4, but not 2E1 or 2B6, had high activity at 0.5 mM, while 2E1 and 2B6 displayed considerable
activity at 10 mM. Using the average levels and the turnover numbers of various CYPs in
human liver, they concluded that fuel oxygenate ethers were predominantly metabolized by
CYP2A6, with considerable contribution from CYP3A4. CYP2E1, they concluded, did not play
a significant role in human metabolism of these substances.
3.3.2. Metabolism in Animals
3.3.2.1. Metabolism of ETBE in Animals In Vivo
Bernauer et al. (1998) studied the metabolism and excretion of [13C]-ETBE, MTBE, and
TBA in rats. F344 rats, 2/sex, were exposed via inhalation to 2,000 ppm ETBE or MTBE for
6 hours, or three male F344 rats received 250 mg/kg TBA by gavage. Urine was collected for
48 hours. The metabolic profiles for ETBE and MTBE were essentially identical, with excretion
of MPD > HBA > TBA-sulfate > TBA-glucuronide. Oral administration of TBA produced a
similar metabolite profile, with HBA > TBA-sulfate > MPD » TBA-glucuronide ~ TBA. TBA
could not be detected in urine when ETBE or MTBE were administered by inhalation. Traces of
acetone were also detected in urine. Amberg et al. (2000) exposed F344 NH rats, 5/sex/dose, to
ETBE in the same exposure chamber coincident with the volunteers (see Section 3.1). Urine was
collected for 72 hours following exposure. Blood samples were drawn from the tail vein every
6 hours up to 48 hours. Peak blood levels of ETBE and TBA were much lower than in humans,
5.3 ±1.2 and 21.7 ± 4.9 |iM at 40 ppm and 1.0 ± 0.7 and 5.7 ± 0.8 |iM at 4 ppm, respectively.
Similar to humans, rats excreted mostly HBA in urine, followed by MPD and TBA. The molar
ratios for total urinary excretion of TBA:MPD:HBA were 1:2.3:15 after exposure to 40 ppm and
1:1.5:11 after 4 ppm. Parent ETBE was not identified in rat urine in this study.
In a review covering mostly their own work on fuel oxygenate metabolism, Dekant et al.
(2001b) focused on aspects of metabolism of MTBE and ETBE in humans and rats. They
reported that, at a high exposure level (2,000 ppm), rats predominantly excreted the glucuronide
of TBA in urine, which, at low levels (4 or 40 ppm) had been barely detectable. They concluded
07/14/2009 14 DRAFT - DO NOT CITE OR QUOTE
-------
that, at high exposure levels, the normally rapid metabolism of TBA to MPD and HBA became
saturated, forcing more of the initial metabolite of ETBE or MTBE through the glucuronidation
pathway. The apparent final metabolite of ETBE was HBA, although this substance can undergo
further metabolism to acetone. The latter process appears to play a minor role in the overall
metabolism of ETBE or MTBE. The authors also pointed out that many metabolites of the fuel
oxygenate ethers, such as formaldehyde, acetaldehyde, TBA, HBA, or acetone, occur naturally in
normal mammalian physiology, providing a highly variable background that needs to be
corrected for in metabolic experiments.
3.3.2.2. Metabolism of ETBE in Animal Tissues In Vitro
Using isolated rat liver microsomes, Hong et al. (1997a) found that metabolism occurred
only in the presence of an NADPH-regenerating system and that the metabolic activity was
inhibited by 80% after treating the microsomal preparation with carbon monoxide, indicating
CYP involvement. In another study investigating potential target tissues for ETBE toxicity,
Hong et al. (1997b) studied the metabolic activities of olfactory mucosa, respiratory epithelium,
liver, lung, and olfactory bulb from rats. They prepared microsomes, added an NADPH-
regenerating system, and evaluated enzyme kinetics at various substrate concentrations. In
olfactory mucosa, the authors derived Km values of 125 and 111 |iM for ETBE and MTBE, with
corresponding Vmax values of 11.7 and 10.3 nmol/minute-mg protein, respectively. Addition of
TAME to the reaction mixture exerted a concentration-dependent inhibition of ETBE or MTBE
metabolism. Coumarin, a CYP2A6 substrate, also inhibited ETBE metabolism. These results
indicated that rat olfactory mucosa, on a per-weight basis, has 37 times the capacity of liver to
metabolize fuel oxygenate ethers, and hence, has the capacity for first-pass metabolism.
Hong et al. (1999b) used CYP2E1 knockout mice to investigate whether this enzyme
plays a major role in fuel oxygenate ether metabolism. They compared the ether-metabolizing
activity of liver microsomes (30 minutes at 37°C and 1 rnM ether) between the CYP2E1
knockout mice and their parental lineage strains using four or five female mice (7 weeks of age)
per group. The ETBE metabolizing activities (nmol/minute-mg protein) were 0.51 ± 0.24 for
CP2E1 knockout mice, 0.70 ± 0.12 for C57BL/6N mice, and 0.66 ± 0.14 for 129/Sv mice. The
MTBE metabolizing activities (nmol/minute-mg protein) were 0.54 ± 0.17 for CP2E1 knockout
mice, 0.67 ± 0.16 for C57BL/6N mice, and 0.74 ± 0.14 for 129/Sv mice. The TAME
metabolizing activities (nmol/minute-mg protein) were 1.14 ± 0.25 for CP2E1 knockout mice,
1.01 ± 0.26 for C57BL/6N mice, and 0.76 ± 0.25 for 129/Sv mice. Mice that did not express any
CYP2E1 did not differ from wild-type animals in their ability to metabolize ETBE, MTBE, or
TAME, suggesting that CYP2E1 is unlikely to be important in the metabolism of ETBE. Turini
et al. (1998) investigated the influence of ETBE exposure on hepatic microsomal enzyme
activities (as measured using CYP isozyme-specific substrates) and the effects of specific
enzyme induction on ETBE metabolism in male Sprague-Dawley rats. Moderate doses of ETBE
07/14/2009 15 DRAFT - DO NOT CITE OR QUOTE
-------
(200 or 400 mg/kg) administered intraperitoneally (i.p.) for 4 days did not induce any hepatic
CYPs. However, ETBE (2 mL/kg) administered by gavage as a 50% corn oil solution for 2 days
almost doubled activities of 3A1/2 and 2B1, doubled 2E1, and induced CYP2B1/2 sixfold.
CYP1 Al/2 activity was slightly reduced after 2 days of ETBE (2 mL/kg) by gavage. The
authors also estimated kinetic constants for various CYPs in rats and found the following Km or
Vmax values: controls (2C forms predominant), 6.3 mM/0.93 nmol/minute-mg protein; 2A/2B
induced, 4.1/3.8; 2E1 induced, 4.7/1.6; 3A induced, 4.4/1.4; and 1A induced, not determined/0.9.
Using a system with reconstituted CYPs, the authors found that CYP2B1 displayed the lowest
Km (2.3 mM), and the highest turnover number (56 nmol/minute-nmol CYP), and concluded that
this isoform was the principal CYP to metabolize ETBE in the rat.
The enzymes that metabolize TB A to MPD, HB A, and even acetone, have not been fully
characterized. However, it is clear that TBA is not subject to metabolism by alcohol
dehydrogenases (Dekant et al., 2001a).
3.4. ELIMINATION
3.4.1. Elimination in Humans
Nihlen et al. (1998a) exposed eight healthy male volunteers (average age: 29 years) to 5,
25, and 50 ppm ETBE by inhalation for 2 hours. ETBE, TBA, and acetone were measured in
urine for up to 22 hours after exposure. The blood profiles of parent compound and metabolites
were similar at all three exposure levels, and reflected exposure concentrations. The authors
estimated the ETBE amount to the volunteers to be 0.58, 2.9, and 5.8 mmol for the 5, 25, and
50 ppm exposure levels, respectively. Based on blood AUC values for ETBE and metabolites,
the authors calculated that respiratory uptake was 32-34% in humans, and net uptake (which
excludes ETBE exhaled during exposure) was calculated to be 26% of the dose at all three
exposure levels. During the 24 hours following the start of inhalation exposure, respiratory
excretion was calculated at 45—50% of the inhaled ETBE (respiratory uptake) and net respiratory
excretion was 31% (of the net respiratory uptake), of which TBA accounted for only 1.4-3.8%).
Urinary excretion of parent ETBE accounted for even less, 0.12, 0.061, and 0.056% of the dose
retained after 5, 25, and 50 ppm exposures, respectively. The authors identified four phases of
elimination of ETBE from blood, with half-lives of about 2 and 20 minutes and 1.7 and 28 hours.
Only one phase for elimination of TBA from blood was identified with a half-life of 12 hours
(10 hours in another study with volunteers: Johanson et al., 1995). In urine, ETBE displayed two
phases of elimination, with half-lives of about 8 minutes and 8.6 hours. The half-life of TBA in
urine was determined to be 8 hours (Johanson et al., 1995).
These data suggest complex toxicokinetics for ETBE in humans. The first phase of
elimination from blood likely indicate uptake into highly perfused tissues. The other phases may
indicate uptake into less perfused tissues and fat as well as metabolism events. The apparent
total body clearance of ETBE (based upon the net respiratory uptake) was 0.57 L/hour-kg
07/14/2009 16 DRAFT - DO NOT CITE OR QUOTE
-------
(average of the three exposure levels). The metabolic clearance was calculated as
0.39 L/hour-kg and the exhalation clearance as 0.35 L/hour-kg.
Amberg et al. (2000) exposed six volunteers (three males and three females, 28 ± 2 years
old) to 4 and 40 ppm of ETBE, respectively (actual exposure concentrations were 4.5 and
40.6 ppm, respectively). The exposures lasted 4 hours, and the two concentrations were
administered to the same volunteers 4 weeks apart. Urine was collected at 6-hour intervals for
72 hours. Blood was drawn immediately and at 4 or 6 hours after exposure, and thereafter every
6 hours for 48 hours. Parent ETBE and TBA were determined in blood and urine. Two further
metabolites of TBA, HBA and MPD, were also determined in urine.
At 40 ppm, the peak concentration of ETBE in blood was 12.1 ± 4.0 |iM, while that for
TBA was 13.9 ± 2.2 |iM. The corresponding values at 4 ppm were 1.3 ± 0.7 and 1.8 ± 0.2 |iM,
respectively. At the high exposure concentration, two elimination half-lives were found for
ETBE, 1.1 ±0.1 and 6.2 ±3.3 hours. TBA displayed only one half-life, 9.8 ± 1.4 hours. At the
low exposure concentration, only the short half-life for ETBE could be measured at 1.1 ±
0.2 hours, while that for TBA was 8.2 ± 2.2 hours. The predominant urinary metabolite
identified was HBA, excreted in urine at 5-10 times the amount of MPD and 12-18 times the
amount of TBA (note: urine samples had been treated with acid before analysis to cleave
conjugates). Excretion of unchanged ETBE in urine was minimal. The time courses of urinary
elimination after 40 and 4 ppm, respectively, were similar, but relative urinary levels of HBA
after 4 ppm were higher, while those for MPD were lower, as compared to 40 ppm. HBA in
urine showed a broad maximum at 12-30 hours after exposure to both concentrations, with a
slow decline thereafter. MPD in urine peaked at 12 and 18 hours after 40 and 4 ppm,
respectively, while TBA peaked at 6 hours after both concentrations. The time to peak of the
three metabolites reflected the sequence of their formation and interconversion as ETBE is
metabolized. Individual variations were large, but the authors did not report gender differences
in the toxicokinetics of ETBE. Based on the dose estimates presented in Section 3.3.1, Amberg
et al. (2000) calculated that 43 ± 12% of the 40 ppm dose and 50 ± 20% of the 4 ppm dose had
been excreted in urine by 72 hours. Respiratory elimination was not monitored.
3.4.2. Elimination in Animals
Amberg et al. (2000) exposed F344 NH rats, 5/sex/dose concurrent with the volunteers in
the same exposure chamber. Urine was collected for 72 hours following exposure. Similar to
humans, rats excreted mostly HBA in urine, followed by MPD and TBA. Parent ETBE was not
identified in rat urine. The half-life for TBA in rat urine was 4.6 ± 1.4 hours at 40 ppm, but
could not be calculated at 4 ppm. Corresponding half-lives were 2.6 ± 0.5 and 4.0 ± 0.9 hours
for MPD, and 3.0 ± 1.0 and 4.7 ± 2.6 hours for HBA. The authors concluded that rats eliminated
ETBE considerably faster than humans. Urinary excretion accounted for 53 ± 15 and 50 ± 30%
07/14/2009
17
DRAFT - DO NOT CITE OR QUOTE
-------
of the estimated dose at 40 and 4 ppm exposures, respectively, with the remainder of the dose
being eliminated via exhalation, as suggested by the authors.
Bernauer et al. (1998) studied the excretion of [13C]-ETBE and MTBE in rats. F344 rats,
2/sex, were exposed via inhalation to 2,000 ppm ETBE or MTBE for 6 hours, or three male
F344 rats received 250 mg/kg TBA by gavage. Urine was collected for 48 hours. The metabolic
profiles for ETBE and MTBE were essentially identical, with relative excreted amounts of MPD
> HBA > TBA-sulfate > TBA-glucuronide. Oral administration of TBA produced a similar
metabolite profile, with relative amounts of HBA > TBA-sulfate > MPD » TBA-glucuronide ~
TBA.
Although there are several unpublished reports relevant to the elimination of ETBE
following inhalation exposure, no additional peer-reviewed publications were identified.
Unpublished reports have not gone through the public peer-review process and are of unknown
quality. They are included here as additional information only.
Sun and Beskitt (1995a, unpublished report) investigated the pharmacokinetics of
[14C]-ETBE in F344 rats (3/sex/dose) exposed by nose-only inhalation at target concentrations of
500, 750, 1,000, 1,750, 2,500, and 5,000 ppm for a single 6-hour period (the true doses differed
by less than 10% from the targets). Specific activity of the administered [14C]-ETBE and
localization of the label were not reported. Note, that in the absence of the specific activity and
localization of the label, it is not clear how the "mg ETBE equivalents" were calculated in the
Sun and Beskett (1995a, b, unpublished reports) for "Total" column in Table 3-3 or for the
specific tissues in Table 3-4. Of the three animals per sex exposed concurrently, two were used
in the further study, while the third was kept as a spare. One animal/sex was placed into a
metabolic cage and monitored for up to 118 hours. Exhaled organic volatiles were trapped in
charcoal filters. Exhaled CO2 was trapped in aqueous 1 M KOH. Samples from the 5,000 ppm
treated animals were collected at 3, 6, 12, 18, 24, 48, 72, 96, and 118 hours after termination of
exposure. At the lower exposure concentrations listed above, samples were collected at fewer
time points; generally, at full-day intervals up to 96 hours. Animals were euthanized either
immediately after exposure or after being removed from the metabolic cages, and blood and
kidneys were collected. Cages were washed and the wash fluid collected. Charcoal traps were
eluted with methanol. Urine, cage wash, trapped 14C02, and charcoal filter eluates were
measured directly by liquid scintillation spectrometry. Blood and kidney tissue were combusted
in a sample oxidizer and analyzed by liquid scintillation spectrometry.
07/14/2009
18
DRAFT - DO NOT CITE OR QUOTE
-------
Table 3-3. Elimination of [14C]-ETBE-derived radioactivity from rats and
mice within 96 hours following a single 6-hour inhalation exposure
Exposure level (ppm)
Volatile organics3
Exhaled CO/
Urine3
Feces"
Totalb
F344 Ratc
500
37
1
60
2
9.92
750
36
1
62
2
17.5
1,000
42
1
56
2
22.1
1,750
58
2
38
3
56.9
2,500
52
2
45
2
56.2
5,000d
63
2
34
1
97.5
(51)
(1)
(44)
(3)
(116)
CD-I Mousee
500
10
1
74
16
6.38
750
28
2
60
10
7.94
1,000
29
2
64
6
12.8
1,750
42
2
46
10
13.7
2,500
42
2
47
10
22.7
5,000d
44
5
39
12
18.9
(37)
(2)
(57)
(2)
(28)
aPercent of total eliminated radioactivity; mean of one male and one female.
bIn mg [14C]-ETBE equivalents.
Sources: °Sun and Beskitt (1995a, unpublished report); dvalues in parentheses: Borghoff (1996, unpublished report);
eSun and Beskitt (1995b, unpublished report)
Table 3-4. Radioactivity in blood and kidney of rats and blood and liver of
mice, following 6 hours of [14C]-ETBE inhalation exposure
Exposure level
(ppm)
F344 Rat3'
CD-I Mouse3'
Bloodb
Kidney0
Bloodb
Liver0
500
0.037
0.074
0.154
0.208
750
0.062
0.094
0.340
0.348
1,000
0.080
0.116
0.336
0.540
1,750
0.124
0.152
0.481
0.724
2,500
0.156
0.185
0.474
0.628
5,000
0.114
0.182
0.408
0.592
aMean values of one male and one female.
bIn mg [14C]-ETBE equivalents per gram blood.
°In mg [14C]-ETBE equivalents.
Sources: Sun and Beskitt (1995a, 1995b, unpublished reports).
During 96 hours in metabolic cages, approximately 60% of the eliminated radioactivity
was recovered from urine and approximately 38% was recovered from exhaled organic volatiles.
07/14/2009 19 DRAFT - DO NOT CITE OR QUOTE
-------
This pattern was maintained at an exposure concentration of 1,000 ppm; above that, urinary
excretion of radioactivity decreased to 34% of the recovered radioactivity, while exhalation of
organic volatiles increased to 63%. Exhalation of 14C02 increased marginally, from 1% at
500 ppm to 2% at 5,000 ppm, while fecal elimination remained rather constant at about 2%
throughout the exposure concentrations. A compilation of these results, together with results
from mice from a parallel study (Sun and Beskitt, 1995b, unpublished report), is given in Table
3-3. The authors concluded that the metabolic pathways leading to urinary excretion of ETBE
degradation products became saturated at an exposure concentration of approximately
1,750 ppm.
The time course of elimination indicated that exhalation of organic volatiles was
essentially complete by 24 hours, while urinary excretion of ETBE-derived radioactivity
displayed a broad peak at 12-48 hours. The bulk of each dose was eliminated within 48 hours
after the end of exposure. At 5,000 ppm, 14CC>2 exhalation and fecal excretion of radioactivity
remained rather constant from 12 to 118 hours. Levels of radioactivity in blood and kidneys
after increasing exposure concentrations of [14C]-ETBE are shown in Table 3-4 (again combined
with the mouse data from the parallel study). The major finding was that radioactivity levels
increased up to 2,500 ppm, but leveled off in kidney and fell considerably in blood at 5,000 ppm.
To the authors, these data were indicative of saturation of the absorption pathway at around
2,500 ppm. However, it is noteworthy that total elimination of ETBE-derived radioactivity
increased steadily from 500 to 5,000 ppm (Table 3-3). The authors reported no deaths following
6 hours of ETBE exposure. The findings of Sun and Beskitt (1995a, unpublished report) at
5,000 ppm were essentially confirmed by Borghoff (1996, unpublished report) in a pilot study
that used the identical species, experimental protocol, materials, and methods, but was conducted
at a different laboratory at a later time point.
In a parallel study with an identical experimental protocol, Sun and Beskitt (1995b,
unpublished report) exposed CD-I mice (3/sex/dose) to 500, 750, 1,000, 1,750, 2,500, and
5,000 ppm [14C]-ETBE. The only difference from the rat study (Sun and Beskitt, 1995a,
unpublished report) was that, instead of kidneys, livers were harvested from mice. The
corresponding results from this study are shown in Tables 3-3 and 3-4, jointly with the results
from the rat study.
Noteworthy differences between the two species were that, in general, mice eliminated a
smaller percentage of the dose in the form of volatile organics and a higher amount in urine, at
least up to 1,000 ppm (Table 3-3), and excreted about 5 times as much [14C]-ETBE-derived
radioactivity via feces than did rats. The total amounts of eliminated radioactivity were
considerably higher, as reported, in rats than in mice; however, the values in the respective
columns of Table 3-3 are not corrected for body weight. When normalized to body weight, it is
apparent that mice absorbed a higher dose than rats and/or had a higher metabolic capacity.
However, the total eliminated radioactivity at 5,000 ppm showed no further increase over the
07/14/2009 20 DRAFT - DO NOT CITE OR QUOTE
-------
values at 2,500, indicating that the absorptive and metabolic capacities of mice had become
saturated. Judging from the data in Table 3-4, saturation of blood and liver had occurred already
at 1,750 ppm. The authors reported no deaths following 6 hours of ETBE exposure. It may be
noted here that Sun and Beskitt (1995a, b, unpublished reports) did not state any estimates for
absorbed dose. The data in Table 3-3, however, indicate that, given the rapid exhalation of
[14C]-ETBE-derived material, any attempt to estimate a level of inhalation absorption following
a 6-hour exposure without respiratory elimination control would be futile.
Borghoff (1996, unpublished report) conducted studies to establish experimental
conditions for future bioassays of ETBE, based on the two studies previously conducted by Sun
and Beskitt (1995a, b, unpublished reports). The experimental protocol and materials were
identical to the ones used by Sun and Beskitt (1995a, b, unpublished reports); however, in this
pilot study, only three male F344 rats and three male CD-I mice were used per experiment, with
the only exposure level of 5,000 ppm. Also, only blood was collected from the animals, while
the whole carcasses were liquefied and assayed for retained radioactivity immediately after
exposure and after the end of the animals' stay in metabolic cages. Radioactive ETBE was
obtained by mixing [14C]-ETBE with unlabeled material in the gas phase for a specific activity of
2.74 |iCi/mmol. It was found that rats, when assayed immediately after exposure, had absorbed
2.57 ± 0.14 |iCi radioactivity, while the balance of radioactivity after 96 hours in metabolic
cages came to 3.17 ± 0.08 |iCi (mean ± standard deviation [SD], n = 3). The authors could not
make any suggestion as to the origin of this discrepancy. Absorbed doses in mice were 0.85 ±
0.08 |iCi immediately after exposure and 0.77 ± 0.16 |iCi for animals placed in metabolism
cages. Elimination values detected in these rats and mice are shown in parentheses in Table 3-3;
the percentage values shown in this table were based on the total body burden of the individual
animals from which the elimination data were obtained, not on group means.
Mice had eliminated most of the dose within 12 hours after exposure, rats within
24 hours. Organic volatiles collected on charcoal filters were analyzed for ETBE and TBA
contents. Rats exhaled 22% of the absorbed ETBE within 1 hour after exposure, 12% during the
following 2 hours, and only another 3% during the next 3 hours. TBA exhalation accounted for
1% of the total during the first hour, 3% during the following 2 hours, and 4% during the last
3 hours of the experimental period. Mice, on the other hand, exhaled 16% of the unmetabolized
ETBE within 1 hour after exposure and 1% during the following 2 hours, with immeasurable
amounts thereafter. TBA exhalation made up 6% of total during the first hour, 8% in the next
2 hours, and 4% during the final 3 hours. Elimination of ETBE, TBA, HBA, and MPD in urine
were assayed. During 24 hours of collection, rats eliminated about 7 times as much TBA as
ETBE in urine; in mice, the ratio was >60. HBA was detected in urine of both species, but could
not be quantified. MPD was not detected. These results may be interpreted as suggesting that
mice metabolize, and hence, eliminate ETBE faster than rats.
07/14/2009
21
DRAFT - DO NOT CITE OR QUOTE
-------
3.5. PHYSIOLOGICALLY BASED TOXICOKINETIC MODELS
A physiologically based toxicokinetic (PBTK) model of ETBE for humans has been
developed (Nihlen and Johanson, 1999), but models are not available for any other species.
Although physiologically based pharmacokinetic (PBPK) models for a structurally related
substance, MTBE, exist that may allow for interspecies extrapolation of dosimetry between
rodents and humans (Borghoff et al., 1996), and for predictions of internal dosimetry in humans
exposed to MTBE in potable water (Rao and Ginsberg, 1997), no models are available for
ETBE. In the Borghoff et al. (1996) model, the MTBE metabolic parameters (Vmax and Km)
were estimated from gas uptake data by retrofitting only the parent compound module. Rao and
Ginsberg (1997) updated the model by fitting these parameters simultaneously with both
modules for MTBE and its metabolite, TBA, against concentrations of both compounds in blood.
The utility of this MTBE model in risk assessment of MTBE was evaluated by performing cross-
species extrapolations of internal dosimetry of the parent compound and TBA, and relating the
animal acute toxicity data to predictions of internal doses in human brain after simulated bathing
and showering in MTBE-contaminated water (Rao and Ginsberg, 1997).
The PBTK model of Nihlen and Johanson (1999) addresses human inhalation exposure
only and describes the pharmacokinetics of ETBE and its main metabolite, TBA, in lungs, liver,
fat, rapidly perfused tissues, and resting and working muscles (Figure 3-2). The authors assumed
that ETBE is metabolized in the liver by a first-order process, and that TBA (the metabolite) is
excreted in the urine also by a first order elimination process. This perfusion-limited model
differs in several ways from conventional PBPK models that usually follow an anatomically
representative, typical description (e.g., Andersen, 1991), as introduced by Ramsey and
Andersen (1984). Thus, in the Nihlen and Johanson (1999) PBTK model, tissue volumes and
blood flows were calculated from individual data on body weight and height, and, moreover, the
blood flows are expressed as functions of physical activity (oxygen uptake above rest, VOD, in
L/minute, linked in turn by an empirical function with workload, W, in watts), unlike the
conventional PBPK models in which tissue volumes, blood flows, and Michaelis-Menten kinetic
constants can be scaled with allometric adjustments solely to body weight. Compartments for
slowly perfused tissues and gastrointestinal tract are not included in this PBTK model, and the
liver perfusion is described as a single blood flow, Qh (in L/minute), without splitting into
arterial and portal circulations. Free fat mass (in kg) and lean body volume (in L) are expressed
in terms of total body water (in L), which in turn is linked, by an empirical function, with body
height (in m) and body weight (in kg). Such a model structure precludes allometric scaling of
variables.
07/14/2009
22
DRAFT - DO NOT CITE OR QUOTE
-------
ETBE Compartments
TBA Compartments
Qalv
Cair-ETBE
.Qalv
Calv-ETBE
Qalv
Cair-TBA
.Qalv
Calv-TBA
Qco
Qco
CLiT
Qwm
Qwm
Qrm
Qrm
Qh
Qh
CLiE
CLiM
Urine
Resting Muscles
Fat
Fat
Resting Muscles
Liver
Liver
Working Muscles
Working Muscles
Rapidly Perfused Tissues
Rapidly Perfused Tissues
Lungs and Arterial Blood
Lungs and Arterial Blood
Symbols of parameters and variables
Qalv - alveolar ventilation (L/min)
Cair-ETBE - concentration of ETBE in ambient air (microM)
Calv-ETBE - concentration of ETBE in alveolar air (microM)
CairTBA - concentration of TBA in ambient air (microM)
CalvTBA - concentration of TBA in alveolar air (microM)
CLiT - clearance of TBA to urine (L/min)
kel - first order excretion rate constant of TBA (l/min)
Qco - Cardiac output (L/min)
Qr - Blood flow to rapidly perfused compartment (L/min)
Qf - Blood flow to fat compartment (L/min)
Qwm - Blood flow to working muscle compartment (L/min)
Qrm - Blood flow to resting muscle compartment (L/min)
Qh - Blood flow to liver compartment (L/min)
CLiE - Intrinsic hepatic clearance of ETBE (L/min)
CLiM - Intrinsic hepatic clearance of TBA (L/min)
Source: Modified from Nihlen and Johansen (1999).
Figure 3-2. Structure of the PBTK model for ETBE and TBA.
07/14/2009
23
DRAFT - DO NOT CITE OR QUOTE
-------
While two partition coefficients for ETBE—water:air and blood:air—were measured in
vitro (Nihlen et al., 1995), the tissue:blood partitioning was calculated for each tissue based on
its water and lipid contents. The metabolism and excretion clearances and elimination rate
constant were also estimated individually by fitting the model to the experimental data from
eight volunteers, exposed to 5, 25, and 50 ppm of ETBE in air (Nihlen et al., 1998a). The same
individual data were used in the PBTK model validation.
Although some limited pharmacokinetic data from rodents exposed to ETBE are
available in the literature (Dekant et al., 2001a, b; Borghoff, 1996, unpublished report), the
human PBTK model from Nihlen and Johanson (1999) cannot be used for rodents. The
structural simplifications of this human PBTK model, single route of exposure, and the same
limited data sets used to calibrate and validate the model, limit its potential for application in
human health risk assessment. Therefore, at this time, sufficient information is not available to
allow interspecies extrapolation of ETBE dosimetry between rodents and humans, or to apply the
existing PBPK models for MTBE to the case of ETBE.
07/14/2009
24
DRAFT - DO NOT CITE OR QUOTE
-------
4. HAZARD IDENTIFICATION
4.1. STUDIES IN HUMANS - EPIDEMIOLY, CASE REPORTS, CLINICAL
CONTROLS
No epidemiologic studies in humans or case reports of accidental exposure to ETBE have
been reported.
4.1.1. Studies in Humans
Nihlen et al. (1998b) exposed eight healthy male volunteers (range 21-41 years, mean
body weight 82 kg) to ETBE vapor for 2 hours during light exercise on a bicycle ergometer. The
study was carried out according to the Declaration of Helsinki after approval by the Regional
Ethical Committee of the institution where the study was performed, and written informed
consent by the volunteers. The ETBE vapor was generated at four nominal levels (0, 5, 25, and
50 ppm) in a 20 m3 exposure chamber with a controlled climate (average temperature 19°C, 43%
humidity, 16 air exchanges per hour). Each subject was exposed at least once to each
concentration, with a 2-week interval between exposures. Measurements of ocular, nasal, and
pulmonary physiological function were conducted prior to exposures, during the exposures, and
afterward. In addition, the subjects rated symptoms of irritation, discomfort, and central nervous
system (CNS) effects in a questionnaire. Significantly, dose-related ratings of solvent smell
(recorded on a 100-mm visual analog scale graded from "not at all" to "almost unbearable";
p = 0.001, repeated measures analysis of variance) occurred as volunteers entered the chamber
containing ETBE vapor. However, the ratings declined slowly with time during exposure and
after the exposure period had been completed. Significantly elevated ratings of discomfort in the
throat and airways were reported during and after exposure to 50 ppm ETBE in comparison to
exposure to clean air, but ratings at lower concentrations, although somewhat higher than control
values, did not differ significantly from control responses. Questions on discomfort in the eyes,
fatigue, nausea, dizziness, and intoxication had the highest average ratings at the 50 ppm
exposure level. However, no exposure concentration vs. effect correlation was seen, and none of
the ratings differed significantly from the clean air ratings. No significant acute effects of ETBE
were seen regarding eye redness, measured or reported tear-film breakup time, or conjunctival
epithelial damage. Increases in eye-blinking frequency of 10-14 blinks per minute by as much
as 50% (p = 0.01) were reported. Increased nasal swelling (p = 0.001, compared with pre-
exposure values) was indicated by a 6—15% decrease in nasal volume using acoustic rhinometry.
Analysis of nasal lavage fluid for total cells and markers of inflammation (albumin, lysozyme,
eosinophilic cationic protein, myeloperoxidase, interleukin 8) showed some sporadic changes,
but these were not related to exposure levels (p > 0.05). Slightly impaired pulmonary function
(vital capacity -3.2, -3.4%, and forced vital capacity -3.6, -4.4% at 25 and 50 ppm, respectively)
was observed compared with values measured 35-50 minute after exposures. Single breath
07/14/2009 25 DRAFT - DO NOT CITE OR QUOTE
-------
carbon monoxide diffusing capacity was reduced with borderline significance after exposure to
25 ppm ETBE, but not to 50 ppm. Some individuals reported a "bad taste in their mouth" after
exposure to 25 and 50 ppm. Thus, healthy subjects exposed for 2 hours to 25 or 50 ppm ETBE
experienced irritation in throat and airways, nasal swelling, a bad taste in the mouth, and slightly
impaired lung function. However, the low number of experimental subjects reduced the
statistical power of this study. Amberg et al. (2000) and Bernauer et al. (1998) also conducted
studies in humans, presented in detail in Chapter 3, but these studies focused on metabolism of
ETBE in humans, not on health effects.
Vetrano (1993, unpublished report) evaluated odor and taste thresholds for ETBE (99.0-
99.5% purity) and MTBE (99.9% purity). Using six or seven subjects (six women and
sometimes a single man), the average calculated detection and recognition threshold values for
aerosolized ETBE were determined. Test agent samples (0.6 |j,L) were vaporized in
hydrocarbon-free air in a Tedlar® bag and subsequently diluted; airborne concentrations were
confirmed by gas chromatographic analysis. Each airborne sample was presented to each test
panel member using a dynamic triangle olfactometer operated according to American Society for
Testing and Materials Standard Practice E 679. Each concentration was presented 3 times for
evaluation. Variability of panel responses was tabulated in the report, but was not summarized.
Both the detection threshold, defined as the minimum airborne concentration at which 50% of
the test subjects could differentiate between a test sample and odor free air, and the recognition
threshold, defined as the minimum concentration at which 50% of subjects recognized or
identified the odorant, were determined. For ETBE, detection and recognition thresholds were
0.013 and 0.024 ppm, respectively, and for MTBE they were 0.053 and 0.08 ppm, respectively.
Thus, ETBE was detected at approximately a fourfold lower concentration as an airborne
olfactory stimulus than MTBE. In other experiments, odor threshold values for various
concentrations of MTBE and ETBE in water were measured. The average odor detection and
recognition limits for ETBE in water were determined to be 0.049 and 0.106 ppm, respectively.
Average detection and recognition limits for MTBE in water were determined to be 0.095 and
0.193 ppm, respectively; therefore, ETBE was detected at approximately twice as low of a
concentration as MTBE in water. Substances with odor thresholds of less than 1 ppm are
generally categorized as highly odorous. Finally, taste detection thresholds for ETBE and
MTBE in water were measured. The average taste detection threshold values for ETBE and
MTBE in drinking water were reported to be 0.047 and 0.134 ppm, respectively. Thus, ETBE
was detected at approximately a threefold lower concentration than MTBE. The taste of both
oxygenates was described as highly objectionable.
In an investigation conducted on behalf of the American Petroleum Institute, TRC
Environmental Corp. (1993) repeated work for MTBE and examined the effects of various
oxygenate additions on the odor of gasoline blends using methods identical to those described by
Vetrano (1993, unpublished report). Odor detection threshold and odor recognition values for
07/14/2009 26 DRAFT - DO NOT CITE OR QUOTE
-------
97% pure MTBE in air were determined to be 0.053 and 0.125 ppm, similar to previously
reported values (Vetrano, 1993, unpublished report). Odor detection threshold and odor
recognition values for MTBE in water were 0.045 and 0.055 ppm, respectively, considerably
less than earlier reported values. The taste threshold for 97% MTBE in water was found to be
0.039 ppm, 3 times lower than the value reported in Vetrano (1993, unpublished report). The
taste of MTBE was described as bitter, nauseating, like rubbing alcohol, etc. In addition to
studies with ETBE and MTBE alone, the oxygenates MTBE, ETBE, and TAME were added to
one or more of three gasoline blends to evaluate their effects on gasoline odor detection and
recognition thresholds. When the summer blend of gasoline was mixed with 15% ETBE (99%
purity), the average odor detection and recognition threshold values were 0.064 ppm and
0.139 ppm, respectively, a reduction of 89% compared with the odor threshold for the summer
blend of gasoline alone.
Durand and Dietrich (2007) evaluated odor threshold and intensity of ETBE standards as
part of a study measuring ETBE leaching from a PEX pipe using a utility quick test designed to
assess taste, odor, migration, and leaching of materials in water distribution systems. ETBE was
observed leaching into tap water with and without the addition of free chlorine or
monochloramine using solid phase microextration/GS-MS. Aqueous concentrations of ETBE in
the leachate ranged from 23>140 [j.g/L and decreased with increased flushing. A team of
10 panelists were recruited and trained for several weeks in flavor profile analysis in a research
protocol approved according to the standards of the Virginia Tech Institutional Review Board for
human subjects. Four to six panel members were present for all tests. ETBE for quantification
was purchased from Chem Service, Inc. (purity not reported). Panelists were asked to identify
and describe the odor of known concentrations (5-50 |ig/L) of ETBE in experimental tap water
alone or with 2 mg/L Cb or 4 mg/L NH2C1 as CI2. Panelists reported a chemical or solvent odor
and a burning sensation during the flavor profile analysis of ETBE samples that was experienced
by most panelists in the absence or presence of chlorine disinfectants. The ability of panel
members to detect ETBE was reduced in the presence of chlorinous odor from free chlorine, but
not monochloramine. Panelists were able to smell ETBE at a concentration of 5 |ig/L, the lowest
concentration tested (Durand and Dietrich, 2007).
4.2. SUBCHRONIC AND CHRONIC STUDIES AND CANCER BIOASSAYS IN
ANIMALS—ORAL AND INHALATION
4.2.1. Subchronic Studies—Oral
No subchronic studies of oral exposure to ETBE were found in the literature.
4.2.2. Chronic Studies—Oral
As part of series of carcinogenicity studies, Maltoni et al. (1999) carried out chronic
exposure studies of petroleum components and additives, including ETBE (BT959). The authors
07/14/2009 27 DRAFT - DO NOT CITE OR QUOTE
-------
indicated that the publication was preliminary in nature; however, no further explanation was
given to this characterization despite the fact that as of 2008, this was still the only publication of
these data. Male and female CRC/RF Sprague-Dawley rats, 60/sex/group, were dosed by gavage
with ETBE (purity >94%) at dose levels of 0, 250, and 1,000 mg/kg for 104 weeks. Impurities
in the test solution included ethyl alcohol (2.88%), TBA (1.59%), MTBE (1%), 2-ethoxy-butane
(0.12%>), olefin C8 (0.11%>), ter-butyl-isopropyl ether (0.09%>), and methyl alcohol (0.01%>).
ETBE was administered in olive oil 4 days/week (Monday, Tuesday, Thursday, and Friday).
The study authors state that administration of a dose of 1,000 mg/kg for 5 or more days/week
would not have been tolerated, suggesting that the maximum tolerated dose may have been
exceeded under standard dosing regimes. Rats were 8 weeks of age at study initiation; they were
weighed and examined for gross lesions weekly for the first 13 weeks, then biweekly for the
remainder of the bioassay. No major effects of dosing on food and water intake or body weights
were observed.
Starting at week 40, a dose-related increase in mortality for both male and female rats
was reported. The authors present two survival curves (Figures 7 and 8 from Maltoni et al.,
1999) and state that "there was a dose-correlated increase of [sic] the mortality rate in males
starting from the 40th week until the end of the experiment... and in females from the 40th to the
88th week of the biophase." Other than this statement, the authors did not identify a lowest-
observed-adverse-effect level (LOAEL) for mortality, did not indicate if the increased mortality
was statistically significant, and only presented the data graphically. Attempts to digitize the
data from the survival curves support the increased mortality of high-dose females relative to
control animals from weeks 56 to 88 by approximately 8-20% with increased survival relative to
control females from weeks 104 to 136 by approximately 4-1%. Mortality of the low-dose
females was similar to the controls (± 2-3%) with the exception that at 88 weeks of age,
mortality was approximately 11% higher than controls and at 120 weeks of age, mortality was
approximately 5% lower than controls. Mortality in the high dose males was approximately 8-
30%o higher than controls from weeks 56 to 120, whereas the low-dose males exhibited higher
mortality by approximately 7—17% from weeks 56 to 88 and were similar to controls at other
time points. Using the digitized numbers from the survival curves in Maltoni et al. (1999),
ETBE-exposure is associated with increased mortality relative to controls in animals of both
sexes 56 to 88 weeks of age and increased survival relative to controls in females >104 weeks of
age. At 56 weeks of age, 250 mg/kg is the LOAEL in males for a 7% increase in mortality and
1,000 mg/kg is the LOAEL in females for an 8%> increase in mortality. At 104 weeks in age,
1,000 mg/kg is the LOAEL in males for a 13%> increase in mortality and ETBE-exposure in
females is associated with decreased mortality. The treated-related effects on mortality are
relative to considerable mortality in control animals. Less than approximately 30% of control
rats of either sex remained alive by 104 weeks of age. No explanation was provided for the low
survival rate displayed among control animals in the study. Historical control data on mortality
07/14/2009 28 DRAFT - DO NOT CITE OR QUOTE
-------
were not provided by the authors. Average mortality data observed in gavage studies in female
Sprague-Dawley rats are available from the National Toxicology Program (NTP, 2005) and are
provided here for purposes of comparison. Survival of female rats in the NTP studies treated
with corn-oil gavage averaged 42% by 104 weeks (range 28-47%), suggesting that the animals
from the Maltoni et al. (1999) study with olive oil gavage were at the low end of survival based
on the limited NTP data set. Rats in the Maltoni et al. (1999) study were allowed to live out their
natural lives or until week 137, when the study was terminated. Upon death, rats were
necropsied and tissue/organs were taken for microscopic examination.
The incidence and multiplicity of tumors were determined in all treatment groups
(Table 4-1). The data were presented according to dose group, tumor site, and histiotype. In
addition, the data were tabulated as incidence and number of benign tumors/treatment group and
incidence and number of malignant tumors/treatment group. Some grouped totals included only
tumors (e.g., total malignant tumors of the uterus), whereas others included precancers and
tumors (e.g., defined by the authors as pathologies of oncological interest of the mouth
epithelium and forestomach included acanthomas, dysplasias, and carcinomas). Total benign
and malignant tumor incidences at both doses were comparable with the control group in both
sexes. The total number of malignant tumors per 100 animals was significantly greater in female
rats dosed with 250 mg/kg-day ETBE (55 per 100) than the total number of malignant tumors per
100 animals in female control rats (25 per 100; p < 0.05).
07/14/2009
29
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1. Tumor incidences resulting from 2-year gavage exposure of
Sprague-Dawley rats to ETBE
Tumor formation
Administered dose (mg/kg-day)
0
250
1,000
0
250
1,000
Males
Females
Benign (total)
40/60
40/60
32/60
50/60
53/60
49/60
Malignant (total)
11/60
14/60
14/60
9/60
21/60
19/60
Mouth epithelium (total)
6/60
14/60
15/60a
14/60
16/60
18/60
acanthomas
1/60
0/60
2/60
1/60
2/60
3/60
squamous cell dysplasias
5/60
14/60
11/60
11/60
11/60
12/60
squamous cell dysplasias with in situ
carcinoma
0/60
0/60
1/60
2/60
1/60
2/60
squamous cell carcinomas
0/60
0/60
1/60
0/60
2/60
1/60
Forestomach (total)
13/60
24/60
13/60
12/60
10/60
11/60
acanthomas
5/60
7/60
4/60
5/60
3/60
6/60
squamous cell dysplasias
8/60
14/60
9/60
7/60
4/60
5/60
squamous cell carcinomas
0/60
3/60
0/60
0/60
3/60
0/60
Uterus (total malignant)
2/60
10/60^
2/60
carcinomas
1/60
im
0/60
sarcomas
1/60
8/60°
2/60
Hemolymphoreticular (total)
3/60
8/60
6/60
3/60
6/60
5/60
lymphoblastic lymphoma
0/60
1/60
0/60
0/60
1/60
0/60
lymphocytic lymphoma
0/60
0/60
0/60
0/60
1/60
0/60
lymphoimmunoblastic lymphoma
2/60
4/60
5/60
1/60
3/60
2/60
histiocytic sarcoma
1/60
1/60
1/60
2/60
0/60
2/60
myeloid leukaemia
0/60
2/60
0/60
0/60
1/60
0/60
Statistically significant (p < 0.05), as calculated by the authors.
bNot statistically significant by yl test (the same test used by the authors on the grouped data) when the vaginal
schwannomas are removed from the "uterus".
Identified as including four malignant schwannomas of the uterus-vagina
Source: Maltoni et al. (1999).
Two significantly increased tumor types or combined tumors and precancers were
reported. In each case, only the total tumors or combined tumors and precancers were
significantly increased and there was no significant increase in individual histiotypes for tumors
or precancers. In the first example, the incidence of total pathologies of oncological interest,
which includes tumors and precancers (i.e., acanthomas, squamous cell dysplasia, squamous cell
dysplasia borderline with carcinoma in situ, and squamous cell carcinoma of oral cavity, tongue,
and lips) was significantly increased in male rats of the 1,000 mg/kg group (25 vs. 10% in the
male control group; p < 0.05). In the second example, total malignant tumors of the uterus
(carcinomas and sarcomas) were noted in 250 mg/kg females (16.7% incidence,/* < 0.05 that
07/14/2009 30 DRAFT - DO NOT CITE OR QUOTE
-------
appears to also include vaginal tumors), but uterine malignancies were diagnosed in only 3.3% of
females in the 1,000 mg/kg group and in 3.3% of females in the control group.
Two examples of nonstatistically increased tumors were also reported. There was a
nonsignificant increase in the incidence of total pathologies of oncological interest in the
forestomach for males in the 250 mg/kg-day group (24/60) but not females (10/60) or animals of
either sex at the high dose (13/60 male and 11/60 female), which is similar to controls
(13/60 males and 12/60 females). In addition, hemolymphoreticular neoplasias (lymphomas,
sarcomas, and leukemias) and, in particular, lymphoimmunoblastic lymphomas (8.3% incidence
in the 1,000 mg/kg males), were increased overall in the male (3/60, 8/60, and 6/60 at 0, 250, and
1,000 mg/kg-day respectively) and female (3/60, 6/60, and 5/60) treatment groups, but none of
the increases noted were statistically significant. The authors attributed the lack of adequate
dose-response to the relatively high mortality in the treatment groups. However, a survival
analyses was not presented, nor were data provided on tumor incidence for individual animals by
week or month that would allow for a time-to-tumor analysis. A summary of total tumor and
precancer incidences is given in Table 4-1.
The number of acanthomas of the mouth epithelium in both males (2/60) and females
(3/60) of the 1,000 mg/kg group was about twice that in corresponding control rats (1/60).
Squamous cell carcinoma of the forestomach occurred in 5% of both male and female rats in the
250 mg/kg group (not statistically significant), but no squamous cell carcinomas of the
forestomach were found in either sex of the control group or in the 1,000 mg/kg treatment group.
The authors identified three limitations of the study, including the use of only two dose
groups and a single animal species and increased mortality of the animals exposed to ETBE.
However, the authors concluded that in spite of the study limitations, the study results show a
statistically significant increased incidence of total pathologies of oncological interest of the
mouth epithelium in males and total malignant tumors of the uterus in females and total
malignant tumors (which were only increased in females at the low dose when tabulated as the
number of tumors per 100 animals), and a nonsignificant increase in total pathologies of
oncological interest of the forestomach in males and of hemolymphoreticular neoplasias.
4.2.3. Subchronic Studies—Inhalation
4.2.3.1. Subchronic Inhalation Studies—Rats
In subchronic inhalation experiments, ETBE was administered as a vapor at target
chamber concentrations of 0 (filtered air control), 500, 2,000, and 4,000 ppm (mean analytical
concentrations: 0, 501, 2,090, and 3,910 ppm) to groups of Sprague-Dawley rats. The rats
(10/sex/group and 9 weeks of age at initial exposure) were exposed for 6 hours/day,
5 days/week, for 4 weeks (published as White et al., 1995; IIT Research Institute, 1991,
unpublished report). The rats were observed daily, and salivation and redness around the
nose/mouth/face were occasionally reported for test animals during exposures. A functional
07/14/2009
31
DRAFT - DO NOT CITE OR QUOTE
-------
observation battery (FOB) was administered 1 week prior to the exposures and about 60 minutes
after 1, 5, or 20 exposures to evaluate neuromuscular function and sensory perception (described
in detail in Section 4.4.1). Ataxia and sedation, which are overt signs of CNS depression, were
seen following exposure termination in the 4,000 ppm group, but ETBE-exposed rats appeared
normal within 15 minutes of the end of exposure. Mean body temperature was reduced 2.00-
2.14% in 4,000 ppm males after the fifth exposure, and a trend for increased hind limb splay in
both sexes of the high concentration group occurred. These effects were described by the
authors as being associated with transient CNS depression. No other indications of CNS
depression or neurotoxicity were detected. No premature mortality occurred, and no statistically
significant effect of treatments on weekly body weights was observed. Necropsy was performed
18 hours after the final exposure to ETBE, at which time blood for serum chemistry and
hematology was taken, and the following tissues were weighed and prepared for histological
examination: brain, adrenal glands, gonads, heart, kidneys, liver, lungs, and spleen.
Approximately 31 additional tissues were also collected and prepared for histological
comparison between the high-dose group (4,000 ppm) and controls. At termination, no
significant effects of ETBE exposure were seen in serum chemistry (liver function enzymes
[creatine kinase, alanine aminotransferase, asparatate aminotransferase, and alkaline
phosphatase], electrolytes [sodium, potassium, and chloride], glucose, triglycerides, cholesterol,
creatinine, blood urea nitrogen, total serum protein, and albumin) or on hematology evaluations
(red cell count, hemoglobin, mean corpuscular volume [MCV], total and differential leukocyte
counts, and platelet count). The only exception was a significant increase in white blood cell
count (leukocytes) in females exposed to 2,000 and 4,000 ppm ETBE. This finding was noted to
be of questionable toxicological significance because it was not accompanied by changes in
histopathology. In rats exposed to 4,000 ppm, absolute and relative liver weights in males at
termination were increased 16.8 and 16.1%, respectively; absolute and relative liver weights in
females were increased 9.5 and 12.5%, respectively. Relative liver weights were also increased
10%) in female rats exposed to 2,000 ppm ETBE. In addition, absolute kidney and adrenal
weights were increased 12.8 and 13.7%, respectively, in male rats exposed to 4,000 ppm of
ETBE. No observations attributed to ETBE exposures were recorded at necropsy or upon
histological examination of any tissues from the high-dose animals, including gonads, adrenal
glands, kidneys, and liver.
Subchronic 13-week ETBE inhalation studies using both rats and mice were conducted
by the Chemical Industry Institute of Toxicology (Medinsky et al., 2006 [erratum]; Medinsky et
al., 1999; Bond et al., 1996a, b, unpublished reports). Male and female F344 rats (6.5 weeks old)
and male and female CD-I mice (7.5 weeks old; described in Section 4.2.3.2 below) were
exposed in whole-body chambers to 0 (control), 500, 1,750, or 5,000 ppm ETBE (97.5% pure)
for 6 hours/day, 5 days/week, for 13 weeks. For each exposure level group of F344 rats, the total
number of 48 rats/sex was subdivided into a series of subgroups: a basic core subgroup
07/14/2009
32
DRAFT - DO NOT CITE OR QUOTE
-------
(11 rats/sex), a neurotoxicology subgroup (12 rats/sex), an interim clinical pathology (chemistry
and hematology) subgroup (10 rats/sex), and a cell replication subgroup (15 rats/sex).
Special attention was given to the assessment of effects in rat kidneys and mouse livers,
including serum enzyme assays, on the basis of effects previously noted in chronic oncogenicity
studies with MTBE (reviewed in Ahmed, 2001; Cal EPA, 1999; Mennear, 1997). In addition,
studies of cell turnover and induced cell proliferation were evaluated in kidneys from male and
female rats and livers from male and female mice (five animals/sex/group), using 5-bromo-
2'-deoxyuridine (BrdU) labeling after 1, 4, and 13 weeks of exposure. A wide range of clinical
chemistry and hematological parameters was evaluated after 6 and 13 weeks in rats, but only
after 13 weeks in mice. At termination, a broad suite of tissues and organs was examined in
control and high-concentration animals; potential target organs (lungs, liver, kidneys) and gross
lesions were examined in all groups. Mallory-Heidenhain staining for possible accumulation of
hyaline droplets in renal tubules, and immunohistochemical staining for renal tubular alpha2U-
globulin were conducted on thin sections of rat kidneys. In addition, testicular seminiferous
tubules of male rats from all treatment groups were analyzed for degenerative changes.
Significant findings in rats that are related to ETBE exposure are summarized in Table 4-2 (see
Table 4-6 in Section 4.2.3.2 for the summary of significant findings in mice).
07/14/2009
33
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-2. Summary of significant results from the 13-week subchronic ETBE inhalation study in F344 rats
Endpoint
Measurement frequency
Observations
Basic core subgroup"
Mortality
Twice daily
No exposure-related mortality
Growth rate
Weekly
Females: increased, 5,000 ppm only
Clinical signs
Weekly
Males: ataxia postexposure, 5,000 ppm
Organ weights
Study termination
Males: increased kidney, liver weights at 1,750, 5,000 ppm; increased adrenal weights at 5,000 ppm
Females: increased adrenal, liver weights at 5,000 ppm; increased kidney weights at 1,750,
5,000 ppm; increased heart weights at 500, 5,000 ppm
Gross pathology
Study termination
Not exposure-related
Histopathology
Study termination
Males: renal effects at 500, 1,750, 5,000 ppm; increased percentage of seminiferous tubules with
degeneration of spermatocytes at 1,750, 5,000 ppm
Females: bone marrow congestion at 1,750, 5,000 ppm
Clinical pathology subgroup"
Hematology
Interim (6 wks)
Males: increased platelets and decreased mean corpuscular hemoglobin concentration (MCHC) and
increased MCV at 5,000 ppm
Females: decreased white blood cells and increased13 MCV at 5,000 ppm
Study termination
Males: decreased MCHC at 1,750, 5,000 ppm; increased platelets at 5,000 ppm
Females: increased MCV at 1,750, 5,000 ppm
Clinical chemistry
Interim (6 wks)
Males: increased creatinine, total protein and albumin; decreased chloride and sodium at 5,000 ppm
Females: decreased bilirubin and phosphorus at 5,000 ppm
Study termination
Males: decreased chloride at 1,750, 5,000 ppm; increased total protein 5,000 ppm
Females: decreased bilirubin at 500 ppm
Cell replication subgroupc
Renal labeling index
Wks 1, 4, and 13
Males: increased LI wks 1, 4, 13 at 5,000 ppm; 13 wks at 500, 1,750 ppm
(LI)
Females: increased LI wk 1 at 500, 1,750, 5,000 ppm; wk 4 at 5,000 ppm; no effect at 13 wks
Neurotoxicology subgrouph'd
FOB, neuropathology
18 hrs after exposure days
1,6, 10, 20, 42, 65
No significant findings
"10 animals/sex/exposure group
bMCV is listed as decreased in table 1 from Medinsky et al. (1999), but increased in the text. Note: Table 1 from Medinsky et al. (1999) contain errors that can
be resolved by reading text and referring to Bond et al. (1996a, unpublished report).
°15 animals/sex/exposure group
d12 animals/sex/exposure group
Sources: Medinsky et al. (2006, 1999); Dorman et al. (1997); Bond et al. (1996a, unpublished report).
07/14/2009
34
DRAFT - DO NOT CITE OR QUOTE
-------
Mean analytical concentrations for the ETBE exposures were 0, 505 ± 13, 1,748 ± 59,
and 4,971 ±155 ppm for rats. Transient ataxia (lasting <1 hour) was sometimes observed in
male rats exposed to 5,000 ppm shortly following daily exposure, and decreases (-25%) in body
weight gain were observed during the first week of exposure in male and female rats exposed to
1,750 and 5,000 ppm ETBE. At termination, body weights of female rats in the 5,000 ppm
group were significantly higher than controls, but body weights of other groups, both male and
female, did not differ significantly from those of controls. Significant increases in absolute mean
kidney and liver weights occurred in male rats exposed to 1,750 and 5,000 ppm ETBE compared
to controls (9.8 and 19.3% increases for kidneys; 14.2 and 32.4% increases for liver), and mean
adrenal weights were significantly increased 34.3% in male rats at 5,000 ppm. Significantly
increased absolute weights of kidney (21.3%), adrenal (17.8%), and liver (25.8%) were also
noted in female rats exposed to 5,000 ppm ETBE, and increased kidney weight (12.2%) was
noted in female rats exposed to 1,750 ppm ETBE. Increased heart weight was found in female
rats exposed to 500 ppm (10.1%) and 5,000 ppm (12.3%) ETBE, but not 1,750 ppm. No
significant findings were noted in the histopathology of adrenal glands or livers of high-dose
animals.
Slight but statistically significant increases in various clinical chemistry parameters
(Table 4-2) were seen, but these effects were reported to be of uncertain toxicological
significance. At both the interim and 13-week time points, 3—6% decreases in levels of serum
chloride and 9-11% increases in total protein were observed in male rats exposed to 5,000 ppm
ETBE. No consistent changes in clinical chemistry parameters compared to controls were seen
in female rats exposed to ETBE. The changes seen in peripheral hematology parameters
(Table 4-2) were not considered to be clinically significant by the authors. There was a 1%
increase1 in the MCV in female rats exposed to 5,000 ppm ETBE (58.33 ± 0.31 compared to
57.62 ± 0.40 in controls at 6 weeks) at the interim sampling point and to females exposed to both
1,750 (57.97 ± 0.49) and 5,000 ppm (58.11 ± 0.60 compared to 57.26 ± 0.73 in controls) at the
13-week time point. There was a 4% decrease in mean corpuscular hemoglobin concentration
(MCHC) in male rats exposed to 5,000 ppm (35.13 ± 0.38 compared to 36.46 ± 1.09 in controls)
ETBE at the interim sampling point and a 2.5% decrease in males exposed to both 1,750
(35.00 ± 0.30) and 5,000 ppm (35.01 ± 0.65 compared to 35.91 ± 0.67 in controls) at the
13-week time point.
Kidneys and testes of males and femoral bone marrow of females were the only tissues
with significant histological findings in ETBE-exposed rats. Renal effects occurred in male rats
exposed to ETBE and consisted of a higher incidence and mean number of regenerative foci and
increased hyaline droplet severity in all three ETBE exposure groups (Tables 4-3 and 4-4).
1 Note: Table 1 from Medinsky et al. (1999) contains discrepancies that are not consistent with the text (e.g., MCV
is listed as decreased in female rats at 6 weeks in the table and increased in the text). The text is consistent with the
unpublished study results and data tables from the unpublished report (Bond et al., 1996a, unpublished report).
07/14/2009 35 DRAFT - DO NOT CITE OR QUOTE
-------
Regenerative foci are associated with repair of damaged tubules. Therefore, the presence of
regenerative foci represents indirect evidence of necrosis, as some cells of the proximal tubule
would have had to have died and the remaining cells undergone regenerative proliferation for
regenerative foci to be observed. The increased incidence and mean number of regenerative foci
was exposure- and time-related, but the appearance of hyaline droplets in kidney sections stained
with Mallory-Heidenhain stain was only exposure-related. No regenerative foci were observed
in any treatment group after 1 week of exposure, or in control or 500 ppm treatment groups after
4 weeks of exposure. The mean number of regenerative foci was 2 and 4 in the 1,750 and
5,000 ppm dose groups, respectively, after 4 weeks of exposure. After 13 weeks of exposure
there was a dose-response relationship in the mean number of regenerative foci observed with 2,
11, 17, and 34 foci for 0, 500, 1,750, and 5,000 ppm ETBE respectively. Although Medinsky et
al. (1999) presented hyaline droplet severity (expressed as a mean grade scored with 1 for
minimal, 2 for less than 10%, 3 for 10-25%, 4 for 25-50%, and 5 for greater than 50% of the
cortex involved) as concentration-dependent (Table 4-4), there were no statistics presented.
Alpha2u-globulin immunoreactivity was observed in the hyaline droplets of the renal proximal
tubular epithelium of male rats from all exposure groups. Medinsky et al. (1999) stated that
accumulation of alpha2U-globulin-containing droplets was directly related to treatment-dependent
increases in renal effects and the cell proliferation or labeling index (LI). In male rats exposed to
ETBE, after 1 or 4 weeks of exposure, a greater than twofold increase in the LI occurred at the
high dose only (Table 4-5). After 13 weeks of exposure, there was a greater than twofold
increase in the LI for all doses of ETBE relative to control, but there were no differences in LI
between ETBE doses. The authors characterized the differences in LI as time- and ETBE-
concentration-dependent (Medinsky et al., 2006; Medinsky et al., 19992; Bond et al., 1996a,
unpublished report). In female rats exposed to ETBE, after 1 week of exposure, a statistically
significant, but less than twofold increase in the LI occurred at the 500, 1,750, and 5,000 ppm
levels and after exposure to 5,000 ppm for 4 weeks (Medinsky et al., 2006; Medinsky et al.,
1999; Bond et al., 1996a, unpublished report). No effect on the renal LI in female rats was seen
at any exposure concentration after 13 weeks of exposure to ETBE. The higher LI in control rats
at weeks 1 and 4 was characterized as growth-related, and the lower LI at the end of the study
was considered typical of a mature kidney. Nephropathy was not observed in female rats
exposed to ETBE, consistent with the absence of alpha2U-globulin immunoreactivity in F344
females. The increased cell replication, hyaline droplet accumulation, and presence of
alpha2U-globulin immunoreactivity in males led the authors to conclude that the observed renal
effects were due to an alpha2u-globulin mode of action and, therefore, were not relevant for
humans exposed to ETBE through inhalation. An evaluation of human relevance of the alpha2U-
2 Note: Tables 1 and 3 from Medinsky et al. (1999) contain errors (e.g., data from males and females are reversed)
and the reversed data from males and females were resolved in a published erratum (Medinsky et al., 2006).
07/14/2009 36 DRAFT - DO NOT CITE OR QUOTE
-------
globulin accumulation as discussed in the Risk Assessment Forum Technical Panel Report (U.S.
EPA, 1991b) is discussed in Section 4.5.3.
Table 4-3. Incidence of lesions in kidney, seminiferous tubules, and bone
marrow of F344 rats from 13-week subchronic ETBE inhalation study
Exposure group (ppm)
Sex
Tissue
Control
500
1,750
5,000
Male
Kidney
Unaffected
7/11 (64%)
1/11 (9%)
0/11 (0%)
0/11 (0%)
Kidneys with regenerative foci
4/11 (36%)
10/1 la (91%)
i i/i r (100%)
ll/lla (100%)
Testes
Unaffected
0/11 (0%)
0/11 (0%)
0/11 (0%)
1/10 (10%)
Testes with degenerated
spermatocytes
11/11 (100%)
11/11 (100%)
11/11 (100%)
10/11 (91%)
Seminiferous tubules with some
degenerated spermatocytes'3
2.1%
2.4%
7.8%a
12.7%a
Testes with sloughed epithelium
7/11 (64%)
3/11 (27%)
3/11 (27%)
7/11 (64%)
Seminiferous tubules with
lumenal debrisb
2.1%
0.7%
2.8%
1.0%
Female
Kidney
Unaffected
10/10 (100%)
11/11 (100%)
11/11 (100%)
11/11 (100%)
Femoral bone marrow
Unaffected
5/10 (50%)
8/11 (73%)
4/11 (36%)
0/11 (0%)
Congestion
0/10 (0%)
0/11 (0%)
5/1 la (45%)
ll/lla (100%)
Necrosis
5/10 (50%)
3/11 (27%)
3/11 (27%)
2/11 (18%)
aSignificantly different to control (p < 0.05).
bThe incidence of tubules with spermatocyte degeneration or lumenal debris was quantified by counting the number
of affected tubules out of 100 total tubules in a cross-section of testes from each rat.
Sources: Medinsky et al. (1999); Bond et al. (1996a, unpublished report).
07/14/2009
37
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-4. Mean number of regenerative foci and hyaline droplet severity in
kidneys from male F344 rats exposed to ETBE in a 13-week subchronic
inhalation study
Exposure Group
Regenerative foci"
Hyaline droplet severity gradeb
1 wk
4 wks
13 wks
1 wk
4 wks
13 wks
Control
0
0
2
1.2
1.8
1.8
500 ppm ETBE
0
0
11
3.4
2.6
3.0
1,750 ppm ETBE
0
2
17
4.0
3.4
3.2
5,000 ppm ETBE
0
4
34
4.6
3.8
3.8
aMean number of regenerative foci per two kidney sections/rat.
bData expressed as mean grade using the following Mallory Heidenhain scoring scale: 1 for minimal, 2 for <10% of
the cortex involved, 3 for 10-25% of the cortex involved, 4 for 25-50% of the cortex involved, and 5 for >50% of
the cortex involved.
Sources: Medinsky et al. (1999); Bond et al. (1996a, unpublished report).
Table 4-5. Cell division or LI in proximal tubule cells from male and female
F344 rats exposed to ETBE in a 13-week subchronic inhalation study
Exposure Group
LI (%) in males
LI (%) in females
1 wk
4 wks
13 wks
1 wk
4 wks
13 wks
Control
3.52
3.27
0.91
2.65
1.38
0.59
500 ppm ETBE
4.90
4.04
2.16b
4.25a
1.42
1.02
1,750 ppm ETBE
4.34
2.80
3.40°
4.97a
1.59
0.97
5,000 ppm ETBE
7.12b
8.98°
2.47°
3.96a
1.81a
0.87
"Difference from control atp< 0.05.
V<0.01
><0.001
Sources: Medinsky et al. (2006, 1999); Bond et al. (1996a, unpublished report).
A treatment-related increase in the incidence of congestion in the bone marrow of female
rats was observed (Table 4-3), but, in spite of the congestion, the hematopoietic cell population
of bone marrow appeared to be unaffected. No additional description of the observation was
provided by the authors. The presence of bone marrow congestion in the absence of
hematopoietic changes was considered insignificant and of no clinical relevance by the authors.
Degenerated spermatocytes were found in the testes of rats from all treatment groups,
including controls. Male rats exposed to ETBE displayed a dose-related increase in the mean
percentage of testicular seminiferous tubules with spermatocytes that displayed signs of
degeneration (Table 4-3). Significant increases occurred in rats exposed to 1,750 and 5,000 ppm,
with seminiferous tubules containing degenerated spermatocytes observed at 7.8 and 12.7%,
respectively, relative to 2.1% in controls. The occurrence of debris in the lumen of seminiferous
07/14/2009
38
DRAFT - DO NOT CITE OR QUOTE
-------
tubules was not affected by ETBE treatment. A no-observed-adverse-effect level (NOAEL) of
500 ppm was suggested by the authors on the basis of the dose-response data for the percent of
seminiferous tubules with degenerated spermatocytes.
4.2.3.2. Subchronic Inhalation Studies—Mice
Subchronic 13-week ETBE inhalation studies using both rats and mice were conducted
by the Chemical Industry Institute of Toxicology (Medinsky et al., 2006 [erratum]; Medinsky et
al., 1999; Bond et al., 1996a, b, unpublished reports). Male and female F344 rats (6.5 weeks old,
described above in Section 4.2.3.1) and male and female CD-I mice (7.5 weeks old) were
exposed in whole-body chambers to 0 (control), 500, 1,750, or 5,000 ppm ETBE (97.5% pure)
for 6 hours/day, 5 days/week, for 13 weeks. Each exposure group of 40 mice/sex was
subdivided into a basic core and clinical chemistry subgroup (15 mice/sex), a hematology
subgroup (10 mice/sex), and a cell replication subgroup (15 mice/sex).
Mean analytical concentrations for the ETBE exposures to CD-I mice were 0, 501 ± 14,
1,754 ± 50, and 4,962 ± 140 ppm (Medinsky et al., 1999; Bond et al., 1996b, unpublished
report). In male and female mice, transient ataxia was sometimes observed shortly after daily
exposure in 5,000 ppm groups. No ETBE-related alterations in body weights were observed in
male or female mice. However, for both male and female mice, significant increases in absolute
liver weights occurred in the 1,750 and 5,000 ppm groups (13 and 18% for males; 19 and 33%
for females) compared with controls. Other organ weights were comparable to control values. A
dose-related increase in the LI was seen in the livers of male mice exposed to 1,750 and
5,000 ppm after 1 and 4 weeks of exposure, but the LI had returned to control values at
13 weeks. In female mice, a similar dose-related increase was seen after exposure to ETBE for
1 week and 13 weeks to 1,750 ppm and 5,000 ppm; however, effects on liver LI were not
significant at 4 weeks. The increases in the LI were thought to be consistent with a mitogenic
response of the liver to ETBE.
A statistically significant increase in serum total protein (15%) was observed in male
mice exposed to 5,000 ppm ETBE. Statistically significant increases in serum albumin levels
(6%) and total protein (12%) were noted in female mice exposed to 5,000 ppm ETBE as well.
The only significant histopathological lesion in exposed mice was centrilobular hypertrophy in
livers of both male (8/10) and female (9/14) mice exposed to 5,000 ppm ETBE for 13 weeks.
The incidences of centrilobular hypertrophy in males (2/15) and females (1/15) exposed to
1,750 ppm, or in males (0/15) and females (2/15) exposed to 500 ppm ETBE, were not
statistically significant. Minimal hepatocellular necrosis was occasionally reported in all groups,
including controls, and was not exposure-related. A synopsis of results is provided in Table 4-6.
Bond et al. (1996b, unpublished report) suggested a NOAEL of 500 ppm in mice.
4.2.4. Chronic Studies—Inhalation
No chronic studies by the inhalation route were found in the literature.
07/14/2009 39 DRAFT - DO NOT CITE OR QUOTE
-------
4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES—ORAL AND INHALATION
Berger and Horner (2003) assessed in vivo effects of gasoline oxygenates on gamete
quality. Female Simonson rats (a Sprague-Dawley-derived strain), 3-4 rats per test agent, were
treated with 0.3% ETBE, MTBE, TAME, or MPD, a metabolite of MTBE and ETBE, in
drinking water for 2 weeks. Oocytes from the rats were then recovered after injection of donors
with pregnant mare serum gonadotropin and human chorionic gonadotropin to stimulate
ovulation. Sperm obtained from untreated males was used to fertilize the oocytes in vitro. The
percentages of oocytes fertilized and number of sperm heads and attached sperm per oocyte were
counted using a fluorescent microscope after 20 hours of incubation at 37°C with 5 x 103 sperm
in 120 |j,L medium. The fertilization efficiency of oocytes from treated females was compared
with that of oocytes from untreated control rats (n = 6). There were no effects of treatment on
the percentage of females ovulating or the number of oocytes per ovulating female. The authors
noted that oocytes from females treated with MPD tended to be more fragile during removal of
the zona pellucida (45% of oocytes were intact compared to 57% for controls). No significant
differences in the mean percentage of oocytes fertilized were observed (84, 82, 80, and 78%,
respectively) for control, ETBE-, MTBE-, and MPD-treated rats. A significant reduction to 65%
fertilization efficacy occurred in oocytes from TAME-exposed females. The mean number of
penetrated sperm per oocyte for each of the treatment groups was comparable with the mean
value for oocytes from untreated control rats. The data indicate that ETBE and MTBE did not
affect female gamete quality.
The potential effects of ETBE on reproduction, the male and female reproductive systems
in general and a specific examination of the potential effects of ETBE on spermatogenesis were
assessed in F344 and Sprague-Dawley rats (CIT, 2003, unpublished report). The report states
that it was conducted under GLP. Five groups of male F344 rats (12/dose group) were
administered daily oral (gavage) doses of vehicle alone (corn oil, 4 mL/kg-day), or ETBE at 50,
250, 500, or 1000 mg/kg-day in corn oil for 12 weeks. Five groups of male (12/dose group)
Sprague-Dawley rats were administered daily oral (gavage) doses of vehicle alone (corn oil, 4
mL/kg-day), or ETBE at 50, 250, 500, or 1000 mg/kg-day in corn oil for 12 weeks (10 pre-
mating and 2 weeks during mating). Five groups of female (24/dose group) Sprague-Dawley
rats were administered daily gavage doses of vehicle alone (corn oil, 4 mL/kg-day), or ETBE at
50, 250, 500, or 1,000 mg/kg-day in corn oil during a 2-week premating period, a 2 week mating
period, and pregnancy through gestation day (GD) 19 or until the end of lactation (postnatal day
[PND] 21). In-life clinical signs and mortality were checked daily and body weight and food
consumption were monitored at scheduled intervals for both rat strains. The mating index, pre-
coital time, and fertility index were calculated and the males were euthanized at the end of the
mating period. Male rats of each strain were subjected to a macroscopic examination. The testes
07/14/2009
40
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-6. Summary of significant results from the 13-week subchronic ETBE inhalation study in CD-I mice
Endpoint
Measurement frequency
Observations
Basic core subgroup "
Mortality
Twice daily
No exposure-related mortality
Growth rate
Weekly
No changes
Clinical signs
Weekly
Males and females: ataxia postexposure, 5,000 ppm
Organ weights
Study termination
Males and females: increased liver weights at 1,750, 5,000 ppm
Gross pathology
Study termination
Not exposure-related
Histopathology
Study termination
Males and females: centrilobular hypertrophy at 5,000 ppm (53% males, 60% females, 0% controls)
Clinical pathology subgrouph
Hematology
Interim (6 wks)
No changes
Study termination
Males: increased hemoglobin and hematocrit at 1,750 ppm; females: no effect
Clinical chemistry
Interim (6 wks)
No changes
Study termination
Males: increased total protein at 5,000 ppm; decreased chloride at 500, 1,750 ppm
Females: increased total protein and albumin at 5,000 ppm; increased blood urea nitrogen at 1,750 ppm
Cell replication subgroupc
Hepatic LI
Wks 1, 4, and 13
Males: increased LI, wks 1, 4 at 1,750, 5,000 ppm
Females: increased LI, wks 1, 13 at 1,750, 5,000 ppm
a15 animals/sex/exposure group.
b10 animals/sex/exposure group.
°15 animals/sex/exposure group.
Sources: Medinsky et al. (1999); Bond et al. (1996b, unpublished report).
07/14/2009
41
DRAFT - DO NOT CITE OR QUOTE
-------
and epididymides were weighed separately and sperm were sampled from one testis and one
epididymis for evaluation of spermatozoa count and viability. One testis and one epididymis
from F344 and Sprague-Dawley males of the control and high dose (1,000 mg/kg-day) groups
were fixed in Bouin's fluid and examined histopathologically. Half of the females were
euthanized on GD 20 and subjected to gross pathological examination. The fetuses were
removed by hysterectomy, and the following parameters were recorded: weight of gravid uterus,
number of corpora lutea, implantation sites, early and late resorptions, and dead and live fetuses.
The fetuses were weighed, sexed, and submitted to external examination. The remaining females
were allowed to deliver normally and gestational duration was calculated. The litters were
observed for clinical signs daily during the lactational period and body weight and sex ratio were
recorded.
No changes in body weight were observed in males of either strain at any dose of ETBE.
Increased food consumption was noted in F344 males from week 4 to week 12 of 10-18% and in
Sprague-Dawley males in weeks 4 and 5 of 8-12%, food consumption was not different for
either strain when calculated over the entire treatment period. Minor changes in body weight
gain were noted at all doses during the pre-mating period; these changes were not dose-related
and were therefore considered unrelated to treatment by the authors. No difference in body
weight or weight gain was noted at 50 or 250 mg/kg-day ETBE in dams during pregnancy. A
nonsignificant decrease (6 %,p> 0.05) in body weight gain was recorded for pregnant females
exposed to 500 mg/kg-day ETBE and an 11% (p < 0.05) decrease in body weight gain was noted
in females exposed to 1,000 mg/kg-day. No changes were noted in body weight or weight gain
in females at any dose during the lactation period. No changes in food consumption were
observed at any dose for females during any period.
Excessive salivation (ptyalism) was noted in male rats of both strains at the higher doses
as follows (F344 males at 50 [1/12], 250 [1/12], 500 [1/12], and 1,000 mg/kg-day [6/12];
Sprague-Dawley males at 250 [1/12], 500 [1/12], and 1,000 mg/kg-day [10/12]). Ptyalism was
also observed in Sprague-Dawley female rats at the higher doses, and data were collected
separately by reproductive time-period as follows (pre-mating at 1,000 mg/kg-day only [3/12];
pregnancy at 500 [1/12] and 1,000 mg/kg-day [8/12]; and lactation at 500 [1/12] and
1,000 mg/kg-day [6/12]). The study authors stated that ptyalism was not observed every
treatment day, but specific data on frequency of symptoms were not provided. The study
concludes that there were no effects of ETBE treatment at doses up to 1,000 mg/kg-day on
gonadal function, mating behavior, fertility, embryo-fetal development, or parturition and,
therefore, reports a no-observed-effect level (NOEL) of 1,000 mg/kg-day for parental and fetal
toxicity in F344 rats and Sprague-Dawley rats. The authors also report a NOEL of
500 mg/kg-day for maternal toxicity based on the lower body weight gain in Sprague-Dawley
dams observed at 1,000 mg/kg-day during pregnancy.
07/14/2009
42
DRAFT - DO NOT CITE OR QUOTE
-------
The potential effects of ETBE on pregnancy and embryo-fetal development were
assessed in Sprague-Dawley rats (CIT, 2004a, unpublished report). As described below, because
endpoints from the CIT (2004a, b, unpublished reports) studies were considered for benchmark
dose (BMD) modeling and the derivation of an RfD, the study reports were externally peer-
reviewed by EPA in November 2008. The study was conducted under EPA's testing guidelines
(OPPTS 870.3700; U.S. EPA, 1998c). Female rats (24/dose group) that had been impregnated
by unexposed males were administered daily gavage doses of vehicle alone (corn oil, 4 mL/kg-
day), or ETBE at 250, 500, or 1,000 mg/kg-day in corn oil on GDs 5-19. Scheduled
hysterectomy and necropsy were performed on GD 20. In-life clinical signs and mortality were
checked daily. On GD 20, the dams were euthanized and subjected to gross pathological
examination. The fetuses were removed by hysterectomy, and the following parameters were
recorded: weight of gravid uterus, number of corpora lutea, implantation sites, early and late
resorptions, and dead and live fetuses. The fetuses were weighed, sexed, and submitted to
external examination. Half of the fetuses from each treatment group were fixed in Harrison's
fluid and subjected to a detailed, serial-section examination of soft tissues, while the remainder
underwent a detailed skeletal examination following staining of bone with alizarin red and
cartilage with alcian blue.
Significantly lower maternal body weights (- ll%,/7 < 0 .05) and decreased net weight
gains since the start of treatment (-11%, p < 0.05) occurred in the dams dosed with 1,000 mg/kg-
day ETBE when compared with untreated controls, and the decreased weight change was not
accompanied by decreased food consumption. No effects on embryo-fetal development were
recorded at this dose level. No significant treatment-related effects on maternal weights or on
embryo-fetal development were observed at ETBE dose levels of 500 or 250 mg/kg-day. The
study authors reported a NOAEL for maternal ETBE toxicity of 500 mg/kg-day when
administered via gavage in the rat, and a NOAEL of 1,000 mg/kg-day with regards to embryo-
fetal development.
A two-generation reproductive toxicity study of ETBE was conducted in rats by CIT
(2004b; unpublished report). As described below, because endpoints from the CIT (2004a, b,
unpublished reports) studies were considered for BMD modeling and the derivation of an RfD,
the study reports were externally peer-reviewed by EPA in November 2008. The study was
conducted under EPA's testing guidelines (OPPTS 870.3800; U.S. EPA, 1998d). Sprague-
Dawley rats (25/sex/dose group) were administered ETBE via gavage at dose levels of 0 (corn
oil vehicle), 250, 500, or 1,000 mg/kg-day. Reproductive parameters evaluated included gonadal
function, the estrous cycle, mating behavior, conception, gestation, parturition, lactation, and
weaning, as well as growth and development of offspring of treated rats. Dosing of all females
in the F0 generation groups commenced 10 weeks before mating, continued during a 2-week
mating period, throughout gestation, and until the end of lactation (PND 21) for a total of
18 weeks; corresponding males were dosed for an identical length of time. Direct gavage
07/14/2009
43
DRAFT - DO NOT CITE OR QUOTE
-------
treatment of the progeny of the FO generation (F1 pups) began at weaning and continued under
the same experimental conditions described for the FO generation. Progeny of the F1 generation
(F2 pups) were treated from weaning until sexual maturity. For both F1 and F2 generations, on
PND 4, litters were culled to eight pups per litter (four males, four females). On PND 22, 25 F1
rats/sex from each dose group were selected for subsequent mating. Time and acquisition of
sexual milestones for each animal were noted, and neurobehavioral and reflex development tests
were conducted at designated intervals. Testicular and epididymal sperm parameters were
evaluated for FO and F1 males. The estrous cycle of females was monitored 3 weeks prior to
mating and during the mating period. Histological examinations were conducted for gross
lesions, reproductive organs, adrenal glands, and pituitary glands from all parental rats; organs
with gross abnormalities were also examined.
For all generations, ptyalism was observed in a dose-related trend in most treated males
and some females. The incidence of ptyalism resolved within 1 hour of dosing in all females and
most males (all but one male of the FO generation, most males of the F1 generation, and all of the
males of the F2 generation). There were no apparent effects of ETBE on mating, fertility,
gestation, fecundity, or delivery, and no significant effects were observed on the progeny from
birth until weaning in any of the treatment groups. Sperm parameters were not affected in any of
the male treatment groups.
In the FO generation at 1,000 mg/kg-day, males showed significantly lower body weight
gains in the last quarter of the treatment period (-22%,p < 0.01 compared to controls over days
85-113; Table 4-7). The decreased weight gain was not different from controls when analyzed
over the entire treatment period (days 1-113), and was not accompanied by a decrease in food
intake. F0 females at this dose consumed 10% more food during the lactation period (PND 1-
21, p < 0.001). Absolute and relative liver weights in 1,000 mg/kg-day males were increased by
17 and 24% (p < 0.01), respectively, apparently related to the slight to moderate centrilobular
hypertrophy in liver tissue of high-dose parental males (3/3) and not seen in the one control male
subjected to histopathological examination. Only the livers of those four males were examined
out of all of the rats from the F0 generation. Absolute and relative kidney weights were
significantly increased in 1,000 mg/kg-day males by 21 and 28% (p < 0.01), respectively, and
correlated with the appearance of acidophilic globules in renal tissue from 5/6 males examined.
Only the kidneys of those six high-dose males were examined out of all of the rats from the F0
generation. In addition, tubular basophilia (4/6), peritubular fibrosis (3/6), and proteinaceous
casts (1/6) were observed in the male rat's kidneys at the high dose. In male rats in the
500 mg/kg-day group, significantly lower body weight gain was noted at the end of the treatment
period (-29%,p <0.001) and absolute and relative kidney weights were increased by 15 and 18%
(p < 0.01), respectively. In the 250 mg/kg-day F0 generation males, absolute (+11 %, p < 0.05)
and relative (+1 .01) kidney weights were increased, but no such effects were found in
07/14/2009
44
DRAFT - DO NOT CITE OR QUOTE
-------
females. In the 250 mg/kg-day group, transient ptyalism was observed in a few males and
females, but no other abnormal effects were noted.
07/14/2009
45
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-7. Effects of oral ETBE treatment on parental Sprague-Dawley rats in a two-generation reproduction and
fertility study
Dose
Mean body weight gain
(percent change as percent of control)
Mean liver weight
(percent of control)
Centrilobular
hypertrophy
(incidence)3
Mean kidney weight
(percent of control)
Renal acidophilic
globules
Daysb
1-113
/7-Value
Days0
85-113
/7-Value
Absolute
Relative
/7-Value
Absolute
Relative
/7-Value
Incidence"
Severity
F0 Generation
1,000 mg/kg-
day
Males
-5
>0.05
-22
<0.01
+17
+24
<0.01
3/3
+21
+28
<0.01
5/6
slight-
moderate
Females
0
>0.05
NA
+6
+4
>0.05
+5
+3
>0.05
500 mg/kg-day
Males
-3
>0.05
-29
<0.001
+2
+6
>0.05
+15
+18
<0.01
Females
+2
>0.05
NA
+4
+8
>0.05
+2
+5
>0.05
250 mg/kg-day
Males
0
>0.05
-3
>0.05
+2
+3
>0.05
+11
+11
<0.01
Females
+1
>0.05
NA
+1
+10
>0.05
+1
+9
>0.05
F1 Generation
1,000 mg/kg-
day
Males
+1
>0.05
-13
>0.05
+27
+25
<0.01
2/2
+58
+58
<0.01
3/4
slight-
marked
Females
0
>0.05
NA
+10
+9
<0.05
+11
+10
<0.01
500 mg/kg-day
Males
+4
>0.05
-7
>0.05
+14
+11
<0.05
+22
+19
<0.01
1/1
marked
Females
-2
>0.05
NA
+3
+6
>0.05
+3
+6
>0.05
07/14/2009
46
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-7. Effects of oral ETBE treatment on parental Sprague-Dawley rats in a two-generation reproduction and
fertility study
Dose
Mean body weight gain
(percent change as percent of control)
Mean liver weight
(percent of control)
Centrilobular
hypertrophy
(incidence)3
Mean kidney weight
(percent of control)
Renal acidophilic
globules
Daysb
1-113
/7-Value
Days0
85-113
/7-Value
Absolute
Relative
/7-Value
Absolute
Relative
/7-Value
Incidence"
Severity
250 mg/kg-day
Males
0
>0.05
-2
>0.05
0
0
>0.05
+10
+11
<0.05,
<0.01
Females
-1
>0.05
NA
+1
+3
>0.05
+4
+6
>0.05
F2 Generation
All doses
No effects noted at any dose level; rats killed prior to mating
"Number observed/number examined.
bBody weight of F1 generation females was only tracked for 105 days (64 days of premating, 20 days of pregnancy, and 21 days of lactation).
°NA - Not applicable; the corresponding period (days 85-113) in females includes a portion of pregnancy (day 14 to 20) and all of lactation.
Source: CIT (2004b, unpublished report).
07/14/2009
47
DRAFT - DO NOT CITE OR QUOTE
-------
Body weight, body weight gain, and food consumption were not affected by exposure in
either sex of the F1 parental dose groups. Transient ptyalism was observed in most males and in
a majority of females. Increases in absolute (+27%,/? < 0.01) and relative (+25%,/? < 0.01) liver
weights occurred in males dosed with 1,000 mg/kg-day, with accompanying centrilobular
hypertrophy in males (2/2) examined. Only the livers of those two high-dose males were
examined out of all of the rats from the F1 generation. Kidney weights, both absolute (+58%,
p < 0.01) and relative (+58%,/? < 0.01), were significantly increased in high-dose males, and
slight to moderate acidophilic globules were seen in the kidney tissue of high-dose males (3/4)
examined microscopically as a result of observed macroscopic lesions. Histology was also
performed on kidneys from one control male and one mid-dose (500 mg/kg-day) male due to the
presence of macroscopic lesions in these animals. Only the kidneys of those six males were
examined out of all of the rats from the F1 generation. Tubular basophilia was observed in the
control male, the mid-dose male, and two of the four high-dose males. Peritubular fibrosis was
also observed in the control male, the mid-dose male, and two of the four high-dose males. The
one mid-dose male examined also had "sloughed degenerated/necrotic cells" in the tubular
lumen. Similar but less striking increases in absolute (+10%,/? < 0.05) and relative (+9%,
/? < 0.05) liver and absolute (+11%,/? < 0.01) and relative (+10%,/? < 0.01) kidney weights
appeared in high-dose females (Table 4-7), although no macroscopic effects were noted and no
histology was performed. In the 1,000 mg/kg-day group, pup body weight gains were slightly,
but not significantly lower during the first 4 days of lactation. Two pups born to 1,000 mg/kg-
day F1 females exhibited gross external malformations (absence of tail with anal atresia also
observed in one pup); however, the authors state that the incidence of these malformations was
comparable to laboratory or external historical control data. In the 500 mg/kg-day F1 generation,
absolute (+14%,/? < 0.05) and relative (+11%,/? < 0.01) liver and absolute (+22%,/? < 0.01) and
relative (+19%,/? < 0.01) kidney weights were increased in parental males, but no such effects
were found in parental females. In the 250 mg/kg-day F1 generation, relative (+1 1%,/? < 0.01)
kidney weights were increased in parental males, but no such effects were found in parental
females. No other macroscopic or histological effects were noted in parental males and females
or in their progeny. No other effects were observed in the F1 parental rats or their progeny at
250 mg/kg-day, but transient ptyalism was observed in a majority of males and some females.
In the F2 generation, transient ptyalism was seen in approximately half of the high-dose
males and females and in a few rats of lower-dose groups. Significant effects on body weight,
body weight gain, food consumption, or liver and kidney weights were not observed. There were
no adverse macroscopic or histological findings.
In summary, significant decreases in body weight gain in male parental rats were
recorded for the F0 generation dosed with 1,000 or 500 mg/kg-day ETBE. Absolute and relative
kidney weights were increased in high- and mid-dose males, which was associated with the
presence of acidophilic globules in the limited number of animals examined histopathologically.
07/14/2009
48
DRAFT - DO NOT CITE OR QUOTE
-------
Increases in absolute and relative liver weights, with concomitant centrilobular hypertrophy,
were also recorded. Histological examination of the kidney or liver was only performed when
lesions were observed macroscopically. Similar macroscopic effects in females were not
observed, and therefore, no histology of the kidney or liver in females was performed. In the F1
generation, both males and females dosed with 1,000 mg/kg-day showed increases in absolute
and relative liver and kidney weights, but the effects were far less pronounced than in males.
Increases in absolute and relative kidney and liver weights were seen in males, but not in
females, dosed with 500 mg/kg-day.
The study authors suggested a NOAEL of 250 mg/kg-day based on systemic toxicity
(specifically on the increased incidence of reduced body weight gain at the end of the treatment
period in F0 males and the increased liver and kidney weights in high dose F0 females and mid-
and high-dose F0 and F1 males as described above). For fertility, gonadal function, reproductive
performance, parturition, and lactation in the parental generations, and development of the
offspring to weaning or sexual maturity, the authors suggested a NOAEL of 1,000 mg/kg-day
(highest dose tested). The authors note that ptyalism is common following gavage of unpalatable
chemicals, and, therefore, the transient ptyalism observed was not considered to represent an
adverse effect of ETBE exposure; a lowest-observed-effect level of 250 mg/kg-day was noted.
Although the increased relative kidney weights observed at 250 mg/kg-day in the F0 and F1
generation males and the increased absolute kidney weights in F0 generation males represent a
LOAEL, these effects were not identified in the study summary. Data tables from CIT (2004b,
unpublished report) report relative kidney weights in males treated at 250 mg/kg-day ETBE were
significantly increased relative to controls as determined by Dunn's test at 1% for the F1 males
and by Dunnett's test at 1% based on pooled variances. Data tables from CIT (2004b,
unpublished report) also report increased absolute kidney weights as determined by Dunnett's
test at 5% in F0 generation males treated at 250 mg/kg-day ETBE relative to controls.
4.4. OTHER DURATION- OR ENDPOINT-SPECIFIC STUDIES
4.4.1. Acute Studies
4.4.1.1. Oral
Roudabush (1966, unpublished report) reported the oral dose to kill half of the animals
(LD50) for ETBE to be >1,600 mg/kg in rats, but no experimental details were provided. An
acute oral study with ETBE was performed in which four groups containing two male and two
female albino rats were administered single doses of 500, 1,000, 2,500, or 5,000 mg/kg ETBE
(no vehicle was employed) by gavage (MB Research Laboratories, Inc., 1988a, unpublished
report). The rats were observed for mortality and toxicity 1, 2, and 4 hours post-dose and twice
daily for 14 days. No deaths or abnormal physical signs were noted. Body weight gains were
normal except for a single female (2,500 mg/kg) that lost weight during the second week of the
07/14/2009
49
DRAFT - DO NOT CITE OR QUOTE
-------
study. Necropsy results were normal. On the basis of data presented, the acute oral LD50 is
>5,000 mg/kg. No other acute oral studies of ETBE were found.
4.4.1.2. Inhalation
The comparative anesthetic activity of a series of commonly available aliphatic ethers
with 2-10 carbon atoms (31 compounds), including ETBE, was determined in inhalation studies
using white mice (Marsh and Leake, 1950). The test agents were placed in a 20 L Pyrex™ jar
and volatilized. Four mice (18-24 g) were then placed in the jar and the experiment was run five
times for a total of 20 animals. The concentration of test agent producing anesthesia within
30 minutes in 9-11 of the 20 mice, the AC50 (concentration in mmol/L required to anesthetize
half of mice), was calculated using the jar volume and the amount of test agent volatilized. The
median lethal concentration (LC50), was determined in a corresponding fashion. The AC50 for
ETBE was determined to be 0.7 mmol/L (17,112 ppm), while the LC50 was 1.2 mmol/L
(29,334 ppm). The Therapeutic Index (LC50/AC50) was thus calculated to be 1.7, and the Certain
Safety Factor (LC, 5/AC95 99) was determined to be 1.25. For comparison, the Therapeutic Index
and Certain Safety Factor for diethyl ether are 3.0 and 1.8, respectively. The AC50 and LC50
values for MTBE were found to be 1.2 and 1.5 mmol/L, respectively, using identical
experimental methods.
In an acute inhalation study (IIT Research Institute, 1989a, unpublished report), ETBE
vapor was administered nose-only to a group of five CD rats/sex at a concentration of 5.88 mg/L
(1,407 ppm). Since none of the rats died after a single 4-hour exposure, it was concluded that the
LC50 is >5.88 mg/L (1,407 ppm).
4.4.2. Direct Administration Studies
4.4.2.1. Dermal Administration
ETBE was applied neat (without vehicle) at a dose of 2 g/kg to the shaved backs of five
male and five female rabbits (IIT Research Institute, 1989b, unpublished report). The test article
was left in contact with the skin under occlusion for 24 hours and then removed. The rabbits
were observed during that period and for 14 days thereafter. Since no deaths occurred during the
study, the dermal LD50 was estimated to be >2 g/kg. Dermal irritation, edema, erythema, and
eschar formation were observed at the sites of application of all rabbits. No gross pathological
lesions were observed in any of the rabbits at necropsy.
Similar results were found in a second acute toxicity study designed to determine the
dermal LD50 performed by MB Research Laboratories, Inc. (1988b, unpublished report). A
2 g/kg dose of ETBE was applied neat to the shaved backs of five male and five female healthy
New Zealand albino rabbits. The test article was kept in contact with the skin under occlusion
for 24 hours. Rabbits were observed 1, 2, and 4 hours post-dosing, and then twice daily for
14 days. Nine of the 10 rabbits survived the dermal dose of 2 g/kg. Physical signs noted in
07/14/2009 50 DRAFT - DO NOT CITE OR QUOTE
-------
surviving rabbits included diarrhea, few feces, yellow nasal discharge, lacrimation, and soiling of
the anogenital area. Five of the rabbits lost weight during the study, while body weight changes
in four rabbits were normal. Dermal reactions at the site of application were well-defined to
moderate on day 1, slight to severe on day 7, and absent to severe on day 14. Instances of poor
hair regrowth, indicative of dermal injuries in depth, were noted on day 14. Necropsy of
survivors revealed abnormalities in skin at the site of treatment and isolated instances of
gastrointestinal and lung abnormalities. Yellow staining of the nose/mouth area and soiling of
the anogenital area were also noted during necropsy. Based on the findings from this study, the
dermal LD50 for ETBE in rabbits was estimated to be >2.0 g/kg.
Roudabush (1966, unpublished report) applied ETBE (dose not reported) to the shaved
skin of guinea pigs under occlusion, which resulted in only slight skin irritation being observed.
No information on dermal absorption was provided. In another dermal irritation study (MB
Research Laboratories, Inc., 1988c, unpublished report), six New Zealand albino rabbits were
dosed with 0.5 mL ETBE at two intact and two abraded sites. ETBE was kept in contact with
the skin for 4 hours under occlusion, and then the wrappings were removed. Dermal reactions
were scored at 4, 24, 48, and 72 hours after application of the test article, and also after 7 and
14 days. Although absent at 4 hours, erythema was well-defined at the site of application at 24
and 48 hours and slight to well-defined at 72 hours. By day 7, erythema was absent at some
sites, but well-defined at other sites, and by day 14, no erythema was detected. The Primary
Irritation Index was 3.08 out of a possible 8.0, and abraded sites were no more severely affected
than intact sites.
A subchronic dermal study was conducted in Sprague-Dawley rats by UBTL Inc. (1994,
unpublished report). Groups of 10 male and 10 female young adult rats were administered
ETBE (no carrier) to clipped areas on their backs 5 days/week for 4 weeks (28 days) at 0
(control), 0.05, 0.25, and 1.0 mL/kg-day. The application site was occluded for at least 6 hours
following dosing. Rats were observed twice daily for viability and once daily for signs of
toxicity. Irritation was evaluated just prior to dosing and 24 hours after the fifth dose each week.
At necropsy, blood samples were taken for clinical chemistry evaluation and selected tissues
were collected for histopathological evaluation. ETBE caused slight to moderate dermal
irritation in both males and females at 1.0 mL/kg-day; there were visible lesions and
histopathological changes in the skin. Less irritation was observed in rats in the mid-dose group,
and very slight dermal irritation was observed in the 0.05 mL/kg-day dose group. There was an
increase in the incidence and severity of lymphoid hyperplasia of the axillary lymph nodes in
males and females of the 0.25 and 1.0 mL/kg-day groups, findings considered to be secondary to
the dermal irritation observed. Under the conditions of the study, the authors reported the dermal
NOEL to be <0.05 mL/kg-day in male and female rats; the systemic NOEL for dermal exposure
was reported to be 1.0 mL/kg-day for both male and female rats.
07/14/2009
51
DRAFT - DO NOT CITE OR QUOTE
-------
4.4.2.2. Ocular Administration
Eye irritation characteristics were evaluated using the Draize test (MB Research
Laboratories, Inc., 1988d, unpublished report). ETBE, 0.1 mL, was placed in the conjunctival
sac of one eye of each of nine healthy New Zealand rabbits. Three eyes were washed with
lukewarm water for 1 minute at 20-30 seconds after dosing and the treated eyes of the remaining
six animals remained unwashed. The eyes were examined and scored at intervals for corneal
opacity, iritis, and conjunctival irritation. For unwashed eyes, slight corneal opacity that cleared
by day 7 occurred in 1/6 eyes; iritis in 1/6 eyes that cleared by day 2, and conjunctival irritation
in 6/6 eyes that cleared by day 7. In the washed eyes, corneal opacity was noted in 1/3 eyes; this
opacity was first observed on day 1, and it had not cleared by day 14. However, no iritis was
observed in the washed eyes, and conjunctival irritation that was observed in 1/3 eyes had
cleared by day 7. Mean Draize scores were 12.2, 8.3, 4.0, 0, and 0 for unwashed eyes on days 1,
2, 3, 7, and 14, respectively, and 6.7, 5.3, 4.0, 3.3, and 3.3 for washed eyes at these same time
points.
4.4.3. Neurological Studies
The ability of ETBE and other oxygenated fuel additives to inhibit the binding of a
convulsant ligand ([3H]t-butylbicycloorthobenzoate) to the y-aminobutyric acidA (GABAa)
receptor was tested in vitro in Sprague-Dawley rat brain membrane preparations (Martin et al.,
2002). Membrane preparations (100 |iL) were incubated on ice for 60 minutes with 20 nM of
[3H]t-butylbicycloorthobenzoate and one of the six chemicals (MTBE, ETBE, TAME, TBA,
TAA, or ethanol) to generate concentration-response curves for binding inhibition. Chemicals
were tested over a concentration range from 10"4 to 1 M and incubations were performed in
triplicate. Subsequent incubations were performed on ice for 90 minutes to generate saturation
curves. Parallel incubations were included with picrotoxin to establish nonspecific binding.
Concentration loss due to evaporation was monitored and was less than 5% for all agents tested.
The GABAa receptor has distinct recognition sites for GAB A, depressants, and convulsants
(Mehta and Ticku, 1999). There was a general correlation of the potentiation of GABAa
receptor responses with their ability to induce general anesthesia (Krasowski and Harrison,
1999). Martin et al. (2004, 2002) suggested that acute neurological symptoms associated with
short-term exposure to oxygenates such as ETBE (e.g., headache, dizziness, and nausea) may
reflect the participation of the GABAa receptor. The potency of the inhibition of convulsant
ligand binding was in the rank order: TAA > TAME > ETBE > TBA > MTBE > ethanol. In a
follow up study, the uptake of 36C1" was measured in synaptoneurosomes (which included pre-
and postsynaptic membranes) from adult Sprague-Dawley rat cerebral cortex (Martin et al.,
2004). The oxygenates and oxygenate metabolites tested produced concentration-dependent
enhancement of muscimol-stimulated uptake of 36C1". The potency of enhancement was as
follows MTBE = TAME > TAA = ETBE > TBA > ethanol. Concentrations that facilitated
07/14/2009
52
DRAFT - DO NOT CITE OR QUOTE
-------
muscimol-stimulated 36C1" uptake ranged from 0.06 to 3 nM, which were described by the
authors as being in the range of blood concentrations reported in experimental animals following
exposures known to induce CNS effects and ataxia (Martin et al., 2004).
In subchronic inhalation experiments, ETBE was administered as a vapor at target
chamber concentrations of 0, 500, 2,000, and 4,000 ppm to groups of Sprague-Dawley rats for
4 weeks as described in Section 4.2.5.1 (published as White et al., 1995; IIT Research Institute,
1991, unpublished report). A FOB (which included tail pinch, rotorod performance, body
temperature, righting reflex, auditory response, hind limb extension, foot splay, grip strength,
home-cage observation, hand-held observation, open-field observation, extensor thrust,
catalepsy, visual placing, tactile placing, negative geotaxis, vision, eye-blink, and pupil response)
was administered 1 week prior to the exposures and about 60 minutes after 1, 5, or 20 exposures
(6 hours/day, 5 days/week) to evaluate neuromuscular function and sensory perception. Ataxia
and sedation, which are overt signs of CNS depression, were seen following exposure
termination in the 4,000 ppm group, but rats appeared normal 15 minutes after the end of
exposure. Mean body temperature was reduced 2.00-2.14% in 4,000 ppm males after the fifth
exposure, and a trend for increased hind limb splay in both sexes of the high concentration group
occurred. These effects were described by the authors as being associated with transient CNS
depression. No other indications of CNS depression or neurotoxicity were detected.
Dorman et al. (1997) performed an evaluation of possible ETBE neurotoxicity with male
and female F344 rats from the 13-week inhalation study described in Section 4.2.3.1 (Medinsky
et al., 1999). Rats from the neurotoxicology subgroup (12/sex/concentration), 8 weeks of age at
study initiation, were exposed to 500, 1,750, or 5,000 ppm ETBE (>97% pure) for 6 hours/day
and 5 days/week for a total of 65 exposures over a 13-week period. Details of the experimental
design and nonneurological effects were provided earlier (Section 4.2.3.1). On exposure days,
all rats were observed for mortality and overt clinical signs of toxicity prior to exposure and
shortly following exposure, while on non-exposure days, the rats were observed once daily. The
following observations of individual rats were conducted during daily cage-side examinations:
body appearance, piloerection, fur appearance, facial crust, skin temperature and color, breathing
pattern, salivation, and an evaluation of eyes, lacrimation, and mucous membranes. Effects on
the respiratory, circulatory, autonomic (e.g., diarrhea, salivation), and central nervous systems
(e.g., ataxia, head tilt, seizures) were evaluated, in addition to specific assessments of
somatosensory activity (e.g., photophobia) and behavior patterns (e.g., circling, aggressiveness).
Motor activity assessments and an FOB were also performed on all rats. These latter tests were
conducted 4 days prior to initial ETBE exposure, and at least 18 hours after the end of days 1, 6,
10, 20, 42, and 65 of exposure, to detect possible persistent neurological effects. The FOB
consisted of observations of spontaneous activity and behavior in an open field, assessment of
visual approach response, auditory startle response, tail pinch response, surface righting reflex,
visual placing response, forelimb and hindlimb grip strength, hind leg splay, and pupillary reflex.
07/14/2009
53
DRAFT - DO NOT CITE OR QUOTE
-------
Each animal was also evaluated for posture, tremors, spasms, convulsions, palpebral closure,
handling reactivity, and muscle tone.
At termination, rats in the 5,000 ppm and control groups were anesthetized and then
perfused with 1.5% glutaraldehyde/4% formaldehyde for preparation of nervous tissue. Brains
were weighed and measured, and paraffin sections of Gasserian ganglia, dorsal spinal root
ganglia, spinal root fibers, brain, spinal cord, eye, and optic nerve were stained with
hematoxylin/eosin to provide a direct evaluation of possible structural neuropathy. Additionally,
portions of sciatic, tibial, and sural nerves were embedded in glycol methacrylate, sectioned, and
stained with Lee's methylene blue basic fuchsin for more extensive histological examination.
The only noteworthy clinical finding during daily observations was transient ataxia,
typically observed in male rats immediately following a 5,000 ppm exposure, and lasting less
than 1 hour. Such effects are typically seen with various ethers, including MTBE and diethyl
ether. Ataxia was only noted during the first 5 weeks of the 13-week study. There were no
significant effects on body appearance, and no significant effects on motor activity during the
post-exposure test sessions were noted. No evidence of sensorimotor dysfunction (altered
acoustic response, tail pinch response, approach response, or visual placing response) or
neuromuscular dysfunction was observed in control or ETBE-exposed rats. No signs of ataxia,
piloerection, excessive vocalization, muscle tremors or spasms, clonic or tonic seizures,
increased salivation, abnormal respiration, or abnormal pupillary reflex were observed in ETBE-
exposed rats. No gross or microscopic abnormalities were observed in the central, peripheral, or
autonomic nervous systems of rats exposed to 5,000 ppm ETBE. No significant effects of
exposures on brain weight or size were observed. The authors concluded that in spite of
transient ataxia in rats following high-level exposure, there was no indication that ETBE is a
neurotoxicant under the conditions tested.
Neurobehavioral endpoints were assessed as part of a two-generation gavage study
conducted in Sprague-Dawley rats as described in Section 4.3 (CIT 2004b, unpublished report).
The study was conducted under EPA's testing guidelines OPPTS 870.3800 (U.S. EPA, 1998d).
Groups of Sprague-Dawley rats (25/sex/dose group) were administered ETBE at dose levels of 0,
250, 500, or 1,000 mg/kg-day by gavage in corn oil. Dosing of the F0 generation groups
commenced 10 weeks before mating, continued during a 2-week mating period, and throughout
the time period required for gestation and lactation for a total of 18 weeks. Treatment of the
progeny of the F0 generation (F1 pups) began at weaning and continued under the same
experimental conditions described for the F0 generation. The following neurobehavioral and
reflex development tests were conducted in rats of the F1 generation: acoustic startle response
and pupil constriction were assessed at 4 weeks of age, and spontaneous locomotor activity was
evaluated when the animals were between 7 and 8 weeks of age. There was no effect of
treatment on neurobehavioral parameters.
07/14/2009
54
DRAFT - DO NOT CITE OR QUOTE
-------
4.5. MECHANISTIC DATA AND OTHER STUDIES IN SUPPORT OF THE MODE OF
ACTION
4.5.1. Genotoxicity
4.5.1.1. In Vitro Bacterial Assays
Zeiger et al. (1992) conducted Salmonella mutagenicity test using different Salmonella
typhimurium strains for 311 chemicals, including ETBE, both in the absence and presence of
metabolic activation (S9). Preincubation protocol was followed as described in Haworth et al.
(1983). Chemicals known or suspected to be volatile, such as ETBE, were incubated in capped
tubes. ETBE (99% purity) did not show any mutagenic activity both in the presence and absence
of S9.
4.5.1.2. In Vitro Mammalian Assays
The potential clastogenicity of ETBE (>98% purity; containing 13 ppm A022, an
antioxidant stabilizer) was evaluated in the in vitro Chinese hamster ovary (CHO) chromosome
aberration assay (Vergnes, 1995, unpublished report). Concentrations of ETBE ranging from 0.1
to 5.0 mg/mL culture medium, both in the presence and absence of rat liver S9 metabolic
activation system, were tested on cultures of CHO K1-BH4 cells. No significant cell cycle
delays were noted in preliminary experiments; consequently, mitotic cells from the highest four
concentrations were harvested 10 hours after treatment and scored for chromosomal aberrations.
Treatment of cultured CHO cells with ETBE did not result in statistically significant or
concentration-related increases in the frequency of chromosome aberrations in the presence or
absence of the S9 metabolic activation system. Treatment of positive control cells with
Mitomycin C demonstrated significantly positive numbers of chromosomal aberrations. ETBE
was, therefore, not considered clastogenic under the conditions of this assay. Although the study
description notes that glass bottles were used because of the solvent properties of ETBE, the
methods do not state that the volatility of ETBE was controlled for; therefore, exposure
concentrations may have been significantly reduced by evaporation of ETBE. The effect of the
antioxidant stabilizer, A022, was not discussed by the authors, but this compound has the
potential to decrease the sensitivity of the assay for compounds that work through oxidative
stress (e.g., antioxidant inhibition of iodoacetic acid genotoxicity measured in CHO cells in vitro
as described in Cemeli et al., 2006). CHO K1-BH4 cells were used to evaluate the mutagenicity
of ETBE by using the hypoxanthine-guanine phosphoribosyl transferase (HGPRT) forward
mutation assay (Vergnes and Kubena, 1995a, unpublished report). The assay detects mutations
in the HGPRT gene, which are scored after killing all nonmutated cells by addition of the purine
analog, 6-thioguanine, to cell cultures. Duplicate cultures of CHO cells were treated with five
concentrations of ETBE (>98% purity; 13 ppm A022-see above for potential effect of A022)
ranging from 0.1 to 5.0 mg/mL ETBE, both in the presence and absence of S9 activation. No
cytotoxicity was observed. No statistically significant, concentration-related increase in the
07/14/2009 55 DRAFT - DO NOT CITE OR QUOTE
-------
HGPRT mutation frequency was observed at any of the ETBE concentrations tested. Mutation
frequencies in the positive control plates treated with 200 |j,g/mL ethylmethanesulfonate were
highly significantly elevated compared with dimethyl sulfoxide vehicle-treated control cells.
ETBE was thus considered nonmutagenic in the HGPRT forward mutation assay. However, as
above, the methods do not state that the volatility of ETBE was controlled for and, therefore,
exposure concentrations may have been significantly reduced by evaporation of ETBE.
4.5.1.3. In Vivo Assays
Vergnes and Kubena (1995b, unpublished report) conducted an in vivo micronucleus
(MN) test in mice in response to ETBE exposure. Male and female CD-I mice (five/sex/group)
were exposed to ETBE by inhalation (>98% purity; 13 ppm A022) at target concentrations of 0
(air-only control), 400, 2,000, and 5,000 ppm ETBE for 6 hours/day for 5 days. Following
treatment, mice were killed and femurs were removed. Aspirated bone marrow cells were
smeared on slides and treated with Giemsa stain. At least 2,000 polychromatic erythrocytes per
mouse were scored in bone marrow smears. No statistically significant increases in the mean
percentages of micronucleated polychromatic erythrocytes were observed in mice of either sex
when exposed to ETBE. Mice of both sexes exposed to the positive control agent
cyclophosphamide (15 mg/kg) showed significant increases in the percentage of polychromatic
erythrocytes with micronuclei.
4.5.2. Studies with ETBE-Gasoline Mixtures
There are unpublished toxicology reports for inhalation exposure to gasoline/ETBE vapor
condensate (G/ETBE). The interpretation of the results of these mixture studies for toxicity of
individual components is confounded by co-exposure to other chemicals.
Huntingdon Life Sciences (2002, unpublished report) conducted a subchronic inhalation
toxicity study in rats using G/ETBE that included a neurotoxicity assessment, a 4-week
immunotoxicity assessment (White, 2002, unpublished report), and a 4-week in vivo-in vitro
genotoxicity assessment (Gudi and Brown, 2002, unpublished report; Mason, 2002, unpublished
report), each of which will be discussed in further detail below. The test material was prepared
to simulate the composition of headspace vapor from an automotive fuel tank at near maximum
in-use temperatures. The G/ETBE sample contained 16.3 area percent ETBE prepared from a
baseline gasoline sample that contained 2.1 area percent benzene and other low boiling-point
hydrocarbons (Daughtrey et al., 2004, abstract only). Conventional unleaded gasoline often
contains MTBE at a low percentage, although MTBE was not reported as a contaminant by
Daughtrey et al. (2004, abstract only).
Male and female Sprague-Dawley rats were exposed by inhalation to the G/ETBE at
target concentrations of 0, 2,000, 10,000, and 20,000 mg/m3 for 6 hours/day, 5 days/week, for
13 weeks (Huntington Life Sciences, 2002, unpublished report). Twenty rats/sex/concentration
07/14/2009 56 DRAFT - DO NOT CITE OR QUOTE
-------
were used for the high concentration and air control groups, while 10 rats/sex/concentration were
included in the mid- and low-dose groups. Possible neurobehavioral effects were monitored
using an FOB and motor activity tests that were performed prior to initiation of exposure, and
after 4, 8, and 13 weeks of exposure. Ten rats/sex/group were killed at 13 weeks; the remaining
rats in the high concentration and control groups were maintained without subsequent exposures
for 4 weeks of untreated recovery. At the week 13 necropsy, 5 rats/sex/group were transcordally
perfused and the nervous system tissue was prepared for neuropathological evaluation; tissues of
all other rats were prepared for routine microscopic pathology examination. Mean analytical
exposure concentrations of G/ETBE were 0, 2027 ± 193, 10,060 ± 691, and 19,930 ±
1,031 mg/m3. Mean body weight gains were decreased in females exposed to 20,000 mg/m3
during study weeks 4-12, but the differences abated during the recovery period; similar effects
did not occur in males. There were no apparent exposure-related effects of G/ETBE on FOB or
motor activity tests. Neuropathology evaluations were negative in all animals. Increased
absolute and relative kidney weights occurred in males, but not females, exposed to
20,000 mg/m3 (data not shown). This effect was reversible, or nearly so, during the 4-week
recovery period. Microscopic findings attributed to G/ETBE exposure occurred only in the
kidneys of male rats exposed to 20,000 mg/m3, and consisted of eosinophilic hyaline granules
within the cytoplasm of renal proximal tubular epithelial cells and evidence of tubular
regeneration with corticomedullary intralumenal tubular casts. These changes were thought to
be consistent with hyaline droplet nephropathy, and the authors suggested that the observations
were potentially attributable to accumulation of alpha2U-globulin within renal tubular cells.
Whether or not the available data on G/ETBE nephrotoxicity are consistent with the alpha2u-
globulin accumulation mode of action as discussed in the Risk Assessment Forum Technical
Panel Report (U.S. EPA, 1991b) has not been determined for the limited data in these
unpublished reports. Unleaded gasoline is one of the model chemicals cited by that report to
produce renal toxicity, and therefore, the exposure to this mixture would prevent attribution of
the observed renal effects to ETBE. The concentration level of 10,000 mg/m3 was considered by
the study authors to be the NOAEL for this study.
White (2002, unpublished report) described the immunotoxicity portion of the
Huntington Life Sciences (2002, unpublished report) study of G/ETBE. The immunotoxicity
study used the same exposure concentrations (i.e., target concentrations of 0, 2,000, 10,000, and
20,000 mg/m3 G/ETBE) and dosing regimen (i.e., 6 hours/day, 5 days/week) as the larger study,
with the exception that only female rats (10/dose group) were exposed to the test agents and for a
duration of only 4 weeks. Mean analytical exposure concentrations were 0, 2040 ± 161, 9978 ±
632, and 19,710 ± 1,044 mg/m3. Rats were sensitized by intravenous injection of 2 x 108 sheep
red blood cells (SRBCs) 4 days prior to the end of the G/ETBE exposure. As a positive control
for immunotoxicity, 50 mg/kg cyclophosphamide was given i.p. on the last 4 days of the study.
At study termination, 1 day after the last exposure, body weight was recorded, animals were
07/14/2009 57 DRAFT - DO NOT CITE OR QUOTE
-------
killed, and the thymus and spleen were removed and weighed. The spleen was placed in Earle's
balanced salt solution for later determination of total splenocytes, and use in the plaque-forming
cell (PFC) assay to measure the SRBC-specific antibody response. G/ETBE had no effect on the
terminal body weight, the absolute or relative weights of the spleen and thymus, or the total
number of splenocytes/animal. Exposure to G/ETBE at 10,000 and 20,000 mg/m3 did result in a
suppression of the PFC response. When evaluated as activity per 106 splenocytes, reductions of
76 and 72% were observed in the PFC for the 10,000 and 20,000 mg/m3 groups, respectively.
When evaluated as total activity per spleen, reductions of 74 and 70% were observed in the PFC
for the 10,000 and 20,000 mg/m3 groups, respectively. The cyclophosphamide positive control
was effective, reducing spleen and thymus weight, total splenocytes, and the PFC response.
Although not stated by the authors, the findings of this study support consideration of
2,000 mg/m3 of G/ETBE as a NOAEL for elicitation of immunotoxicity in the rat.
The genotoxicity data from the Huntington Life Sciences (2002, unpublished report)
study of G/ETBE were reported separately by Mason (2002, unpublished report) (MN study) and
Guidi and Brown (2002, unpublished report) (sister chromatid exchange [SCE] study). As in the
main study, Sprague-Dawley rats (5/sex/group) were exposed to target concentrations of 0,
2,000, 10,000, and 20,000 mg/m3 of G/ETBE 6 hours/day, 5 days/week. However, as in the
immunotoxicity study, the exposure duration was only for 4 weeks. Positive control groups
(5 rats/sex/group) were given cyclophosphamide i.p. (5 mg/kg for the SCE study and 40 mg/kg
for the MN study) 24 hours prior to study termination. At termination, blood from the abdominal
aorta was collected and lymphocytes were cultured for each SCE analysis. Approximately
21 hours after initiation of cultures, lymphocytes were exposed to 5 |j,g/mL BrdU, and colcemide
was added at 68 hours of culture. The cells were harvested at 72 hours and processed for
evaluation of SCE. No significant effects were seen in tests for induction of SCE by G/ETBE.
For the micronucleus analysis, femurs were removed from rats at termination; smears of femoral
bone marrow were placed on slides and stained using a modified Feulgen method. At least
2,000 immature erythrocytes were examined from each animal; the proportion of immature
erythrocytes containing satellite micronuclei was tabulated for each rat. There was an increase in
the number of micronucleated immature erythrocytes in female rats at the 10,000 mg/m3
G/ETBE dose (3.8) relative to control (0.8), although no significance was found at the
20,000 mg/m3 G/ETBE dose (2.0) or at any dose in the male rats. Trend analysis detected a
significant responsiveness in the frequency of MN as a function of G/ETBE exposure, but only if
all doses tested were included in the evaluation. For both the 10,000 and 20,000 mg/m3 groups,
the mean values for the frequency of micronucleated immature erythrocytes were greater than
the historical negative control range. The author concluded that G/ETBE may cause an increase
in the frequency of micronuclei in immature erythrocytes; however, the authors also concluded
that data from this study do not support bone marrow cell toxicity, in part, because of the lack of
significant effect at low and high doses. The utility of genotoxicity results from a mixture study
07/14/2009 58 DRAFT - DO NOT CITE OR QUOTE
-------
is limited as noted above. The presence of 2.1% benzene and other possibly genotoxic agents in
gasoline condensate (Daughtrey et al., 2004, abstract only) confound a determination of whether
ETBE is acting alone, in concert with some other constituent, or not at all, to cause increases in
MN. However, the positive response at 10,000 mg/m3 should be noted and suggests the need for
confirmation in a study of ETBE alone.
4.5.3. Structure-Activity Relationship Evaluations
The structure-activity relationships (SAR) for both MTBE and ETBE were evaluated
using two computer-based systems (computer automated structure evaluator or CASE and
TOPKAT). These software programs were developed to relate structural fragments of test
molecules to structural determinants previously recognized as being associated with
carcinogenicity in rodents, mutagenicity in the S. typhimurium (Ames) test, induction of SCEs
and/or chromosomal aberrations in cultured mammalian cells, or other structural alerts for
genotoxicity (Zeiger et al., 1996; Rosenkranz and Klopman, 1991). Structural alerts by an expert
chemist were used to test the programs. Using the structures of 100 chemicals, the three methods
predicted mutagenicity at 71-76% concordance with the S. typhimurium results. The results of
the software analysis of the various endpoints indicated that, except for a marginal induction of
SCEs in cultured mammalian cells by ETBE, the two ethers were not predicted to be either
genotoxicants or carcinogens.
Zhang et al. (1997) used CASE and multiple computer automated structure evaluation
(MULTICASE) computer models to predict the toxicity of ETBE, MTBE, TAME, DIPE, and
metabolites of ETBE. One of the predicted metabolites, ethylene oxide, was predicted to be a
rodent carcinogen, and one was predicted to have marginal carcinogenicity (N-hydroxy-
N-acetylglycine). All parent ethers were predicted to be negative for rodent carcinogenicity,
salmonella mutagenicity, sensory irritation, eye irritation, contact sensitization, and
developmental toxicity. The principal metabolite, TBA, was also predicted to be negative in all
areas of toxicity, and no predictions for toxic actions of acetaldehyde were provided because the
molecule was too small for the model. Results observed for MTBE, which in contrast to model
predictions, appears to be a multi-site and multi-species carcinogen in animals (reviewed by Cal
EPA 1999; Mennear, 1997) suggest that the CASE, MULTICASE, or TOPKAT programs used
by (Zhang et al., 1997; Zeiger et al., 1996; Rosenkranz and Klopman, 1991) may be of limited
value in predicting potential carcinogenicity associated with exposure to ETBE.
4.6. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS
Liver and kidney toxicity are the primary noncancer health effects associated with
exposure to ETBE based on limited available animal data. Increased liver and kidney weights
were observed following oral exposure in both the parent and offspring generations and both
sexes of rats in a two-generation reproductive toxicity study (CIT 2004b, unpublished report).
07/14/2009 59 DRAFT - DO NOT CITE OR QUOTE
-------
Increased kidney weight was also observed following subchronic inhalation exposure to ETBE in
male and female F344 rats (Medinsky et al., 1999) and male Sprague-Dawley rats (White et al.,
1995). The increased kidney weight was associated with histopathological changes in male rats
(including regenerative foci, which are histological formations that demonstrate response to
cellular necrosis) in males and sustained increase of greater than twofold in cellular proliferation
(Medinsky et al., 1999). The increased kidney weight was not associated with histopathological
changes in female rats, although an increased cellular proliferative response of less than twofold
was observed at early time points (after 1 and 4, but not 13 weeks of exposure) (Medinsky et al.,
1999). Increased liver weight was also observed following subchronic inhalation exposure to
ETBE in mice and rats of both sexes (Medinsky et al., 1999; White et al., 1995). No changes in
histopathology or serum levels of hepatic enzymes were observed in rats. Dose-related increases
in hepatic proliferation were observed in male and female mice at some time periods during a
13-week exposure, and not observed at others (Medinsky et al., 1999). The results of subchronic
and chronic studies are discussed below.
4.6.1. Oral
There is limited information concerning noncancer effects of ETBE following oral dosing
and almost all of the studies were designed to investigate reproductive or developmental toxicity.
There is a single chronic oral ETBE exposure carcinogenicity study that did not report any
noncancer effects with the exception of mortality (i.e., Maltoni et al., 1999), a 2-week oral
exposure study of potential ETBE effects on oocyte quality (Berger and Horner, 2003), and a
series of unpublished oral exposure studies on prenatal developmental toxicity and
two-generation reproduction and fertility effects (CIT, 2004a, b, 2003, unpublished reports).
ETBE treatment had no apparent effects on oocyte quality in the only published reproductive
toxicology study. Oral exposure to ETBE had no effect on sperm parameters, mating, fertility,
gestation, fecundity, or delivery in the CIT (2004a, b, 2003, unpublished reports) studies.
Data from the two-generation reproductive toxicity study (CIT 2004b, unpublished
report) suggest effects of ETBE on the liver and kidneys. Absolute and relative liver weights
were increased in FO-generation male rats exposed to 1,000 mg/kg-day ETBE, F1 generation
animals of both sexes exposed to 1,000 mg/kg-day ETBE, and F1 generation males exposed to
500 mg/kg-day ETBE. However, histology was only performed if abnormal morphology was
detected at necropsy, and, therefore, there are limited histological data to support hepatotoxicity.
Slight to moderate centrilobular hypertrophy was observed in all of the high-dose animals
examined (five total males [3/3 F0 and 2/2 F1 ]) and was not seen in the single control F0
generation male examined. Only the livers from these six males were examined histologically
out of all of the rats in all three generations. Absolute and relative kidney weights were also
significantly increased in F0 generation males and F1 generation males and females. Kidney
weight was increased at lower doses (i.e., at 250 mg/kg compared to 1,000 mg/kg) and to a
07/14/2009
60
DRAFT - DO NOT CITE OR QUOTE
-------
greater extent (e.g., for F1 generation rats exposed to 1,000 mg/kg-day kidney weight was up by
58% for both absolute and relative weight in males and up by 11 and 10% for absolute and
relative weight in females) in males compared to females. Acidophilic globules were observed
in 5/6 F0 and 3/4 F1 generation males at the high dose. In addition, tubular basophilia (4/6),
peritubular fibrosis (3/6), and proteinaceous casts (1/6) were observed in the F0 males at the high
dose (1,000 mg/kg-day). Histology was also performed on kidneys from six F1 males (four
high-dose, one mid-dose [500 mg/kg-day], and one control) due to the presence of macroscopic
lesions in these animals. Tubular basophilia was observed in the control male, the mid-dose
male, and two of the four high-dose males. Peritubular fibrosis was also observed in the control
male, the mid-dose male, and two of the four high-dose males. The one mid-dose male
examined also had sloughed degenerated/necrotic cells in the tubular lumen. Only the kidneys
from these 12 males were examined histologically out of all of the rats in all three generations. It
is difficult to determine conclusions with this limited and select histology. No data are available
to determine the presence or absence of acidophilic globules in females or if the presence of
acidophilic globules was dose-responsive in males because histology was only performed if
abnormal morphology was detected at necropsy.
For all generations, transient ptyalism (excessive salivation) was observed in a dose-
related trend in most treated males and some females. The ptyalism was associated with dosing
and was resolved within 1 hour of gavage for all females and the majority of males. Although
reduced weight gain was observed in some of the parental animals as part of the developmental
toxicity and reproductive toxicity studies over short-term exposure, there was no effect of ETBE
exposure on weight gain over longer treatment periods. For example, pregnant Sprague-Dawley
rats treated with ETBE for 2 weeks (i.e., GDs 5-19) at 1,000 mg/kg-day exhibited an 11%
decrease in maternal body weight gain and a 17% decrease in net weight gain (CIT, 2004a,
unpublished report). However, there was no effect of 18 weeks of ETBE exposure at
1,000 mg/kg-day on body weight in males or females when body weight change was calculated
over the entire treatment period (CIT, 2004b, unpublished report).
ETBE is cleaved to acetaldehyde and TBA by microsomal CYPs, most likely by
CPY2A6, 3A4, or 2B6 in humans and by CYP2B1 in rats. However, acetaldehyde appears to be
primarily a point-of-contact toxicant, and it is unlikely that humans will ingest ETBE in
quantities sufficient to elicit acetaldehyde-related toxicity in the gastrointestinal tract. There are
a number of subchronic and chronic studies of TBA (Archarya et al., 1997, 1995; NTP, 1995;
Lindamood et al., 1992).
F344/N rats were treated for 2 years with TBA via drinking water at estimated doses of 0,
90, 200, and 420 mg/kg-day (males) and 0, 180, 330, and 650 mg/kg-day (females) (NTP, 1995).
Final body weights were reduced at the high dose to a similar extent in males and females,
despite the different dose levels. Mineralization in the kidney was found in males even at the
lowest dose and was thought to be related to alpha2u-globulin-induced nephropathy. Focal renal
07/14/2009
61
DRAFT - DO NOT CITE OR QUOTE
-------
tubule hyperplasia and adenomas were observed in males. Other significant pathology was
generally related to malignancies. B6C3Fi mice were treated at 0, 540, 1,040, and 2,070 mg/kg-
day (males) and 0, 510, 1,020, and 2,110 mg/kg-day (females). Thyroid follicular gland cell
hyperplasia was observed in males at all doses and in females at the two highest doses, although
thyroid effects have not been corroborated in other studies.
Acharya et al. (1997, 1995) treated rats for 10 weeks with 0.5% TBA in drinking water.
They observed hepatic centrilobular necrosis, vacuolation in hepatocytes, and loss of hepatic
architecture. They also detected considerable kidney pathology, such as degeneration of renal
tubules, degeneration of the basement membrane of Bowman's capsule, and vacuolation of
glomeruli. Lindamood et al. (1992) conducted a 90-day subchronic study of TBA in drinking
water in rats and mice at concentrations of up to 4%. The pathologic findings were
predominantly in the urinary tract: calculi, renal pelvic dilatation, thickening of the bladder
mucosa, and transitional epithelial hyperplasia and inflammation.
The limited published data on oral exposure to ETBE are insufficient to determine if the
observed effects of ETBE-exposure, such as increased liver and kidney weight, are related to the
parent compound or its metabolites.
An overview of animal studies with oral exposure to ETBE is provided in Table 4-8.
07/14/2009
62
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8. Oral toxicity studies for ETBE
NOAEL
LOAEL
Species
Dose /duration
(mg/kg-day)
(mg/kg-day)
Effect
Reference
Subchronic studies
Rat (Sprague-
0, 250, 500,
500
1,000
Reduced maternal bodv weisht and
CIT, 2004a,
Dawley)
(24 pregnant
1,000 mg/kg-day,
gavage in corn oil
weisht sain
[-11 and -17% net gain]
unpublished
females/group)
on GDs 5-19
1,000
Not detected
Fetal toxicity or developmental
toxicity
Rat (Sprague-
0, 250, 500,
250
500
Reduced bodv weisht sain
CIT, 2004b,
Dawley)
(two
generations of
25/sex/group)
1,000 mg/kg-day;
gavage in corn oil
for 18 wks,
including
[-22% inFO males at 1,000 mg/kg-
day and -29% in F0 males at
500 mg/kg-day in last '/i of
treatment (day 85 to 113)]
unpublished
premating,
250
500
Increase in liver weishts
mating, gestation,
[(+17% absolute +24% relative in
and weaning
F0 males, +27% absolute +25%
relative inFl males, +10%
absolute +9% relative in F1
females at 1,000 mg/kg-day) and
(+14% absolute +11% relative in
F1 males at 500 mg/kg-day)]; with
hepatic centrilobular hypertrophy
in F0 and F1 males at
1,000 mg/kg-day
Not detected
250
Increase in kidnev weishts
[(+21% absolute +28% relative in
F0 males, +58% absolute +58%
relative inFl males, +11%
absolute +10% relative in F1
females at 1,000 mg/kg-day) and
(+15% absolute +18% relative in
F0 males, +22% absolute +19%
relative inFl males at 500 mg/kg-
day) and +11% relative in F1 and
F0 males at 250 mg/kg-day]; with
acidophilic globules in F0 and F1
males at 1,000 mg/kg-day
1,000
Not detected
Reproductive and developmental
performance
Chronic studies
Rat (Sprague-
0, 250,
Not detected
250
Tumor incidence in uterus
Maltoni et
Dawley)
1,000 mg/kg-day,
(malignant)
al., 1999
(60/sex/group)
4 days/week for
104 weeks;
gavage in olive
oil
250
1,000
Pathologies of oncological interest of
the mouth epithelium
(total) in males
4.6.2. Inhalation
There is limited information on the noncancer toxicity of ETBE by the inhalation route in
humans and animals. Nihlen et al. (1998b) exposed eight volunteers for 2 hours to ETBE
07/14/2009
63
DRAFT - DO NOT CITE OR QUOTE
-------
concentrations of up to 50 ppm (the highest acute dose considered safe in the country where the
study was conducted). The exposed volunteers experienced irritation in the throat and airways,
nasal swelling, a bad taste in the mouth, and slightly impaired lung function. The authors found
some markers of inflammation, but there was no correlation with ETBE exposure levels.
There is evidence for kidney-related effects following inhalation exposure to ETBE, with
the majority of the data coming from the 13-week inhalation study by Medinsky et al. (1999) in
F344 rats and CD-I mice. Absolute kidney weight was increased in male and female F344 rats,
but not CD-I mice. The increased kidney weight in females was accompanied by negative
histopathology and was not associated with additional evidence of nephrotoxicity, with the
exception of a less than twofold increase in cellular proliferative response at early time points
(after 1 and 4, but not 13 weeks of exposure). In male rats, the increased kidney weight was
associated with hyaline droplet accumulation with alpha2U-globulin immunoreactivity, a dose-
response in a sustained increase of greater than twofold in cellular proliferation (after 1, 4, and
13 weeks of ETBE exposure) determined by the increased LI, and a dose-response in
regenerative foci (Medinsky et al., 1999). The Medinsky et al. (1999) kidney data included a
semi-quantitative measure of hyaline droplets, and a weak dose-response in cellular proliferation.
Support for these data for a male rat-specific alpha2U-globulin-mediated mode of action as
presented in the Risk Assessment Forum Technical Panel Report (U.S. EPA, 1991b) is discussed
in Section 4.6.3. There is only one additional study that provides evidence relative to the effect
of inhalation exposure to ETBE on the kidney. White et al. (1995) found an increase in the
absolute kidney weight of male, but not female, Sprague-Dawley rats after 4 weeks of exposure
to 4,000 ppm ETBE. Histopathology of the kidney of both males and females was negative.
Several studies demonstrated an effect of inhalation exposure to ETBE on the liver, an
effect that was largely restricted to an increase in absolute or relative liver weight. Medinsky et
al. (1999) exposed F344 rats and CD-I mice for 13 weeks to 500-5,000 ppm ETBE. Absolute
liver weight was increased in male rats and mice of both sexes at 1,750 ppm, and in mice and
rats of both sexes at 5,000 ppm. Absolute liver weight was also increased in male and female
Sprague-Dawley rats after 4 weeks of inhalation exposure to 4,000 ppm ETBE (White et al.,
1995). Although serum levels of hepatic enzymes were measured and histological observations
were made of the livers in both rat studies, there was no additional support for hepatotoxicity in
the rat data. The incidence of centrilobular hypertrophy in the livers of both male and female
mice exposed to 5,000 ppm was significantly higher than in controls. Hepatic proliferation was
also measured in the CD-I mice as part of the Medinsky et al. (1999) study, although it was not
measured in rats. Dose-related increases in the LI were seen in the livers of male mice exposed
to 1,750 and 5,000 ppm after 1 and 4 weeks of exposure (but not after 13 weeks) and female
mice exposed to 1,750 and 5,000 ppm after 1 and 13 weeks of exposure (but not after 4 weeks);
the effect was thought to be consistent with a mitogenic response of the liver to ETBE. Small
statistically significant changes in peripheral hematology parameters were also measured in the
07/14/2009 64 DRAFT - DO NOT CITE OR QUOTE
-------
Medinsky et al. (1999) study in rats that were not considered to be clinically significant by the
authors. There was a 1% increase in MCV in female rats exposed to 5,000 ppm ETBE at the
interim sampling point and females exposed to both 1,750 and 5,000 ppm at the 13-week time
point. There was a 4% decrease in MCHC in male rats exposed to 5,000 ppm ETBE at the
interim sampling point and a 2.5% decrease in males exposed to both 1,750 and 5,000 ppm at the
13-week time point. The range of historical values is not discussed by the authors and no
changes in MCV or MCHC were observed in the parallel study in CD-I mice at any dose. There
was no pattern to support a relationship between the small changes in MCV or MCHC and
increased liver weight or any other liver data in rats and mice from the available studies. Such a
relationship would support acetaldehyde-mediated hepatotoxicity (i.e., via one of the main
metabolites of ETBE), as elevated MCV is a good predictor of liver disease in alcoholics
(Conigrave et al., 1993). Therefore, in the absence of any histopathological evidence or serum
enzyme evidence of hepatotoxicity, and with no evidence of a relationship between the 1%
increase in MCV in female rats and observed minimal hepatic effects (e.g., increased liver
weight in female rats at 5,000 ppm only compared the 1% MCV at 1,750 and 5,000 ppm or the
absence of elevated MCV in male rats or mice of either sex despite increased liver weight in
male rats and CD-I mice of both sexes), the small increase in MCV in female rats only was not
considered support for hepatotoxicity.
There was no evidence of neurotoxicity from two studies that performed a FOB in rats
after 4 or 13 weeks inhalation exposure at concentrations up to 5,000 ppm (Dorman et al., 1997;
White et al., 1995). However, high doses of ETBE (4,000-5,000 ppm) were associated with
ataxia and sedation, described by the authors as being associated with transient CNS depression.
In both experiments, signs associated with CNS depression were absent within 15 minutes to
1 hour following exposure.
No typical reproductive or developmental toxicity studies of inhalation exposure to
ETBE were identified, although the Medinsky et al. (1999) 13-week exposure study of F344 rats
did include histology of the gonads and some reproductive tissue. Degenerated spermatocytes
were found in the testes of rats from all treatment groups, including controls. The percentage of
seminiferous tubules containing degenerated spermatocytes was significantly increased at the
two high doses. The authors did not observe any other effects on the testes or reproductive tissue
following ETBE exposure from the other data collected (e.g., epididymal histopathology and
examination of the seminiferous tubules for a shift in developmental stage or the presence of
lumenal debris). However, standard endpoints of spermatocyte function that would be included
in a two-generation reproduction and fertility study (e.g., epididymal sperm motility, epididymal
sperm count, and epididymal sperm morphology) were not included in the Medinsky et al.
(1999) study.
There is no evidence of other systemic effects resulting from inhalation exposure to
ETBE from the limited published data available. However, Medinsky et al. (1999) reported a
07/14/2009 65 DRAFT - DO NOT CITE OR QUOTE
-------
treatment-related increase in the incidence of congestion in the bone marrow of female rats. No
additional description of the observation was provided by the authors. The bone marrow
histology was considered not clinically relevant by the authors, as it was not supported by any
changes in hematopoietic cell populations.
ETBE is metabolized to acetaldehyde and TBA. Acetaldehyde, primarily a point-of-
contact toxicant, is an effective respiratory tract irritant. Although results from inhalation studies
with ETBE in humans (Nihlen et al., 1998a, b) provide some evidence for irritant action, data are
unavailable to determine if the effect is related to the parent compound or metabolites.
In a subchronic inhalation study of TBA in F344/N rats and B6C3Fi mice (0, 135, 270,
540, 1,080, or 2,100 ppm, 6 hours/day, 5 days/week), Mahler (1997) reported outward symptoms
that were consistent with alcohol toxicity (rough coat, hypoactivity, ataxia, prostration).
Absolute and relative kidney weights were increased in both sexes of rats, as were relative liver
weights in female rats, at the two highest doses. Male rats displayed dose-related increases in the
severity of chronic nephropathy. Gonad histology was unaffected in both sexes.
There are few studies of the toxicity of ETBE following inhalation exposure. Available
data suggest hepatic and renal effects. The database is insufficient to determine if the observed
effects of ETBE exposure, such as increased liver and kidney weight, are related to the parent
compound or its metabolites.
An overview of animal studies with inhalation exposure to ETBE is provided in
Table 4-9.
07/14/2009
66
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-9. Subchronic inhalation toxicity studies for ETBE
Species
Dose /duration
NOAEL'
(ppm)
LOAEL"
(ppm)
Effect
Reference
Rat (Sprague-
Dawley)
(10/sex/group)
0, 501, 2,090, and
3,910 ppm,
6 hrs/day,
5 days/wk, for
4 wks
501
2,090
Increase in liver weisht
[+16.8% absolute 16.1% relative in
males and +9.5% abs. +12.5% rel.
in females at 3,910 ppm; +10% rel.
in females at 2,090 ppm]
White et
al., 1995
2,090
3,910
Increase in kidnev weisht
[+12.8% absolute in males]
2,090
3,910
Increase in adrenal weisht
[+13.7% absolute in males]
Rat (F344)
(48/sex/group)
0, 505, 1,748, and
4,971 ppm,
6 hrs/day,
5 days/wk, for
13 wks
505
1,748
Increase in absolute liver weisht
[+32.4% in males and +25.8% in
females at 4,971 ppm; +14.2% in
males at 1,748 ppm]
Medinsky
etal., 1999
505
1,748
Increase in absolute kidnev weisht
[+19.3% in males and +21.3% in
females at 4,971 ppm; +9.8% in
males and +12.2% in females at
1,748 ppm]
Not detected
505
Renal effects in males, i.e.,
regenerative foci, hyaline droplet
accumulation with presence of
alpha2u-globulin
505
1,748
Increase in % seminiferous tubules
with SDcrmatic deseneration
[+12.7% at 4,971 ppm and +7.8% at
1,748 ppm]
1,748
4,971
Increase in absolute adrenal weisht
[+34.3% in males and +17.8% in
females]
Mice (CD-I)
(40/sex/group)
0, 501, 1,754, and
4,962 ppm,
6 hrs/day,
5 days/wk, for
13 wks
501
1,754
Increase in absolute liver weisht
[+18% in males and +32.6% in
females at 4,962 ppm; +12.9% in
males and 19% in females at
1,754 ppm]; with hepatic
centrilobular hypertrophy in both
sexes at 4,962 ppm
Medinsky
etal., 1999
Rat (F344)
(12 sex/group
from Medinsky et
al., 1999; Bond et
al., 1996a)
0, 505, 1,748, and
4,971 ppm,
6 hrs/day,
5 days/wk, for
13 wks
4,971
Not
detected
Transient ataxia immediately after
exposure
Dorman et
al., 1997
"Reported concentrations are non-duration-adjusted animal exposures.
07/14/2009 67 DRAFT - DO NOT CITE OR QUOTE
-------
4.6.3. Mode of Action Information
There are only a few studies that have been conducted to evaluate the effects of exposure
to ETBE by the oral or inhalation route. There is a single inhalation-exposure study (Medinsky
et al., 1999) that was designed to provide information relevant to the determination of
alpha2U-globulin-associated nephropathy under the assumption that ETBE would lead to renal
cytotoxicity based on its structural similarity to MTBE, a compound that has been shown to be a
renal toxicant in male rats (Cal EPA, 1999; Prescott-Mathews, et al., 1997). Detailed studies on
the potential toxicity of ETBE to any organ system have not been conducted.
Kidney Effects:
Several studies (see Sections 4.2.3.1 and 4.3) have demonstrated that oral or inhalation
exposure to ETBE results in kidney effects. For example, ETBE exposure was associated with
increased kidney weight in several studies: in male Sprague-Dawley rats in a 4-week inhalation
study (White et al., 1995); in male and female F344 rats, but not CD-I mice, in a 13-week
inhalation study (Medinsky et al., 1999); and in male and female Sprague-Dawley rats in an oral
two-generation reproduction and fertility study (CIT 2004b, unpublished report). The Medinsky
et al. (1999) study focused on the potential role of alpha2U-globulin based on the assumption that
there would be overt renal effects associated with ETBE because of the renal toxicity found with
the methyl analog (MTBE). The data from Medinsky et al. (1999) also showed increased
hyaline droplet formation, the presence of alpha2U-globulin in the proximal tubules, a twofold
increase in LI, and the presence of regenerative foci in male F344 rats but not female rats.
Regenerative foci are associated with repair of damaged tubules. Therefore, the presence of
regenerative foci represents indirect evidence of necrosis, as some cells of the proximal tubule
would have had to have died and the remaining cells undergone regenerative proliferation for
regenerative foci to be observed.
General issues concerning the determination of alpha2U-globulin-associated nephropathy
Alpha2U-globulin derived from hepatic synthesis is unique to male rats; it is not generally
found in female rats or in mice or humans of either sex (U.S. EPA, 1991b; Alden, 1986).
Although young male rats may have some hyaline droplets in proximal tubules with alpha2u-
globulin, chemically-induced accumulation of alpha2U-globulin in the proximal tubule is
restricted to mature male rats. Therefore, data from mice and female rats can be used to
demonstrate that chemical-associated nephrotoxicity is male-rat specific. The increased
alpha2u-globulin accumulation is proposed to result from reduced renal catabolism of the
alpha2U-globulin-chemical complex. The resulting accumulation is thought to initiate a sequence
of events leading to chronic proliferation of the renal tubule epithelium, as well as an
exacerbation of chronic progressive nephropathy. The histopathological sequence in mature
male rats, potentially leading to the formation of renal tumors consists of the following:
07/14/2009
68
DRAFT - DO NOT CITE OR QUOTE
-------
• Excessive accumulation of hyaline droplets containing alpha2U-globulin in renal
proximal tubules
• Subsequent cytotoxicity and single-cell necrosis of the tubule epithelium
• Sustained regenerative tubule cell proliferation, providing exposure continues
• Development of intralumenal granular casts from sloughed cell debris associated with
tubule dilation and papillary mineralization
• Foci of tubule hyperplasia in the convoluted proximal tubules, and
• Renal tubule tumors
In the absence of information demonstrating that the alpha2U-globulin processes is
operative (as in criteria 1-3 below), it should be assumed that male rat nephropathy is relevant
for risk assessment purposes. Additional categorical guidelines available from the Risk
Assessment Forum Technical Panel Report (U.S. EPA, 1991b) outline the data necessary to
determine the involvement of alpha2U-globulin and the full report should be consulted rather than
the summary presented here.
The following information from adequately conducted studies of male rats is used for
demonstrating that the alpha2u-globulin process may be a factor in any observed renal effects-an
affirmative response in each of the three categories is required. If data are lacking for any of the
criteria in any one category, the available renal toxicity data should be considered relevant to
humans and analyzed in accordance with standard risk assessment principles. The three
categories of information and criteria are as follows:
(1) Increased number and size of hyaline droplets in the renal proximal tubule cells of
treated male rats. The abnormal accumulation of hyaline droplets in the proximal tubule
helps differentiate alpha2U-globulin inducers from chemicals that produce renal tubule
toxicity by other modes of action.
(2) Accumulating protein in the hyaline droplets is alpha2U-globulin. Hyaline droplet
accumulation is a nonspecific response to protein overload and, thus, it is necessary to
demonstrate that the protein in the droplet is, in fact, alpha2U-globulin.
(3) Additional aspects of the pathological sequence of lesions associated with alpha2U-
globulin nephropathy are present. Typical lesions include single-cell necrosis,
exfoliation of epithelial cells into the proximal tubular lumen, formation of granular casts,
linear mineralization of papillary tubules, and tubule hyperplasia. If the response is mild,
not all of these lesions may be observed. However, some elements consistent with the
pathological sequence must be demonstrated to be present.
The Risk Assessment Forum Technical Panel Report (U.S. EPA, 1991b) also suggests
additional information that may be useful for the analysis including sustained cell division in the
proximal tubule of the male rat. This relates to a sustained increase in cell replication of the
renal tubule at doses used in the studies and a dose-related increase in atypical hyperplasia of the
07/14/2009
69
DRAFT - DO NOT CITE OR QUOTE
-------
renal tubule is consistent with the alpha2U-globulin process, especially if other laboratory animals
were tested and did not show similar responses.
The report specifically states: "If a compound induces alpha2u-globulin accumulation in
hyaline droplets, the associated nephropathy in male rats is not an appropriate endpoint to
determine noncancer (systemic) effects potentially occurring in humans."
ETBE and alpha2U-globulin-associated renal effects
Criteria (1): Data from Medinsky et al. (1999) demonstrated a dose-dependent increase
in hyaline droplet formation in the proximal tubule of male rats following inhalation exposure to
ETBE for 1, 4, and 13 weeks (Table 4-4). Hyaline droplet severity was expressed as a mean
grade scored with 1 for minimal, 2 for <10%, 3 for 10-25%, 4 for 25-50%, and 5 for >50%) of
the cortex involved with Mallory-Heidenhain staining. The Mallory-Heidenhain staining of
female rats was negative for all time points. Only the mean scores are presented, without
standard errors or statistics. The authors state that there is a concentration-dependent increase in
hyaline droplet accumulation in the male rat kidney, although the differences are small on this
semi-quantitative scale after 1 week of exposure (Table 4-4: severity of 1.2 in controls, 3.4 at
500 ppm, 4.0 at 1,750 ppm, and 4.6 at 5,000 ppm). A similar difference is also reported after
4 weeks of exposure (Table 4-4: severity of 1.8 in controls, 2.6 at 500 ppm, 3.4 at 1,750 ppm,
and 3.8 at 5,000 ppm) and 13 weeks of exposure (Table 4-4: severity of 1.8 in controls, 3.0 at
500 ppm, 3.2 at 1,750 ppm, and 3.8 at 5,000 ppm). Although the data minimally support the first
criteria, a quantitative measure of hyaline droplet number and size would represent stronger data.
Criteria (2): Alpha2u-globulin immunoreactivity was observed in the hyaline droplets of
the renal proximal tubular epithelium of male rats from all exposure groups, and not in female
rats (Medinsky et al., 1999). There was no quantitative determination of the alpha2U-globulin in
the study. The authors state that there is an ETBE-related increase in hyaline droplets and that
these droplets are immunoreactive for alpha2u-globulin (Medinsky et al., 1999).
Criteria (3): There are two additional aspects of the pathological sequence associated
with alpha2U-globulin-related effects that are present in male rats from the Medinsky et al. (1999)
study: sustained cell proliferation and the presence of regenerative foci. In male rats exposed to
ETBE, after 1 or 4 weeks of exposure, a greater than twofold increase in the cell proliferation or
LI in the renal tubule occurred at the high dose only (Table 4-5). The increased cell proliferation
in high-dose males was sustained through 13 weeks of exposure as determined by the LI. After
13 weeks of exposure, the LI was statistically increased by more than twofold in males relative
to control for all doses of ETBE, and, therefore, at all doses demonstrating elevated hyaline
droplet severity. However, the value of the LI at 13 weeks, does not demonstrate a dose-
response (LI = 0.91, 2.16, 3.40, and 2.47 for 0, 500, 1,750, and 5,000 ppm ETBE). The
difference between the LI for the low dose (500 ppm) and the control was significant atp< 0.05,
while the difference between the two higher doses and the control were significant at/? < 0.001.
07/14/2009
70
DRAFT - DO NOT CITE OR QUOTE
-------
Tubule cell proliferation was not sustained in female rats exposed to ETBE, although cell
proliferation was statistically increased in female rats after 1 week of exposure. The less than
twofold increase was observed at all doses with no evidence of a dose-response in female rats.
The LI remained increased by less than twofold in female rats at the high dose (5,000 ppm) at the
4 week time point. No effect on the renal LI in female rats was seen at any exposure
concentration after 13 weeks of exposure to ETBE.
The presence of regenerative foci is also an indication of alpha2u-globulin-associated
effects. The number of regenerative foci was increased in male rat kidneys in the 1,750 and
5,000 ppm dose groups, respectively, after 4 weeks of exposure (Table 4-4). After 13 weeks of
exposure, there was a dose-response relationship in the mean number of regenerative foci
observed with 2, 11, 17, and 34 foci for 0, 500, 1,750, and 5,000 ppm ETBE, respectively. The
available data do not indicate other aspects of the pathological sequence of lesions associated
with alpha2U-globulin-related effects such as single-cell necrosis, exfoliation of the epithelial
cells into the proximal tubular lumen, granular casts, or linear mineralization of papillary tubules.
The Risk Assessment Forum Technical Panel Report (U.S. EPA, 1991b) notes that if the
alpha2U-globulin response is mild, not all of these lesions may be observed, but states that
elements consistent with the pathological sequence must be demonstrated to be present. The
absence of any of these typical lesions in the pathological sequence in the available data for
ETBE presents considerable uncertainty in the determination of alpha2u-globulin as a mode of
action for observed effects. It is important to note that the presence of regenerative foci
represent indirect evidence of necrosis because regenerative foci are associated with repair of
damaged tubules, and therefore, some cells of the proximal tubule would have had to have died
and the remaining cells undergone regenerative proliferation for this to occur. However, direct
evidence of lesions associated with alpha2U-globulin-related pathological sequence would
represent stronger data.
There is not enough information to determine if some or all of ETBE-induced kidney
effects are caused by the parent compound or its metabolite, TBA. Although there is no
evidence of renal cancer in the single study of ETBE carcinogenicity, the nephropathy and
subsequent cancer incidence in male rats exposed to TBA has been postulated to be mediated by
alpha2u-globulin accumulation (Williams and Borghoff, 2001; Takahashi et al., 1993). However,
the dose associated with an increase in the severity of chronic nephropathy in male rats in a
13-week inhalation study (Mahler, 1997) is below doses associated with accumulation of protein
droplets in the kidney either in that study (Mahler, 1997) or in a 10-day exposure (Borghoff et
al., 2001), suggesting that alpha2u-globulin may not be the mechanism or the only mechanism for
TBA-associated nephropathy. More detailed evaluation of nephrotoxicity associated with TBA
and data on the relationship between ETBE metabolism and toxicity will be required to answer
this question.
07/14/2009
71
DRAFT - DO NOT CITE OR QUOTE
-------
ETBE and renal effects in female rats
As stated above, ETBE exposure was associated with increased kidney weight in female
F344 rats in a 13-week inhalation study (Medinsky et al., 1999) and female Sprague-Dawley rats
from an unpublished oral two-generation reproduction and fertility study (CIT 2004b,
unpublished report), but not in female Sprague-Dawley rats in a 4-week inhalation study (White
et al., 1995), or in female CD-I mice in a 13-week inhalation study (Medinsky et al., 1999).
There was no evidence of renal lesions, and both Mallory-Heidenhain staining and alpha2u-
globulin immunoreactivity of kidneys from female rats were negative for all time points (1, 4,
and 13 weeks) in the 13-week inhalation study (Medinsky et al., 1999). Although renal tubule
cell proliferation was statistically increased in all female rats exposed to ETBE after 1 week of
exposure, the increase was less than twofold, and the increase was not sustained in female rats
exposed to ETBE. The LI remained elevated by less than twofold in female rats at the high dose
(5,000 ppm) at the 4 week time point and no effect on renal cell proliferation was seen at any
exposure concentration in female rats after 13 weeks of exposure to ETBE. The short-term
increase in proliferation in females (and in the absence of alpha2u-globulin) suggests that a non-
alpha2U-globulin mechanism may be responsible for this short-term effect in females. No
increase in LI was seen in female rats after similar short-term (10-day) inhalation exposure to
TBA (Borghoff et al., 2001) at concentrations up to 1,750 ppm or MTBE (Prescott-Mathews et
al., 1997) at concentrations up to 3,013 ppm.
Summary
There are data indicating renal toxicity in male and female rats following ETBE
exposure. ETBE exposure causes an increase in renal weight in both male and female rats and a
semi-quantitative increase in hyaline droplet formation in the proximal tubule of male rats (and
not female rats). The presence of alpha2U-globulin in the hyaline droplets has been confirmed in
male rats and not female rats by immunoreactivity studies, sustained cell proliferation in cells of
the renal tubule has been shown to occur in male rats with a transient increase observed in
female rats, and regenerative foci have been identified in male rats. The presence of regenerative
foci indicates that cellular repair mechanisms are responding to cell death. Typical lesions in the
pathological sequence associated with alpha2u-globulin nephropathy include: evidence of single
cell necrosis, exfoliation of epithelial cells into the proximal tubule lumen, the presence of
granular casts in the renal tubule, linear mineralization of tubules within the renal papilla, and
tubule hyperplasia. These effects have not been reported in the available studies. Female rats
have been shown to exhibit an increase in kidney weight and an increase in renal tubule cell
proliferation that was not sustained following ETBE exposure. Given the available data, a
determination cannot be made as to whether alpha2U-globulin accumulation is the mode of action
or the only mode of action for renal effects observed in male and female rats associated with
ETBE exposure.
07/14/2009 72 DRAFT - DO NOT CITE OR QUOTE
-------
All of the data supporting alpha2U-globulin-associated kidney effects are from a single
subchronic study (Medinsky et al., 1999). A well-conducted chronic study to examine the
effects of ETBE exposure on the kidney of rats and mice of both sexes would be beneficial to
determine the role of alpha2U-globulin in observed renal toxicity associated with ETBE.
4.7. EVALUATION OF CARCINOGENICITY
4.7.1. Summary of Overall Weight-of-Evidence
No epidemiological evidence is available that allows for the assessment of the
carcinogenic potential of ETBE in humans. ETBE has not been shown to act as a genotoxicant
in any of the available assays of ETBE, although some studies did not control for potential
evaporative loss of ETBE during exposure. In addition, a G/ETBE mixture exposure study
reported equivocal results in the micronucleus test. In the only available animal cancer bioassay,
Maltoni et al. (1999) reported an increased incidence of four tumors or combined tumors and
precancers reported as pathologies of oncological interest in Sprague-Dawley rats exposed to 0,
250, and 1,000 mg/kg-day of ETBE by gavage 4 days/week for 104 weeks. The authors reported
a statistically significant increase in the total pathologies of oncological interest of the mouth
epithelium in males at the high dose of ETBE and total malignant uterine tumors (a classification
that included vaginal schwannomas) in females that was restricted to the low dose of ETBE only.
Nonsignificant increases in hemolymphoreticular neoplasias in both sexes and total pathologies
of oncological interest of the forestomach in males that was restricted to the low dose of ETBE
only were also reported. No other tumor types were found to be elevated in this study. The
Maltoni et al. (1999) study utilized two dose levels and both doses tested caused increased
mortality at early time points.
The Maltoni et al. (1999) study showed early mortality associated with both doses of
ETBE. The ETBE-associated mortality presents a complication on the ability to interpret
increased tumor incidences that were reported to be restricted to the low-dose group (i.e., total
malignant uterine tumors in females and total pathologies of oncological interest of the
forestomach in males) based on the possibility that the increased mortality in high-dose animals
could potentially mask an observation of increased tumor incidence that might otherwise be seen
at the high dose. Individual animal data were not provided, and therefore, survival analyses or a
time to tumor analysis could not be conducted. It is not known if the increased mortality was
secondary to toxic or carcinogenic effects.
No criteria were defined for the determination of dysplasia or for potential grading of
lesions, which introduces uncertainty in the reported increase in the total pathologies of
oncological interest for the forestomach and the mouth epithelium where dysplasias represent
high portions of those totals (i.e., 58 and 69% of the total pathologies in the forestomach were
squamous cell dysplasias and 100 and 73% of the total pathologies in the mouth epithelium were
squamous cell dysplasias at 250 and 1,000 mg/kg respectively) (Table 4-1). The grade or rating
07/14/2009 73 DRAFT - DO NOT CITE OR QUOTE
-------
of dysplasia is particularly important to the predictive value of these precancers, as the majority
of lesions with mild to moderate dysplasia do not progress into cancer (Rosin et al., 2000;
Schepman et al., 1998). Precancerous lesions identified as dysplasias represent the majority of
lesions that the authors include within their categories of total pathologies of oncological interest
for both the mouth epithelium and forestomach, which limits the ability to interpret the relevance
of the increased incidences of these two categories of lesions.
Maltoni et al. (1999) did not indicate the pathological information used to delineate
tumors associated with the tongue from other oral pathologies, or the grouping of tissues of the
uterus and vagina. These tissues were also listed separately and when the vaginal schwannomas
are removed from the uterine tumors, the total malignant tumors of the uterus (Table 4-1) were
not statistically increased relative to control at either dose by % test (the same test used by the
authors on the grouped data).
The mode(s) of action of ETBE for the increased incidence of tumors or combined
tumors and precancers reported in Maltoni et al. (1999) are unknown. It is unclear whether or
not observed effects are associated with the parent compound or the metabolites. A limited
number of unpublished in vivo and in vitro genotoxicity tests demonstrate no DNA damage
resulting from exposure to ETBE and all available assays of genotoxicity of ETBE were
negative. Carcinogenicity data from the metabolites of ETBE (TBA and acetaldehyde) or
MTBE, the methyl analog of ETBE, indicate dissimilar tumor types for these compounds. The
available carcinogenicity data for MTBE support the possibility that ETBE may cause an
increase in lymphohemoreticular neoplasias, as there is evidence that exposure to MTBE is
associated with an increase in lymphohematopoietic cancer in female rats (Belpoggi et al., 1998,
1997, 1995). However, as discussed in Sections 4.2.3 and 4.6.2, the increase in these tumors
following ETBE exposure is slight, not statistically significant, and shows no evidence of dose
response (3/60, 8/60, and 6/60 in males and 3/60, 6/60, and 5/60 in females, at 0, 250, and
1,000 mg/kg-day respectively).
The data reported in Maltoni et al. (1999) indicated that exposure to ETBE is associated
with a statistically significant increased incidence in total uterine tumors in females and a
combined total pathologies of oncological interest of the mouth epithelium in males. Although
the total pathologies of oncological interest in the mouth epithelium includes precancers as well
as tumors, the total uterine tumors does not include any precancers. Maltoni et al. (1999) also
report a nonstatistical increase in hemolymphoreticular neoplasias in both sexes and total
pathologies of oncological interest of the forestomach in males. The increased incidence of
uterine tumors in female rats was restricted to the low dose of ETBE, and the individual animal
data are not presented to allow a time to tumor analysis to evaluate if the increased mortality in
the high-dose females prevented a dose-response in uterine tumors. In addition, uterine tumors
alone are not statistically increased, and it is only when tumors from additional tissues (i.e.,
vaginal schwannomas) are included that the low dose increase is statistically significant. The
07/14/2009 74 DRAFT - DO NOT CITE OR QUOTE
-------
reported increase in pathologies of oncological interest of the mouth epithelium in males at the
high dose of ETBE contains considerable uncertainty because of the dependence on the inclusion
of squamous cell dysplasias (which represent 73% of the total at the high dose) without the
histopathological criteria necessary to evaluate the predictive value of these precancers. The
increase in lymphohemoreticular neoplasias and total pathologies of oncological interest of the
forestomach in males are not statistically significant and showed no evidence of a dose response.
Thus, under EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), there
is suggestive evidence of human carcinogenicity of ETBE based on the only oral cancer bioassay
in Sprague-Dawley rats (Maltoni et al., 1999). The classification is consistent with an example
provided in EPA's Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a) which
suggests suggestive evidence of human carcinogenicity when a statistically significant increase is
seen at one dose only, but no significant response at the other doses, and no overall trend.
4.7.2. Synthesis of Human, Animal, and Other Supporting Evidence
No studies are available that assess the carcinogenic potential of ETBE in humans. In the
single chronic cancer bioassay evaluating the carcinogenic potential of ETBE, Maltoni et al.
(1999) exposed 8-week-old Sprague-Dawley rats (60/group) to 0, 250, and 1,000 mg/kg-day
ETBE by gavage 4 days/week for 104 weeks. Animals were allowed to die a natural death; upon
death, animals were necropsied and tissues were preserved in 70% ethyl alcohol for later
histopathological evaluation and classification. The tumor data were presented according to dose
group, tumor site, and histiotype, and in some cases, malignant tumors were distinguished from
benign or total tumors (Table 4-1). There was a statistically significant increase in the incidence
of total pathologies of oncological interest of the mouth epithelium (including oral cavity,
tongue, and lips) in males at the high dose (1,000 mg/kg-day) of ETBE and total malignant
uterine tumors (a classification that included vaginal schwannomas) in females at the low dose
(250 mg/kg-day) of ETBE only. No other tumor types were found to be significantly elevated in
this study; however, the authors also reported nonsignificant increases in hemolymphoreticular
neoplasias in both sexes and total pathologies of oncological interest of the forestomach in males
at the low dose of ETBE.
The authors reported limitations associated with the design and conduct of the study,
including the use of only two dose levels, testing only a single animal species, and the fact that
both test doses of ETBE caused increased mortality of ETBE-exposed rats.
The increased mortality of animals at both doses of ETBE presents a limitation in the
study and ability to interpret the results from a quantitative perspective. Mortality differed for
each dose and sex such that, for some weeks of the study, mortality was increased in treated
animals relative to controls, whereas for other weeks, there was no difference in mortality, or
treated ETBE-exposed animals had lower mortality. Digitization of the data from the survival
curves in Maltoni et al. (1999) shows an approximately 8-20% increase in mortality of high dose
07/14/2009
75
DRAFT - DO NOT CITE OR QUOTE
-------
females relative to control animals from weeks 56 to 88. A similar increased in mortality (8-
30%) was also observed in the high-dose males relative to controls from weeks 56 to 120. Low-
dose males and females exhibited a smaller increase in mortality (e.g., mortality in males was
increased by 7-17% in low-dose males and 11% in low-dose females). The increased mortality
in low dose males and females was observed at fewer time points than animals exposed to the
high dose of ETBE. For example, low-dose females only displayed increased mortality at
88 weeks of age relative to controls. Low-dose males exhibited increased mortality relative to
controls from weeks 56 to 88 and were similar to controls at other time points. The authors
suggested that due to the increased mortality in the ETBE exposed rats, some of the carcinogenic
effects were only observed at the low dose and not at the high dose.
In the absence of that data on tumor incidence and survival of individual animals, a
quantitative time to tumor analysis cannot be performed. In one of the examples where an effect
is reported at the low exposure dose only, the authors describe an increase in the total incidence
of pathologies of oncological interest (acanthomas, squamous cell dysplasias, and squamous cell
carcinomas) of the forestomach in males at the low dose (250 mg/kg-day), while the incidence at
the high dose (1,000 mg/kg-day) was identical to controls (13/60, 24/60, and 13/60, at 0, 250,
and 1,000 mg/kg-day respectively). The increased mortality associated with high-dose males
after 40 weeks of age, provides some support for the possibility that an effect at the high dose of
ETBE is obscured by increased mortality in this group; however, that suggestion cannot be
evaluated without individual animal data on mortality and tumor incidences. Similarly, Maltoni
et al. (1999) reported a statistical increase in total malignant tumors (carcinomas and sarcomas)
of the uterus (that includes tumors of the uterus and vagina) at the low dose (250 mg/kg-day),
while the tumor incidence at the high dose was identical to controls (2/60, 10/60, 2/60,
respectively). Based on the survival curves, the appearance of malignant uterine tumors would
need to be correlated to weeks 56-88 to support a dose response because the mortality of the
high dose females was only increased relative to controls during that time period. The decreased
mortality of females in the high-dose group relative to controls after 88 weeks of age would
make the detection of tumors more likely in females exposed to the high dose during this time
period, not less likely.
Maltoni et al. (1999) reported dysplasias as one of the pathologies of oncological interest
for both the mouth epithelium and the forestomach. The increase in total pathologies of
oncological interest was only statistically significant for the grouped lesions the mouth
epithelium, not for any individual tumor type or precancer. The authors report the pathologies of
oncological interest of the forestomach as nonsignificantly increased. The criteria for the
determination of dysplasia or for potential grading of lesions were not described for either the
forestomach or the mouth epithelium. No rating of dysplasias was reported for the forestomach,
and two grades of dysplasia were included within the total pathologies of oncological interest of
the mouth epithelium. The inclusion of a separate, more severe classification for squamous cell
07/14/2009 76 DRAFT - DO NOT CITE OR QUOTE
-------
dysplasias borderline with in situ carcinoma in the mouth epithelial data as a distinct category
from squamous cell dysplasias suggests that the criteria for general squamous cell dysplasias
used by Maltoni et al. (1999) for both the mouth epithelium and forestomach may have included
mild to moderate dysplasias, which, as described below, may not be predictive for cancer. This
more severe lesion was only reported in one of the male rats at any dose and the majority of the
total pathologies of oncological interest of the mouth epithelium were general squamous cell
dysplasias (which represented 83, 100, and 73% of the total at 0, 250, and 1,000 mg/kg-day
respectively). The rating of dysplasia is particularly important to the predictive value of these
precancers. Although there is a good correlation between severe dysplasia and eventual
carcinoma, the majority of lesions with mild to moderate dysplasia do not progress into cancer
(Rosin et al., 2000; Schepman et al., 1998). The use of squamous cell dysplasia data without
explicit criteria or the relative number of mild to moderate dysplasia for both the total
pathologies of oncological interest of the mouth epithelium and the forestomach is problematic
to the interpretation of data from Maltoni et al. (1999). In both cases, that the study authors
suggested that there was an increase in total pathologies of interest in male rats, dysplasias
represent high portions of that total (i.e., 58 and 69% of the total pathologies in the forestomach
were squamous cell dysplasias and 100 and 73% of the total pathologies in the mouth epithelium
were squamous cell dysplasias at 250 and 1,000 mg/kg respectively) (Table 4-1). Therefore, the
fact that the precancerous lesions identified as dysplasias represent the majority of lesions that
the authors include within their categories of total pathologies of oncological interest for both the
mouth epithelium and forestomach, limits the ability to interpret the relevance of the increased
incidences of these two categories of lesions reported in Maltoni et al. (1999).
As discussed above, the pathologies of oncological interest in both the mouth epithelium
and forestomach included both tumors and precancers. The pathologies of oncological interest
of the mouth epithelium listed by the authors consisted of several organs including the tongue,
lips, and oral cavity. Of note in this grouping is the issue that tongue cancer is different from
other cancers of the oral cavity and should be considered separately (Sathyan et al., 2006). In
addition, tissues of the uterus and vagina were combined within the category of total malignant
tumors of the uterus, one of the two tumor types listed as statistically increased by Maltoni et al.
(1999). Although, the authors indicated that the incidence of total malignant tumors of the uterus
were increased at the low dose of 250 mg/kg-day (and not at 1,000 mg/kg-day), it appears that
four malignant schwannomas of the uterus-vagina were included in the eight total uterine
sarcomas at 250 mg/kg-day. When the vaginal schwannomas are removed from the uterine
tumors, the total malignant tumors of the uterus (Table 4-1) were not statistically increased
relative to control at either dose by % test (the same test used by the authors on the grouped
data).
There is some evidence of carcinogenic potential for the two primary metabolites of
ETBE; TBA, as indicated in a 2-year study conducted by the National Toxicology Program
07/14/2009
77
DRAFT - DO NOT CITE OR QUOTE
-------
(NTP, 1995; Cirvello et al., 1995), and acetaldehyde (WHO, 1995). The NTP study (NTP, 1995;
Cirvello et al., 1995) concludes that chronic exposure to TBA is associated with an increased
incidence of combined adenoma or carcinoma in the kidney of male F344 rats and thyroid
follicular cell adenoma in female B6C3Fi mice with equivocal evidence of thyroid follicular cell
adenoma or carcinoma in male mice. The World Health Organization (WHO) assessment
suggests that the carcinogenicity of acetaldehyde (increased incidence of nasal tumors in male
and female rats and laryngeal tumors in hamsters) is associated with the irritancy of
acetaldehyde, as it is only seen at concentrations above the NOEL for irritation.
There is also evidence for the carcinogenicity of MTBE, the methyl analogue of ETBE,
which appears to be a multi-site and multi-species carcinogen (Cal EPA, 1999; Mennear, 1997).
Exposure to MTBE is associated with increased incidence of tumors in both sexes of two species
of animals. Specifically, hepatocellular tumors in both male and female mice (Bird et al., 1997;
Burleigh-Flayer et al., 1992), renal tumors in male rats (Bird et al., 1997; Chun et al., 1992),
Ley dig cell tumors in male rats (Belpoggi et al., 1998, 1997, 1995; Bird et al., 1997; Chun et al.,
1992), and lymphohematopoietic cancer in female rats (Belpoggi et al., 1998, 1997, 1995) have
been observed following exposure to MTBE. The Belpoggi et al. (1998, 1997, 1995) studies
were conducted in the same laboratory as the single, preliminary report of results of the ETBE
cancer bioassay (Maltoni et al., 1999). The mode(s) of action for carcinogenicity for ETBE are
unknown for any of the tumor types reported in Maltoni et al. (1999). It is also unclear whether
or not observed effects are associated with the parent compound or the metabolites. The small
database for ETBE contributes to the uncertainty, but carcinogenicity data from the metabolites
of ETBE (TBA and acetaldehyde) or MTBE, the methyl analog of ETBE, are not similar to
either of the tumor types that were statistically increased in Maltoni et al. (1999) (i.e., total
malignant uterine tumors in females at the low dose and total pathologies of oncological interest
of the mouth epithelium in males) as they were not observed with these structurally related
chemicals. The evidence for the carcinogenicity of MTBE supports the possibility that ETBE
may cause an increase in lymphohemoreticular neoplasias, as there is evidence that exposure to
MTBE is associated with an increase in lymphohematopoietic cancer in female rats (Belpoggi et
al., 1998, 1997, 1995). However, as discussed above and in Section 4.2.3, the increase in these
tumors following ETBE exposure is slight, not statistically significant, and shows no evidence of
dose response.
4.7.3. Mode of Action Information
ETBE has not been shown to act as a genotoxicant in most of the tests conducted, with
equivocal results in the micronucleus test. Medinsky et al. (1999) concluded that the increased
hepatocyte labeling indices observed at several time points and all doses tested during a 13-week
subchronic inhalation study in mice might indicate a mitogenic response to ETBE in the liver.
07/14/2009
78
DRAFT - DO NOT CITE OR QUOTE
-------
However, there was no evidence of hepatic tumors in the only chronic ETBE study (i.e., Maltoni
etal., 1999).
4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES
4.8.1. Possible Childhood Susceptibility
The limited data on ETBE (presented in Section 4.3) are insufficient to determine if
ETBE is a teratogen or a developmental toxicant. General patterns of enzymes levels associated
with development suggest that humans at early age may have diminished ability to metabolize
ETBE because of decreased levels or reduced activity of the CYP enzymes that metabolize
ETBE. Direct evidence of enzyme differences from ETBE exposure studies in humans is not
available. Therefore, the limited available developmental toxicity data on ETBE are inadequate
to determine if children are more susceptible to potential ETBE toxicity.
The primary enzymes likely to be involved in ETBE metabolism in humans based on in
vitro studies using liver microsomes are CYP2A6, 3 A4, and 2B6, in rank order of their
contribution to the initial step in metabolism of ETBE from the parent compound to TBA and
acetaldehyde. Much of the knowledge concerning childhood levels of CYPs comes from studies
on drug clearance. Dome and coworkers (Dome et al., 2005; Dome, 2004) have published
evaluations of factors associated with drug clearance vs. ethnicity or age. These data indicate
that clearance of CYP3 A4-metabolized drugs should be threefold lower in neonates than in
adults, while in children 1-16 years of age, it should be about 1.4-fold higher. Data on the
activity of CYP3 A4 from the drug clearance data may be relevant to potential age-associated
susceptibility to ETBE, as CYP3 A4 is one of the three main CYPs likely to contribute to ETBE
metabolism; however, other enzyme activities relevant to ETBE metabolism were not presented
by Dome et al. (2005). Alcorn and McNamara (2002) also used a mathematical approach to
assess hepatic and renal clearance in children. They produced scaling factors derived from
known enzyme activities and physical parameters (body, liver weight). According to their
calculations, clearance of CYP3A4-metabolized drugs in fetuses of less than 30 weeks should be
approximately 4% that of adults, climbing to about 17% immediately after birth, and to 50%
within the first 2 months of life. This was essentially confirmed by findings of de Wildt et al.
(1999). However, these findings may be inconsistent with a report by Blanco et al. (2000), who
investigated CYP3A4 activities in human liver microsomes between birth and 75 years of age. A
linear regression of their data revealed no change of CYP3 A4 activity with age, but it is not
evident that any of their microsome preparations came from donors less than 1 year of age.
Hines and McCarver (2002) assembled data on metabolic activities of human CYPs with
developmental age. They reported that CYP2A6 activity in nasal mucosa of children 13-
18 weeks old was readily detectable, but that liver-specific activity was only 1-5% of adult
values. Age-related differences in CYP2A6 may be particularly relevant to ETBE because data
from Le Gal et al. (2001) from liver microsome samples suggested that CYP2A6 may account
07/14/2009 79 DRAFT - DO NOT CITE OR QUOTE
-------
for the majority of ETBE metabolism in the human liver. Hines and McCarver (2002) concluded
that in general, there was no evidence that isozymes from the CYP2A family were expressed in
fetal liver. By 1 year of age, however, specific activities had come close to adult levels. There
are limited data on the developmental profile of CYP2B6, one of the secondary enzymes that
may catalyze the metabolism of ETBE to TBA and acetaldehyde in humans based on data from
liver microsomes from Le Gal et al. (2001). Hines and McCarver (2002) reported that CYP2B6
was not detectable in 11- or 24-week-old fetal livers, but that CYP2B6 levels reached about 10%
of adult levels during the first year of life. Similarly, CYP3A4, one of the other secondary
enzymes that may catalyze the metabolism of ETBE based on liver microsome data from Le Gal
et al. (2001), is not expressed in fetal liver; although there is a fetal form for the CYP3A family,
CYP3 A7. However, its catalytic activity has not been well researched and nothing is known
about whether it may be involved in the metabolism of ETBE. It appears that the change from
CYP3 A7 to CYP3 A4 begins immediately after birth. While CYP3 A7 is still expressed for some
time during early postnatal life, CYP3 A4 attains adult activity levels by 1 year of age.
In summary, general patterns of developmental changes in cytochrome P-450 levels
indicate that humans at early age may have diminished ability to metabolize ETBE; however,
direct evidence of enzyme differences from ETBE exposure studies in humans is not available.
The limited available developmental toxicity data on ETBE are inadequate to determine if
children are at an increased risk from ETBE exposure. The generation and publication of
developmental toxicity data from oral and inhalation studies of ETBE in multiple species would
provide data necessary to predict if children are at an increased risk from ETBE exposure.
4.8.2. Possible Gender Differences
The various studies in humans conducted by Nihlen and coworkers (Nihlen et al.,
1998a, b), and Dekant and coworkers (Dekant et al., 2001a, b; Amberg et al., 2000; Bernauer et
al., 1998), include a low number of individuals and are inadequate to assess gender differences in
human susceptibility to ETBE-induced toxicity. Reports of studies in animals (e.g., CIT, 2004b,
unpublished report; Medinsky et al., 1999) suggest that male rats are more sensitive to renal
effects than females. The available data also suggest greater sensitivity for hepatic effects in rats
exposed to ETBE, but not in mice. For example, as part of an oral two-generation reproduction
and fertility study, liver weight increases were observed in male rats in the F0 and F1 generation
at lower doses (500 mg/kg-day) than females (1,000 mg/kg-day in F1 generation only) (see
Table 4-7; CIT, 2004b, unpublished report). However, in an inhalation exposure study in CD-I
mice, hepatic effects including increased liver weight, centrilobular hypertrophy, and increased
cellular proliferation were observed in males and females at the same doses (Medinsky et al.,
1999). In addition, several studies in animals reported lesser responses in females at the same
dose as given to males (e.g., 34% increased adrenal weight in male rats compared to 18%
07/14/2009
80
DRAFT - DO NOT CITE OR QUOTE
-------
increase in females after inhalation exposure in Medinsky et al., 1999). It is not clear if this
observation is related to differences in metabolism between males and females.
Rademaker (2001) stated that women have a 40% higher activity of CYP3A4 than men,
making them potentially more sensitive to toxicants, such as ETBE, that are activated by this
enzyme. Studies with human microsomes, however, have not indicated that this is the case
(Bebia et al., 2004; Parkinson et al., 2004).
4.8.3. Self-reported Sensitive Individuals
The existence of individuals who are sensitive to MTBE has been suggested in several
studies and this may also apply to ETBE. However, most of the data for MTBE are anecdotal
and no studies of potential ETBE-sensitive individuals were found in the literature. In one of the
few controlled exposure studies of self-reported MTBE-sensitive individuals, Fiedler et al.
(2000) compared the performance of 12 subjects that were self-reported as sensitive to MTBE
exposure with 19 controls in psychophysiologic and neurobehavioral responses as well as their
rating of 42 symptoms previously reported to be associated with exposure to MTBE in gasoline,
solvent exposure, anxiety, and depression. Subjects were exposed for 15 minutes to clean air,
gasoline, gasoline with 11% MTBE (G/l 1MTBE), and gasoline with 15% MTBE (G/15MTBE)
and symptoms and psychophysiologic and neurobehavioral response were measured before,
during, and after exposure. Individuals reported significantly more symptoms than controls
during the pre-exposure period, suggesting heightened reporting by these individuals
independent of exposure. Individuals reported significantly more symptoms than controls when
exposure to a G/15MTBE mixture was contrasted to clean air or G/l 1MTBE (p = 0.02).
Although nonsignificant, MTBE-sensitive individuals tended to report more symptoms than
controls when exposure to G/15MTBE was contrasted to gasoline alone (p = 0.08). No MTBE-
related changes in psychophysiologic or neurobehavioral responses were observed. Neither
group could distinguish whether MTBE was present at 11 or 15% and a majority of both groups
indicated that MTBE was present when they were exposed to gasoline only. Separate analysis of
the 42 symptom scores by subclass (i.e., MTBE-, anxiety-, depression-, breathing-, solvent-, and
environment-related symptoms) did not confirm the symptom specificity for MTBE exposure
suggested by the epidemiologic literature. The data from Fiedler et al. (2000) support the
possibility of a MTBE-sensitive population, but did not support a dose-response to MTBE
exposure. MTBE-sensitive individuals may also be sensitive to ETBE or there may be a similar
ETBE-sensitive population. The physiologic basis for the MTBE-sensitive individuals is
unknown; however, Hong et al. (2001) looked for polymorphisms in the enzymes responsible for
MTBE metabolism as a possible explanation (see Section 4.8.4 below).
07/14/2009
81
DRAFT - DO NOT CITE OR QUOTE
-------
4.8.4. Other—Aging; Gene Polymorphisms
It is generally assumed that human's drug metabolizing capacity is diminished with age.
Loss of the activities of three main enzymes that are most likely to be responsible for the
metabolism of ETBE to TBA and acetaldehyde (CYP2A6, 3A4, and 2B6) might be expected to
occur in the elderly, but no relevant information was found. Dome and coworkers (Dome et al.,
2005; Dome, 2004) estimated that renal clearance of drugs metabolized by CYP2A6 or 3A4
would be decreased by 20-30 or 20-50%, respectively, in the elderly (i.e., >70 years of age).
However, Blanco et al. (2000) did not find evidence that CYP3A4 activity in human liver
decreased with age up to 75 years. These authors did not present any data on CYP2A6 or 2B6.
The knowledge of polymorphic genes has been emerging rapidly with the introduction of
molecular biology and gene sequencing techniques. A gene mutation can affect the coding or
noncoding region of a gene. Substitution of one nucleotide, if within the limits of the degenerate
genetic code, would establish a variant for a gene expression without affecting biological
activity. If the base substitution results in an amino acid substitution, the affected protein may
have any amount of change in activity from negligible to totally inactive, and it may also be
more or less stable towards degradation. If the mutation affects the noncoding region of a gene,
in particular the promoter region, the result may be a change in the level of transcription, a
frameshift mutation with an entirely inactive gene product, or the introduction of an early stop
codon and release of an incomplete gene product (Hong and Yang, 1997). Multiple nucleotide
substitutions also have been identified. The nomenclature for polymorphic genes mostly uses the
common gene name, such as CYP2A6 for the cytochrome P-450 2A6 gene locus, and attaches an
asterisk with a subsequent numeral. If one variant exists with subvariants, a capital letter follows
the numeral, beginning with A. Given the large number of possible heterozygous combinations
of polymorphic CYPs involved in ETBE metabolism and the preliminary stage of current
knowledge, a definitive assessment cannot be made to what extent gene polymorphism affects
the sensitivity of humans toward ETBE exposure. Activity and variants for the main enzymes
that are most likely to be responsible for the metabolism of ETBE and for the metabolism of the
ETBE metabolite, acetaldehyde, are discussed below.
As pointed out in Section 3.3, CYP2A6 is likely to be the lead enzyme in humans to
cleave the ether bond in ETBE. It exists in an array of variants, and although not all of the
variants have been characterized with respect to their biological activity, it is clear that at least
one variant (2A6*4) has no catalytic activity (Fukami et al., 2004). Hong et al. (2001) identified
three novel CYP2A6 gene variants in a total of 23 individuals who self-identified as sensitive to
MTBE exposure. The activity of two of the variants was reduced (2.1 and 2.0 relative to
2.6 pmol/minute/pmol CYP for the wildtype CYP2A6) and the third variant showed a total loss
in ether-metabolizing activity. An overview of some of the CYP2A6 variants, with ethnic
frequencies and catalytic activities, where available, is presented in Table 4-10. As described in
Section 3.3, CYP3A4 generally has lower catalytic activity toward ETBE than CYP2A6;
07/14/2009 82 DRAFT - DO NOT CITE OR QUOTE
-------
however, CYP3 A4 has the highest abundance of all CYPs in human liver. In humans, the
activity of CYP3 A4 varies as much as 20-fold. Detailed data on the frequency of the individual
CYP3A4 alleles were not available. It may be concluded from the report by Hsieh et al. (2001)
that the low- or zero-activity variants are rare, and thus, CYP3 A4 polymorphism may not play a
role in ETBE toxicity unless it is paired with a low- or zero-activity variant of CYP2A6.
CYP2B6 may have the lowest catalytic activity toward ETBE among the three enzymes most
likely to be responsible for metabolism of ETBE in humans (CYP2A6, 3A4, and 2B6) (see
Section 3.3). At least 13 single nucleotide polymorphisms have been described for this enzyme,
making for eight variants, CYP2B6*2-*9, with several subvariants (Jacob et al., 2004). Jinno et
al. (2003) reported (in percent of wild-type activity): CYP2B6*2, 160%; *3, 150%; *4, 200%;
and *5, 170%. With respect to ETBE toxicity, higher catalytic activity would signify potentially
higher risk. Polymorphisms in aldehyde dehydrogenase (ALDH), the enzyme that oxidizes
acetaldehyde to acetic acid, may also affect potential ETBE toxicity. The virtually inactive form,
ALDH2*2, is responsible for alcohol intolerance and is found in about one-half of all Asians
(Ames et al., 2002). This variant is associated with slow metabolism of acetaldehyde, and hence,
extended exposure to a possible human carcinogen. With respect to ETBE exposure, the
ALDH2*2 variant should increase any type of risk associated with acetaldehyde exposure, since
it would allow prolonged exposure to the ETBE metabolite, acetaldehyde.
07/14/2009
83
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-10. Frequencies of gene polymorphisms of human CYP2A6
Percent of population affected
Catalytic
activity
Variant
Caucasian
Turkish
African-
American
Asian
Black
African
Brazilian3
2A6*1A
66
43
80.5
Wild-type
2A6*1B
30
40
11.9
29.9
t
2A6*1F
1.8
0
2A6*1G
1.2
13.3
2A6*1H
3.1
5.2
2A6*2
1-3
3
n.d.-0.7
n.d.
1.7
u
2A6*4A
3
n.d.
15-20
1.9
0.5
none
2A6*4D
n.d.
0.5
2A6*5
n.d.
n.d.-O.l
n.d.
2A6*6
0.4
n.d.
u
2A6*7
n.d.
2A6*8
n.d.
2A6*9
5.2
7.2
5.7
5.7
i
2A6*10
n.d.
2A6*11
n.d.
2A6*17
n.d.
9.4
i
aAverage value; according to the authors, frequency of * IB was white > mixed race > black.
n.d. = Variant not detected.
Empty cell = no data.
T = Increased catalytic activity.
I = Reduced catalytic activity.
II = Strongly reduced catalytic activity.
Sources: Gyamfi et al. (2005); Vasconcelos et al. (2005); Fukami et al. (2004); Nakajima et al. (2004, 2001); von
Richter et al. (2004); Yoshida et al. (2003); Oscarson (2001); Paschke et al. (2001); Pitarque et al. (2001); Chen et
al. (1999).
07/14/2009
84
DRAFT - DO NOT CITE OR QUOTE
-------
5. DOSE RESPONSE ASSESSMENTS
5.1. ORAL REFERENCE DOSE (RfD)
5.1.1. Choice of Principal Study and Critical Effect - with Rationale and Justification
The database for oral exposure to ETBE is limited in the number and scope of available
studies. There are no available human occupational or epidemiological studies of oral exposure
to ETBE. The animal toxicity data associated with oral exposure to ETBE include a single
chronic, gavage cancer bioassay in Sprague-Dawley rats (Maltoni et al., 1999), a 2-week oral
exposure study evaluating oocyte quality in Simonson rats (Berger and Horner, 2003), and
several reproductive and developmental toxicity studies from CIT (2004a, b, 2003 unpublished
reports) in Sprague-Dawley and F344 rats. Most of the oral toxicity data on ETBE come from
the CIT (2004a, b, 2003 unpublished reports) studies that were specifically designed to assess the
reproducibility of histopathological changes in the testes of F344 rats (increased percent of
seminiferous tubules with degenerated spermatocytes) observed in a subchronic inhalation study
(Medinsky et al., 1999). Therefore, the database for oral ETBE toxicity is unusual in that it
contains a two-generation reproduction and fertility study, but lacks a subchronic or chronic oral
exposure study with the full range of basic toxicological endpoints such as histology from a
standard array of organs and organ systems.
Liver and kidney organ weight increases are the primary noncancer health effects
associated with oral exposure to ETBE based on limited available animal data. Increased liver
and kidney weights were observed following oral exposure in both the parent and offspring
generations and both sexes of rats in a two-generation reproductive toxicity study (CIT 2004b,
unpublished report). Additional effects observed in the available studies also included mortality
and decreased body weight gain. Further consideration was given to these endpoints as potential
critical effects for the determination of the point of departure (POD) for derivation of the oral
RfD. BMD modeling, if the data were amenable, was performed and is discussed in detail in
Section 5.1.2 and Appendix B.
Data suggesting kidney toxicity in the available oral ETBE exposure studies include the
demonstration of increased kidney weight in Sprague Dawley rats of both sexes in the parental
(F0) and F1 generations from a two-generation reproductive toxicity study (CIT 2004b,
unpublished report). Relative kidney weights were increased in F0- and F1-generation male rats
exposed to 250, 500, and 1,000 mg/kg-day ETBE and F1 generation female rats exposed to
1,000 mg/kg-day ETBE. Absolute kidney weights were increased in F0 and F1 generation male
rats exposed to 500 and 1,000 mg/kg-day ETBE, F1 males exposed to 250 mg/kg-day ETBE, and
F1 generation female rats exposed to 1,000 mg/kg-day ETBE. The LOAEL for increased
relative kidney weight was 250 mg/kg-day for oral exposure to ETBE and both values apply to
males of either the parental (F0) generation of F1 generation. As 250 mg/kg-day was the lowest
dose utilized in the study, a NOAEL was not available for increased relative kidney weight in
07/14/2009
85
DRAFT - DO NOT CITE OR QUOTE
-------
either the FO or F1 generation males. Slight to moderate centrilobular hypertrophy was observed
in all of the high-dose animals examined (five total males [3/3 FO and 2/2 Fl]) and was not seen
in the single control FO generation male examined. Only the livers from these six males were
examined histologically out of all of the rats in all three generations. Histology was only
performed if abnormal morphology was detected at necropsy, and, therefore, there are limited
histological data to support nephrotoxicity (histology was only performed on the kidneys from
12 males out of all of the rats in all three generations in the study).
Data suggesting liver toxicity in the available oral ETBE exposure studies were limited to
the demonstration of increased liver weight in Sprague-Dawley rats of both sexes in the parental
(FO) and Fl generations from a two-generation reproductive toxicity study (CIT 2004b,
unpublished report). Absolute and relative liver weights were increased in FO-generation male
rats exposed to 1,000 mg/kg-day ETBE, Fl generation animals of both sexes exposed to
1,000 mg/kg-day ETBE, and Fl generation males exposed to 500 mg/kg-day ETBE (LOAEL in
male rats). The NOAEL for increased liver weight was 250 mg/kg-day for oral exposure to
ETBE during development as the NOAEL for males in the Fl generation was lower than the
NOAEL for the parental generation. Acidophilic globules were observed in 5/6 F0 and 3/4 Fl
generation males at the high dose. In addition, tubular basophilia (4/6), peritubular fibrosis (3/6),
and proteinaceous casts (1/6) were observed in the F0 males at the high dose (1,000 mg/kg-day).
Histology was also performed on kidneys from six Fl males (four high-dose, one mid-dose
[500 mg/kg-day], and one control) due to the presence of macroscopic lesions in these animals.
Tubular basophilia was observed in the control male, the mid-dose male, and two of the four
high-dose males. Peritubular fibrosis was also observed in the control male, the mid-dose male,
and two of the four high-dose males. The one mid-dose male examined also had sloughed
degenerated/necrotic cells in the tubular lumen. Histology was only performed if abnormal
morphology was detected at necropsy, and, therefore, there are limited histological data to
support hepatotoxicity (histology was only performed on the livers from six males out of all of
the rats in all three generations in the study).
As discussed in Section 4.7, ETBE-exposure of Sprague-Dawley rats in a chronic cancer
bioassay was associated with increased mortality relative to controls in animals of both sexes 56-
88 weeks of age and increased survival relative to controls in females >104 weeks of age
(Maltoni et al., 1999). At 56 weeks of age, 250 mg/kg is the LOAEL in males for a 7% increase
in mortality. Although all rats were necropsied and tissue/organs were taken for microscopic
examination, no additional information on related toxicity or pathology associated with the
observed increase in mortality was provided. Given the frank effect, lack of quantitative
mortality data, and a consistent dose-response trend, this endpoint was not considered ideal for
the derivation of the RfD and was not amenable to BMD modeling. However, a LOAEL of
250 mg/kg-day for mortality in males at 56 weeks exposure was considered as a possible POD.
07/14/2009
86
DRAFT - DO NOT CITE OR QUOTE
-------
There were several examples of reduced body weight gain from the prenatal and
two-generation reproductive toxicity studies in Sprague-Dawley rats (CIT, 2004a, b, unpublished
reports). In a prenatal toxicity study (CIT, 2004a, unpublished report), pregnant Sprague-
Dawley dams in the 1,000 mg/kg-day ETBE group (LOAEL in female rats) had an 11%
decreased maternal body weight gain and a 17% decrease in net weight gain when animals were
euthanized on GD 20. The NOAEL for maternal toxicity was 500 mg/kg-day based on
decreased maternal weight gain. There was no effect of ETBE treatment on body weight of
females in the two-generation reproduction and fertility study (CIT, 2004b, unpublished report)
at the same doses over longer treatment periods). Male rats in the F0 generation at 500 and
1,000 mg/kg-day showed significantly lower body weight gains (-29 and -22%, respectively) in
the last quarter of the treatment period. The authors suggest a NOAEL of 250 mg/kg-day for this
endpoint, although the weight gain in F0 generation males was not different from controls when
analyzed over the entire treatment period (days 1-113) (for additional details see Section 4.5.1 of
the Toxicological Review).
Oral exposure to ETBE had no effect on oocyte quality, sperm parameters, mating,
fertility, gestation, fecundity, or delivery in rats (CIT 2003, 2004a, b, unpublished reports;
Berger and Horner, 2003).
CIT (2004b, unpublished report) was chosen as the principal study for derivation of the
RfD because the increased kidney and liver weights reported in this study represent the most
sensitive effects identified in the database. The LOAEL for increased relative kidney weight was
250 mg/kg-day for oral exposure to ETBE and both values apply to males of either the parental
(F0) generation of F1 generation. The LOAEL for increased liver weight was 500 mg/kg-day
and the NOAEL for increased liver weight was 250 mg-kg-day for males in the F1 generation.
There are limited histological data to support nephrotoxicity or hepatotoxicity in the oral
two-generation reproductive toxicity study because histology was only performed if abnormal
morphology was detected at necropsy (CIT, 2004b, unpublished report). Increased kidney
weight was also observed in F1 generation females at higher doses. Increased liver weight was
also observed at higher doses in F1 generation females and F0 generation males. The data for
increased organ weight of the kidneys and livers in males were subjected to BMD modeling
(Section 5.1.2 and Appendix B) because the effects in males were more pronounced and
occurred at lower doses than in females. The data for increased organ weight of the kidneys in
females were subjected to BMD modeling for comparison purposes. Data on both absolute and
relative organ weights were modeled. The ETBE database contains additional support for the
kidney and liver as target organs as determined by increased kidney and liver weights and some
related effects seen in inhalation toxicity studies of ETBE (Medinsky et al., 1999; White et al.,
1995). Data suggesting kidney toxicity associated with oral exposure to ETBE are limited to
increased kidney weights in male and female rats (at higher doses), with limited
histopathological support as described above. A number of kidney effects are also reported after
07/14/2009
87
DRAFT - DO NOT CITE OR QUOTE
-------
inhalation exposure to ETBE, suggesting that the kidney is a target organ of ETBE exposure.
Additional kidney effects observed after inhalation exposure to ETBE include: histological
evidence of cellular necrosis (increased incidence of regenerative foci) in the kidneys of male
rats, sustained increase in kidney cellular proliferation in male rats, increased kidney cellular
proliferation in female rats after 1 and 4 weeks, and increased kidney weights in male and female
rats (Medinsky et al., 1999; White et al., 1995). A similar situation is presented by the data on
ETBE-related liver effects. Oral exposure to ETBE is associated with increased liver weights in
male and female rats (at higher doses), with limited histopathological support as described
above. A number of hepatic effects are also reported after inhalation exposure to ETBE,
suggesting that the liver is a target organ of ETBE exposure. Additional hepatic effects observed
after inhalation exposure to ETBE include: hepatic centrilobular hypertrophy in male and female
mice, sustained increase in hepatic cellular proliferation in female mice, increased hepatic
cellular proliferation in male mice after 1 and 4 weeks, increased liver organ weight in male and
female mice, and increased liver organ weight in male and female rats (Medinsky et al., 1999).
Table 5-1 summarizes the BMD modeling results of the available data (see Section 5.1.2
for complete discussion). The benchmark response (BMR) levels and the PODs are identified in
Table 5-1 for each effect. The BMR levels represent a change of one SD from the control mean
for continuous variables and are presented as the percent difference from controls. The range of
the PODs (approximately 100-900 mg/kg-day) is less than a factor of 10.
07/14/2009
88
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1. Summary of BMD modeling results of ETBE oral toxicity studies for selection of principal study in
Sprague-Dawley rats
Organ
Endpoint
Sex
Dosing
duration
(days)
"Best-fit" model
Goodness
of fit
/j-valuc
AIC
BMDa
(mg/kg-day)
BMDL
(mg/kg-day)
Study
reference
BMD
analyses
Body
Dam body weight
change
F
14
Power
(non-constant variance13)
0.35
554.3
1,015
879
CIT,2004a
Net dam body
weight change
F
14
Linear
(constant variance)
0.49
476.6
1,063
696
CIT, 2004a
F0 body weight
change
(days 85-113)
M
120
Power - Highest Dose Dropped
(non-constant variance13)
0.73
479.0
492
385
CIT, 2004b
Liver
F1 liver weight
(absolute)
M
120
Hill
(non-constant variance)
0.96
329.7
482
294
CIT,2004b
F1 liver weight
(relative)
M
120
All continuous models
exhibited significant lack-of-fit
No data0
No data0
No data0
No data0
CIT, 2004b
Kidney
F0 kidney weight
(absolute)
M
120
Hill
(constant variance)
0.81
-40.1
381
167
CIT,2004b
F0 kidney weight
(relative)
M
120
Hill
(constant variance)
0.94
-453.0
227
143
CIT, 2004b
F1 kidney weight
(absolute)
M
120
All continuous models
exhibited significant lack-of-fit
No data0
No data0
No data0
No data0
CIT, 2004b
F
120
Linear
(constant variance)
0.30
-180.2
1,016
687
CIT, 2004b
F1 kidney weight
(relative)
M
120
All continuous models
exhibited significant lack-of-fit
No data0
No data0
No data0
No data0
CIT, 2004b
F
120
Linear
(non-constant variance13)
0.10
-412.3
898
562
CIT, 2004b
07/14/2009
89
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1. Summary of BMD modeling results of ETBE oral toxicity studies for selection of principal study in
Sprague-Dawley rats
LOAEL
NOAEL/
Analyses
Organ
Endpoint
Sex
Dosing
duration
(days)
LOAEL
(mg/kg-day)
NOAEL
(mg/kg-day)
Study
reference
Mortality
M
728d
250
NAe
Maltoni et
al., 1999
Liver
F1 liver weight
(relative)
M
120
500
250
CIT,2004b
Kidney
F1 kidney weight
(absolute)
M
120
500
250
CIT,2004b
F1 kidney weight
(relative)
M
120
250
NAf
CIT,2004b
aAll data were modeled as a continuous variable, and a BMR of a change of one SD from the control mean was employed.
bVariance model employed (i.e., variance modeled as a power function of the mean) failed to adequately address nonconstant variance.
°No data because all models in the software exhibited significant lack-of-fit.
increased mortality determined at 56 weeks of a 104-week treatment.
eNA = not applicable as both doses of ETBE (1,000 and 250 mg/kg-day) resulted in increased mortality and therefore a NOAEL was not determined (Maltoni et al.,
1999).
fNA = not applicable as all doses of ETBE (1,000, 500, and 250 mg/kg-day) resulted in increased relative kidney weight and therefore, a NOAEL was not determined
(CIT 2004b, unpublished report).
BMDL = lower 95% confidence limit on the benchmark dose; F= female; M = Male
07/14/2009
90
DRAFT - DO NOT CITE OR QUOTE
-------
5.1.2. Methods of Analysis
The increased liver and kidney weights in Sprague-Dawley rats from the two-generation
reproductive toxicity study of ETBE were treated as continuous variables for dose-response
modeling (CIT, 2004b, unpublished report). The reduced body weight gain in Sprague-Dawley
rats from the prenatal developmental and two-generation reproductive toxicity studies of ETBE
were also treated as continuous variables for dose-response modeling (CIT, 2004a, b,
unpublished reports). All available models for continuous variables in U.S. EPA's Benchmark
Dose Software (BMDS, version 1.4.1) were fit to the data in accordance with U.S. EPA (2000b)
BMD methodology. The BMR was defined as a change of one SD from the control mean. For
these datasets, this BMR corresponds to an approximate 7-13% change in organ weights and a
14-21% change in body weights. The lower 95% confidence limit on the benchmark dose
(BMDL) estimates for the best fitting models for increased liver and kidney weights and reduced
body weight gain from CIT (2004a, b, unpublished report) are presented in Table 5-1, and
detailed discussion of the modeling of each endpoint is presented in Appendix B. For each
endpoint modeled, overall goodness-of-fit tests (x2) and determined the Akaike's Information
Criterion (AIC) were determined. The chi-square goodness-of-fit test is a measure of how well
the model fits the observed data. Models with xV-values >0.1 were considered to have
adequate fits. The AIC is a measure of the model fit based on the log-likelihood at the maximum
likelihood estimates for the parameters. Within the subset of models that exhibit adequate fit,
models with lower AIC values are preferred. The "best-fit" model selection criteria are
presented in Appendix B and described in detail in EPA's Benchmark Dose Technical Guidance
Document (U.S. EPA, 2000b).
The increased liver and kidney weights of male rats following oral exposure to ETBE
were evaluated as potential PODs. As described above and in Section 4.3, liver and kidney
weight was increased in males and females, and the organ weight in males increased at lower
doses and to a greater degree. Absolute and relative liver weights of F1 generation males and
absolute and relative kidney weights of F0 and F1 generation males were subjected to BMD
modeling.
As shown in Appendix B, a somewhat limited set of the available continuous models in
BMDS provided adequate fits to the data for absolute and relative liver weights of F1 generation
males and absolute and relative kidney weights of F0 and F1 generation males (CIT 2004b,
unpublished report). For liver, the Hill model provided the best fit to the increase in absolute
liver weight in F1 males. None of the continuous models available in BMDS adequately fit the
relative liver weight in F1 males, so a NOAEL of 250 mg/kg-day was retained as a potential
POD. For kidney, the Hill model provided the best fit to both absolute and relative kidney
weights in F0 males. None of the continuous models available in BMDS adequately fit the
absolute or relative kidney weight in F1 males so a NOAEL of 250 mg/kg-day was retained as a
potential POD for absolute kidney weight in F1 males, and a LOAEL of 250 mg/kg-day was
07/14/2009 91 DRAFT - DO NOT CITE OR QUOTE
-------
retained as a potential POD for relative kidney weight in F1 males. The linear model provided
the best fit to both absolute and relative kidney weights in F1 females. Additionally, the models
that exhibited adequate fit also showed good fits to the incidence data at low doses (i.e., in the
vicinity of the BMR) as evidenced by examining the %2 scaled residuals, and the visual fit of the
model to the data in the plots from the BMDS output. Thus, the Hill model was selected to
estimate the BMDL or POD for absolute liver weight in F1 males and for absolute and relative
kidney weights in FO males. The BMDL associated with a change of one SD from the control
mean for an increase in absolute liver weight in F1 males was 294 mg/kg-day. The BMDLs
associated with a change of one SD from the control mean for an increase in absolute and
relative kidney weights in FO males were 167 and 143 mg/kg-day, respectively. The BMDLs
associated with a change of one SD from the control mean for an increase in absolute and
relative kidney weights in F1 females were 687 and 562 mg/kg-day, respectively.
As shown in Appendix B, several of the available continuous models in BMDS provided
adequate fits to the data for decreased body weight gain and net body weight gain in dams in CIT
(2004a, unpublished report) and decreased body weight gain in FO males during the last quarter
of the treatment period in CIT 2004b, unpublished report). The power model provided the best
fit to the body weight gain in pregnant dams, the linear model provided the best fit to the net
body weight gain in pregnant dams, and the power model (with the highest dose group dropped)
provided the best fit to the body weight gain FO generation males during the last quarter of
treatment period, as assessed by AIC. Thus, the power model was selected to estimate the
BMDL or POD for body weight gain in the pregnant dams, the linear model was selected to
estimate the POD for net body weight gain in the pregnant dams, and the power model (with the
highest dose group dropped) was selected to estimate the POD for body weight gain in the FO
generation males during the last quarter of treatment from CIT studies (2004a, b, unpublished
reports). The BMDL associated with a change of one SD from the control mean for body weight
gain in pregnant dams was 879 mg/kg-day. The BMDL associated with a change of one SD
from the control mean for net body weight gain in pregnant dams was 696 mg/kg-day. The
BMDL associated with a change of one SD from the control mean for net body weight gain in F0
generation males during the last quarter of treatment was 385 mg/kg-day.
As discussed in Section 5.1.1, the increased mortality relative to controls observed in
Sprague-Dawley rats exposed to ETBE by gavage in a chronic cancer bioassay (Maltoni et al.,
1999) was not amenable to BMD modeling. Mortality represents a frank effect, there are a lack
of quantitative mortality data from Maltoni et al. (1999), and the data also do not display a
consistent dose-response trend. However, a LOAEL of 250 mg/kg-day for mortality in males at
56 weeks exposure was considered as a possible POD.
CIT (2004b, unpublished report) was selected as the principal study (Section 5.1.1) and
increased relative kidney weight in F0 generation males as the critical effect because it resulted
in the lowest BMDL, 143 mg/kg-day. Data suggesting kidney toxicity associated with oral
07/14/2009 92 DRAFT - DO NOT CITE OR QUOTE
-------
exposure to ETBE are limited to increased kidney weights in males and females (at higher
doses), with limited histopathological support as described in Section 5.1.1. Increased kidney
weight may be considered an adverse effect and may also be an early indicator of more overt
kidney toxicity. The designation of kidney toxicity as the critical effect is supported by a
number of kidney effects reported after inhalation exposure to ETBE. In addition to increased
kidney weights in male and female rats, additional effects observed after inhalation exposure to
ETBE include: histological evidence of cellular necrosis (increased incidence of regenerative
foci) in the kidneys of male rats, sustained increase in kidney cellular proliferation in male rats,
and increased kidney cellular proliferation in female rats after 1 and 4 weeks (Medinsky et al.,
1999; White et al., 1995). Based on application of the criteria from the Benchmark Dose
Technical Guidance Document (U.S. EPA, 2000b) as described above, the Hill model provided
the best fit to the kidney weight data.
5.1.3. RfD Derivation—Including Application of Uncertainty Factors (UFs)
Considering the uncertainties in the ETBE database, which are described in Appendix C,
the total composite UF is 10,000, consisting of four areas of maximum uncertainty. In the report,
A Review of the Reference Dose and Reference Concentration Processes (U.S. EPA, 2002), the
technical panel concluded that, in cases where maximum uncertainty exists in four or more areas
of extrapolation, or when the total UF is 10,000 or more, it is unlikely that the database is
sufficient to derive a reference value. Therefore, in lieu of an RfD, Appendix C contains a
derivation of an oral minimal data value for ETBE using an UF of 10,000. Use of this minimal
data value is not recommended except in limited circumstances, for example, in screening level
risk assessments or to rank relative risks. Any use of this value should include a discussion of
the uncertainty associated with its derivation.
5.1.4. Previous RfD
An oral assessment for ETBE was not previously available on IRIS.
5.2. INHALATION REFERENCE CONCENTRATION (RfC)
5.2.1. Choice of Principal Study and Critical Effect - with Rationale and Justification
The inhalation toxicity database for ETBE is from a limited number of studies which
generally report relatively mild effects such as increased organ weights. Data on the effects of
ETBE in humans is limited to several 2-hour inhalation studies at doses up to 50 ppm (Nihlen et
al., 1998b; Vertrano, 1993, unpublished report; TRC Environmental Corp., 1993). Healthy
subjects exposed to ETBE experienced irritation in throat and airways, nasal swelling, a bad taste
in the mouth, and slightly impaired lung function (about 3% reduction in vital capacity and
forced vital capacity). No chronic inhalation studies are available, although there are several
subchronic studies in mice and rats (Medinsky et al., 1999; White et al., 1995; see Section 4.5.2
07/14/2009 93 DRAFT - DO NOT CITE OR QUOTE
-------
and Table 4-9). Data from these inhalation toxicity studies of ETBE indicated that among the
minimal effects associated with ETBE exposure, the kidney and liver appear to be target organs
for ETBE toxicity as determined by increased kidney and liver weights and some additional
effects described below. Concordant support for kidney and liver effects resulting from exposure
to ETBE is gained through oral exposure studies, which are described more fully in Sections 4.3;
4.5.1; and 5.1.1 (CIT, 2004b, unpublished report).
In a subchronic inhalation study, CD-I mice and F344 rats were exposed to 0, 500, 1,750,
and 5,000 ppm for 13 weeks (Medinsky et al., 1999). Increased liver weight was observed in
female rats at 5,000 ppm and in male rats at 1,750 and 5,000 ppm. Increased liver weight was
also observed in male and female mice at 1,750 and 5,000 ppm. Hepatic centrilobular
hypertrophy was associated with ETBE-induced liver weight increases in the mice, but not the
rats. Mice were also tested for a hepatic mitogenic response to ETBE exposure. Dose-related
increases in the hepatic LI were seen in male and female mice at 1,750 and 5,000 ppm at
different time points. The LI was increased in males at 1 and 4 weeks (not 13 weeks) and in
females at 1 and 13 weeks (not 4 weeks). The authors did not find any histological evidence of
hepatic lesions or elevated serum enzymes characteristic of hepatotoxicity. A 1% increase in
MCV was observed in female rats and a 2.5% decrease in MCHC was noted in male rats exposed
to 1,750 and 5,000 ppm ETBE after 13 weeks of exposure, but there was no pattern to support
hepatotoxicity as there was no relationship between the small changes in MCV or MCHC and
increased liver weight or any other liver data in rats and mice from the available studies
(Medinsky et al., 1999). No additional evidence of irregular pathology in the liver was observed
in this study. However, there are a number of hepatic endpoints that could constitute a LOAEL:
(1) the observed increased liver weights in male and female CD-I mice and male rats at
1,750 ppm, (2) the increased incidence of centrilobular hypertrophy at 5,000 ppm in male and
female CD-I mice, and (3) the increased LI which was sustained in female CD-I mice at
13 weeks at 1,750 ppm. Increased liver weight was also observed in Sprague-Dawley rats in a
4-week inhalation study (White et al., 1995).
In a 4-week inhalation study, Sprague-Dawley rats were exposed to 0, 500, 2,000, and
4,000 ppm ETBE (White et al., 1995). The only treatment associated responses from the 4-week
subchronic study were increased liver weight in females at 2,000 and 4,000 ppm and increased
liver, kidney, and adrenal weight in males at 4,000 ppm. No histopathological findings were
reported in the over 40 tissues examined, including the liver, kidney, and adrenal gland. No
indications of neurotoxicity were detected in a FOB consisting of 19 parameters covering
sensory perception, reflex response, body temperature, and neuromuscular function. Transient
ataxia and sedation, overt signs of CNS depression, were seen following exposure termination in
the 4,000 ppm group, but ETBE-exposed animals appeared normal within 15 minutes. There
were also no significant effects of ETBE exposure observed in serum chemistry (including
hepatic enzymes) or hematology evaluations.
07/14/2009 94 DRAFT - DO NOT CITE OR QUOTE
-------
In addition to the liver, Medinsky et al. (1999) observed effects of inhalation ETBE
exposure in the kidney of male and female rats. Specifically, increased kidney weight was
detected in male rats exposed to 1,750 and 5,000 ppm and in female rats exposed to 5,000 ppm
as part of the same 13-week inhalation study (Medinsky et al., 1999). In male rats, the increased
kidney weight was associated with hyaline droplet accumulation with alpha2U-globulin
immunoreactivity, the presence of regenerative foci, and greater than twofold increase in cellular
proliferation as determined by the increased LI after 1 and 4 weeks of exposure at 5,000 ppm and
at all doses after 13 weeks of exposure (Medinsky et al., 1999). Regenerative foci are associated
with repair of damaged tubules, and some cells of the proximal tubule would have had to have
died and the remaining cells undergone regenerative proliferation for regenerative foci to be
observed. However, no other renal effects were noted in this study or in the available subchronic
studies. Alpha2U-globulin derived from hepatic synthesis is unique to male rats; it is not found in
female rats or in mice or humans of either sex (U.S. EPA, 1991b). Based on the available data
for ETBE, alpha2U-globulin accumulation may be associated with the observed kidney effects in
male rats; however, there is considerable uncertainty because of the lack of evidence of typical
lesions in the pathological sequence of lesions associated with alpha2U-globulin nephropathy.
Therefore, a determination cannot be made as to whether alpha2U-globulin accumulation is the
mode of action or the only mode of action for renal effects associated with ETBE exposure (U.S.
EPA, 1991b). The increased kidney weight observed in females was not accompanied by
significant histopathology except for a less than twofold increase in LI (i.e., proliferation) at
early time points (after 1 and 4 weeks of exposure, not after 13 weeks). Although statistically
significant, this small increase in LI that was not sustained over the study period was of
questionable biological significance. It is also useful to note that there is insufficient information
to determine if some or all of ETBE-induced renal effects are caused by the parent compound or
its metabolite, TBA. Some authors have proposed that the nephropathy and subsequent cancer
incidence in male rats exposed to TBA is mediated by an excessive accumulation of alpha2u-
globulin in proximal tubular cells (Williams and Borghoff, 2001; Takahashi et al., 1993).
However, the dose of TBA associated with an increase in the severity of chronic nephropathy in
male rats in a 13-week inhalation study (Mahler, 1997) is below doses associated with
accumulation of protein droplets in the kidney either in that study (Mahler, 1997) or in a 10-day
exposure (Borghoff et al., 2001), suggesting that alpha2U-globulin may not be the mechanism or
the only mechanism for TBA-associated nephropathy. Although some data support alpha2U-
globulin accumulation as a mode of action for renal effects associated with ETBE exposure in
male rats, sufficient data are not available to determine if alpha2u-globulin accumulation is the
only mode of action for renal effects associated with ETBE exposure (U.S. EPA, 1991b).
The 13-week inhalation exposure study (Medinsky et al., 1999) also included histological
evidence that the percentage of seminiferous tubules with spermatocyte degeneration was
slightly increased at 1,750 and 5,000 ppm (+7.8 and +12.7%, respectively) in F344 rats.
07/14/2009
95
DRAFT - DO NOT CITE OR QUOTE
-------
Degenerated spermatocytes were also found in the testes of control rats. Spermatocyte
degeneration is a widely examined histopathological outcome in the testis, and these data are
often presented as a percentage of total tubules; however, quantitation of this type provides crude
information about the nature and severity of potential spermatogenic damage (Russell et al.,
1990).
Several additional effects were also observed in F344 rats in the 13-week inhalation
exposure study of ETBE (Medinsky et al., 1999) including increased adrenal weight in males and
females, increased heart weight in females, and increased incidence of bone marrow congestion
in females. An increase in the adrenal weight was observed in both male and female F344 rats at
5,000 ppm, but there were no associated histopathological findings. A similar increase in
adrenal weight was observed in male Sprague-Dawley rats following 4-week inhalation exposure
to 4,000 ppm ETBE (White et al., 1995), also without any associated pathological findings in the
histology performed on the adrenal gland. Increased heart weight was observed in female F344
rats exposed by inhalation to 500 and 5,000 ppm ETBE, but not to an intermediate dose of
1,750 ppm for 13 weeks. Histological evidence of bone marrow congestion was seen in tissue
from female rats at 1,750 and 5,000 ppm. The reported increase in bone marrow congestion was
not supported by any changes in hematopoietic cell populations and was, thus, judged to be not
clinically relevant by the authors. Also, there were no effects of ETBE exposure in serum
chemistry or hematology judged to be clinically relevant by the authors.
In summary, the Medinsky et al. (1999) study was adequately designed with several
acceptable dose groups and an adequate number of animals. Sensitive endpoints identified in
this study included a number of effects in the liver of mice and rats (i.e., sustained proliferation
in the liver indicated by a sustained increase in the LI in mice, increased centrilobular
hypertrophy in male and female mice, and increased liver weight in mice and rats). A number of
effects were identified in the kidney including increased kidney weight in male and female rats
as well as increased incidence and mean number of regenerative foci and sustained proliferation
in the kidney indicated by a sustained increase in the LI in male rats. Additional effects were
identified in rats including an increased percentage of seminiferous tubules with degenerated
spermatocytes in males, increased adrenal gland weight in males and females, and increased
heart weight and incidence of bone marrow congestion in females. After consideration of all
available endpoints, the mean number of regenerative foci in the kidneys was determined to be
the most sensitive and biologically significant effect detected in these studies. As described
previously, regenerative foci are associated with repair of damaged kidney tubules and some
cells of the proximal tubule would have had to have died and the remaining cells undergone
regenerative proliferation for regenerative foci to be observed. Therefore, EPA considers
regenerative foci to be indicators of cellular necrosis and a biomarker of an adverse effect.
Furthermore, the ETBE database includes additional evidence for the liver and kidney as target
organs of ETBE toxicity as determined by increased liver and kidney weights seen in male and
07/14/2009 96 DRAFT - DO NOT CITE OR QUOTE
-------
female Sprague-Dawley rats in an oral two-generation reproduction and fertility study (CIT,
2004b, unpublished report).
Therefore, Medinsky et al. (1999) was chosen as the principal study for derivation of the
RfC because the kidney effects (i.e., regenerative foci) observed in this study represent the most
sensitive effects identified in the database evaluating exposure through the inhalation route to
ETBE. White et al. (1995) and CIT (2004b, unpublished study; oral exposure route only)
provided supporting data for this endpoint; however, kidney effects were observed in these
studies at higher doses.
5.2.2. Methods of Analysis
As described in Section 4.5.2, the database of inhalation toxicity studies on ETBE is
limited with the majority of the data from a 4-week inhalation study in Sprague-Dawley rats
(White et al., 1995) and a 13-week inhalation study in F344 rats and CD-I mice (Medinsky et al.,
1999). The effects observed in these studies included organ weight changes and effects in the
liver, kidney, heart, adrenal, and testes. Further consideration was given to these endpoints for
the selection of the critical effect as described below.
Selected endpoints that were treated as continuous variables (i.e., LI of liver, liver weight,
LI of kidney, kidney weight, mean number of regenerative foci in kidney, percent seminiferous
tubules with degenerated spermatocytes, adrenal gland weight, and heart weight) were modeled
employing all available continuous models in U.S. EPA's BMDS (version 1.4.1) in accordance
with U.S. EPA (2000b) BMD methodology. The BMR was defined as a change of one SD from
the control mean. For these datasets, this BMR corresponds to an approximate 6-19% change in
organ weights. All available dichotomous models in the U.S. EPA's BMDS (version 1.4.1) were
fit to endpoints treated as quantal (i.e., incidence of centrilobular hypertrophy in the liver of male
and female CD-I mice, incidence of regenerative foci in the kidney of male rats, and incidence
of bone marrow congestion in female F344 rats). For dichotomous models, the BMR was
defined as a 10% increase in extra risk because there was no clear biological rationale for
selecting an alternate BMR for these data. The lower 95% confidence limit on the benchmark
concentration (BMCL) estimates for the best fitting models for the selected data sets are
presented in Table 5-2, and a detailed discussion of the modeling of each endpoint is presented in
Appendix B.
07/14/2009
97
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-2. Summary of BMD modeling results of ETBE inhalation toxicity studies for selection of principal
study
Organ
Endpoint
Species
strain
Sex
Dosing
duration(
wks)
"Best-fit" model
Goodness
of fit
/j-valuc
AIC
BMC
(ppm)
BMCLC
(ppm)
BMCLhec"
(mg/m3)
Reference
Liver
Lf
CD-I Mice
F
13
Linear
(constant variance)
0.57
31.7
2040
1307
975
Medinsky et al..
1999
Centrilobular
hypertrophy13
CD-I Mice
M
13
Log-Probit
0.998
25.8
1602
957
714
Medinsky et al..
1999
F
13
Logistic
0.24
44.3
1826
1255
937
Medinsky et al.,
1999
Liver weight3
(absolute)
CD-I Mice
M
13
2° Polynomial
(constant variance)
0.99
-76.8
1754
936
699
Medinsky et al..
1999
F
13
2° Polynomial
(constant variance)
0.32
- 111.5
1109
709
529
Medinsky et al.,
1999
F344 Rats
M
13
Linear
(non-constant variance6)
0.39
27.9
1648
1260
932
Medinsky et al.,
1999
F
13
Linear
(constant variance)
0.99
- 19.7
1663
1300
962
Medinsky et al.,
1999
Sprague-
Dawley
M
4
Linear
(constant variance)
0.55
49.7
2309
1616
1196
White et al., 1995
Rats
F
4
2° Polynomial
(constant variance)
0.97
-22.4
1593
779
576
White et al., 1995
Liver weight3
(relative)
Sprague-
Dawley
M
4
2° Polynomial
(constant variance)
0.80
-81.7
2678
1619
1198
White et al., 1995
Rats
F
4
Linear
(non-constant variance6)
0.30
-89.4
1600
997
738
White et al., 1995
07/14/2009
98
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-2. Summary of BMD modeling results of ETBE inhalation toxicity studies for selection of principal
study
Organ
Endpoint
Species
strain
Sex
Dosing
duration(
wks)
"Best-fit" model
Goodness
of fit
/j-valuc
AIC
BMC
(ppm)
BMCLC
(ppm)
BMCLhec"
(mg/m3)
Reference
Kidney
Kidney weight3
(absolute)
F344 Rats
M
13
Linear
(constant variance)
0.23
-130.4
2,169
1,632
1,208
Medinsky et al.,
1999
F
13
2° Polynomial
(constant variance)
0.83
-191.2
717
494
366
Medinsky et al.,
1999
Sprague-
Dawley
Rats
M
4
Linear
(constant variance)
0.87
-54.7
3,010
1,960
1,450
White et al., 1995
Lf
F344 Rats
M
13
2° Polynomial
(non-constant variance0)
0.12
15.4
160
81
60
Medinsky et al.,
1999
Regenerative
foci3b
F344 Rats
M
13
Hill3
(non-constant variance6)
0.14
82.4
40
23
17
Medinsky et al.,
1999
F344 Rats
M
13
Logisticb
1.0
25.1
47
24
18
Medinsky et al.,
1999
Testes
Percent
seminiferous
tubules with
degenerated
spermatocytes1
F344 Rats
M
13
Linear
(non-constant variance6)
0.41
145.7
397
268
198
Medinsky et al.,
1999
Adrenal
gland
Adrenal gland
weight3
(absolute)
F344 Rats
M
13
Linear
(constant variance)
0.40
-387.3
3,214
2,223
1,645
Medinsky et al.,
1999
F
13
Linear
(constant variance)
0.54
-406.7
3,576
2,394
1,771
Medinsky et al.,
1999
Sprague-
Dawley
Rats
M
4
Power
(constant variance)
0.14
-351.3
3,367
2,220
1,643
White et al., 1995
07/14/2009
99
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-2. Summary of BMD modeling results of ETBE inhalation toxicity studies for selection of principal
study
Organ
Endpoint
Species
strain
Sex
Dosing
duration(
wks)
"Best-fit" model
Goodness
of fit
/j-valuc
AIC
BMC
(ppm)
BMCLC
(ppm)
BMCLhec"
(mg/m3)
Reference
Bone
marrow
Bone marrow
congestion13
F344 Rats
F
13
3° Multistage
0.98
17.5
987
401
297
Medinsky et al.,
1999
Heart
Heart weight3^
(absolute)
F344 Rats
F
13
All continuous models
exhibited significant lack-
of-fit
No data
No data
No data
No data
No data
Medinsky et al.,
1999
'Continuous data.
bDichotomous data.
Tor continuous variables, a BMR of a change of one SD from the control mean was employed, while for dichotomous variables, a BMR of a 10% change
relative to controls was used.
dThe BMCL (in ppm) was converted to a BMCLHec (in mg/m3) using the following three steps: (1) the BMCL (in ppm) was converted to standard units of
mg/m3 using the equation, mg/m3 = (ppm x molecular weight)/24.45, where molecular weight for ETBE = 102.18; (2) the BMCL (in mg/m3) was duration-
adjusted to a 24 hour/day, 7 day/week exposure by multiplying by 6 hours/24 hours and 5 days/7 days; and (3) assuming ETBE is a Category 3 gas, this
24 hour/day, 7 day/week BMCL (in mg/m3) was converted to a HEC for extra-respiratory effects by multiplying it by the ratio of animal to human blood:gas
partition coefficients, which was 1.0 for mice and 11.6/11.7 or 0.99 for rats.
eVariance model employed (i.e., variance modeled as a power function of the mean) failed to adequately address non-constant variance.
fHeart weight was increased in female rats exposed to 500 or 5,000 ppm, not 1,750 ppm; therefore, the response is not amenable to BMD modeling or the
LOAEL/NOAEL approach.
07/14/2009
100
DRAFT - DO NOT CITE OR QUOTE
-------
For each endpoint modeled, BMDS performed a %2 goodness-of-fit test and determined
the AIC. The goodness-of-fit test is a measure of how well the model fits the observed data.
Models with xV-values >0.1 were considered to exhibit adequate fits. The AIC is a measure of
the model fit based on the log-likelihood at the maximum likelihood estimates for the
parameters. Within the subset of models that exhibit adequate fit, models with lower AIC values
are preferred. The "best-fit" model selection criteria are presented in Appendix B and described
in detail in EPA's Benchmark Dose Technical Guidance Document (U.S. EPA, 2000b).
Medinsky et al. (1999) was selected as the principal study (Section 5.2.1) and
regenerative foci in the kidneys of male rats as the critical effect. Based on application of the
criteria from the Benchmark Dose Technical Guidance Document (U.S. EPA, 2000b) as
described above the Hill model provided the best fit to the mean number of regenerative foci
data.
5.2.2.1. Adjustment to a Human Equivalent Exposure Concentration
Because the RfC is a standard applicable to a continuous lifetime human exposure but
derived from animal studies featuring intermittent, subchronic exposure, EPA guidance (U.S.
EPA, 1994a) provides mechanisms for: (1) adjusting experimental exposure concentrations to a
value reflecting continuous exposure duration and (2) determining a human equivalent
concentration (HEC) from the animal exposure data. The former employs an inverse
concentration-time relationship to derive a health-protective duration adjustment to time-weight
the intermittent exposures used in the studies. The BMCL for each endpoint in ppm was
converted to standard units (mg/m3) using the equation mg/m3 = (ppm x molecular
weight /24.45, where molecular weight for ETBE = 102.18. This animal exposure in standard
units is then adjusted to reflect a continuous exposure by multiplying it by
(6 hours/day)/(24 hours/day) and (5 days/week)/(7 days/week) as follows:
BMCLadj = BMCL (mg/m3) x 6/24 x 5/7
The RfC methodology provides a mechanism for deriving a human equivalent
concentration from the duration-adjusted POD (BMCLadj) determined from the animal data.
The approach takes into account the extra-respiratory nature of the toxicological responses and
accommodates species differences by considering blood:air partition coefficients for ETBE in
the laboratory animal (rat or mouse) and humans. According to the RfC guidelines (U.S. EPA,
1994b), ETBE is a Category 3 gas because it is largely inactive in the respiratory tract, is rapidly
transferred between the lungs and blood, and the toxicological effects observed are extra-
respiratory. Therefore, the duration-adjusted BMCLadj is multiplied by the ratio of
animal/human blood:air partition coefficients (LA/LH). As detailed in Section 3.2, the values
07/14/2009
101
DRAFT - DO NOT CITE OR QUOTE
-------
reported in the literature for these parameters include an LA of 11.6 for Wistar rats (Kaneko et
al., 2000) and an LH in humans of 11.7 (Nihlen et al., 1995). No data were available on
blood:gas partitioning of ETBE in the mouse; therefore, the default ratio of 1.0 (U.S. EPA,
1994b) was used for the mouse data. This allowed a BMCLhec to be derived as follows:
For rat data:
BMCLhec = BMCLadj (mg/m3) x (LA/LH) (interspecies conversion)
= BMCLadj (mg/m3) x (11.6/11.7)
= BMCLadj (mg/m3) x (0.1368)
For mouse data:
BMCLhec = BMCLadj (mg/m3) x 1 (interspecies conversion)
Table 5-2 shows the BMCLhec values calculated for each of the endpoints modeled. The
BMCLhec value of 17 mg/m3 for mean number of regenerative foci in the kidneys of male F344
rats was used as the POD to derive the RfC for ETBE.
5.2.3. RfC Derivation—Including Application of Uncertainty Factors (UFs)
The BMCLhec of 17 mg/m3 for the mean number of regenerative foci in the kidneys of
male F344 rats exposed to ETBE for 13 weeks was used as the POD for the RfC.
A total UF of 3,000 was applied to the POD of 17 mg/m3: 3 for interspecies extrapolation
from animals to humans (UFA); 10 for human intraspecies variability (UFH); 10 to extrapolate
from a subchronic to a chronic study (UFS); and 10 to account for database deficiencies (UFD).
A threefold UF was used to account for uncertainties in extrapolating from laboratory
animals to humans. This value is adopted by convention where an adjustment from an animal-
specific BMCLadj to a BMCLhec has been incorporated. Application of a full UF of 10 would
depend on two areas of uncertainty (i.e., toxicokinetic and toxicodynamic uncertainties). In this
assessment, the toxicokinetic component is mostly addressed by the determination of a HEC as
described in the RfC methodology (U.S. EPA, 1994b). The toxicodynamic uncertainty is also
accounted for to a certain degree by the use of the applied dosimetry method.
A 10-fold UF was used to account for variation in susceptibility among members of the
human population (i.e., interindividual variability). Insufficient information is available to
predict potential variability in human susceptibility.
A 10-fold UF was used to extrapolate from subchronic to chronic exposure. The
BMCLhec of 17 mg/m3 for the mean number of regenerative foci from Medinsky et al. (1999)
was observed in the kidneys of male F344 rats exposed to ETBE for 13 weeks. Therefore UFS of
10 was applied to extrapolate from subchronic to chronic exposure.
A UFd of 10 was used to account for deficiencies in the toxicity database of inhalation
exposure to ETBE. Data on the effects of ETBE in humans is limited to several 2-hour
inhalation studies. The inhalation database contains several subchronic studies in mice and rats.
The database for ETBE lacks both a developmental toxicity study and a multigeneration
07/14/2009 102 DRAFT - DO NOT CITE OR QUOTE
-------
reproductive toxicity study by inhalation exposure, although oral studies of developmental and
reproductive toxicity have been conducted (CIT, 2004a, b, unpublished reports). The lack of an
immunotoxicity study of inhalation exposure to ETBE also contributes to the database
uncertainty in light of the potential for suppression of the antibody response suggested by the
unpublished G/ETBE mixture study (White, 2002, unpublished report).
A UF for LOAEL to NOAEL extrapolation was not used because the current approach is
to address this factor as one of the considerations in selecting a BMR for BMD modeling. In this
case, a BMR of a change of one SD from the control mean was selected under the assumption
that it represents a minimal biologically significant change. Therefore the RfC from the data in
Medinsky et al. (1999) was calculated as follows:
RfC = BMCLhec - UF
= 17 mg/m3- 3,000
= 0.0057 or 6.0 x "3 mg/m3
5.2.4. Previous RfC
An inhalation assessment for ETBE was not previously available on IRIS.
5.2.5. Reference Value Comparison Information
Figure 5-1 presents the POD, applied UFs, and derived chronic reference values (RfVs)
for additional effect endpoints that were modeled using EPA BMDS (version 1.4.1) and appear
in Table 5-2. This comparison is intended to provide information on additional health effects
associated with ETBE exposure.
PODs and chronic RfCs that could be derived from the additional health effects identified
in Table 5-2 are presented in Figure 5-1 to allow a comparison with the critical effect.
Consideration of the available dose-response data to determine an estimate of inhalation
exposure that is likely to be without an appreciable risk of adverse health effects over a lifetime
has led to the selection of the Medinsky et al. (1999) and mean number of regenerative foci in
male rat kidneys as the principal study and critical effect for deriving the chronic RfC for ETBE.
07/14/2009
103
DRAFT - DO NOT CITE OR QUOTE
-------
10000.00
1000.00
100.00
=¦ 10.00 --
¦E- i.oo
0.10
0.01
0.00
I
fihrt
JL
HI
ft
Mb
JL
v-
_Ll
////// # ? / / / ? ? / / / / * / / /
/ / // / / / / / / / / / / / / / / / / /
/////////////////////
/ //
3 $
Erelpoint [Sot, Strain, Sped as]
OPOD
in Aihiahto-limai
~ Himai uaidttji
Os ibcl r t> c k roi t
0 Datto35 e defIcfe i cles
~ PIC
Figure 5-1. RfV comparison array for alternative PODs for inhalation data.
07/14/2009
104
DRAFT - DO NOT CITE OR QUOTE
-------
There are a variety of renal effects seen in rats and hepatic effects observed in rats and
mice, as well as increased adrenal weight found in rats. Effects observed in the kidney included
increased kidney weight in male and female rats, as well as increased mean number and
incidence of regenerative foci and sustained proliferation in the kidney indicated by a sustained
increase in the LI in male rats. A number of hepatic effects were considered as a potential
critical effect including sustained proliferation in the liver indicated by a sustained increase in the
LI in mice, increased centrilobular hypertrophy in male and female mice, and increased liver
weight in mice and rats. Additional effects considered as potential critical effects included an
increased percentage of seminiferous tubules with degenerated spermatocytes in male rats,
increased adrenal gland weight in male and female rats, increased heart weight in female rats,
and increased incidence of bone marrow congestion in female rats. PODs and chronic RfV that
could be derived from the additional health effects identified in Table 5-2 are presented in Figure
5-1 to allow a comparison with the critical effect. For hepatic LI, centrilobular hypertrophy,
increased liver weight, increased kidney weight, renal LI, regenerative foci in the kidneys,
percentage of seminiferous tubules with degenerated spermatocytes, increased adrenal gland
weight, and increased incidence of bone marrow congestion, the total UF applied was 3,000-fold;
threefold UF to account for uncertainty in extrapolating from laboratory animals to humans, 10-
fold UF to account for variation in susceptibility among members of the human population, 10-
fold UF to account for sub chronic-to-chronic extrapolation, and 10-fold UF for database
deficiencies.
5.3. UNCERTAINTIES IN THE INHALATION REFERENCE CONCENTRATION
(RfC)
Risk assessments need to portray associated uncertainty. The following discussion
identifies uncertainties associated with the chronic RfC for ETBE. As presented earlier (Sections
5.1.2 and 5.1.3; 5.2.2 and 5.2.3), the UF approach, following EPA practices and RfC and RfD
guidance (U.S. EPA, 1994b), was applied to a POD, a BMDLrec for the chronic RfC. Factors
accounting for uncertainties associated with a number of steps in the analyses were adopted to
extrapolating from an animal bioassay to human exposure, a diverse population of varying
susceptibilities, and to account for database deficiencies. These extrapolations are carried out
with default approaches given the paucity of experimental ETBE data to inform individual steps.
The database of animal toxicity studies available for the hazard assessment of ETBE, as
described throughout the previous section (Chapter 4), is limited. The database of oral toxicity
studies includes a prenatal developmental toxicity study, two-generation reproductive toxicity
study, and mating and fertility reproductive toxicity study. A single chronic cancer bioassay is
available, but the authors did not evaluate or report any noncancer endpoints except mortality.
There are no standard subchronic or chronic toxicity studies available for oral exposure to ETBE.
Toxicity associated with oral exposure to ETBE is observed as increased organ weight in the
07/14/2009
105
DRAFT - DO NOT CITE OR QUOTE
-------
liver and kidneys and decreased body weight gain. The database of inhalation toxicity studies in
animals includes two subchronic studies in rats and one subchronic study in mice. No chronic
inhalation studies are available. Data from these inhalation toxicity studies of ETBE also
indicated that the kidney and liver are target organs for ETBE toxicity as determined by
increased kidney and liver weights and additional effects described below. Effects associated
with inhalation exposure to ETBE were observed in the liver of mice and rats, kidney and
adrenal gland in rats of both sexes, heart and bone marrow in female rats, and testes of male rats.
In the liver and kidney, an increase in organ weights, LI, regenerative foci (kidney only), and
hepatic centrilobular hypertrophy were reported. Heart and adrenal weights were also increased
in rats, but no histopathological lesions were present. Bone marrow congestion was observed in
female rats, in the absence of additional hematopoietic effects. An increased percentage of
degenerated spermatocytes was also observed in male rats. In addition to the oral and inhalation
data are numerous absorption, distribution, metabolism, and excretion references. Critical data
gaps have been identified and uncertainties associated with data deficiencies are more fully
discussed below.
The critical effect selected for the derivation of the chronic RfC is the mean number of
regenerative foci in the kidneys of male rats. Although an increase in liver weights was apparent
rats and mice, lesions and serum enzyme levels indicative of liver damage were not evident in
rats. Hepatic centrilobular hypertrophy and an increased LI were observed in CD-I mice. The
bone marrow congestion observed in female rats was not considered adverse, as there was no
change in the clinical chemistry and hematology parameters.
The selection of the BMD models for the quantitation of the chronic RfC does not lead to
significant uncertainty in estimating the PODs since benchmark effect levels were within the
range of experimental data. However, the selected model for the RfC, the Hill model, does not
represent all possible models one might fit, and other models could be selected to yield more
extreme results, both higher and lower than those included in this assessment.
Extrapolating from animals to humans embodies further issues and uncertainties. The
effect and the magnitude associated with the concentration at the POD in rodents are
extrapolated to human response. Pharmacokinetic models are useful to examine species
differences in pharmacokinetic processing; however, dosimetric adjustment using
pharmacokinetic modeling was not possible for the toxicity observed following oral and
inhalation exposure to ETBE. For the chronic RfC, a factor of 3 was adopted by convention
where an adjustment from an animal specific BMCLadj to a BMCLrec has been incorporated.
Application of a full UF of 10 would depend on two areas of uncertainty (i.e., toxicokinetic and
toxicodynamic uncertainties). In this assessment, the toxicokinetic component is mostly
addressed by the determination of a human equivalent concentration as described in the RfC
methodology (U.S. EPA, 1994b). The toxicodynamic uncertainty is also accounted for to a
07/14/2009
106
DRAFT - DO NOT CITE OR QUOTE
-------
certain degree by the use of the applied dosimetry method and a UF of 3 is retained to account
for this component.
Heterogeneity among humans is another uncertainty associated with extrapolating doses
from animals to humans. Uncertainty related to human variation also needs consideration in
extrapolating dose from a subset or smaller sized population, say of one sex or a narrow range of
life stages typical of occupational epidemiologic studies, to a larger, more diverse population. In
the absence of ETBE-specific data on human variation, a factor of 10 was used to account for
uncertainty associated with human variation in the derivation of the chronic RfC. Human
variation may be larger or smaller; however, ETBE-specific data to examine the potential
magnitude of over- or under-estimation are unavailable.
The database of inhalation studies is also of concern due to the lack of a chronic toxicity
study, a developmental study, and a multigenerational reproductive toxicity study. The
inhalation database contains several subchronic studies in mice and rats. The database for ETBE
lacks both a developmental toxicity study and a multigenerational reproductive toxicity study by
inhalation exposure, although oral studies of developmental and reproductive toxicity have been
conducted (CIT, 2004a, b, unpublished reports). The lack of an immunotoxicity study of
inhalation exposure to ETBE also contributes to the database uncertainty in light of the potential
for suppression of the antibody response suggested by the unpublished G/ETBE mixture study
(White, 2002, unpublished report).
5.4. CANCER ASSESSMENT
Available data indicate there is suggestive evidence of carcinogenic potential (U.S. EPA,
2005a) following exposure to ETBE. One oral animal cancer bioassay in rats (Maltoni et al.,
1999) is available. Maltoni et al. (1999) exposed Sprague-Dawley rats (60/sex/group) to 0, 250,
and 1,000 mg/kg-day of ETBE by gavage 4 days/week for 104 weeks. Statistically significant
increases in two tumor types were identified in this study: total pathologies of oncological
interest of the mouth epithelium at the high dose (1,000 mg/kg-day) in males and total malignant
uterine tumors only at the low dose (250 mg/kg-day) in females. Nonsignificant increases in
total pathologies of oncological interest of the forestomach in males (at the lower of the two test
doses) and hemolymphoreticular system were reported in both sexes. In all four cases, the total
tumors including precancers were listed as increased, and no individual tumor type was reported
as increased. This study was reported as a preliminary study and did not include criteria used for
the histopathological classification. This is especially relevant to the inclusion of dysplasias as
one of the pathologies of oncological interest for both the mouth epithelium and the forestomach.
Although there is an association between severe dysplasia and eventual carcinoma, the majority
of lesions with mild to moderate dysplasia do not progress to cancer. The significance of the
reported increase in total pathologies of oncological interest of the forestomach and mouth
epithelium is, therefore, confounded as dysplasias represent high portions (i.e., 58-100%) of the
07/14/2009
107
DRAFT - DO NOT CITE OR QUOTE
-------
reported data. The authors grouped tissues, such as total pathologies of oncological interest in
the mouth epithelium (including the tongue, lips, and oral cavity), without providing individual
tumor information. Tumors of the uterus and vagina were also combined, and the total
malignant tumors of the uterus would not be statistically increased relative to control if the
vaginal tumors were removed. Knowledge of the historical incidence of tumors in this
laboratory would provide further context for the concurrent controls. Historical controls may be
particularly relevant to the incidence of hemolymphoreticular neoplasias, which are listed as
nonsignificantly increased in the ETBE study. An examination of other oral gavage studies by
the same lab suggests that the 5% incidence in control males from Maltoni et al. (1999) may be
low relative to historical controls. The total pathologies of oncological interest of the mouth
epithelium in males were the only pathologies that exhibited a dose-response or positive dose-
related trend. The increased mortality of animals at both doses of ETBE presents a limitation in
the study and the ability to interpret the results from a quantitative perspective. For these
reasons, an estimate of cancer risks was not quantified.
07/14/2009
108
DRAFT - DO NOT CITE OR QUOTE
-------
6. MAJOR CONCLUSIONS IN THE CHARACTERIZATION OF
HAZARD AND DOSE RESPONSE
6.1. HUMAN HAZARD POTENTIAL
ETBE is a colorless liquid with the chemical formula of C6Hi40 and a characteristic
strong, gasoline-like odor, and a taste that has been described as highly objectionable. It is an
oxygenate used as a gasoline additive at amounts up to 17% by weight to reduce the emission of
carbon monoxide and ozone into the atmosphere. ETBE is released into the environment
through the use and storage of gasoline. Environmental concern surrounding fuel oxygenates is
associated with automotive emissions, inhalation and/or dermal exposure while refueling, and
oral or dermal exposure from groundwater contamination (largely from leaking underground fuel
storage tanks).
Absorption data are only available for exposure to ETBE via inhalation. The percent of
the respired dose retained was 32-34%, with a portion of the absorbed dose exhaled, to result in
a net uptake via the respiratory tract of about 26% (Nihlen et al., 1998a). Distribution data from
ETBE are not available, although data from MTBE, the methyl analog of ETBE, are available
along with in vitro experiments comparing MTBE and ETBE partition coefficients in human
blood samples. The higher oil: water partition coefficient of ETBE suggests a greater distribution
to fat and lipid-rich tissues than MTBE. The major metabolites of ETBE are TBA and
acetaldehyde. TBA can be sulfated, glucuronidated, or oxidized to MPD, and subsequently to
HBA.
Data on the effects of ETBE in humans are limited to several 2-hour inhalation studies at
doses up to 50 ppm. Healthy subjects exposed to ETBE experienced irritation in throat and
airways, nasal swelling, a bad taste in the mouth, and slightly impaired lung function (about 3%
reduction in vital capacity and forced vital capacity). No chronic inhalation study was available,
although there are several subchronic studies in mice and rats.
Liver and kidney toxicity were the primary noncancer health effects of subchronic oral
and inhalation exposure to ETBE based on the limited available animal data. Increased liver and
kidney weights were observed following oral exposure in rats of both sexes in a two-generation
reproductive toxicity study (CIT, 2004b, unpublished report). However, histology was only
performed if abnormal morphology was detected at necropsy, and, therefore, there are limited
histological data to support nephrotoxicity or hepatotoxicity. Oral exposure to ETBE was also
associated with decreased body weight gain in rats (CIT 2004 a, b, unpublished reports).
Additional evidence of ETBE-associated kidney toxicity is provided by the ETBE inhalation
exposure database. Increased kidney weight was also observed following subchronic inhalation
exposure to ETBE in male and female F344 rats (Medinsky et al., 1999) and male Sprague-
Dawley rats (White et al., 1995). Data from the inhalation studies demonstrated that increased
kidney weight was associated with histopathological changes in male rats (e.g., regenerative foci
07/14/2009
109
DRAFT - DO NOT CITE OR QUOTE
-------
indicative of cellular necrosis) in males and sustained increase of greater than twofold in cellular
proliferation (Medinsky et al., 1999). The increased kidney weight was not associated with
histopathological changes in female rats, although an increased cellular proliferative response of
less than twofold was observed at early time points (after 1 and 4, but not 13 weeks of exposure)
(Medinsky et al., 1999). Based on the available data for ETBE, there is the suggestion that
alpha2U-globulin accumulation is the mode of action; however, there is considerable uncertainty
because of the lack of evidence of typical lesions in the pathological sequence of lesions
associated with alpha2U-globulin nephropathy and, therefore, a determination cannot be made as
to whether alpha2U-globulin accumulation is the mode of action or the only mode of action for
renal effects associated with ETBE exposure. Increased liver weight was also observed
following subchronic inhalation exposure to ETBE in mice and rats of both sexes (Medinsky et
al., 1999; White et al., 1995). No changes in histopathology or serum levels of hepatic enzymes
were observed in rats. Dose-related increases in hepatic proliferation were observed in male and
female mice at some time periods during a 13-week exposure, and not observed at others
(Medinsky et al., 1999).
Inhalation exposure to ETBE was also associated with increased adrenal gland weights in
male and female rats, increased percentage of seminiferous tubules with degenerated
spermatocytes in male rats, increased heart weight in female rats, and increased incidence of
bone marrow congestion in female rats (Medinsky et al., 1999). No sensorimotor dysfunction,
neuromuscular dysfunction, or microscopic evidence of neuropathy were detected in F344 rats
after 13 weeks of inhalation exposure (Dorman et al., 1997). Evidence of suppressed antibody
production from a G/ETBE mixture inhalation immunotoxicity study (White, 2002, unpublished
report) suggests that ETBE may have immunosuppressive activity, but there are no data from an
immunotoxicity study of ETBE alone.
Under the Guidelines for Carcinogen Risk Assessment (U.S. EPA 2005a) available data
for ETBE can be classified as suggestive evidence of carcinogenic potential, based on a single
oral animal cancer bioassay in rats (Maltoni et al., 1999). Although there is evidence of
carcinogenicity of both TBA, the primary metabolite of ETBE (NTP, 1995; Cirvello et al.,
1995), and the other primary metabolite, acetaldehyde (WHO, 1995), as well as MTBE (the
methyl substituted analog of ETBE reviewed in Ahmed, 2001; Cal EPA, 1999; Mennear, 1997),
there is a lack of data on ETBE, its mode(s) of action, and whether the parent compound or
metabolites are responsible for observed effects. The genotoxicity data for ETBE were
essentially negative.
6.2. DOSE RESPONSE
6.2.1. Noncancer / Oral
Considering the uncertainties in the ETBE database, which are described in Appendix C,
the total composite UF is 10,000, consisting of four areas of maximum uncertainty. In the report,
07/14/2009 110 DRAFT - DO NOT CITE OR QUOTE
-------
A Review of the Reference Dose and Reference Concentration Processes (U.S. EPA, 2002), the
RfD/RfC technical panel concluded that, in cases where maximum uncertainty exists in four or
more areas of extrapolation, or when the total UF is 10,000 or more, it is unlikely that the
database is sufficient to derive a reference value. Appendix C contains a derivation of an oral
minimal data value for ETBE using an UF of 10,000. Use of this minimal data value is not
recommended except in limited circumstances, for example, in screening level risk assessments
or to rank relative risks. Any use of this value should include a discussion of the uncertainty
associated with its derivation.
6.2.2. Noncancer / Inhalation
The RfC of 6 x 10"3 mg/m3 was derived based on increased mean number of regenerative
foci in the kidneys in male rats exposed to ETBE for 13 weeks by inhalation (Medinsky et al.,
1999). This study was chosen as the principal study because the kidney effects (i.e., regenerative
foci) observed in this study represent the most sensitive effects identified in the database
evaluating exposure through the inhalation route to ETBE. In addition, EPA considers
regenerative foci to be indicators of cellular necrosis and a biomarker of an adverse effect. The
RfC is derived by dividing the BMCL of 17 mg/m3 by a composite UF of 3,000 (factor of 3 for
interspecies variability, factors of 10 for interindividual variability, sub chronic-to-chronic
extrapolation, and database deficiencies). A factor of 3 was selected to account for uncertainties
in extrapolating from rats to humans, which is adopted by convention where an adjustment from
an animal specific BMCLadj to a BMCLrec has been incorporated. Insufficient information is
available to predict the potential variability in human susceptibility among the population; thus,
the human variability UF of 10 was applied. A 10-fold UF was used to account for uncertainty
in extrapolating from a subchronic to chronic exposure duration. A 10-fold UF was used to
account for deficiencies in the database. The database for ETBE lacks both a developmental
toxicity and a multigenerational reproductive toxicity study by inhalation exposure.
The overall confidence in this RfC assessment is medium. Confidence in the principal
study (Medinsky et al., 1999) is medium. Confidence in the database is low-to-medium due to
the lack of both a developmental toxicity and a multigenerational reproductive toxicity study by
inhalation exposure. Reflecting medium confidence in the principal study and low-to-medium
confidence in the database, confidence in the RfC is low-to-medium.
6.2.3. Cancer / Oral
Available data that indicate there is suggestive evidence of carcinogenic potential (U.S.
EPA, 2005a) following exposure to ETBE. One oral animal cancer bioassay in rats (Maltoni et
al., 1999) is available. Statistically significant increases in two tumor types were identified in
this study: total pathologies of oncological interest of the mouth epithelium at the high dose
(1,000 mg/kg-day) in males and total malignant uterine tumors only at the low dose (250 mg/kg-
07/14/2009
111
DRAFT - DO NOT CITE OR QUOTE
-------
day) in females. Nonsignificant increases in total pathologies of oncological interest of the
forestomach in males (at the lower of the two test doses) and hemolymphoreticular system were
reported in both sexes. In all four cases, the total tumors including precancers were listed as
increased, and no individual tumor type was reported as increased. This study did not include
the criteria used for the histopathological classification, which is especially relevant to the
inclusion of dysplasias as one of the pathologies of oncological interest for both the mouth
epithelium and the forestomach. Although there is an association between severe dysplasia and
eventual carcinoma, the majority of lesions with mild to moderate dysplasia do not progress to
cancer. The significance of the reported increase in total pathologies of oncological interest of
the forestomach and mouth epithelium is, therefore, confounded as dysplasias represent high
portions (i.e., 58-100%) of the reported data. The authors grouped tissues, such as total
pathologies of oncological interest in the mouth epithelium (including the tongue, lips, and oral
cavity) without providing individual tumor information. Tumors of the uterus and vagina were
also combined, and the total malignant tumors of the uterus would not be statistically increased
relative to control if the vaginal tumors were removed. Knowledge of the historical incidence of
tumors in this laboratory would provide further context for the concurrent controls. Historical
controls may be particularly relevant to the incidence of hemolymphoreticular neoplasias, which
are listed as nonsignificantly increased in the ETBE study. An examination of other oral gavage
studies by the same lab suggests that the 5% incidence in control males from Maltoni et al.
(1999) may be low relative to historical controls. The total pathologies of oncological interest of
the mouth epithelium in males were the only pathologies that exhibited a dose-response or
positive dose-related trend. The increased mortality of animals at both doses of ETBE presents a
limitation in the study and in the ability to interpret the results from a quantitative perspective.
For these reasons, an estimate of cancer risks was not quantified.
07/14/2009
112
DRAFT - DO NOT CITE OR QUOTE
-------
7. REFERENCES
Acharya, S; Mehta, K; Rodriguez, S; et al. (1995) Administration of subtoxic doses of t-butyl alcohol and
trichloroacetic acid to male Wistar rats to study the interactive toxicity. Toxicol Lett 80(l-3):97-104.
Acharya, S; Mehta, K; Rodriguez, S; et al. (1997) A histopathological study of liver and kidney in male Wistar rats
treated with subtoxic doses of t-butyl alcohol and trichloroacetic acid. Exp Toxicol Pathol 49(5):369-373.
Ahmed, FE. (2001) Toxicology and human health effects following exposure to oxygenated or reformulated
gasoline. Toxicol Lett 123(2—3):89—l 13.
Alcorn, J; McNamara, PJ. (2002) Ontogeny of hepatic and renal systemic clearance pathways in infants. Part II.
Clin Phannacokinet 41(13):1077-1094.
Alden, CL. (1986) A review of unique male rat hydrocarbon nephropathy. Toxicol Pathol 14:109-111.
Amberg, A; Rosner, E; and Dekant, W. (2000) Biotransformation and kinetics of excretion of ethyl tert-butyl ether
in rats and humans. Toxicol Sci 53:194-201.
Ames, BN; Elson-Schwab, I; Silver, EA. (2002) High-dose vitamin therapy stimulates variant enzymes with
decreased coenzyme binding affinity (increased Km): relevance to genetic disease and polymorphisms. Am J Clin
Nutr 75:616-658.
Andersen, ME. (1991) Physiological modeling of organic compounds. Ann Occup Hyg 35(3):309-321.
Bebia, Z; Buch, SC; Wilson, JW; et al. (2004) Bioequivalence revisited: influence of age and sex on CYP enzymes.
Clin Pharmacol Ther 76:618-627.
Belpoggi, F; Soffritti, M; Maltoni, C. (1995) Methyl-tertiary-butyl ether (MTBE) - a gasoline additive - causes
testicular and lymphohaematopoietic cancers in rats. Toxicol Ind Health 11(2): 119-149.
Belpoggi, F; Soffritti, M; Filippini, F; et al. (1997) Results of long-term experimental studies on the carcinogenicity
of methyl tert-butyl ether. Ann NY Acad Sci 837:77-95.
Belpoggi, F; Soffritti, M; Maltoni, C. (1998) Pathological characterization of testicular tumours and lymphomas-
leukaemias, and of their precursors observed in Sprague-Dawley rats exposed to methyl-tertiary-butyl-ether
(MTBE). Eur J Oncol 3(3):201-206.
Berger, T; Horner, CM. (2003) In vivo exposure of female rats to toxicants may affect oocyte quality. Reprod
Toxicol 17(3):273-281.
Bernauer, U; Amberg, A; Scheutzow, D; et al. (1998) Biotransformation of 12C- and 2-13C-labeled methyl tert-butyl
ether, ethyl tert-butyl ether, and tert-butyl alcohol in rats: identification of metabolites in urine by 13C nuclear
magnetic resonance and gas chromatography/mass spectrometry. Chem Res Toxicol 11(6):651-658.
Bird, MG; Burleigh-Flayer, HD; Chun, JS; et al. (1997) Oncogenicity studies of inhaled methyl tertiary butyl ether
(MTBE) in CD-I mice and F344 rats. J Appl Toxicol 17(Suppl 7):S45-S55.
Blanco, JG; Harrison, PL; Evans, WE; et al. (2000) Human cytochrome P450 maximal activities in pediatric versus
adult liver. Drug Metab Dispos 28:370-382.
Blue Ribbon Panel on Oxygenates in Gasoline. (1999) Achieving clean air and clean water: the report of the Blue
Ribbon Panel on Oxygenates in Gasoline. U.S. EPA, Washington, DC. EPA420-R-99-021. Available online at
http://www.epa.gov/otaq/consumer/fuels/oxypanel/r99021 .pdf.
Bond, JA; Medinsky, MA; Wolf, DC; et al. (1996a) Ethyl tertiary butyl ether (ETBE): ninety-day vapor inhalation
toxicity study with neurotoxicity evaluations in Fischer 344 rats. Chemical Industry Institute of Toxicology under
07/14/2009
113
DRAFT - DO NOT CITE OR QUOTE
-------
contract to ARCO Chemical Company, Research Triangle Park, NC; Laboratory Project ID 95029, 1-90.
Unpublished report.
Bond, JA; Medinsky, MA; Wolf, DC; et al. (1996b) Ethyl tertiary butyl ether (ETBE): ninety-day vapor inhalation
toxicity study in CD-I mice. Chemical Industry Institute of Toxicology under contract to ARCO Chemical
Company, Research Triangle Park, NC; Laboratory Project ID 95030, 1-69. Unpublished report.
Borghoff, SJ. (1996) Ethyl tertiary butyl ether (ETBE): a pilot/methods development pharmacokinetic study in male
F344 rats and male CD-I mice after a single nose-only inhalation exposure. Chemical Industry Institute of
Toxicology under contract to ARCO Chemical Company, Research Triangle Park, NC; Laboratory Protocol Number
CIIT-95025. Unpublished report.
Borghoff, SJ; Murphy, JE; Medinsky, MA. (1996) Development of physiologically based pharmacokinetic model
for methyl tertiary-butyl ether and tertiary-butanol in male Fischer-344 rats. Fundam Appl Toxicol 30:264-275.
Borghoff, SJ; Prescott, JS; Janszen, DB; et al. (2001) alpha 2u-Globulin nephropathy, renal cell proliferation and
dosimetry of inhaled fcrt-biityl alcohol in male and female F-344 rats. Toxicol Sci 61(1): 176—186.
Burleigh-Flayer, HD; Chun, JS; Kintigh, WJ. (1992) Methyl tertiary butyl ether: vapor inhalation oncogenicity study
in CD-mice. Report No. 91N0013A. October 15. Submitted to U.S. EPA under TSCA Section 4 Testing Consent
Order 40 CFR 799.5000. EPA/OPTS#42098: Union Carbide Chemicals and Plastics Co., Bushy Run Research
Center, Export, PA.
Cal EPA (California Environmental Protection Agency). (1999) Public health goal for methyl tertiary butyl ether
(MTBE) in drinking water. Office of Environmental Health Hazard Assessment. Cal EPA, Sacramento, CA.
Cemeli, E; Wagner, ED; Anderson, D; et al. (2006) Modulation of the cytotoxicity and genotoxicity of the drinking
water disinfection byproduct lodoacetic acid by suppressors of oxidative stress. Environ Sci Technol 40(6): 1878-
1883.
CFDC (Clean Fuels Development Coalition). (2001) The ETBE fact book: a compilation of information on ethanol
ethers. Clean Fuels Development Coalition, Bethesda, MD. Available online at www.cleanfuelsdc.org.
CFDC (Clean Fuels Development Coalition). (2007) The ethanol fact book. Clean Fuels Development Coalition,
Washinton, DC. 07CFDC004 1107 7.5M. Available online at
www.cleanfuelsdc.org/pubs/documents/CFDC_Fact%20Book_1107.pdf.
Chen, GF; Tang, YM; Green, B; et al. (1999) Low frequency of CYP2A6 gene polymorphism as revealed by a one-
step polymerase chain reaction method. Pharmacogenetics 9(3):327-332.
Chun, JS; Burleigh-Flayer, HD; Kintigh, WJ. (1992) Methyl tertiary butyl ether: vapor inhalation oncogenicity study
in Fischer 344 rats. Report No. 91N0013B. Submitted to U.S. EPA under TSCA Section 4 Testing Consent Order
40 CFR 799.500. EPA/OPTS #42098. Bushy Run Research Center, Union Carbide Chemicals and Plastics Co.
Export, PA.
Cirvello, JD; Radovsky, A; Heath JE; et al. (1995) Toxicity and carcinogenicity of t-butyl alcohol in rats and mice
following chronic exposure in drinking water. Toxicol Ind Health 11(2): 151-165.
CIT (Centre International de Toxicologie). (2003) Ethyl tertiary butyl ether (ETBE), cas no. 637-92-3:
Reproductive/developmental toxicity dose-range finding/probe study by the oral route (gavage) in two strains of rat.
CIT under contract for TOTAL France S.A., Evreux, France. Study No. 24168 RSR. Unpublished report.
CIT (Centre International de Toxicologie). (2004a) Prenatal developmental toxicity study by the oral route (gavage)
in rats: ethyl tertiary butyl ether (ETBE). CIT under contract to TOTAL France S.A., Evreux, France; Study No.
24860 RSR. Unpublished report. [An external peer review was conducted by EPA in November 2008 to evaluate the
accuracy of experimental procedures, results, and interpretation and discussion of the findings presented. A report of
this peer review is available through the EPA's IRIS Hotline, at (202) 566-1676 (phone), (202) 566-1749 (fax), or
hotline.irisVv epa.gov (e-mail address) and on the IRIS website (www.epa.gov/iris).]
07/14/2009
114
DRAFT - DO NOT CITE OR QUOTE
-------
CIT (Centre International de Toxicologie). (2004b) Two-generation study (reproduction and fertility effects) by oral
route (gavage) in rats: ethyl tertiary butyl ether (ETBE). CIT under contract to TOTAL France S.A., Evreux,
France; Study No. 24859 RSR. Unpublished report. [An external peer review was conducted by EPA in November
2008 to evaluate the accuracy of experimental procedures, results, and interpretation and discussion of the findings
presented. A report of this peer review is available through the EPA's IRIS Hotline, at (202) 566-1676 (phone),
(202) 566-1749 (fax), or hotlinc.irisV7cpa.gov (e-mail address) and on the IRIS website (www.epa.gov/iris).]
Conigrave, KM; Saunders, JB; Reznik, RB; et al. (1993) Prediction of alcohol-related harm by laboratory test
results. Clin Chem 39(11):2266-2270.
Daughtrey, WC; Henley, M; Burnett, DM; et al. (2004) Inhalation toxicity of gasoline & fuel oxygenate: test sample
preparation. Toxicologist 78:146-147.
de Wildt, SN; Kearns, GL; Leeder, JS; et al. (1999) Cytochrome P450 3A: ontogeny and drug disposition. Clin
Phannacokinet 376:485-505.
Deeb, RA; Hu, HY; Hanson, JR; et al. (2001) Substrate interactions in BTEX and MTBE mixtures by an MTBE-
degrading isolate. Environ Sci Technol 35(2):312-317.
Dekant, W; Bernauer, U; Rosner, E; et al. (2001a) Biotransformation of MTBE, ETBE, and TAME after inhalation
or ingestion in rats and humans. Health Eff Inst Res Rep 102:29-71.
Dekant, W; Bernauer, U; Rosner, E; et al. (2001b) Toxicokinetics of ethers used as fuel oxygenates. Toxicol Lett
124(l-3):37-45.
Donnan DC; Strove, MF; Wong, BA; et al. (1997) Neurotoxicological evaluation of ethyl tertiary-butyl ether
following subchronic (90-day) inhalation in the Fischer 344 rat. J Appl Toxicol 17(4):235-242.
Dome, JLCM. (2004) Impact of inter-individual differences in drug metabolism and pharmacokinetics on safety
evaluation. Fundam Clin Pharmacol 18:609-620.
Dome, JLCM; Walton, K; Renwick, AG. (2005) Human variability in xenobiotic metabolism and pathway-related
uncertainty factors for chemical risk assessment: a review. Food Chem Toxicol 43:203-216.
Drogos, DL; Diaz, AF. (2002) Appendix A: physical properties of fuel oxygenates and additives. In: Diaz, AF;
Drogos, DL, eds. Oxygenates in gasoline: enviromnental aspects. Washington, DC: American Chemical Society/
Oxford University Press; pp. 258-279.
Durand, ML; Dietrich, AM. (2007) Contributions of silane cross-linked PEX pipe to chemical/solvent odours in
drinking water. Water Sci Technol 55(5): 153-160.
Fayolle, F; Vandecasteele, JP; Monot, F. (2001) Microbial degradation and fate in the enviromnent of methyl tert-
butyl ether and related fuel oxygenates. Appl Microbiol Biotechnol 56(3-4):339-349.
Fiedler, N; Kelly-McNeil, K; Molir, S; et al. (2000) Controlled human exposure to methyl tertiary butyl ether in
gasoline: symptoms, psychophysiologic and neurobehavioral responses of self-reported sensitive persons. Environ
Health Perspect 108(8):753-763.
Fukami, T; Nakajima, M; Yoshida, R; et al. (2004) A novel polymorphism of human CYP2A6 gene CYP2A6*17
has an amino acid substitution (V365M) that decreases enzymatic activity in vitro and in vivo. Clin Pharmacol Ther
76(6):519-527.
Gudi, R; Brown, CM. (2002) In vivo-in vitro rat peripheral lymphocyte sister chromatid exchange assay in gasoline
ETBE vapor condensate. BioReliance under contract to Huntingdon Life Sciences, Rockville, MD;
AA40NY.130.BTL, Sponsor No. 00-6129. Unpublished report.
Gyamfi, MA; Fujieda, M; Kiyotani, K; et al. (2005) High prevalence of cytochrome P450 2A6*1A alleles in a black
African population of Ghana. Eur J Clin Pharmacol 60(12):855-857.
07/14/2009
115
DRAFT - DO NOT CITE OR QUOTE
-------
Haworth S; Lawlor T; Morelman K; et al. (1983). Salmonella mutagenicity results for 250 chemicals. Environ
Mutagen 5(suppl 1):3—142.
Hines, RN; McCarver, DG. (2002) The ontogeny of human drug-metabolizing enzymes: phase I oxidative enzymes.
J Pharmacol Exp Ther 300(2):355-360.
Hong, JY; Yang, CS. (1997) Genetic polymorphism of cytochrome P450 as a biomarker of susceptibility to
enviromnental toxicity. Environ Health Perspect 105(Suppl 4):759-762.
Hong, JY; Yang, CS; Lee, M; et al. (1997a) Role of cytochromes P450 in the metabolism of methyl tert-butyl ether
in human livers. Arch Toxicol 71(4):266-269.
Hong, JY; Wang, YY; Bondoc, FY; et al. (1997b) Rat olfactory mucosa displays a high activity in metabolizing
methyl tert-butyl ether and other gasoline ethers. Fundam Appl Toxicol 40(2):205-210.
Hong, JY; Wang, YY; Bondoc, FY; et al. (1999a) Metabolism of methyl tert-butyl ether and other gasoline ethers by
human liver microsomes and heterologously expressed human cytochromes P450: identification of CYP2A6 as a
major catalyst. Toxicol Appl Pharmacol 160(l):43-48.
Hong, JY; Wang, YY; Bondoc, FY; et al. (1999b) Metabolism of methyl tert-butyl ether and other gasoline ethers in
mouse liver microsomes lacking cytochrome P450 2E1. Toxicol Lett 105(l):83-88.
Hong, JY; Wang, YY; Molir, SN; et al. (2001) Human cytochrome P450 isozymes in metabolism and health effects
of gasoline ethers. Res Rep Health Eff Inst (102):7-27.
Hsieh, KP; Lin, YY; Cheng, CL; et al. (2001) Novel mutations of CYP3A4 in Chinese. Drug Metab Disp 29:268-
273.
Huntingdon Life Sciences. (2002) Gasoline ETBE vapor condensate: a 13-week whole-body inhalation toxicity
study in rats with neurotoxicity assessments and 4-week in vitro genotoxicity and immunotoxicity assessments.
Huntingdon Life Sciences under contract to American Petroleum Institute, East Millstone, NJ; Study No. 00-6129;
Sponsor No. 211-ETBE-S. Unpublished report.
IIT Research Institute (Illinois Institute of Technology Research Institute). (1989a) Acute inhalation toxicity study
of ethyl-t-butyl ether (ETBE) in rats. IIT Research Institute, Life Sciences Research under contract to Amoco
Corporation, Chicago, IL; Study No. 1496. Unpublished report.
IIT Research Institute (Illinois Institute of Technology Research Institute). (1989b) Acute dermal toxicity study of
ethyl-tert-butyl ether (ETBE) in rabbits. IIT Research Institute, Life Sciences Research under contract to Amoco
Corporation, Chicago, IL; Study No. 1495. Unpublished report.
IIT Research Institute (Illinois Institute of Technology Research Institute). (1991) Four-week inhalation toxicity
study of ethyl tert-butyl ether (ETBE) in rats. IIT Research Institute, Life Sciences Research under contract to
Amoco Corporation, Chicago, IL; Study No. 1544. Unpublished report.
Jacob, RM; Johnstone, EC; Neville, MJ; et al. (2004) Identification of CYP2B6 sequence variants by use of
multiplex PCR with allele-specific genotyping. Clin Chem 50:1372-1377.
Jinno, H; Tanaka-kagawa, T; Olino, A; et al. (2003) Functional characterization of cytochrome P450 2B6 allelic
variants. Drug Metab Disp 31:398-403.
Johanson, G; Nihlen A; Lof, A. (1995) Toxicokinetics and acute effects of MTBE and ETBE in male volunteers.
Toxicol Lett 82-83:713-718.
Kaneko, T; Wang, PY; Sato, A. (2000) Partition coefficients for gasoline additives and their metabolites. J Occup
Health 42(2):86-87.
07/14/2009
116
DRAFT - DO NOT CITE OR QUOTE
-------
Krasowski, MD; Harrison, NL. (1999) General anaesthetic actions on ligand-gated ion channels. Cell Mol Life Sci
55(10): 1278-1303.
Le Gal, A; Dreano, Y; Gervasi, PG; et al. (2001) Human cytochrome P450 2A6 is the major enzyme involved in the
metabolism of three alkoxyethers used as oxyfuels. Toxicol Lett 124(l-3):47-58.
Lindamood, C, III; Farnell, DR; Giles, HD; et al. (1992) Subchronic toxicity studies of t-butyl alcohol in rats and
mice. Fundam Appl Toxicol 19(1): 91-100.
Mahler, J. (1997) NTP technical report on toxicity studies of t-butyl alcohol (CAS No. 75-65-0) administered by
inhalation to F344/N rats and B6C3F1 mice. Public Health Service, U.S. Department of Health and Human
Services; NTP Technical Report Series TOX-53:l-56, A1-D9. Available from: National Institute of Enviromnental
Health Sciences, Research Triangle Park, NC.
Maltoni, C; Belpoggi, F; Soffritti, M; et al. (1999) Comprehensive long-term experimental project of carcinogenicity
bioassays on gasoline oxygenated additives: plan and first report of results from study of ethyl-tertiary-butyl-ether
(ETBE). Eur J Oncol 4:493-508.
Marsh, DF; Leake, CD. (1950) Comparative anesthetic activity of the aliphatic ethers. Anesthesiology 11:455-463.
Martin, JV; Bilgin, NM; Iba, MM. (2002) Influence of oxygenated fuel additives and their metabolites on the
binding of a convulsant ligand of the gamma-aminobutyric acid(A) (GABA(a) receptor in rat brain membrane
preparations. Toxicol Lett 129(3):219-226.
Martin, JV; Iyer, SV; Mcllroy, PJ; et al. (2004) Influence of oxygenated fuel additives and their metabolites on
gamma-aminobutyric acidA (GABAa) receptor function in rat brain synaptoneurosomes. Toxicol Lett 147(3):209-
217.
Mason, CE. (2002) Satellite procedure gasoline ETBE vapor condensate rat micronucleus test. Huntingdon Life
Sciences under contract to American Petroleum Institute, East Millstone, NJ; Study No. 00-6129; ERC Report No:
APT 007/022682. Unpublished report.
MB Research Laboratories, Inc. (Millennium Bioresearch Research Laboratories). (1988a) Single dose oral toxicity
in rats/LD50 in rats. MB Research Laboratories, Inc. under contract to ARCO Chemical Company, Spinnerstown,
PA; Laboratory Project ID MB 88-9137 A. Unpublished report.
MB Research Laboratories, Inc. (Millennium Bioresearch Research Laboratories). (1988b) Acute dermal toxicity in
rabbits/LD50 in rabbits. MB Research Laboratories, Inc. under contract to ARCO Chemical Company,
Spinnerstown PA; Laboratory Project ID MB 88-9107 B. Unpublished report.
MB Research Laboratories, Inc. (Millennium Bioresearch Research Laboratories). (1988c) Primary dermal irritation
in rabbits. MB Research Laboratories, Inc. under contract to ARCO Chemical Company, Spinnerstown, PA;
Laboratory Project ID MB 88-9107 C. Unpublished report.
MB Research Laboratories, Inc. (Millennium Bioresearch Research Laboratories). (1988d) Eye irritation in rabbits.
MB Research Laboratories, Inc. under contract to ARCO Chemical Company, Spinnerstown, PA; Laboratory
Project ID MB 88-9107 D. Unpublished report.
Medinsky, MA. (2006) ERRATUM: Effects of a thirteen-week inhalation exposure to ethyl tertiary butyl ether on
Fischer-344 rats and CD-I mice. Toxicol Sci 94(2):441.
Medinsky, MA; Wolf, DC; Cattley, RC; et al. (1999) Effects of a thirteen-week inhalation exposure to ethyl tertiary
butyl ether on Fischer-344 rats and CD-I mice. Toxicol Sci 51(1): 108—118.
Mehta, AK; Ticku, MK. (1999) An update on GABAa receptors. Brain Res Brain Res Rev 29(2-3): 196-217.
Mennear, JH. (1997) Carcinogenicity studies on MTBE: critical review and interpretation. Risk Anal 17(6):673-
681.
07/14/2009
117
DRAFT - DO NOT CITE OR QUOTE
-------
Montgomery, CR. (1994) n-Octanol/water partition co-efficient (Kow) for ethyl tertiary butyl ether (ETBE) at 25 °C
+/- 1 °C. BIODEVELOPMENT Laboratories, Inc. under contract to ARCO Chemical Company, Cambridge, MA;
Project Identification 57374-2. Unpublished report.
Nakajima, M; Yoshida, R; Fukami, T; et al. (2004) Novel human CYP2A6 alleles confound gene deletion analysis.
FEBS Lett 569(l-3):75-81.
National Cancer Institute Carcinogenesis. (1976) Guidelines for carcinogen bioassay in small rodents. Technical
Report Series NCI-CG-TR-1. Available online at http://ntp.niehs.nih.gov/ntp/htdocs/LT-rpts/tr001.pdf.
Nihlen, A; Johanson, G. (1999) Physiologically based toxicokinetic modeling of inhaled ethyl tertiary-butyl ether in
humans. Toxicol Sci 51(2): 184-194.
Nihlen A; Lof, A; Johanson, G. (1995) Liquid/air partition coefficients of methyl and ethyl t-butyl ethers, t-amyl
methyl ether, and t-butyl alcohol. J Expo Anal Environ Epidemiol 5(4):573-582.
Nihlen, A; Lof, A; Johanson, G. (1998a) Controlled ethyl tert-butyl ether (ETBE) exposure of male volunteers I.
Toxicokinetics. Toxicol Sci 46(1): 1—10.
Nihlen, A; Lof, A; Johanson, G. (1998b) Controlled ethyl tert-butyl ether (ETBE) exposure of male volunteers II.
Acute effects. Toxicol Sci 46(1):143-150.
Nihlen, A; Lof, A; Johanson, G. (1998c) Experimental exposure to methyl tertiary-butyl ether I. Toxicokinetics in
humans. Toxicol Appl Pharmacol 148:274-280.
NRC (National Research Council). (1983) Risk assessment in the federal government: managing the process.
Washington, DC: National Academy Press.
NRC (National Research Council). (1996) Toxicological and performance aspects of oxygenated motor vehicle
fuels. National Academy Press, Washington DC. Available online at http://www.nap.edu/books/0309055458/litml/.
NTP (National Toxicology Program). (1995) NTP toxicology and carcinogenicity studies of t-butyl alcohol (CAS
No. 75-65-0) in F344/N rats and B6C3F1 mice (drinking water studies). NTP Technical Report Series TR-436:1-
305.
NTP (National Toxicology Program). (2005) NTP Historical Controls for NTP-2000 Diet. Available online at
http://ntp.niehs.nih.gov/ntpweb/index.cfm?obiectid=92E705C7-FlF6-975E-72D23026B1645EB9.
Oscarson, M. (2001) Genetic polymorphisms in the cytochrome P450 2A6 (CYP2A6) gene: implications for
interindividual differences in nicotine metabolism. Drug Metab Disp 29:91-95.
Parkinson, A; Mudra, DR; Johnson, C; et al. (2004) The effect of gender, age, ethnicity, and liver cirrhosis on
cytochrome P450 enzyme activity in human liver microsomes and inducibility in cultured human hepatocytes.
Toxicol Appl Pharmacol 199:193-209.
Paschke, T; Riefler, M; Schuler-Metz, A; et al. (2001) Comparison of cytochrome P450 2A6 polymorphism
frequencies in Caucasians and African-Americans using a new one-step PCR-RFLP genotyping method.
Toxicology 168(3):259-268.
Pitarque, M; von Richter, O; Oke, B; et al. (2001) Identification of a single nucleotide polymorphism in the TATA
box of the CYP2A6 gene: impairment of its promoter activity. Biochem Biophys Res Commun 284(2):455-460.
Potts, RO; Guy, RH. (1992) Predicting skin permeability. PhannRes 9:663-669.
Prah, J; Ashley, D; Blount, B; et al. (2004) Dermal, oral, and inhalation pharmacokinetics of methyl tertiary butyl
ether (MTBE) in human volunteers. Toxicol Sci 77(2): 195-205.
07/14/2009
118
DRAFT - DO NOT CITE OR QUOTE
-------
Prescott-Mathews, JS; Wolf, DC; Wong, BA; et al. (1997) Methyl tert-butyl ether causes a2u-globulin nephropathy
and enhanced renal cell proliferation in male Fischer-344 rats. Toxicol Appl Pharmacol 143:302-314.
Rademaker, M. (2001) Do women have more adverse drug reactions? Am J Clin Dermatol 2:349-351.
Ramsey, JC; Andersen, ME. (1984) A physiologically based description of the inhalation pharmacokinetics of
styrene in rats and humans. Toxicol Appl Pharmacol 73:159-175.
Rao, HV; Ginsberg, GL. (1997) A physiologically based pharmacokinetic model assessment of methyl t-butyl ether
in groundwater for bathing and showering determination. Risk Anal. 17:583-598.
Rosenkranz, HS; Klopman, G. (1991) Prediction of the lack of genotoxicity and carcinogenicity in rodents after two
gasoline additives: methyl- and ethyl-t-butyl ethers. In Vitro Toxicol 4:49-54.
Rosin MP; Cheng, X; Poh C; et al. (2000) Use of allelic loss to predict malignant risk for low-grade oral epithelial
dysplasia. Clin Cancer Res 6(2):357-362.
Rothenstein, C. (2004) Report. FY 2004 Semi-annual mid-year activity report [memo to EPA UST/LUST Regional
Division Directors, Regions 1-10], U.S. Enviromnental Protection Agency, Office of Solid Waste and Emergency
Response, Office of Underground Storage Tanks, Washington, DC. Available online at
http://www.epa.gov/oust/cat/ca_04_12.pdf.
Roudabush, RL. (1966) Toxicity and health hazard summary. Laboratory of Industrial Medicine of Eastman Kodak
Company, Rochester, NY; ACC No. 907320; Lab No. 59-516. Unpublished report.
Russell, LD; Ettlin, RA; Sinlia Hikim, AP; et al. (1990) Histological and Histoapthological Evaluation of the Testis.
Cache River Press, Clearwater, FL.
Sathyan, KM; Sailasree, R; Jayasurya, R; et al. (2006) Carcinoma of tongue and the buccal mucosa represent
different biological subentities of the oral carcinoma. J Cancer Res Clin Oncol 132(9) :601-609.
Schepman, KP; van der Meij, EH; Smeele, LE; et al. (1998) Malignant transformation of oral leukoplakia: a follow-
up study of a hospital-based population of 166 patients with oral leukoplakia from The Netherlands. Oral Oncol
34(4):270-275.
Schuetzle, D; Siegl, WO; Jensen, TE; et al. (1994) The relationship between gasoline composition and vehicle
hydrocarbon emissions: a review of current studies and future research needs. Environ Health Perspect
102(Suppl)4:3-12.
Sliih T; Rong, Y; Harmon, T; et al. (2004) Evaluation of the impact of fuel hydrocarbons and oxygenates on
groundwater resources. Environ Sci Technol 38(l):42-48.
Sun, JD; Beskitt, JL. (1995a) Ethyl tertiary butyl ether (ETBE): pharmacokinetics after single and repeated
inhalation exposures in rats. Bush Run Research Center, Union Carbide Corporation under contract to ARCO
Chemical Company, Export, PA; Laboratory Project ID 94N1454. Unpublished report.
Sun, JD; Beskitt, JL. (1995b) Ethyl tertiary butyl ether (ETBE): pharmacokinetics after single and repeated
inhalation exposures in mice. Bush Run Research Center, Union Carbide Corporation under contract to ARCO
Chemical Company, Export, PA; Laboratory Project ID 94N1455. Unpublished report.
Takahashi, K; Lindamood, C, III; Maronpot, RR. (1993) Retrospective study of possible alpha-2u-globulin
nephropathy and associated cell proliferation in male Fischer 344 rats dosed with t-butyl alcohol. Environ Health
Perspect 101(Suppl 5):281-285.
TRC Enviromnental Corporation (Travelers Research Corporation Enviromnental Corporation). (1993) Odor
threshold studies performed with gasoline and gasoline combined with MTBE, ETBE and TAME. TRC
Enviromnental Corporation under contract to American Petroleum Institute, Windsor, CT; API Publication Number
4592.
07/14/2009
119
DRAFT - DO NOT CITE OR QUOTE
-------
Turini, A; Amato, G; Longo, V; et al. (1998) Oxidation of methyl- and ethyl- tertiary-butyl ethers in rat liver
microsomes: role of the cytochrome P450 isofonns. Arch Toxicol 72(4):207-214.
UBTL Inc. (Utah Biomedical Testing Laboratory). (1994) Twenty-eight (28) day dermal toxicity study in rats
administered test article F-266. UBTL, Inc. under contract to ARCO Chemical Company, Salt Lake City, UT;
UBTL Study No. 66894; Protocol No. ATX-92-0114, 1-174. Unpublished report.
U.S. DOE (Department of Energy). (2007a) EIA-819M Monthly oxygenate report. Available online at
http://www.eia.doe.gov/oil_gas/petroleum/data_publications/montlily_oxygenate_telephone_report/motr.html.
U.S. DOE (Department of Energy). (2007b) Weekly Imports and Exports. Available online at
http://tonto.eia.doe.gov/dnav/pet/pet_move_wkly_dc_nus-zOO_mbblpd_w.htm.
U.S. EPA (Enviromnental Protection Agency). (1986a) Guidelines for the health risk assessment of chemical
mixtures. Federal Register 51(185):34014-34025. Available online at http://www.epa.gov/iris/backgr-d.htm..
U.S. EPA (Enviromnental Protection Agency). (1986b) Guidelines for mutagenicity risk assessment. Federal
Register 51(185):34006-34012. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (1988) Recommendations for and documentation of biological values
for use in risk assessment. Prepared by the Enviromnental Criteria and Assessment Office, Office of Health and
Enviromnental Assessment, Cincinnati, OH for the Office of Solid Waste and Emergency Response, Washington,
DC; EPA 600/6-87/008. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (1991a) Guidelines for developmental toxicity risk assessment.
Federal Register 56(234):63798-63826. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (1991b) Alpha2u-globulin: association with chemically induced renal
toxicity and neoplasia in the male rat. Risk Assessment Forum, Washington, DC; EPA/625/3-91/019F. Available
from: National Technical Information Service (NTIS), Springfield, VA; NTIS PB92143668.
U.S. EPA (Enviromnental Protection Agency). (1992) Alternative fuels research strategy [external review draft].
Office of Research and Development, Washington, DC. EPA 600/AP-92-002. Available online at
http://www.epa.gov/ncea/pdfs/mtbe/altfuel.pdf.
U.S. EPA (Enviromnental Protection Agency). (1994a) Interim policy for particle size and limit concentration issues
in inhalation toxicity studies. Federal Register 59(206):53799. Available online at http://www.epa.gov/iris/backgr-
d.htm.
U.S. EPA (Enviromnental Protection Agency). (1994b) Methods for derivation of inhalation reference
concentrations and application of inhalation dosimetry. Office of Research and Development, Washington, DC;
EPA/600/8-90/066F. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (1995) Use of the benchmark dose approach in health risk
assessment. Risk Assessment Forum, Washington DC; EPA/630/R-94/007. Available online at
http://cfpub.epa.gov/ncea/raf/recordisplay.cfm?deid=42601.
U.S. EPA (Enviromnental Protection Agency). (1996) Guidelines for reproductive toxicity risk assessment. Federal
Register 61(212):56274-56322. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (1998a) Guidelines for neurotoxicity risk assessment. Federal
Register 63(93):26926-26954. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (1998b) Oxygenates in water: critical information and research
needs. Office of Research and Development, Washington, DC; EPA/600/R-98/048. Available online at
http://www.epa. gov/ncea/ pdfs/oxy_li2o .pdf.
07/14/2009
120
DRAFT - DO NOT CITE OR QUOTE
-------
U.S. EPA (Environmental Protection Agency). (1998c) Health effects test guidelines: OPPTS 870.3700 prenatal
developmental toxicity study. Office of Prevention, Pesticides and Toxic Substances, Washington, DC. EPA/712/C-
98/207. Available online at
http://www.epa.gOv/opptsfrs/OPPTS_Hannonized/870_Health_Effects_Test_Guidelines/Series/870-3700.pdf.
U.S. EPA (Enviromnental Protection Agency). (1998d) Health effects test guidelines: OPPTS 870.3800
reproduction and fertility effects. Office of Prevention, Pesticides and Toxic Substances, Washington, DC.
EPA/712/C-98/208. Available online at
http://www.epa.gOv/opptsfrs/OPPTS_Hannonized/870_Health_Effects_Test_Guidelines/Series/870-3800.pdf.
U.S. EPA (Enviromnental Protection Agency). (2000a) Science policy council handbook: risk characterization.
Office of Science Policy, Office of Research and Development, Washington, DC; EPA 100-B-00-002. Available
online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2000b) Benchmark dose technical guidance document [external
review draft]. Risk Assessment Forum, Washington, DC; EPA/630/R-00/001. Available online at
http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2000c) Supplementary guidance for conducting for health risk
assessment of chemical mixtures. Risk Assessment Forum, Washington DC; EPA/630/R-00/002. Available online
at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2002) A review of the reference dose and reference concentration
processes. Risk Assessment Forum, Washington DC; EPA/630/P-02/0002F. Available online at
http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2005a) Guidelines for carcinogen risk assessment. Risk Assessment
Forum, Washington, DC; EPA/630/P-03/001B. Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2005b) Supplemental guidance for assessing susceptibility from
early-life exposure to carcinogens. Risk Assessment Forum, Washington, DC; EPA/630/R-03/003F. Available
online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2006a) Science policy council handbook: peer review. Third
edition. Office of Science Policy, Office of Research and Development, Washington, DC; EPA/100/B-06/002.
Available online at http://www.epa.gov/iris/backgr-d.htm.
U.S. EPA (Enviromnental Protection Agency). (2006b) A Framework for Assessing Health Risk of Enviromnental
Exposures to Children. National Center for Enviromnental Assessment, Washington, DC, EPA/600/R-05/093F.
Available online at http://cfpub.epa.gov/ncea/cfm/recordisplay.cfm?deid=158363.
Vasconcelos, GM; Struchiner, CJ; Suarez-Kurtz, G. (2005) CYP2A6 genetic polymorphisms and conelation with
smoking status in Brazilians. Phannacogenomics J 5(l):42-48.
Vergnes, JS. (1995) Ethyl tertiary butyl ether: in vitro chromosome abenations assay in Chinese hamster ovary cells.
Bush Run Research Center, Union Carbide Corporation under contract to ARCO Chemical Company, Export, PA;
Laboratory Project ID 94N1425. Unpublished report.
Vergnes, JS; Kubena, MF. (1995a) Ethyl tertiary butyl ether: bone manow micronucleus test in mice. Bush Run
Research Center, Union Carbide Corporation under contract to ARCO Chemical Company, Export, PA; Laboratory
Project ID 94N1426. Unpublished report.
Vergnes, JS; Kubena, MF. (1995b) Ethyl tertiary butyl ether: mutagenic potential in the CHO/HGPRT forward
mutation assay. Bush Run Research Center, Union Carbide Corporation under contract to ARCO Chemical
Company, Export, PA; Laboratory Project ID 94N1424. Unpublished report.
07/14/2009
121
DRAFT - DO NOT CITE OR QUOTE
-------
Vetrano, KM. (1993) Final report to ARCO Chemical Company on the odor and taste threshold studies performed
with methyl tertiary-butyl ether (MTBE) and ethyl tertiary-butyl ether (ETBE). TRC Enviromnental Corporation
under contract to ARCO Chemical Company, Windsor, CT; Project no. 13442-M31. Unpublished report.
von Richter, O; Pitarque, M; Rodriguez-Antona, C; et al. (2004) Polymorphic NF-Y dependent regulation of human
nicotine C-oxidase (CYP2A6). Pharmacogenetics 14(6):369-379.
White, KL. (2002) Immunological evaluation of gasoline ETBE vapor condensate in female Sprague-Dawley rats
using the plaque forming cell assay. ImmunoTox, Inc. under contract to Huntingdon Life Sciences, Richmond, VA;
Project No. ITI 901. Unpublished report.
White, RD; Daughtrey, WC; Wells, MS. (1995) Health effects of inhaled tertiary amyl methyl ether and ethyl
tertiary butyl ether. Toxicol Lett 82-83:719-724.
WHO (World Health Organization). (1995) Acetaldehyde. In: Enviromnental health criteria. Vol. 167. Geneva,
Switzerland: World Health Organization. Available online at
http://www.inchem.org/documents/ehc/ehc/ehcl67.htm.
Williams TM, Borghoff SJ. (2001) Characterization of tert-butyl alcohol binding to alpha-2u-globulin in F344 rats.
Toxicol Sci 62(2):228-235.
Yoshida, R; Nakajima, M; Nishimura, K; et al. (2003) Effects of polymorphism in promoter region of human
CYP2A6 gene (CYP2A6*9) on expression level of messenger ribonucleic acid and enzymatic activity in vivo and in
vitro. Clin Pharmacol Ther 74(l):69-76.
Zeiger, E; Anderson, B; Haworth, S; et al. (1992) Salmonella mutagenicity tests: V. Results from the testing of 311
chemicals. Environ Mol Mutagen 19(Suppl 21):2—141.
Zeiger, E; Ashby, J; Bakale, G; et al. (1996) Prediction of Salmonella mutagenicity. Mutagenesis 11:471-484
Zhang, YP; Macina, OT; Rosenkranz, HS; et al. (1997) Prediction of the metabolism and toxicological profiles of
gasoline oxygenates. Inlial Toxicol 9(3):237-254.
07/14/2009
122
DRAFT - DO NOT CITE OR QUOTE
-------
1
2 APPENDIX A. SUMMARY OF EXTERNAL PEER REVIEW AND PUBLIC
3 COMMENTS AND DISPOSITION
4
5
6
7
07/14/2009
A-l
DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX B. BMD CALCULATIONS FOR THE ORAL MINIMAL DATA VALUE
AND RfC
B.l. NONCANCER DOSE-RESPONSE ASSESSMENT FOR ORAL EXPOSURE TO
ETBE: BMD MODELING RESULTS
In this appendix, results of BMD modeling are presented for each noncancer endpoint
showing significantly elevated means relative to controls following oral exposure to ETBE. For
each endpoint, a summary of the dose-response data is presented, followed by a table
summarizing the results of the dose-response modeling. Finally, the standard output from the
EPA's BMDS (version 1.4.1), for the best-fitting dose-response model is presented.
For these modeling exercises, all continuous models available in BMDS were fit to the
corresponding data for each endpoint with the BMR set at one SD above the control mean. To
select the "best-fit" model, AIC values were evaluated for all models that did not exhibit a
significant lack of fit (i.e. ,p< 0. 1), according to the %2 goodness-of-fit test. Of these models, the
model with the lowest AIC value was typically selected as the best-fit model unless examination
of the chi-square scaled residuals indicated another model with a similar AIC exhibited a better
fit in the region of the curve where the BMD was estimated. Selection of the BMR and the
procedure for selecting the best-fit model are consistent with the EPA's most current Benchmark
Dose Technical Guidance Document (U.S. EPA, 2000b).
07/14/2009
B-l
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Body
Endpoint: Dam Body Weight Change
Species/Gender: Sprague-Dawley Female Rats
(CIT, 2004a, unpublished report)
Table B-l. Mean dam body weight change (and SD) in Sprague-Dawley
female rats orally exposed to ETBE on GDs 5 through 20
Administered dose (mg/kg)
Mean dam body weight change, g (SD)
Control (n= 21)
135 (22)
250 (n= 19)
132(12)
500 (n = 20)
134(19)
1,000 (n = 22)
120a(15)
aStatistically significantly different from control at p< 0.05 as reported by CIT (2004a, unpublished report).
Source: CIT, (2004a, unpublished report).
07/14/2009
B-2
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-2. A summary of BMDS (version 1.4.1) modeling results based on
mean dam body weight change in Sprague-Dawley female rats orally
exposed to ETBE on GDs 5 through 20
Model" (non-constant variance)
%2/>-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (1°)
0.19
555.6
1,236
796
Power
0.35
554.3
1,015
879
Hill
0.15
556.3
1,045
Computation of the
lower bound failed
Tor all models, the variance model employed (i.e., variance modeled as a power function of the mean) failed to
adequately address the non-constant variance.
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benclunark concentration corresponding to a 1 SD change relative
to the control mean.
Source: CIT (2004a, unpublished report).
07/14/2009
B-3
DRAFT - DO NOT CITE OR QUOTE
-------
Power Model with 0.95 Confidence Level
Power
145
140 r
Mean Resp
135 r <
130
125
120
BMDL
800
BMI
0
200
400
600
1000
dose
14:57 08/07 2007
Source: CIT (2004a, unpublished report).
Figure B-l. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean dam body weight change in Sprague-
Dawley female rats orally exposed to ETBE on GDs 5 through 20.
Power Model. (Version: 2.14; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_FEMALE_DAMS_BODY_WT_CIIT04.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_FEMALE_DAMS_BODY_WT_CIIT04.pit
Tue Aug 07 14:57:57 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = control + slope * doseApower
Dependent variable = MEAN
Independent variable = Dose
The power is restricted to be greater than or equal to 1
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
07/14/2009
B-4
DRAFT - DO NOT CITE OR QUOTE
-------
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha = 5.72308
rho = 0
control = 120
slope = 3.5147 8
power = 0.222392
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 )
lalpha rho control slope
lalpha 1 -1 0.17 -0.31
rho -1 1 -0.17 0.31
control 0.17 -0.17 1 -0.59
slope -0.31 0.31 -0.59 1
Parameter Estimates
Variable
lalpha
rho
control
slope
power
Estimate
-12.3216
3.6953
133.717
-1.37167 e-053
18
Std. Err.
16.6779
3.42706
2.3108
38 616e-054
NA
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-45.0097
-3.02161
129.188
-2.13334 e-053
20.3664
10.4122
138.246
5.09 9 96e-054
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
250
500
1, 000
21
19
20
22
135
132
134
120
134
134
134
120
22
12
19
15
17. 9
17. 9
17. 9
14 .7
0.329
-0.418
0.0708
-2.04e-007
07/14/2009
B-5
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-273.595943
-269.540413
-272.114636
-273.151218
-278.636681
# Param's
5
8
6
4
2
AIC
557.191887
555.080827
556.229273
554 . 302436
561.273362
Explanation of Tests
Test 1:
Test 2
Test 3
Test 4
(Note:
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
18 . 1925
8.11106
5.14845
2.07316
0.005769
0.04377
0.07621
0.3547
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1.
model appears to be appropriate
A non-homogeneous variance
The p-value for Test 3 is less than .1.
different variance model
You may want to consider a
07/14/2009
B-6
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1014 . 9
BMDL = 87 9.018
07/14/2009
B-7
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Body
Endpoint: Net Dam Body Weight Change
Species/Gender: Sprague-Dawley Female Rats
(CIT, 2004a, unpublished report)
Table B-3. Mean net dam body weight change (and SD) in Sprague-Dawley
female rats orally exposed to ETBE on GDs days 5 through 20
Administered dose (mg/kg)
Mean net dam body weight change, g (SD)
Control (n= 21)
61.8(13)
250 (n= 19)
59.4(8.1)
500 (n = 20)
60.0(11.3)
1,000 (n = 22)
51.5a (10.3)
aStatistically significantly different from control at p< 0.05 as reported by CIT (2004a, unpublished
report).
Source: CIT (2004a, unpublished report).
07/14/2009
B-8
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-4. A summary of BMDS (version 1.4.1) modeling results
based on mean net dam body weight change in Sprague-Dawley
female rats orally exposed to ETBE on GDs 5 through 20
Model (constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (1°)
0.49
476.6
1,063
696
Power
0.49
477.7
1,044
748
Hill
NA
479.7
1,044
747
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate
that the model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD
change relative to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0;
therefore, this test is not valid for evaluating lack of fit.
Source: CIT (2004a, unpublished report).
07/14/2009
B-9
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
Linear
65
~T
50
1
k
45
BMDLl
me:
0
200
400
600
800
1000
dose
16:20 08/07 2007
Source: CIT (2004a, unpublished report).
Figure B-2. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean dam body weight change in Sprague-
Dawley female rats orally exposed to ETBE on GDs 5 through 20.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_FEMALE_DAMS_NET_BODY_WT_CIIT04.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_FEMALE_DAMS_NET_BODY_WT_CIIT04.pit
Tue Aug 07 16:20:42 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
Independent variable = Dose
rho is set to 0
07/14/2009
B-10 DRAFT - DO NOT CITE OR QUOTE
-------
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 118.141
rho = 0 Specified
beta 0 = 62.54
beta~l = -0.00997714
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 6e-009 5.9e-009
beta 0 6e-009 1 -0.76
beta 1 5.9e-009 -0.76 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 114.378 17.8629 79.3677 149.389
beta_0 62.55 1.83181 58.9597 66.1403
beta 1 -0.0100599 0.00312435 -0.0161835 -0.00393626
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 21 61.8 62.6 13 10.7 -0.321
250 19 59.4 60 8.1 10.7 -0.259
500 20 60 57.5 11.3 10.7 1.04
1,000 22 51.5 52.5 10.3 10.7 -0.434
07/14/2009
B-ll
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-234.596525
-232.361985
-234.596525
-235.319981
-240.201188
# Param's
5
8
5
3
2
AIC
479. 193049
480.723971
479.193049
476.639962
484 .402375
Explanation of Tests
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (A1 vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
15.6784
4.46908
4.46908
1.44691
0.01559
0.2151
0.2151
0.4851
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1.
be appropriate here
The modeled variance appears to
07/14/2009
B-12
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 10 63.11
BMDL = 696.25
07/14/2009
B-13
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Body
Endpoint: Fn Body Weight Change
Species/Gender: Sprague-Dawley Male Rats
(CIT, 2004b, unpublished report)
Table B-5. Mean F0 body weight change (and SD) in Sprague-Dawley male
rats orally exposed to ETBE in a two-generation reproductive toxicity study
Administered dose (mg/kg)
Mean F0 body weight change, g (SD)
Control (n = 25)
58.0(15)
250 (n= 25)
56.0 (9)
500 (n= 25)
41.0a (19)
1,000 (n= 25)
45.0a (5)
aStatistically significantly different from control at p< 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished report).
07/14/2009
B-14
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-6. A summary of BMDS (version 1.4.1) modeling results based on
mean FO body weight change in Sprague-Dawley male rats orally exposed to
ETBE in a two-generation reproductive toxicity study
Model" (non-constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1sd (mg/kg)
Polynomial (1°)
0.01
486.4
405
276
Power
0.73
479.0
492
385
Hillb
—
—
—
—
Tor all models, the highest dose group was dropped prior to fitting the model, and the variance model
employed (i.e., variance modeled as a power function of the mean) failed to adequately address the non-
constant variance.
bThe Hill model could not be fit to these data as the number of model parameters to be estimated exceeded
the number of observations.
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benclunark concentration corresponding to a 1 SD change
relative to the control mean.
Source: CIT (2004b, unpublished report).
07/14/2009
B-15
DRAFT - DO NOT CITE OR QUOTE
-------
Power Model with 0.95 Confidence Level
65
60
qj 55
CO
£=
t 50
QL
I 45
Zl
40
35
0 100 200 300 400 500
dose
10:17 08/09 2007
Source: CIT (2004b, unpublished report).
Figure B-3. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
power) based on mean FO body weight change in Sprague-Dawley male rats
orally exposed to ETBE in a two-generation reproductive toxicity study.
Power Model. (Version: 2.14; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_MALE_FO_BODY_WT_DAYS 8 5_113_CIIT04. (d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_MALE_FO_BODY_WT_DAYS 8 5_113_CIIT04.pit
Thu Aug 09 10:17:20 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = control + slope * doseApower
Dependent variable = MEAN
Independent variable = Dose
07/14/2009 B-16 DRAFT - DO NOT CITE OR QUOTE
Power
BMDU
BMP
-------
The power is restricted to be greater than or equal to 1
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 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 Parameter Values
lalpha = 5.40418
rho = 0
control = 58
slope = -0.034
power = 1
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -power
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
lalpha rho control slope
lalpha 1 -1 -0.17 0.58
rho -1 1 0.18 -0.58
control -0.17 0.18 1 -0.42
slope 0.58 -0.58 -0.42 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha 15.447 5.11829 5.41539 25.4787
rho -2.58484 1.29189 -5.11691 -0.0527773
control 57 1.71977 53.6293 60.3707
slope -4.19432e-048 1.07511e-048 -6.3015e-048 -2.08714e-048
power 18 NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 25 58 57 15 12.2 0.411
07/14/2009
B-17
DRAFT - DO NOT CITE OR QUOTE
-------
250
500
25
25
56
41
57
41
9
19
12.2
18. 6
-0.411
3.09e-006
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) # Param's AIC
A1 -238.625841 4 485.251682
A2 -232.212019 6 476.424038
A3 -235.450799 5 480.901597
fitted -235.510734 4 479.021468
R -247.578841 2 499.157682
Explanation of Tests
Test 1:
Test
Test
Test
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 30.7336 4 <.0001
Test 2 12.8276 2 0.001639
Test 3 6.47756 1 0.01092
Test 4 0.11987 1 0.7292
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
07/14/2009
B-18
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is less than .1. You may want to consider a
different variance model
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 4 92.436
BMDL = 3 8 4.702
07/14/2009
B-19
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Fi^ Liver Weight (absolute)
Species/Gender: Sprague-Dawlev Male Rats
(CIT, 2004b, unpublished report)
Table B-7. Mean absolute F1 liver weight (and SD) in Sprague-Dawley male
rats orally exposed to ETBE in a two-generation reproductive toxicity study
Administered dose (mg/kg)
Mean F1 liver weight, g (SD)
Control (n = 24)
18.9 (2.45)
250 (n= 25)
18.9(2.32)
500 (n = 24)
21.6a (4.16)
1,000 (n= 25)
23.9a (4.10)
aStatistically significantly different from control at p< 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished report).
07/14/2009
B-20
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-8. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute F1 liver weight in Sprague-Dawley male rats orally exposed
to ETBE in a two-generation reproductive toxicity study
Model (non-constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (1°)
0.04
334.3
462
337
Power
0.01
335.9
514
345
Hill
0.96
329.7
482
294
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change
relative to the control mean.
Source: CIT (2004b, unpublished report).
07/14/2009
B-21
DRAFT - DO NOT CITE OR QUOTE
-------
Model with 0.95 Confidence Level
CD
O
Q.
(fl
-------
Power parameter restricted to be greater than 1
The variance is to be modeled as Var(i) = exp(lalpha + rho * In(mean(i)))
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha =
rho =
intercept =
v =
n =
k =
2 . 43091
0
18 .89
5. 06
3. 18615
515.152
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -n
have been estimated at a boundary point, or have been specified by
the user,
lalpha
rho
intercept
and do not appear in the correlation matrix )
lalpha
1
-1
-0.3
0.57
0.26
rho
-1
1
0.29
-0.58
-0.25
intercept
-0.3
0.29
1
-0.41
0. 028
v
0.57
-0.58
-0.41
1
0.52
k
0.26
-0.25
0. 028
0.52
1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha -14.7294 5.67709 -25.8562 -3.60247
rho 5.60352 1.87657 1.92552 9.28152
intercept 18.8835 0.338055 18.2209 19.5461
v 4.67469 0.939755 2.8328 6.51658
n 18 NA
k 480.83 21.3555 438.974 522.685
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
07/14/2009
B-23
DRAFT - DO NOT CITE OR QUOTE
-------
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
24
18.9
18.9
2.45
2.38
0.0134
250
25
18.9
18.9
2.32
2.38
0.119
500
24
21.6
22
4.16
3.66
-0.577
1, 000
25
23.9
23.6
4.1
4.43
0.443
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
Log(likelihood)
# Param's
AIC
A1
-166.072486
5
342.144972
A2
-158.990348
8
333. 980695
A3
-159.830921
6
331.661841
fitted
-159.832061
5
329. 664122
R
-182.852138
2
369.704276
Explanation of Tests
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (A1 vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
47 . 7236
14.1643
1.68115
0.0022814
<.0001
0.00269
0.4315
0.9619
07/14/2009
B-24
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect
1
Risk Type
Estimated standard deviations from the control mean
Confidence level
0. 95
BMD
481.838
BMDL
294.036
07/14/2009
B-25
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: F1 Liver Weight (relative)
Species/Gender: Sprague-Dawlev Male Rats
(CIT, 2004b, unpublished report)
Table B-9. Mean relative F1 liver weight (and SD) in Sprague-Dawley male
rats orally exposed to ETBE in a two-generation reproductive toxicity study
Administered dose (mg/kg)
Mean F1 liver weight as percent body weight (SD)
Control (n = 24)
3.20 (0.225)
250 (n= 25)
3.21 (0.245)
500 (n = 24)
3.54a (0.317)
1,000 (n= 25)
4.01a (0.389)
aStatistically significantly different from control at p< 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished report).
07/14/2009
B-26
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-10. A summary of BMDS (version 1.4.1) modeling results based on
mean relative F1 liver weight in Sprague-Dawley male rats orally exposed to
ETBE in a two-generation reproductive toxicity study
Model (non-constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (2°)
0.03
-135.4
418
284
Power
0.05
-136.2
414
295
Hill
NA
-138.0
444
341
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change
relative to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this
test is not valid for evaluating lack of fit.
Source: CIT (2004b, unpublished report).
All continuous dose-response models available in BMDS (version 1.4.1) exhibited
significant lack-of-fit (i.e., chi-square /7-value for goodness-of-fit < 0.1); therefore, no POD
could be derived from these data based on the modeling results.
07/14/2009
B-27
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: FO Kidney Weight (absolute)
Species/Gender: Sprague-Dawley Male Rats
(CIT, 2004b, unpublished report)
Table B-ll. Mean absolute FO kidney weight (and SD) in Sprague-Dawley
male rats orally exposed to ETBE in a two-generation reproductive toxicity
study
Administered dose (mg/kg)
Mean F0 kidney weight, g (SD)
Control (n = 25)
3.58(0.413)
250 (n= 25)
3.96a (0.446)
500 (n= 25)
4.12a (0.624)
1,000 (n= 25)
4.34: (0.434)
aStatistically significantly different from control at p < 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished study).
07/14/2009
B-28
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-12. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute FO kidney weight in Sprague-Dawley male rats orally exposed
to ETBE in a two-generation reproductive toxicity study
Model (constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (2°)
0.58
-39.8
404
250
Power
0.20
-38.9
679
513
Hill
0.81
-40.1
381
167
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change
relative to the control mean.
Source: CIT (2004b, unpublished report).
07/14/2009
B-29
DRAFT - DO NOT CITE OR QUOTE
-------
Hill Model with 0.95 Confidence Level
0 200 400 600 800 1000
dose
12:10 08/09 2007
Source: CIT (2004b, unpublished report).
Figure B-5. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
Hill) based on mean absolute FO kidney weight in Sprague-Dawley male rats
orally exposed to ETBE in a two-generation reproductive toxicity study.
Hill Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_MALE_FO_KIDNEY_WT_CIIT04.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_MALE_FO_KIDNEY_WT_CIIT04.pit
Thu Aug 09 12:10:32 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = intercept + v*doseAn/(kAn + doseAn)
Dependent variable = MEAN
Independent variable = Dose
rho is set to 0
07/14/2009 B-30 DRAFT - DO NOT CITE OR QUOTE
-------
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.236804
rho = 0 Specified
intercept = 3.58
v = 0.76
n = 0. 647728
k = 250
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho -n
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha intercept v k
alpha 1 -7e-008 -1.9e-007 -2.2e-007
intercept -7e-008 1 0.036 0.42
v -1.9e-007 0.036 1 0.89
k -2.2e-007 0.42 0.89 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.227462 0.032168 0.164414 0.29051
intercept 3.58236 0.0952609 3.39565 3.76906
v 1.16337 0.440153 0.300691 2.02606
n 1 NA
k 548.322 492.789 -417.527 1514.17
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
07/14/2009
B-31
DRAFT - DO NOT CITE OR QUOTE
-------
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
25
3.58
3.58
0.413
0. 477
-0.0247
250
25
3. 96
3.95
0.446
0. 477
0.14
500
25
4.12
4.14
0.624
0. 477
-0.181
1, 000
25
4.34
4.33
0. 434
0.477
0.0657
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
24 . 067171
26.992591
24 . 067171
24 . 038627
9.481790
# Param's
5
8
5
4
2
AIC
-38.134342
-37.985183
-38.134342
-40.077253
-14.963581
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
35. 0216
5.85084
5. 85084
0.057089
<.0001
0.1191
0.1191
0.8112
07/14/2009
B-32
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect
1
Risk Type
Estimated standard deviations from the control mean
Confidence level
0. 95
BMD
380.964
BMDL
167.157
07/14/2009
B-33
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: FO Kidney Weight (relative)
Species/Gender: Sprague-Dawley Male Rats
(CIT, 2004b, unpublished report)
Table B-13. Mean relative FO kidney weight (and SD) in Sprague-Dawley
male rats orally exposed to ETBE in a two-generation reproductive toxicity
study
Administered dose (mg/kg)
Mean F0 kidney weight, as percent body weight (SD)
Control (n = 25)
0.596 (0.053)
250 (n= 25)
0.6623 (0.052)
500 (n= 25)
0.7063 (0.076)
1,000 (n= 25)
0.7633 (0.063)
aStatistically significantly different from control at p< 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished report).
07/14/2009
B-34
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-14. A summary of BMDS (version 1.4.1) modeling results based on
mean relative FO kidney weight in Sprague-Dawley male rats orally exposed
to ETBE in a two-generation reproductive toxicity study
Model (constant variance)
%2 /'-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (2°)
0.75
-452.9
243
176
Power
0.13
-450.9
382
317
Hill
0.94
-453.0
227
143
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change
relative to the control mean.
Source: CIT (2004, unpublished report).
07/14/2009
B-35
DRAFT - DO NOT CITE OR QUOTE
-------
Hill Model with 0.95 Confidence Level
0.8
0.75
O
Q.
V>
CD
m
0.7
-------
rho is set to 0
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-OOf
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha =
rho =
intercept =
v =
n =
k =
0.0038145
0
0.59628
0.16713
0.221145
649.462
Specified
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho -n
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha
intercept
alpha
1
5. le-009
-1. 6e-008
-1.4e-008
intercept
5. le-009
1
0.27
0.49
v
-1.6e-008
0.27
1
0. 96
-1.4e-008
0.49
0. 96
1
Parameter Estimates
Variable
alpha
intercept
k
Estimate
0.00366216
0.596439
0.345284
1
1070.39
Std. Err.
0.000517907
0.0119542
0.122057
NA
697.553
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
0.00264708
0.573009
0.106057
-296.786
0.00467724
0.619869
0.584512
2437.57
07/14/2009
B-37
DRAFT - DO NOT CITE OR QUOTE
-------
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
25
0.596
0.596
0.053
0.0605
-0.0131
250
25
0.662
0. 662
0.052
0.0605
0. 0533
500
25
0.706
0.706
0.076
0.0605
-0.0566
1, 000
25
0.763
0.763
0. 063
0.0605
0.0164
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) # Param's AIC
A1 230.488384 5 -450.976768
A2 232.931535 8 -449.863070
A3 230.488384 5 -450.976768
fitted 230.485140 4 -452.970280
R 195.370878 2 -386.741756
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
75.1213
4.8863
4.8863
0. 0064882
<.0001
0.1803
0.1803
0.9358
07/14/2009
B-38
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect
1
Risk Type
Estimated standard deviations from the control mean
Confidence level
0. 95
BMD
227 .467
BMDL
143.401
07/14/2009
B-39
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Fi^ Kidney Weight (absolute)
Species/Gender: Sprague-Dawley Male Rats
(CIT, 2004b, unpublished report)
Table B-15. Mean absolute F1 kidney weight (and SD) in Sprague-Dawley
male rats orally exposed to ETBE in a two-generation reproductive toxicity
study
Administered dose (mg/kg)
Mean F1 kidney weight, g (SD)
Control (n = 24)
3.38 (0.341)
250 (n= 25)
3.73 (0.449)
500 (n = 24)
4.13a (0.640)
1,000 (n= 25)
5.34a (5.390)
aStatistically significantly different from control at p < 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished report).
07/14/2009
B-40
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-16. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute F1 kidney weight in Sprague-Dawley male rats orally exposed
to ETBE in a two-generation reproductive toxicity study
Model (non-constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (2°)
0.09
80.0
313
218
Power
0.07
80.5
337
240
Hill
NA
82.5
337
Computation of the
lower bound failed
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change
relative to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this
test is not valid for evaluating lack of fit.
Source: CIT (2004b, unpublished report).
All continuous dose-response models available in BMDS (version 1.4.1) exhibited
significant lack-of-fit (i.e., chi-square /7-value for goodness-of-fit < 0.1); therefore, no POD
could be derived from these data based on the modeling results.
07/14/2009
B-41
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: F1 Kidney Weight (absolute)
Species/Gender: Sprague-Dawley Female Rats
(CIT, 2004b, unpublished report)
Table B-17. Mean absolute F1 kidney weight (and SD) in Sprague-Dawley
female rats orally exposed to ETBE in a two-generation reproductive
toxicity study
Administered dose (mg/kg)
Mean F1 kidney weight, g (SD)
Control (n = 25)
2.24 (0.178)
250 (n = 24)
2.34 (0.242)
500 (n= 25)
2.30 (0.226)
1,000 (n= 23)
2.49a (0.284)
aStatistically significantly different from control at p < 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT (2004b, unpublished report).
07/14/2009
B-42
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-18. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute F1 kidney weight in Sprague-Dawley female rats orally
exposed to ETBE in a two-generation reproductive toxicity study
Model (constant variance)
%2/7-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (1°)
0.30
-180.2
1,016
687
Power
0.14
-178.4
1,033
699
Hill
NA
-176.4
1,033
662
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: CIT, (2004b, unpublished report).
07/14/2009
B-43
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
400 600
dose
15:13 09/17 2007
Source: CIT, (2004b, unpublished report).
~l—1—1—1—
Linear
,BMDL|
Figure B-7. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean absolute F1 kidney weight in
Sprague-Dawley female rats orally exposed to ETBE in a two-generation
reproductive toxicity study.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_FEMALE_F1_KIDNEY_WT_CIIT04.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\ORAL\SD_RATS_FEMALE_F1_KIDNEY_WT_CIIT04.pit
Mon Sep 17 15:13:50 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
07/14/2009
B-44
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = MEAN
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.0549209
rho = 0 Specified
beta 0 = 2.242
beta~l = 0.000229714
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 -9e-010 -2.1e-010
beta_0 -9e-010 1 -0.76
beta 1 -2.le-010 -0.76 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.0539761 0.00775052 0.0387854 0.0691668
beta_0 2.24166 0.0362871 2.17053 2.31278
beta 1 0.000228659 6.44489e-005 0.000102342 0.000354977
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 25 2.24 2.24 0.178 0.232 -0.0356
250 24 2.34 2.3 0.242 0.232 0.868
500 25 2.3 2.36 0.226 0.232 -1.2
1,000 23 2.49 2.47 0.284 0.232 0.406
07/14/2009
B-45
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
94.282678
96.875846
94.282678
93.081873
87.164175
# Param's
5
8
5
3
2
AIC
-178.565356
-177.751692
-178.565356
-180.163745
-170.328351
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
19.4233
5. 18634
5. 18634
2 .40161
0. 003505
0.1587
0.1587
0.301
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-46
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1016.04
BMDL = 687.185
07/14/2009
B-47
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: F1 Kidney Weight (relative)
Species/Gender: Sprague-Dawley Male Rats
(CIT, 2004b, unpublished report)
Table B-19. Mean relative F1 kidney weight (and SD) in Sprague-Dawley
male rats orally exposed to ETBE in a two-generation reproductive toxicity
study
Administered dose (mg/kg)
Mean F1 kidney weight, as percent body weight (SD)
Control (n = 24)
0.574 (0.043)
250 (n= 25)
0.6343 (0.046)
500 (n = 24)
0.6843 (0.068)
1,000 (n= 25)
0.9083 (0.958)
aStatistically significantly different from control at p < 0.05 as reported by CIT (2004b, unpublished
report).
Source: CIT, (2004b, unpublished report).
07/14/2009
B-48
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-20. A summary of BMDS (version 1.4.1) modeling results based on
mean relative F1 kidney weight in Sprague-Dawley male rats orally exposed
to ETBE in a two-generation reproductive toxicity study
Model (non-constant variance)
%2 />-value
AIC
BMD1sd (mg/kg)
BMDL1SD (mg/kg)
Polynomial (2°)
0.004
-318.0
271
194
Power
0.003
-317.8
315
226
Hill
NA
-28.7
13,880
Computation of the
lower bound failed
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: CIT, (2004b, unpublished report).
All continuous dose-response models available in BMDS (version 1.4.1) exhibited
significant lack-of-fit (i.e., chi-square /7-value for goodness-of-fit < 0.1); therefore, no POD
could be derived from these data based on the modeling results.
07/14/2009
B-49
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: F1 Kidney Weight (relative)
Species/Gender: Sprague-Dawley Female Rats
(CIT, 2004b, unpublished report)
Table B-21. Mean relative F1 kidney weight (and SD) in Sprague-Dawley
female rats orally exposed to ETBE in a two-generation reproductive
toxicity study
Administered dose (mg/kg)
Mean F1 kidney weight, as percent body weight (SD)
Control (n = 25)
0.692 (0.061)
250 (n = 24)
0.733 (0.075)
500 (n= 25)
0.731 (0.048)
1,000 (n= 23)
0.7623 (0.097)
aStatistically significantly different from control at p < 0.05 as reported by CIT (2004b, unpublished report).
Source: CIT, (2004b, unpublished report).
07/14/2009
B-50
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-22. A summary of BMDS (version 1.4.1) modeling results based on
mean relative F1 kidney weight in Sprague-Dawley female rats orally
exposed to ETBE in a two-generation reproductive toxicity study
Model" (non-constant variance)
%2/'-value
AIC
BMD1sd (mg/kg)
BMDL1sd (mg/kg)
Polynomial (1°)
0.10
-412.3
898
562
Power
0.0001
-398.4
5
Computation of the lower bound
failed
Hill
0.03
-410.3
856
Computation of the lower bound
failed
Tor all models, the variance model employed (i.e., variance modeled as a power function of the mean) failed to
adequately address the non-constant variance.
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benclunark concentration corresponding to a 1 SD change relative
to the control mean.
Source: CIT, (2004b, unpublished report).
07/14/2009
B-51
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
Linear
0.8
-
T ¦
|
0.78
-
1 J
I
0.76
~T
—¦
|
0.74
-
-------
Dependent variable = MEAN
Independent variable = Dose
Signs of the polynomial coefficients are not restricted
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha = -5.26454
rho = 0
beta 0 = 0.702362
beta~l = 6.20811e-005
Asymptotic Correlation Matrix of Parameter Estimates
lalpha rho beta 0 beta 1
lalpha 1 0.99 -0.13 0.18
rho 0.99 1 -0.13 0.18
beta_0 -0.13 -0.13 1 -0.72
beta_l 0.18 0.18 -0.72 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha -2.71477 1.3171 -5.29625 -0.1333
rho 8.26635 4.13448 0.162911 16.3698
beta_0 0.700759 0.01,00071 0.681145 0.720373
beta 1 6.5 9036e-005 2.14377e-005 2.38865e-005 0.000107921
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 25 0.692 0.701 0.061 0.0592 -0.724
250 24 0.733 0.717 0.075 0.0652 1.21
500 25 0.731 0.734 0.048 0.0716 -0.224
1,000 23 0.762 0.767 0.097 0.0858 -0.259
Model Descriptions for likelihoods calculated
07/14/2009
B-53
DRAFT - DO NOT CITE OR QUOTE
-------
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) # Param's AIC
A1 208.872773 5 -407.745547
A2 215.204783 8 -414.409567
A3 212.407849 6 -412.815699
fitted 210.131654 4 -412.263308
R 203.219183 2 -402.438367
Explanation of Tests
Test 1:
Test
Test
Test
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 23.9712 6 0.0005287
Test 2 12.664 3 0.005422
Test 3 5.59387 2 0.061
Test 4 4.55239 2 0.1027
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is less than .1. You may want to consider a
different variance model
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
07/14/2009
B-54
DRAFT - DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect
1
Risk Type
Estimated standard deviations from the control mean
Confidence level
0. 95
BMD
898.036
BMDL
561.971
B.2. NONCANCER DOSE-RESPONSE ASSESSMENT FOR INHALATION EXPOSURE
TO ETBE: BMD MODELING RESULTS
In this appendix, results of BMD modeling are presented for each noncancer endpoint
showing significantly elevated incidences (for dichotomous data) or means (for continuous data)
relative to controls following inhalation exposure to ETBE. For each endpoint, a summary of the
dose-response data is presented, followed by a table summarizing the results of the dose-
response modeling. Finally, the standard output from the EPA's BMDS (version 1.4.1), for the
best-fitting dose-response model is presented.
For these modeling exercises, all dichotomous or continuous models available in BMDS
(version 1.4.1) were fit to the corresponding data for each endpoint with the BMR set at 0.1 (i.e.,
10% extra risk) for dichotomous data and one SD above the control mean for continuous data.
To select the "best-fit" model, AIC values were evaluated for all models that did not exhibit a
significant lack of fit (i.e. ,p< 0. 1), according to the chi-square goodness-of-fit test. Of these
models, the model with the lowest AIC value was typically selected as the best-fit model unless
examination of the chi-square scaled residuals indicated another model with a similar AIC
exhibited better fit in the region of the curve where the BMD was estimated. Selection of the
BMR and the procedure for selecting the best-fit model are consistent with the EPA's most
current Benchmark Dose Technical Guidance Document (U.S. EPA, 2000b).
07/14/2009
B-55
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Labeling Index
Species/Gender: CD-I Female Mice
(Medinsky et al., 1999)
Table B-23. Mean LI (and SD) from the livers of CD-I female mice exposed
to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean LI (SD)
Control (n = 4)
1.92 (2.45)
500 (n= 4)
1.70 (0.88)
1,750 (n= 5)
3.44a (0.90)
5,000 (n= 3)
4.97a (1.37)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-56
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-24. A summary of BMDS (version 1.4.1) modeling results based on
mean LI data from the livers of CD-I female mice exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/>-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.57
31.7
2,040
1,307
Power
0.57
31.7
2,040
1,307
Hill
NA
34.6
1,699
543
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-57
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
CD
in
o
Q.
in
-------
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 2.27228
rho = 0 Specified
beta_0 = 1.82498
beta 1 = 0.00065332
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 -3.7e-009 5.7e-009
beta_0 -3.7e-009 1 -0.67
beta 1 5.7e-009 -0.67 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 1.82871 0.646547 0.561502 3.09592
beta_0 1.84685 0.45746 0.950245 2.74346
beta 1 0.000662928 0.000191491 0.000287613 0.00103824
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 4 1.92 1.85 2.45 1.35 0.108
500 4 1.7 2.18 0.876 1.35 -0.704
1,750 5 3.44 3.01 0.899 1.35 0.723
5000 3 4.97 5.16 1.37 1.35 -0.245
07/14/2009
B-59
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-12.264824
-9.151449
-12.264824
-12.828887
-17.301515
# Param's
5
8
5
3
2
AIC
34 . 529648
34 . 302898
34 . 529648
31.657773
38.603029
Explanation of Tests
Test 1:
Test 2
Test 3
Test 4
(Note:
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
16.3001
6.22675
6.22675
1.12812
0.01223
0.1011
0.1011
0.5689
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
The p-value for Test 3 is greater than .1.
be appropriate here
The modeled variance appears to
07/14/2009
B-60
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 2039.89
BMDL = 1306.94
07/14/2009
B-61
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Centrilobular Hypertrophy
Species/Gender: CD-I Male Mice
(Medinsky et al., 1999)
Table B-24. Incidence of centrilobular hypertrophy in the livers of CD-I male mice exposed to four different
concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Incidence
Control
0/15
500
0/15
1,750
2/15
5,000
8/10a
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-62
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-25. A summary of BMDS (version 1.4.1) modeling results based on
incidence of centrilobular hypertrophy in the livers of CD-I male mice
exposed to ETBE via inhalation for 13 weeks
Model
%2/'-value
AIC
BMC10 (ppm)
BMCL10 (ppm)
Gamma
0.98
25.9
1,604
901
Logistic
0.62
27.2
1,976
1,304
Log-logistic
0.98
25.9
1,606
943
Multistage (2°)
0.94
24.4
1,351
784
Probit
0.73
26.7
1,847
1,227
Log-probit
0.998
25.8
1,602
957
Quantal-linear
0.20
30.0
623
380
Weibull
0.94
26.0
1,612
865
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCio = Benchmark concentration corresponding to a 10% change relative to controls.
BMCLio = 95% lower confidence limit on the benchmark concentration corresponding to a 10% change relative to
controls.
Source: Medinsky et al. (1999).
07/14/2009
B-63
DRAFT - DO NOT CITE OR QUOTE
-------
Pro bit Model with 0.95 Confidence Level
Probit
Source: Medinsky et al. (1999).
Figure B-10. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
log-probit) based on incidence of centrilobular hypertrophy in the livers of
CD-I male mice exposed to ETBE via inhalation for 13 weeks.
2000 3000 4000 5000
dose
14:05 08/02 2007
Probit Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_MALES_LIVER_HYPER_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_MALES_LIVER_HYPER_MEDINSKY99.pit
Thu Aug 02 14:05:10 2007
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
07/14/2009
B-64
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = Response
Independent variable = Dose
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial (and Specified) Parameter Values
background = 0
intercept = -8.93692
slope = 1.10637
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
intercept slope
intercept 1 -1
slope -1 1
Parameter Estimates
Variable
background
intercept
slope
Estimate
0
-15.0695
1. 86856
Std. Err.
NA
4.49702
0.565754
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
-23.8835
0.759699
-6.25553
2.97742
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model
Full model
Fitted model
Reduced model
Log(likelihood)
-10.8941
-10.8984
-26.0777
Analysis of Deviance Table
#
Param's
4
2
1
Deviance Test d.f.
0.00844147
30.367
P-value
0.9958
<.0001
AIC: 25.7967
07/14/2009
B-65
DRAFT - DO NOT CITE OR QUOTE
-------
Dose
Goodness of Fit
Est._Prob. Expected Observed Size
Scaled
Residual
0.0000
500.0000
1,750.0000
5000.0000
0.0000
0.0003
0.1321
0.8010
0. 000
0. 004
1.982
8.010
15
15
15
10
0.000
-0.064
0. 014
-0.008
Chi^2
0.00
d.f. = 2
P-value
0.9978
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1601. 9
BMDL = 956.583
07/14/2009
B-66
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Centrilobular Hypertrophy
Species/Gender: CD-I Female Mice
(Medinsky et al., 1999)
Table B-26. Incidence of centrilobular hypertrophy in the livers of CD-I
female mice exposed to four different concentrations of ETBE via inhalation
for 13 weeks
Administered dose (ppm)
Incidence
Control
0/13
500
2/15
1,750
1/15
5,000
9/14a
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-67
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-27. A summary of BMDS (version 1.4.1) modeling results based on
incidence of centrilobular hypertrophy in the livers of CD-I female mice
exposed to ETBE via inhalation for 13 weeks
Model
%2/'-value
AIC
BMC10 (ppm)
BMCL10 (ppm)
Gamma
0.38
44.0
3,385
511
Logistic
0.24
44.3
1,826
1,255
Log-logistic
0.10
46.2
801
335
Multistage (2°)
0.21
44.9
1,722
463
Probit
0.22
44.5
1,644
1,149
Log-probit
0.0.17
46.0
3,661
1,119
Quantal-linear
0.29
43.7
675
430
Weibull
0.17
46.0
4,273
511
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCio = Benchmark concentration corresponding to a 10% change relative to controls.
BMCLio = 95% lower confidence limit on the benchmark concentration corresponding to a 10% change relative to
controls.
Source: Medinsky et al. (1999).
07/14/2009
B-68
DRAFT - DO NOT CITE OR QUOTE
-------
Logistic Model with 0.95 Confidence Level
T3
CD
tj
CD
<
£Z
O
tj
0.3
0.6
0.4
0.2
Logistic
f T
T j I
hi—
I:
i
: i—* 1 1
bmdU
i
bryiD
1000
2000 3000
dose
4000
5000
14:41 08/02 2007
Source: Medinsky et al. (1999).
Figure B-ll. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
logistic) based on incidence of centrilobular hypertrophy in the livers of CD-
1 female mice exposed to ETBE via inhalation for 13 weeks.
Logistic Model. (Version: 2.9; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_FEMALES_LIVER_HYPER_MEDINSKY9 9.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_FEMALES_LIVER_HYPER_MEDINSKY9 9.pit
Thu Aug 02 14:41:46 2007
BMDS MODEL RUN
The form of the probability function is:
P[response] = 1/[1+EXP(-intercept-slope*dose)]
07/14/2009
B-69
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = Response
Independent variable = Dose
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
background = 0 Specified
intercept = -2.87922
slope = 0.000663794
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
intercept
slope
intercept
1
-0. 85
slope
-0. 85
1
Parameter Estimates
Variable
intercept
slope
Estimate
-3.24768
0.000754885
Std. Err.
0.760128
0.0 0 0202312
95.0% Wald Confidence Interval
LC'Wer Ccrif. Limit Upp'er Ccrif. Limit
-4.7375
0.000358361
-1.75785
0.00115141
Model
Full model
Fitted model
Reduced model
Analysis of Deviance Table
Log(likelihood)
-18 . 6887
-20.1599
-29.3352
# Param's
4
2
1
Deviance Test d.f.
2.94242
21.2931
P-value
0.2296
<.0001
AIC:
44.3197
Dose
Est. Prob.
Goodness of Fit
Expected Observed Size
Scaled
Residual
0.0000
500.0000
1,750.0000
5000.0000
0.0374
0.0536
0.1271
0.6287
0.486
0. 805
1.907
8.802
13
15
15
14
-0.711
1.370
-0.703
0.109
07/14/2009
B-70
DRAFT - DO NOT CITE OR QUOTE
-------
Chi A2 = 2.89 d.f. = 2
P-value = 0.2360
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 1826.48
BMDL = 1255.48
07/14/2009
B-71
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (absolute)
Species/Gender: CD-I Male Mice
(Medinsky et al., 1999)
Table B-28. Mean absolute liver weight (and SD) in CD-I male mice
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean absolute liver weight, g (SD)
Control (n= 15)
2.16(0.36)
500 (n= 15)
2.26 (0.28)
1,750 (n= 15)
2.44: (0.24)
5,000 (n= 10)
2.55a (0.25)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-72
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-29. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute liver weight in CD-I male mice exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/>-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.99
-76.8
1,754
936
Power
0.30
-76.4
3,781
2,521
Hill
NA
-74.8
1,758
598
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-73
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
EL 2 4
GO
CD
HI „
BMDLj.
1000
EMP
2000
0
3000
4000
5000
dose
09:08 08/03 2007
Source: Medinsky et al. (1999).
Figure B-12. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
2° polynomial) based on mean absolute liver weight in CD-I male mice
exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_MALES_LIVER_WT_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_MALES_LIVER_WT_MEDINSKY99.pit
Fri Aug 03 09:08:17 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
07/14/2009
B-74
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = MEAN
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.0849361
rho = 0 Specified
beta 0 = 2.16369
beta~l = 0.000204224
beta 2 = -2.52315e-008
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1 beta 2
alpha 1 6.3e-010 -7.1e-010 7.4e-010
beta_0 6.3e-010 1 -0.71 0.6
beta_l -7.le-010 -0.71 1 -0.97
beta 2 7.4e-010 0.6 -0.97 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.078759 0.0150188 0.0493228 0.108195
beta_0 2.16369 0.0622113 2.04175 2.28562
beta_l 0.000204223 8.4892e-005 3.7838e-005 0.000370609
beta 2 -2.52309e-008 1.61469e-008 -5.68783e-008 6.41641e-009
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 15 2.16 2.16 0.359 0.281 0.00434
07/14/2009
B-75
DRAFT - DO NOT CITE OR QUOTE
-------
500 15 2.26
1,750 15 2.44
5000 10 2.55
Model Descriptions for
2.26 0.284
2.44 0.243
2.55 0.252
likelihoods calculated
0.281 -0.00675
0.281 0.00267
0.281 -0.000318
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) # Param's AIC
A1 42.387507 5 -74.775013
A2 43.831021 8 -71.662041
A3 42.387507 5 -74.775013
fitted 42.387471 4 -76.774942
R 35.748395 2 -67.496791
Explanation of Tests
Test 1:
Test
Test
Test
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 16.1653 6 0.01289
Test 2 2.88703 3 0.4094
Test 3 2.88703 3 0.4094
Test 4 7.16895e-005 1 0.9932
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
07/14/2009
B-76
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1754.48
BMDL = 936.067
07/14/2009
B-77
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (absolute)
Species/Gender: CD-I Female Mice
(Medinsky et al., 1999)
Table B-30. Mean absolute liver weight (and SD) in CD-I female mice
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean absolute liver weight, g (SD)
Control (n= 13)
1.56 (0.21)
500 (n= 15)
1.59(0.16)
1,750 (n= 15)
1.86a (0.19)
5,000 (n= 14)
2.0T (0.30)
aStatistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-78
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-31. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute liver weight in CD-I female mice exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.32
-111.5
1,109
709
Power
0.14
-110.5
2,113
1,644
Hill
NA
-110.5
1,345
704
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-79
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
2.3
r 1—1—¦—1— ¦—¦—¦—¦—i —¦—¦—¦—¦ ¦—¦—i—¦— —¦—¦— 1—¦—¦—¦—
r Polynomial
~
2.2
T
r
2.1
r
k '
2
—————
1.9
T
I :
1.0
J
-
1.7
1.6
1 1
1.5
V 1 1 |
i i
1.4
\ . BMDL| . mO
0 1000 2000 3000 4000 5000
dose
09:41 08/03 2007
Source: Medinsky et al. (1999).
Figure B-13. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
2° polynomial) based on mean absolute liver weight in CD-I female mice
exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_FEMALES_LIVER_WT_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\CDl_MICE_FEMALES_LIVER_WT_MEDINSKY99.pit
Fri Aug 03 09:41:35 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
Independent variable = Dose
07/14/2009 B-80 DRAFT - DO NOT CITE OR QUOTE
-------
rho is set to 0
Signs of the polynomial coefficients
A constant variance model is fit
are not restricted
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.0477941
rho = 0 Specified
beta 0 = 1.534 8
beta~l = 0.000212599
beta 2 = -2.09645e-008
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1 beta 2
alpha 1 -7.7e-012 4.8e-012 -8.2e-012
beta_0 -7.7e-012 1 -0.73 0.62
beta_l 4.8e-012 -0.73 1 -0.98
beta 2 -8.2e-012 0.62 -0.98 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
alpha 0.0452046 0.0084676 0.0286084 0.0618008
betaJJ 1.53173 0.0496296 1.43446 1.629
beta_l 0.000215557 6.59293e-005 8.63377e-005 0.000344776
beta 2 -2.14312e-008 1.22684e-008 -4.54768e-008 2.61439e-009
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
1, 750
5000
13
15
15
14
1.56
1.59
1.86
2.07
1.53
1.63
1.84
2.07
0.211
0.162
0.189
0.295
0.213
0.213
0.213
0.213
0.53
-0.768
0.304
-0.0306
07/14/2009
B-81
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
60.237958
63.184016
60.237958
59.751845
41.070159
# Param's
5
8
5
4
2
AIC
-110.475915
-110.368033
-110.475915
-111.503689
-78.140319
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
44.2277
5.89212
5.89212
0.972226
<.0001
0. 117
0. 117
0.3241
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
07/14/2009 B-82 DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1108.52
BMDL = 708.952
07/14/2009
B-83
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (absolute)
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-32. Mean absolute liver weight (and SD) in F344 male rats exposed
to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean absolute liver weight, g (SD)
Control (n= 11)
8.86(1.19)
500 (n= 11)
9.38 (0.74)
1,750 (n= 11)
10. V (0.49)
5,000 (n= 11)
11.T (0.68)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-84
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-33. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute liver weight in F344 male rats exposed to ETBE via
inhalation for 13 weeks
Model" (non-constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.39
27.9
1,648
1,260
Power
0.30
27.9
1,648
1,260
Hill
0.57
28.3
1,098
645
Tor all models, the variance model employed (i.e., variance modeled as a power function of the mean) failed to
adequately address the non-constant variance.
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benclunark concentration corresponding to a 1 SD change relative
to the control mean.
Source: Medinsky et al. (1999).
07/14/2009
B-85
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
2.5
Linear
12
1.5
0.5
10
9.5
9
8.5
8
BMDU J3MD
1000
0
2000
3000
4000
5000
dose
10:44 08/03 2007
Source: (Medinsky et al. (1999).
Figure B-14. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean absolute liver weight in F344 male
rats exposed to ETBE via inhalation for 13 weeks
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_MALES_LIVER_WT_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_MALES_LIVER_WT_MEDINSKY99.pit
Fri Aug 03 10:44:41 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
07/14/2009 B-86 DRAFT - DO NOT CITE OR QUOTE
-------
Independent variable = Dose
Signs of the polynomial coefficients are not restricted
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha = -0.406567
rho = 0
beta_0 = 9.01743
beta 1 = 0.000553417
Asymptotic Correlation Matrix of Parameter Estimates
lalpha rho beta 0 beta 1
lalpha 1 -1 0.029 -0.039
rho -1 1 -0.029 0.04
beta_0 0.029 -0.029 1 -0.77
beta_l -0.039 0.04 -0.77 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha 7.20485 4.25062 -1.12621 15.5359
rho -3.37146 1.84605 -6.98964 0.246732
beta_0 9.03722 0.174598 8.69502 9.37943
beta 1 0.000544548 5.27201e-005 0.000441219 0.000647878
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 11 8.86 9.04 1.19 0.897 -0.666
500 11 9.38 9.31 0.744 0.853 0.262
1,750 11 10.1 9.99 0.488 0.758 0.559
5000 11 11.7 11.8 0.677 0.576 -0.173
07/14/2009
B-87
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-10.958703
-6.371795
-8 . 992356
-9. 935317
-34.689491
# Param's
5
8
6
4
2
AIC
31. 917406
28 . 743591
29.984711
27 . 870635
73.378982
Explanation of Tests
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (A1 vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 56.6354 6 <.0001
Test 2 9.17381 3 0.02707
Test 3 5.24112 2 0.07276
Test 4 1.88592 2 0.3895
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is less than .1. You may want to consider a
different variance model
07/14/2009 B-88 DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1647.64
BMDL = 12 5 9.77
07/14/2009
B-89
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (absolute)
Species/Gender: F344 Female Rats
(Medinsky et al., 1999)
Table B-34. Mean absolute liver weight (and SD) in F344 female rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean absolute liver weight, g (SD)
Control (n = 10)
5.19(0.44)
500 (n= 11)
5.29 (0.40)
1,750 (n= 11)
5.64 (0.52)
5,000 (n= 11)
6.53a (0.52)
aStatistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-90
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-35. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute liver weight in F344 female rats exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMCisd (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.99
-19.7
1,663
1,300
Power
0.93
-17.7
1,762
1,301
Hill
NA
-15.7
1,760
888
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-91
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
7
: Linear
T
6.5
7
1
6
7
T
5.5
" """ 1
^1 U
5
: 1 1
1 1
1 1
B.MDLl |3MD,
0 1000 2000 3000 4000 5000
dose
1 1:07 08/03 2007
Source: Medinsky et al. (1999).
Figure B-15. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean absolute liver weight in F344 female
rats exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_FEMALES_LIVER_WT_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_FEMALES_LIVER_WT_MEDINSKY99.pit
Fri Aug 03 11:07:08 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
07/14/2009
B-92
DRAFT - DO NOT CITE OR QUOTE
-------
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.223223
rho = 0 Specified
beta_0 = 5.17267
beta 1 = 0.000270389
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 -4e-009 3.5e-010
beta 0 -4e-009 1 -0.69
beta 1 3.5e-010 -0.69 1
Parameter Estimates
Variable
alpha
beta_0
beta 1
Estimate
0.202595
5.1719
0.0 0 0270586
Std. Err.
0.0436928
0.0947258
3 . 51 9 7 9 e - 0 0 5
95.0% Wald Confidence Interval
Lower Ccrif. Limit Upper Ccrif. Limit
0.116959
4.98624
0.0002 015 9 9
0.288232
5.35756
0.000339573
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
1, 750
5000
10
11
11
11
5.19
5.29
5 . 64
6.53
5.17
5.31
5. 65
6.52
0.44
0.397
0.519
0.519
0.45
0.45
0. 45
0.45
0.127
-0.0899
-0.0621
0.0307
07/14/2009
B-93
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
12 . 840251
13.399966
12.840251
12.825726
-5.766125
# Param's
5
8
5
3
2
AIC
-15.680502
-10.799932
-15. 680502
-19.651451
15.532250
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
38.3322
1.11943
1.11943
0.0290511
<.0001
0.7724
0.7724
0.9856
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-94
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1663.45
BMDL = 1299.55
07/14/2009
B-95
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (absolute)
Species/Gender: Sprague-Dawlev Male Rats
(White et al., 1995)
Table B-36. Mean absolute liver weight (and SD) in Sprague-Dawley male
rats exposed to four different concentrations of ETBE via inhalation for
13 weeks
Administered dose (ppm)
Mean absolute liver weight, g (SD)
Control (n = 10)
10.2(1.18)
500 (n= 10)
10.0 (0.62)
2,000 (n= 10)
10.6(1.24)
4,000 (n= 10)
11.9a (1.19)
aStatistically significantly different from control at p < 0.05 as reported by White et al. (1995).
White et al. (1995).
07/14/2009
B-96
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-37. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute liver weight in Sprague-Dawley male rats exposed to ETBE via
inhalation for 13 weeks
Model(constant variance)
%2/>-value
AIC
BMCisd (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.55
49.7
2,309
1,616
Power
0.68
50.7
2,929
1,728
Hill
NA
52.6
2,189
1,620
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exliibits significant lack-of-fit).
BMCisd = Benclunark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benclunark concentration corresponding to a 1 SD change relative to
the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is not
valid for evaluating lack of fit.
Source: White et al. (1995).
07/14/2009
B-97
DRAFT - DO NOT CITE OR QUOTE
-------
Linear
Linear Model with 0.95 Confidence Level
0 500
11:31 08/03 2007
1000 1500 2000 2500 3000 3500 4000
dose
Source: White et al. (1995).
Figure B-16. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean absolute liver weight in SD male rats
exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_MALES_LIVER_WT_WHITE95.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_MALES_LIVER_WT_WHITE95.pit
Fri Aug 03 11:31:03 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
07/14/2009
B-98
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = MEAN
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 1.18263
rho = 0 Specified
beta_0 = 9.93277
beta 1 = 0.000453677
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 -9.9e-011 -2e-011
beta_0 -9.9e-011 1 -0.72
beta 1 -2e-011 -0.72 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
alpha 1.09693 0.245282 0.616191 1.57768
betaJJ 9.93277 0.239424 9.46351 10.402
beta 1 0.000453677 0.00010641 0.000245117 0.000662238
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
2000
4000
10
10
10
10
10.2
10
10.6
11.9
9.93
10.2
10.8
11.7
1.18
0. 62
1.24
1.19
1.05
1.05
1.05
1.05
0.686
-0.422
-0.634
0.37
07/14/2009
B-99
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-21.247521
-18.658223
-21.247521
-21. 850388
-29.342646
# Param's
5
8
5
3
2
AIC
52 .495041
53.316446
52 .495041
49.700776
62.685293
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
21.3688
5. 1786
5. 1786
1.20574
0. 001575
0.1592
0.1592
0.5472
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-100
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 2308.57
BMDL = 1616.16
07/14/2009
B-101
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (absolute)
Species/Gender: Sprague-Dawlev Female Rats
(White et al., 1995)
Table B-38. Mean absolute liver weight (and SD) in Sprague-Dawley female
rats exposed to four different concentrations of ETBE via inhalation for
13 weeks
Administered dose (ppm)
Mean absolute liver weight, g (SD)
Control (n = 10)
6.0 (0.39)
500 (n= 10)
6.2 (0.49)
2,000 (n= 10)
6.5 (0.40)
4,000 (n= 10)
6.6a (0.46)
aStatistically significantly different from control at p < 0.05 as reported by White et al. (1995).
Source: White et al. (1995).
07/14/2009
B-102
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-39. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute liver weight in Sprague-Dawley female rats exposed to ETBE
via inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.97
-22.4
1,593
779
Power
0.48
-23.0
2,996
1,953
Hill
NA
-20.4
1,469
406
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: White et al. (1995).
07/14/2009
B-103
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
CD
GO
I 64
CD
DL
BMDL
MD
0
500
1000
1500
2000
2500
3000
3500 4000
dose
1 1:50 08/03 2007
Source: White et al. (1995).
Figure B-17. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
2° polynomial) based on mean absolute liver weight in Sprague-Dawley
female rats exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_FEMALES_LIVER_WT_WHITE95.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_FEMALES_LIVER_WT_WHITE95.pit
Fri Aug 03 11:50:46 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
07/14/2009
B-104
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = MEAN
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.19095
rho = 0 Specified
beta_0 = 5.99736
beta_l = 0.000337626
beta 2 = -4.85928e-008
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha
beta 0
beta 1
beta 2
alpha
1
7 . 7e-010
-2 . 3e-010
-8 . le-011
beta 0
7 . 7e-010
1
-0.61
0.54
beta 1
-2 . 3e-010
-0.61
1
-0.91
beta 2
-8 . le-011
0.54
-0.91
1
Parameter Estimates
Variable
alpha
beta_0
beta_l
beta 2
Estimate
0.171862
5.99736
0.000337626
-4.85928e-008
Std. Err.
0.0384294
0.112462
0.000167 08 4
3.9939e-008
95.0% Wald Confidence Interval
Lower Ccrif. Limit Upper Ccrif. Limit
0.0965413
5.77694
1. 014 6 2 e - 0 0 5
-1.26872e-007
0.247182
6.21778
0.000665105
2 . 9 6 8 6 2 e - 0 0 8
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
2000
4000
10
10
10
10
6
6.15
6.48
6.57
6
6.15
6.48
6.57
0.39
0.49
0.4
0.46
0. 415
0. 415
0. 415
0. 415
0.0201
-0.0307
0.0134
-0.00288
07/14/2009
B-105
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
15.222084
15.584990
15.222084
15.221315
9.717433
# Param's
5
8
5
4
2
AIC
-20.444167
-15.169980
-20.444167
-22.442631
-15.434865
Explanation of Tests
Test 1:
Test 2
Test 3
Test 4
(Note:
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
11.7351
0.725812
0.725812
0.00153657
0.06815
0.8671
0.8671
0.9687
The p-value for Test 1 is greater than .05. There may not be a
diffence between responses and/or variances among the dose levels
Modelling the data with a dose/response curve may not be appropriate
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
07/14/2009
B-106
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1593.19
BMDL = 778.988
07/14/2009
B-107
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (relative)
Species/Gender: Sprague-Dawlev Male Rats
(White et al., 1995)
Table B-40. Mean relative liver weight (and SD) in Sprague-Dawley male
rats exposed to four different concentrations of ETBE via inhalation for
4 weeks
Administered dose (ppm)
Mean relative liver weight, g (SD)
Control (n = 10)
2.86 (0.21)
500 (n= 10)
2.84 (0.15)
2,000 (n= 10)
2.96 (0.21)
4,000 (n= 10)
3.32a (0.25)
aStatistically significantly different from control at p < 0.05 as reported by White et al. (1995).
Source: White et al. (1995).
07/14/2009
B-108
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-41. A summary of BMDS (version 1.4.1) modeling results based on
mean relative liver weight in Sprague-Dawley male rats exposed to ETBE
via inhalation for 4 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.80
-81.7
2,678
1,619
Power
0.77
-81.7
2,644
1,624
Hill
NA
-79.7
2,150
1,633
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: White et al. (1995).
07/14/2009
B-109
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
3.4
CD
GO
£=
O
CL
tn
Qj
bmdU
1000 1500 2000 2500 3000 3500 4000
MD
0
500
dose
14:06 08/03 2007
Source: White et al. (1995).
Figure B-18. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
2° polynomial) based on mean relative liver weight in Sprague-Dawley male
rats exposed to ETBE via inhalation for 4 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_MALES_REL_LIVER_WT_WHITE95.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_MALES_REL_LIVER_WT_WHITE95.pit
Fri Aug 03 14:06:06 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
07/14/2009 B-l 10 DRAFT - DO NOT CITE OR QUOTE
-------
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.0433
rho = 0 Specified
beta 0 = 2.8517
beta~l = -1.46048e-005
beta 2 = 3.2994e-008
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1 beta 2
alpha 1 3.3e-010 -5.9e-011 -3.1e-011
beta_0 3.3e-010 1 -0.67 0.54
beta_l -5.9e-011 -0.67 1 -0.97
beta_2 -3.1e-011 0.54 -0.97 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.0390352 0.00872854 0.0219276 0.0561428
beta_0 2.8517 0.0535977 2.74665 2.95675
beta_l -1.46048e-005 7.96297e-005 -0.000170676 0.000141466
beta_2 3.2994e-008 1.90343e-008 -4.31244e-009 7.03005e-008
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 10 2.86 2.85 0.21 0.198 0.133
500 10 2.84 2.85 0.15 0.198 -0.202
2000 10 2.96 2.95 0.21 0.198 0.0886
4000 10 3.32 3.32 0.25 0.198 -0.019
Model Descriptions for likelihoods calculated
07/14/2009
B-lll
DRAFT - DO NOT CITE OR QUOTE
-------
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) # Param's AIC
A1 44.899263 5 -79.798526
A2 46.154309 8 -76.308617
A3 44.899263 5 -79.798526
fitted 44.865825 4 -81.731649
R 31.476069 2 -58.952138
Explanation of Tests
Test 1:
Test
Test
Test
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 29.3565 6 <.0001
Test 2 2.51009 3 0.4735
Test 3 2.51009 3 0.4735
Test 4 0.0668772 1 0.7959
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
07/14/2009
B-112
DRAFT - DO NOT CITE OR QUOTE
-------
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 2 678.38
BMDL = 1618.94
07/14/2009
B-113
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Liver
Endpoint: Liver Weight (relative)
Species/Gender: Sprague-Dawlev Female Rats
(White et al., 1995)
Table B-42. Mean relative liver weight (and SD) in Sprague-Dawley female
rats exposed to four different concentrations of ETBE via inhalation for
4 weeks
Administered dose (ppm)
Mean relative liver weight, g (SD)
Control (n = 10)
2.80 (0.17)
500 (n= 10)
2.93 (0.10)
2,000 (n= 10)
3.08a (0.22)
4,000 (n= 10)
3.15a (0.24)
aStatistically significantly different from control at p < 0.05 as reported by White et al. (1995).
Source: White et al. (1995).
07/14/2009
B-114
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-43. A summary of BMDS (version 1.4.1) modeling results based on
mean relative liver weight in Sprague-Dawley female rats exposed to ETBE
via inhalation for 4 weeks
Model" (non-constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.30
-89.4
1,600
997
Power
0.30
-89.4
1,600
997
Hill
NA
-87.9
704
240
aFor all models, the variance model employed (i.e., variance modeled as a power function of the mean) failed
to adequately address the non-constant variance.
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benclunark concentration corresponding to a 1 SD change
relative to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this
test is not valid for evaluating lack of fit.
Source: White et al. (1995).
07/14/2009
B-115 DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
0 500 1000 1500 2000 2500 3000 3500 4000
dose
14:26 08/03 2007
Source: White et al. (1995).
Linear
BMDL
Figure B-19. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean relative liver weight in Sprague-
Dawley female rats exposed to ETBE via inhalation for 4 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_FEMALES_REL_LIVER_WT_WHITE95.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_FEMALES_REL_LIVER_WT_WHITE95.pit
Fri Aug 03 14:26:27 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
07/14/2009
B-116
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = MEAN
Independent variable = Dose
Signs of the polynomial coefficients are not restricted
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha = -3.31801
rho = 0
beta 0 = 2.85748
beta~l = 8 . 15484e-005
Asymptotic Correlation Matrix of Parameter Estimates
lalpha rho beta 0 beta 1
lalpha 1 -1 0.17 -0.26
rho -1 1 -0.17 0.26
beta_0 0.17 -0.17 1 -0.64
beta 1 -0.26 0.26 -0.64 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
lalpha -13.0534 5.49755 -23.8284 -2.27837
rho 8.78327 5.01654 -1.04897 18.6155
betaJJ 2.84555 0.0355734 2.77583 2.91527
beta 1 9.0 3 5 6 5 e-0 0 5 2.1278e-005 4.86523e-005 0.000132061
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 10 2.8 2.85 0.17 0.145 -0.996
500 10 2.93 2.89 0.1 0.155 0.802
2000 10 3.08 3.03 0.22 0.189 0.897
4000 10 3.15 3.21 0.24 0.244 -0.737
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
07/14/2009
B-117
DRAFT - DO NOT CITE OR QUOTE
-------
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
48 . 467326
52.265071
49.925883
48 . 717299
39.537229
# Param's
5
8
6
4
2
AIC
-86.934652
-88 . 530141
-87.851765
-89.434598
-75.074458
Explanation of Tests
Test 1:
Test
Test
Test
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 25.4557 6 0.0002811
Test 2 7.59549 3 0.05516
Test 3 4.67838 2 0.09641
Test 4 2.41717 2 0.2986
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is less than .1. You may want to consider a
different variance model
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
07/14/2009
B-118
DRAFT - DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 1599.81
BMDL = 997.078
07/14/2009
B-119
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Kidney Weight (absolute)
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-44. Mean absolute kidney weight (and SD) in F344 male rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean absolute kidney weight, g (SD)
Control (n= 11)
1.73 (0.16)
500 (n= 11)
1.85 (0.14)
1,750 (n= 11)
1.90a (0.10)
5,000 (n= 11)
2.07a (0.12)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-120
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-45. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute kidney weight in F344 male rats exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.23
-130.4
2,169
1,632
Power
0.23
-130.4
2,169
1,632
Hill
0.24
-130.0
1,099
396
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
Source: Medinsky et al. (1999).
07/14/2009
B-121
DRAFT - DO NOT CITE OR QUOTE
-------
CD
in
o
Q.
in
-------
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.0170425
rho = 0 Specified
beta 0 = 1.78068
beta~l = 5 . 93491e-005
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 2.4e-010 5.6e-010
beta_0 2.4e-010 1 -0.68
beta 1 5.6e-010 -0.68 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.0165668 0.00353206 0.00964412 0.0234895
beta_0 1.78068 0.0265071 1.72873 1.83263
beta 1 5.934 91e-005 9.96331e-006 3.98214e-005 7.88769e-005
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 11 1.73 1.78 0.155 0.129 -1.23
500 11 1.85 1.81 0.137 0.129 1.02
1,750 11 1.9 1.88 0.1 0.129 0.476
5000 11 2.07 2.08 0.124 0.129 -0.269
07/14/2009
B-123
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
69.681815
70.760620
69.681815
68.207768
55.197968
# Param's
5
8
5
3
2
AIC
-129.363630
-125.521241
-129.363630
-130.415535
-106.395937
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
31. 1253
2 . 15761
2 . 15761
2.94809
<.0001
0.5403
0.5403
0.229
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-124
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 2168.73
BMDL = 1632.4
07/14/2009
B-125
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Kidney Weight (absolute)
Species/Gender: F344 Female Rats
(Medinsky et al., 1999)
Table B-46. Mean absolute kidney weight (and SD) in F344 female rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean absolute kidney weight, g (SD)
Control (n = 10)
1.08 (0.07)
500 (n= 11)
1.13 (0.05)
1,750 (n= 11)
1.21a (0.08)
5,000 (n= 11)
1.3 la (0.06)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-126
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-47. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute kidney weight in F344 female rats exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.83
-191.2
717
494
Power
0.09
-188.4
1,465
1,162
Hill
NA
-189.2
641
346
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the
model exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change
relative to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this
test is not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-127
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
1.35
1.3
1.25
1.2
15
1.1
1.05
bmdU pmq
1000
0
2000
3000
4000
5000
dose
09:30 08/06 2007
Source: Medinsky et al. (1999).
Figure B-21. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
2° polynomial) based on mean absolute kidney weight in F344 female rats
exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_FEMALES_KIDNEY_WT_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_FEMALES_KIDNEY_WT_MEDINSKY9 9.pit
Mon Aug 06 09:30:30 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
Independent variable = Dose
07/14/2009 B-128 DRAFT - DO NOT CITE OR QUOTE
-------
rho is set to 0
Signs of the polynomial coefficients
A constant variance model is fit
are not restricted
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.00394613
rho = 0 Specified
beta_0 = 1.07904
beta 1 = 9.00466e-005
beta"2 = -8.93576e-009
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1 beta 2
alpha 1 1.3e-008 -1.8e-008 1.8e-008
beta_0 1.3e-008 1 -0.72 0.62
beta_l -1.8e-008 -0.72 1 -0.98
beta 2 1.8e-008 0.62 -0.98 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
alpha 0.00358299 0.000772727 0.00206847 0.0050975
betaJJ 1.07919 0.0160403 1.04775 1.11062
beta_l 6.9 9 0 5 e-0 0 5 2.1474 9e-005 4.7815e-005 0.000131995
beta 2 -8.91318e-009 3.99586e-009 -1.67449e-008 -1.08143e-009
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
1, 750
5000
10
11
11
11
1. 08
1.13
1.21
1.31
1.08
1.12
1.21
1.31
0.069
0.048
0. 076
0.055
0.0599
0.0599
0.0599
0.0599
-0.115
0.171
-0.0677
0.00659
07/14/2009
B-129
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
99.602165
100.989890
99.602165
99.578511
75.306055
# Param's
5
8
5
4
2
AIC
-189.204331
-185.979779
-189.204331
-191.157022
-146.612110
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
51.3677
2.77545
2.77545
0.0473084
<.0001
0.4276
0.4276
0. 8278
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-130
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 716.719
BMDL = 4 94.12 6
07/14/2009
B-131
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Kidney Weight (absolute)
Species/Gender: Sprague-Dawley Male Rats
(White et al., 1995)
Table B-48. Mean absolute kidney weight (and SD) in Sprague-Dawley male
rats exposed to four different concentrations of ETBE via inhalation for
4 weeks
Administered dose (ppm)
Mean absolute kidney weight, g (SD)
Control (n = 10)
2.73 (0.35)
500 (n= 10)
2.71 (0.18)
2,000 (n= 10)
2.89 (0.28)
4,000 (n= 10)
3.08a (0.35)
aStatistically significantly different from control at p < 0.05 as reported by White et al. (1995).
Source: White et al. (1995).
07/14/2009
B-132
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-49. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute kidney weight in Sprague-Dawley male rats exposed to ETBE
via inhalation for 4 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.87
-54.7
3,010
1,960
Power
0.65
-52.8
3,186
1,969
Hill
NA
-51.0
2,247
606
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: White et al. (1995).
07/14/2009
B-133
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
CD
Cfl
C
o
Q.
Cfl
iIJ
DL
CD
3.4 F
3.2 :
3 -
2.8 -
2.6 :
2.4
Linear
500
1000 1500
2000
dose
2500 3000 3500 4000
09:53 08/06 2007
Source: White et al. (1995).
Figure B-22. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean absolute kidney weight in Sprague-
Dawley male rats exposed to ETBE via inhalation for 4 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_MALES_KIDNEY_WT_WHITE95.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\SD_RATS_MALES_KIDNEY_WT_WHITE95.pit
Mon Aug 06 09:53:16 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + ...
Dependent variable = MEAN
Independent variable = Dose
07/14/2009
B-134
DRAFT - DO NOT CITE OR QUOTE
-------
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.08895
rho = 0 Specified
beta 0 = 2.69923
beta~l = 9 . 43226e-005
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 3.5e-009 -4.2e-009
beta_0 3.5e-009 1 -0.72
beta 1 -4.2e-009 -0.72 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
alpha 0.0806269 0.0180287 0.0452913 0.115963
betaJJ 2.69923 0.064 9108 2.572 2.82645
beta 1 9.4 3 2 2 6 e-0 0 5 2.884 92e-005 3.777 91e-005 0.000150866
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
2000
4000
10
10
10
10
2.73
2.71
2.89
3. 08
2.7
2 .75
2.89
3.08
0.35
0.18
0.28
0.35
0.284
0.284
0.284
0.284
0.343
-0.405
0.0237
0.0388
07/14/2009
B-135
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
30.500828
32 . 981294
30.500828
30.358450
25.621610
# Param's
5
8
5
3
2
AIC
-51.001655
-49.962588
-51.001655
-54.716900
-47.243219
Explanation of Tests
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (A1 vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
14 .7194
4 . 96093
4 . 96093
0.284755
0. 02256
0. 1747
0. 1747
0.8673
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
07/14/2009
B-136
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 3 010.4
BMDL = 1959.5
07/14/2009
B-137
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Labeling Index
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-50. Mean LI (and SD) in the kidney of F344 male rats exposed to
four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean LI (SD)
Control (n = 5)
0.93 (0.30)
500 (n= 5)
2.26a (0.86)
1,750 (n= 5)
3.42a (0.66)
5,000 (n= 5)
2.59a (1.21)
aStatistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-138
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-51. A summary of BMDS (version 1.4.1) modeling results based on
mean LI in the kidney of F344 male rats exposed to ETBE via inhalation for
13 weeks
Model (non-constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.12
15.4
160
81
Power
<0.0001
30.3
2,300
342
Hill
NA
15.6
406
Computation of the
lower bound failed
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-139
DRAFT - DO NOT CITE OR QUOTE
-------
Polynomial Model with 0.95 Confidence Level
1
0.5
111
eEmd.U Emd
1000 2000 3000 4000 5000
Polynomial
dose
10:55 09/07 2007
Source: Medinsky et al. (1999).
Figure B-23. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
2° polynomial) based on mean LI in the kidney of F344 male rats exposed to
ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_MALES_KIDNEY_LI_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_MALES_KIDNEY_LI_MEDINSKY99.pit
Fri Sep 07 10:55:09 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
07/14/2009 B-140 DRAFT - DO NOT CITE OR QUOTE
-------
Independent variable = Dose
Signs of the polynomial coefficients are not restricted
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha = -0.383191
rho = 0
beta 0 = 1. 09523
beta~l = 0.00197542
beta 2 = -3.35849e-007
Asymptotic Correlation Matrix of Parameter Estimates
lalpha rho beta_0 beta_l beta_2
lalpha 1 -0.88 -0.04 0.025 -0.056
rho -0.88 1 0.0065 -0.069 0.13
beta_0 -0.04 0.0065 1 -0.42 0.34
beta_l 0.025 -0.069 -0.42 1 -0.98
beta_2 -0.056 0.13 0.34 -0.98 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha -2.061 0.693038 -3.41933 -0.702675
rho 1.81929 0.866479 0.121024 3.51756
beta_0 0.957315 0.146009 0.671144 1.24349
beta_l 0.00220546 0.000415615 0.00139087 0.00302006
beta_2 -3.71512e-007 8.46199e-008 -5.37364e-007 -2.0566e-007
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 5 0.93 0.957 0.302 0.343 -0.178
500 5 2.26 1.97 0.857 0.66 0.978
1,750 5 3.42 3.68 0.662 1.17 -0.493
5000 5 2.59 2.7 1.21 0.88 -0.282
07/14/2009
B-141
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-3. 936651
0. 098177
-1.473139
-2 .705971
-13.002695
# Param's
5
8
6
5
2
AIC
17.873301
15.803646
14 . 946277
15.411943
30.005391
Explanation of Tests
Test 1: Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (A1 vs A2)
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 26.2017 6 0.0002042
Test 2 8.06966 3 0.04459
Test 3 3.14263 2 0.2078
Test 4 2.46567 1 0.1164
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
07/14/2009 B-142 DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 159 . 8
BMDL = 80.847
07/14/2009
B-143
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Regenerative Foci (continuous)
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-52. Mean regenerative foci (and SD) in the kidney of F344 male
rats exposed to four different concentrations of ETBE via inhalation for
13 weeks
Administered dose (ppm)
Mean regenerative foci (SD)
Control (n = 5)
2.2 (0.84)
500 (n= 5)
11.0a (4.53)
1,750 (n= 5)
16.803 (7.46)
5,000 (n= 5)
33.803 (6.30)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-144
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-53. A summary of BMDS (version 1.4.1) modeling results based on
mean regenerative foci in the kidney of F344 male rats exposed to ETBE via
inhalation for 13 weeks
Model (non-constant variance)
X2 p-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.02
85.5
62
32
Power
0.001
93.1
554
156
Hill
0.14
82.4
40
23
Chi-square /j-value =/?-valuc from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
Source: Medinsky et al. (1999).
07/14/2009
B-145
DRAFT - DO NOT CITE OR QUOTE
-------
BMDS (version 1.4.1) model output for the best-fit model (i.e., Hill) based on mean
regenerative foci in the kidney of F344 male rats exposed to ETBE via inhalation for
13 weeks (Medinsky et al., 1999)
Hill Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_MALES_KIDNEY_MEAN_REGEN_FOCI_MEDINSKY9 9.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_MALES_KIDNEY_MEAN_REGEN_FOCI_MEDINSKY9 9.pit
Fri Sep 07 11:49:42 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = intercept + v*doseAn/(kAn + doseAn)
Dependent variable = MEAN
Independent variable = Dose
Power parameter restricted to be greater than 1
The variance is to be modeled as Var(i) = exp(lalpha + rho * In(mean(i)))
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha
rho
intercept
v
n
0. 147672
8020.59
3.37245
0
2.2
31. 6
k
07/14/2009
B-146
DRAFT - DO NOT CITE OR QUOTE
-------
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -n
have been estimated at a boundary point, or have been specified by
the user,
lalpha
rho
intercept
v
k
and do not appear in the correlation matrix )
lalpha
1
-0.94
-0.41
0.33
0.3
rho
-0. 94
1
0.36
-0.39
-0.31
intercept
-0.41
0.36
1
-0.089
0. 012
0.33
-0.39
-0.089
1
0. 91
k
0.3
-0.31
0. 012
0.91
1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha -1.80823 1.05076 -3.86769 0.251225
rho 1.85265 0.409579 1.04989 2.65541
intercept 2.16744 0.368446 1.4453 2.88958
v 38.0267 8.92164 20.5406 55.5128
n 1 NA
k 1812.05 797.883 248.224 3375.87
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
1, 750
5000
2.2
11
16. E
33.8
2.17
10.4
20.8
30.1
0.837
4.53
7.46
6.3
0. 829
3.54
6.75
9.48
0.0878
0.385
-1.34
0.878
07/14/2009
B-147
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-41.493114
-33.681268
-35.107830
-36.175922
-60.533595
# Param's
5
8
6
5
2
AIC
92 . 986228
83.362536
82.215660
82 . 351845
125. 067190
Explanation of Tests
Test 1:
Test 2
Test 3
Test 4
(Note:
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
53.7047
15.6237
2 . 85312
2 . 13618
<.0001
0.001354
0.2401
0. 1439
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
07/14/2009
B-148
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 40.3827
BMDL = 22 . 9779
07/14/2009
B-149
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Kidney
Endpoint: Regenerative Foci
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-54. Incidence of regenerative foci in the kidneys of F344 male rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Incidence
Control
4/11
500
10/1 la
1,750
ll/ll3
5,000
ll/ll3
aStatistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-150
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-55. A summary of BMDS (version 1.4.1) modeling results based on
incidence of regenerative foci in the kidneys of F344 male rats exposed to
ETBE via inhalation for 13 weeks
Model
%2/'-value
AIC
BMC10 (ppm)
BMCL10 (ppm)
Gamma
0.999
27.1
175
13
Logistic
0.9996
25.1
47
24
Log-logistic
0.9997
27.1
366
1
Multistage (1°)
0.996
25.1
27
13
Probit
1.00
25.1
49
29
Log-probit
0.9997
27.1
257
16
Quantal-linear
0.996
25.1
27
13
Weibull
0.999
27.1
86
13
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCio = Benchmark concentration corresponding to a 10% change relative to controls.
BMCLio = 95% lower confidence limit on the benchmark concentration corresponding to a 10% change relative to
controls.
Source: Medinsky et al. (1999).
07/14/2009
B-151
DRAFT - DO NOT CITE OR QUOTE
-------
Logistic Model with 0.95 Confidence Level
Logistic
T3
CD
tj
CD
y=
<
1000
2000
3000
4000
5000
dose
1 1:13 08/06 2007
Source: Medinsky et al. (1999).
Figure B-24. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
logistic) based on incidence of regenerative foci in the livers of F344 male rats
exposed to ETBE via inhalation for 13 weeks.
Logistic Model. (Version: 2.9; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_MALES_KIDNEY_REGEN_FOCI_INCIDENCE_MEDINSKY9 9.(d
)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_MALES_KIDNEY_REGEN_FOCI_INCIDENCE_MEDINSKY9 9.pi
t
Mon Aug 06 11:13:21 2007
BMDS MODEL RUN
The form of the probability function is:
P[response] = 1/[1+EXP(-intercept-slope*dose)]
07/14/2009
B-152
DRAFT - DO NOT CITE OR QUOTE
-------
Dependent variable = Response
Independent variable = Dose
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
background = 0 Specified
intercept = 0.95046
slope = 0.000538515
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
intercept slope
intercept 1 -0.51
slope -0.51 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
intercept -0.560447 0.626213 -1.7878 0.666908
slope 0.00573261 0.00243007 0.000969767 0.0104955
Analysis of Deviance Table
Model Log(likelihood) # Param's Deviance Test d.f. P-value
Full model -10.5 613 4
Fitted model -10.5621 2 0.00170557 2 0.9991
Reduced model -20.8621 1 20.6017 3 0.0001274
AIC: 25.1243
Goodness of Fit
Dose Est._Prob. Expected Observed Size
Scaled
Residual
0.0000
500.0000
1,750.0000
5000.0000
0.3634
0.9094
0.9999
1.0000
3. 998
10.003
10.999
11.000
4
10
11
11
11
11
11
11
0.001
-0.003
0. 029
0.000
Chi^2
0.00
d.f. = 2
P-value
0.9996
07/14/2009
B-153
DRAFT - DO NOT CITE OR QUOTE
-------
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 46.5325
BMDL = 23.9564
07/14/2009
B-154
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Adrenal Gland
Endpoint: Adrenal Gland Weight (absolute)
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-56. Mean absolute adrenal gland weight (and SD) in F344 male rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean adrenal gland weight, g (SD)
Control (n= 11)
0.035 (0.005)
500 (n= 11)
0.039 (0.009)
1,750 (n= 11)
0.038 (0.007)
5,000 (n= 11)
0.0473 (0.007)
aStatistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-155
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-57. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute adrenal gland weight in F344 male rats exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.40
-387.3
3,214
2,223
Power
0.19
-385.4
3,759
2,234
Hill
NA
-383.3
3,814
1,408
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-156
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
0.05
| 0.045
o
Q.
in
-------
Independent variable = Dose
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 5.1e-005
rho = 0 Specified
beta 0 = 0.0358301
beta~l = 2 . 16272e-006
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 6.9e-010 -l.le-012
beta_0 6.9e-010 1 -0.68
beta 1 -l.le-012 -0.68 1
Parameter Estimates
Variable
alpha
beta_0
beta 1
Estimate
4 . 8 3101 e - 0 0 5
0.0358301
2.16272e-006
Std. Err.
1. 0 2 9 9 7 e - 0 0 5
0.0014314
5 . 3 8 0 2 6 e - 0 0 7
95.0% Wald Confidence Interval
Lower Ccrif. Limit Upper Ccrif. Limit
2 . 812 3 e - 0 0 5
0.0330246
1.10821e-006
6 . 8 4 9 7 2 e - 0 0 5
0.038 6356
3 . 217 2 3 e - 0 0 6
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
1, 750
5000
11
11
11
11
0.035
0.039
0. 038
0.047
0.0358
0.0369
0.0396
0.0466
0.005
0.009
0. 007
0.007
0. 00695
0.00695
0.00695
0. 00695
-0.396
0.997
-0.771
0.17
07/14/2009
B-158
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
197.537892
199.354746
197.537892
196.633146
189.751789
# Param's
5
8
5
3
2
AIC
-385.075785
-382.709491
-385. 075785
-387.266291
-375.503578
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
19.2059
3. 63371
3.63371
1. 80949
0.00383
0.3038
0.3038
0.4046
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-159
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 3213 . 8
BMDL = 2222.59
07/14/2009
B-160
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Adrenal Gland
Endpoint: Adrenal Gland Weight (absolute)
Species/Gender: F344 Female Rats
(Medinsky et al., 1999)
Table B-58. Mean absolute adrenal gland weight (and SD) in F344 female
rats exposed to four different concentrations of ETBE via inhalation for
13 weeks
Administered dose (ppm)
Mean adrenal gland weight, g (SD)
Control (n = 10)
0.045 (0.004)
500 (n= 11)
0.048 (0.004)
1,750 (n= 11)
0.048 (0.007)
5,000 (n= 11)
0.0533 (0.005)
"Statistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-161
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-59. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute adrenal gland weight in F344 female rats exposed to ETBE
via inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.54
-406.7
3,576
2,394
Power
0.54
-406.7
3,576
2,394
Hill
0.26
-404.7
3,437
750
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
Source: Medinsky et al. (1999).
07/14/2009
B-162
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
0.056
0.054
% 0.052
!Z
S- 0.05
iii
! 0.048
Qj
S 0.046
0.044
0.042
0 1000 2000 3000 4000 5000
dose
14:24 08/06 2007
Source: Medinsky et al. (1999).
Figure B-26. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean absolute adrenal gland weight in
F344 female rats exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F344_RATS_FEMALES_ADRENAL_WT_MEDINSKY99.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_FEMALES_ADRENAL_WT_MEDINSKY9 9.pit
Mon Aug 06 14:24:18 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
Independent variable = Dose
07/14/2009 B-163 DRAFT - DO NOT CITE OR QUOTE
Linear
-------
rho is set to 0
Signs of the polynomial coefficients are not restricted
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 2 . 67692e-005
rho = 0 Specified
beta_0 = 0.0459464
beta 1 = 1. 40886e-006
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha beta 0 beta 1
alpha 1 -2e-008 2.8e-008
beta 0 -2e-008 1 -0.69
beta 1 2.8e-008 -0.69 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 2.4 9 955e-005 5.39067e-006 1.443e-005 3.5561e-005
beta_0 0.0459884 0.00105217 0.0439262 0.0480506
beta 1 1.3 9812e-006 3.90961e-007 6.31853e-007 2.16439e-006
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 10 0.045 0.046 0.004 0.005 -0.625
500 11 0.048 0.0467 0.004 0.005 0.871
1,750 11 0.048 0.0484 0.007 0.005 -0.289
5000 11 0.053 0.053 0.005 0.005 0.0139
07/14/2009
B-164
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
206.956762
209.411887
206.956762
206.331505
200.733556
# Param's
5
8
5
3
2
AIC
-403.913524
-402.823775
-403.913524
-406.663010
-397.467113
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
17.3567
4 . 91025
4 . 91025
1.25051
0. 008058
0. 1785
0. 1785
0.5351
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1.
model appears to be appropriate here
A homogeneous variance
07/14/2009
B-165
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 3575 . 9
BMDL = 2394.08
07/14/2009
B-166
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Adrenal Gland
Endpoint: Adrenal Gland Weight (absolute)
Species/Gender: Sprague-Dawlev Male Rats
(White et al., 1995)
Table B-60. Mean absolute adrenal gland weight (and SD) in Sprague-
Dawley male rats exposed to four different concentrations of ETBE via
inhalation for 4 weeks
Administered dose (ppm)
Mean adrenal gland weight, g (SD)
Control (n = 9)
0.051 (0.004)
500 (n= 10)
0.047 (0.006)
2,000 (n= 10)
0.051 (0.006)
4,000 (n= 10)
0.0583 (0.008)
aStatistically significantly different from control at p < 0.05 as reported by White et al. (1995).
Source: White et al. (1995).
07/14/2009
B-167
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-61. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute adrenal gland weight in Sprague-Dawley male rats exposed to
ETBE via inhalation for 4 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (2°)
0.17
-351.6
3,522
2,341
Power
0.14
-351.3
3,367
2,220
Hill
NA
-348.6
3,912
1,898
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: White et al. (1995).
07/14/2009
B-168
DRAFT - DO NOT CITE OR QUOTE
-------
Power Model with 0.95 Confidence Level
Power
~T
0.06
-------
rho is set to 0
The power is restricted to be greater than or equal to 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 3.86286e-005
rho = 0 Specified
control = 0.047
slope = 6.08741e-008
power = 1.45943
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha control slope power
alpha 1 1.8e-009 -3.9e-009 3.9e-009
control 1.8e-009 1 -0.51 0.5
slope -3.9e-009 -0.51 1 -1
power 3.9e-009 0.5 -1 1
Parameter Estimates
Variable
alpha
control
slope
power
Estimate
3.66 9 96e-005
0.0489671
3.45133e-011
2.33737
Std. Err.
. 31082e-006
0.00140151
. 59598e-010
1.59943
95.0% Wald Confidence Interval
Lower Conf. Limit Upper Conf. Limit
2.04107 e-005
0.0462202
-8.662 82e-010
-0.79745
5.2 98 85e-005
0.051714
9.35308e-010
5. 47219
Table of Data and Estimated Values of Interest
Dose
Obs Mean
Est Mean Obs Std Dev Est Std Dev Scaled Res.
0
500
2000
4000
9
10
10
10
0.051
0.047
0.051
0.058
0. 049
0. 049
0.0508
0. 058
0.004
0.006
0.006
0.008
0.00606
0. 00606
0. 00606
0. 00606
1. 01
-1.06
0.125
-0.0165
Model Descriptions for likelihoods calculated
07/14/2009
B-170
DRAFT - DO NOT CITE OR QUOTE
-------
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log(likelihood) # Param's AIC
A1 180.759773 5 -351.519546
A2 182.906633 8 -349.813266
A3 180.759773 5 -351.519546
fitted 179.648531 4 -351.297063
R 173.330901 2 -342.661801
Explanation of Tests
Test 1:
Test
Test
Test
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test -2*log(Likelihood Ratio) Test df p-value
Test 1 19.1515 6 0.003915
Test 2 4.29372 3 0.2314
Test 3 4.29372 3 0.2314
Test 4 2.22248 1 0.136
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
07/14/2009
B-171
DRAFT - DO NOT CITE OR QUOTE
-------
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 33 66.55
BMDL = 2219.59
07/14/2009 B-172 DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Bone Marrow
Endpoint: Bone Marrow Congestion
Species/Gender: F344 Female Rats
(Medinsky et al., 1999)
Table B-62. Incidence of bone marrow congestion in F344 female rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Incidence
Control
0/10
500
0/11
1,750
5/1 r
5,000
ii/iia
aStatistically significantly different from control at p< 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-173
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-63. A summary of BMDS (version 1.4.1) modeling results based on
incidence of bone marrow congestion in F344 female rats exposed to ETBE
via inhalation for 13 weeks
Model
%2/'-value
AIC
BMC10 (ppm)
BMCL10 (ppm)
Gamma
1.00
17.2
1,305
568
Logistic
1.00
19.2
1,615
753
Log-logistic
1.00
17.2
1,565
678
Multistage (3°)
0.98
17.5
987
401
Probit
1.00
19.2
1,488
684
Log-probit
1.00
19.2
1,418
642
Quantal-linear
0.23
24.8
253
162
Weibull
1.00
19.2
1,456
526
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCio = Benchmark concentration corresponding to a 10% change relative to controls.
BMCLio = 95% lower confidence limit on the benchmark concentration corresponding to a 10% change relative to
controls.
Source: Medinsky et al. (1999).
07/14/2009
B-174
DRAFT - DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
1
0.8
T3
CD
| 0.6
<
£Z
O
B 0.4
ro
Lj_
0.2
0
0 1000 2000 3000 4000 5000
dose
09:14 08/07 2007
Source: Medinsky et al., (1999).
Figure B-28. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
3° multistage) based on incidence of bone marrow congestion in F344 female
rats exposed to ETBE via inhalation for 13 weeks.
Multistage Model. (Version: 2.8; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_FEMALES_BONE_MARROW_CONGEST_INCIDENCE_MEDINSKY9
9. (d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_FEMALES_BONE_MARROW_CONGEST_INCIDENCE_MEDINSKY9
9.pit
Tue Aug 07 09:14:48 2007
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*doseAl-beta2*doseA2-beta3*doseA3)]
Multistage
^ "i
// }
; T /
- / 1 -
: bmdU bmd
07/14/2009
B-175
DRAFT - DO NOT CITE OR QUOTE
-------
The parameter betas are restricted to be positive
Dependent variable = Response
Independent variable = Dose
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta (1) = 0
Beta(2) = 0
Beta(3) = 8.10541e+008
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background -Beta(l) -Beta(2)
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Beta(3)
Beta(3) 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Ccrif. Limit Upper Ccrif. Limit
E'.ackgrO'Urid 0 ^ ^ ^
Beta(l) 0 *
Beta(2) 0 ^ ^ ^
Beta(3) 1.09591e-010 -
- Indicates that this value is nc't calculated.
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
Log(likelihood) # Param's Deviance Test d.f. P-value
-7.5791 4
-7.73218 1 0.306165 3 0.9589
-28.3826 1 41.607 3 <.0001
AIC:
17 .4644
07/14/2009
B-176
DRAFT - DO NOT CITE OR QUOTE
-------
Dose
Est. Prob.
Goodness of Fit
Expected Observed Size
Scaled
Residual
0.0000
500.0000
1,750.0000
5000.0000
ChiA2 = 0.16
0.0000
0.0136
0. 4442
1.0000
0. 000
0.150
4.886
11.000
d.f. = 3
11
P-value = 0.9843
10
11
11
11
0.000
-0.390
0. 069
0.004
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 9 8 6.9 64
BMDL = 400.522
BMDU = 1304 . 05
Taken together, (400.522, 1304.05) is a 90 % two-sided confidence
interval for the BMD
07/14/2009
B-177
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Testes
Endpoint: Degenerated Spermatocytes
Species/Gender: F344 Male Rats
(Medinsky et al., 1999)
Table B-64. Mean degenerated spermatocytes (and SD) in the testes of F344
male rats exposed to four different concentrations of ETBE via inhalation
for 13 weeks
Administered dose (ppm)
Mean degenerated spermatocytes (SD)
Control (n= 11)
2.09 (0.944)
500 (n= 11)
2.36 (1.80)
1,750 (n= 11)
7.82a (3.71)
5,000 (n= 11)
12.703 (10.8)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Sources: Medinsky et al (1999).
07/14/2009
B-178
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-65. A summary of BMDS (version 1.4.1) modeling results based on
mean degenerated spermatocytes in the testes of F344 male rats exposed to
ETBE via inhalation for 13 weeks
Model (non-constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.41
145.7
397
268
Power
0.19
147.7
425
268
Hill
NA
147.9
598
307
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
NA = Degrees of freedom for the chi-square test for goodness-of-fit are less than or equal to 0; therefore, this test is
not valid for evaluating lack of fit.
Source: Medinsky et al. (1999).
07/14/2009
B-179
DRAFT - DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
Linear
20
15
10
5
BMDLj Emd
0
0
1000
2000
3000
4000
5000
dose
13:45 09/07 2007
Source: Medinsky et al., 1999.
Figure B-29. BMDS (version 1.4.1) model output for the best-fit model (i.e.,
1° polynomial or linear) based on mean regenerative spermatocytes in the
testes of F344 male rats exposed to ETBE via inhalation for 13 weeks.
Polynomial Model. (Version: 2.12; Date: 02/20/2007)
Input Data File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_MALES_TESTES_DEGEN_SPERMATOCYTES_MEDINSKY9 9.(d)
Gnuplot Plotting File: G:\ETBE DOSE-RESPONSE
MODELING\INHALATION\F34 4_RATS_MALES_TESTES_DEGEN_SPERMATOCYTES_MEDINSKY9 9.pit
Fri Sep 07 13:45:21 2007
BMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2 + . . .
Dependent variable = MEAN
07/14/2009 B-180 DRAFT - DO NOT CITE OR QUOTE
-------
Independent variable = Dose
Signs of the polynomial coefficients are not restricted
The variance is to be modeled as Var(i) = exp(lalpha + log(mean(i)) * rho)
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
lalpha = 3.51992
rho = 0
beta 0 = 2 . 28527
beta~l = 0.00218743
Asymptotic Correlation Matrix of Parameter Estimates
lalpha rho beta 0 beta 1
lalpha 1 -0.91 -0.091 0.1
rho -0.91 1 0.088 -0.097
beta_0 -0.091 0.088 1 -0.36
beta 1 0.1 -0.097 -0.36 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
lalpha -1.61488 0.506123 -2.60686 -0.622894
rho 2.38634 0.291921 1.81419 2.9585
beta_0 1.9212 0.266417 1.39903 2.44336
beta 1 0.00245095 0.000456859 0.00155553 0.00334638
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 11 2.09 1.92 0.944 0.972 0.579
500 11 2.36 3.15 1.8 1.75 -1.48
1,750 11 7.82 6.21 3.71 3.94 1.35
5000 11 12.7 14.2 10.8 10.6 -0.455
07/14/2009
B-181
DRAFT - DO NOT CITE OR QUOTE
-------
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = exp(lalpha + rho*ln(Mu(i)))
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
-97.341461
-66.381513
-67.972913
-68.858357
-108.026956
# Param's
5
8
6
4
2
AIC
204.682921
148 . 763026
147 . 945826
145.716715
220. 053912
Explanation of Tests
Test 1:
Test 2
Test 3
Test 4
(Note:
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
83.2909
61.9199
3.1828
1.77089
<.0001
<.0001
0.2036
0.4125
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .1. A non-homogeneous variance
model appears to be appropriate
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
07/14/2009
B-182
DRAFT - DO NOT CITE OR QUOTE
-------
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect = 1
Risk Type = Estimated standard deviations from the control mean
Confidence level = 0.95
BMD = 3 9 6.596
BMDL = 267.653
07/14/2009
B-183
DRAFT - DO NOT CITE OR QUOTE
-------
Organ: Heart
Endpoint: Heart Weight (absolute)
Species/Gender: F344 Female Rats
(Medinsky et al., 1999)
Table B-66. Mean absolute heart weight (and SD) in F344 female rats
exposed to four different concentrations of ETBE via inhalation for 13 weeks
Administered dose (ppm)
Mean heart weight, g (SD)
Control (n = 10)
0.495 (0.034)
500 (n= 11)
0.5453 (0.037)
1,750 (n= 11)
0.532 (0.040)
5,000 (n= 11)
0.5563 (0.032)
aStatistically significantly different from control at p < 0.05 as reported by Medinsky et al. (1999).
Source: Medinsky et al. (1999).
07/14/2009
B-184
DRAFT - DO NOT CITE OR QUOTE
-------
Table B-67. A summary of BMDS (version 1.4.1) modeling results based on
mean absolute heart weight in F344 female rats exposed to ETBE via
inhalation for 13 weeks
Model (constant variance)
%2/'-value
AIC
BMC1SD (ppm)
BMCL1SD (ppm)
Polynomial (1°)
0.01
-232.7
4,674
2,873
Power
0.01
-232.7
4,674
2,873
Hill
0.11
-236.7
105
Computation of the
lower bound failed
Chi-square /j-value = /j-value from the chi-square test for goodness-of-fit (/^-values <0.1 indicate that the model
exhibits significant lack-of-fit).
BMCisd = Benchmark concentration corresponding to a 1 SD change relative to the control mean.
BMCLiSD = 95% lower confidence limit on the benchmark concentration corresponding to a 1 SD change relative
to the control mean.
Source: Medinsky et al. (1999).
07/14/2009
B-185
DRAFT - DO NOT CITE OR QUOTE
-------
Hill Model with 0.95 Confidence Level
CD
£Z
O
Q_
(fl
iIJ
DL
c:
ro
-------
rho is set to 0
Power parameter restricted to be greater than 1
A constant variance model is fit
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-OOf
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 0.00129062
rho = 0 Specified
intercept = 0.495
v = 0.061
n = 0.234371
k = 695
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -rho -n
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
alpha intercept v k
alpha 1 -8.8e-010 2.7e-008 4.4e-008
intercept -8.8e-010 1 -0.72 0.07
v 2.7e-008 -0.72 1 0.5
k 4.4e-008 0.07 0.5 1
Parameter Estimates
95.0% Wald Confidence Interval
Variable Estimate Std. Err. Lower Conf. Limit Upper Conf. Limit
alpha 0.00124247 0.000267959 0.000717284 0.00176766
intercept 0.495017 0.0111515 0.47316 0.516873
v 0.0514534 0.0152581 0.021548 0.0813587
n 1 NA
k 48.4598 201.386 -346.25 443.169
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Table of Data and Estimated Values of Interest
Dose N Obs Mean Est Mean Obs Std Dev Est Std Dev Scaled Res.
0 10 0.495 0.495 0.034 0.0352 -0.00152
07/14/2009 B-187 DRAFT - DO NOT CITE OR QUOTE
-------
500 11 0.545 0.542 0.037 0.0352 0.289
1,750 11 0.532 0.545 0.04 0.0352 -1.23
5000 11 0.556 0.546 0.032 0.0352 0.943
Model Descriptions for likelihoods calculated
Model A1: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma(i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = SigmaA2
Model A3 uses any fixed variance parameters that
were specified by the user
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model
A1
A2
A3
fitted
R
Log(likelihood)
123.630904
123.948426
123.630904
122 . 349007
115.878403
# Param's
5
8
5
4
2
AIC
-237.261808
-231. 896853
-237.261808
-236.698014
-227.756806
Test 1:
Test
Test
Test
Explanation of Tests
Do responses and/or variances differ among Dose levels?
(A2 vs. R)
Are Variances Homogeneous? (A1 vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
(Note: When rho=0 the results of Test 3 and Test 2 will be the same.)
Tests of Interest
Test
-2*log(Likelihood Ratio) Test df
p-value
Test 1
Test 2
Test 3
Test 4
16. 14
0.635045
0.635045
2 . 56379
0. 01302
0. 8884
0. 8884
0.1093
The p-value for Test 1 is less than .05. There appears to be a
difference between response and/or variances among the dose levels
It seems appropriate to model the data
The p-value for Test 2 is greater than .1. A homogeneous variance
07/14/2009
B-188
DRAFT - DO NOT CITE OR QUOTE
-------
model appears to be appropriate here
The p-value for Test 3 is greater than .1. The modeled variance appears to
be appropriate here
The p-value for Test 4 is greater than .1. The model chosen seems
to adequately describe the data
Benchmark Dose Computation
Specified effect
1
Risk Type
Estimated standard deviations from the control mean
Confidence level
0. 95
BMD
105.411
BMDL
5e-012
07/14/2009
B-189 DRAFT - DO NOT CITE OR QUOTE
-------
APPENDIX C. DERIVATION OF THE ORAL MINIMAL DATA VALUE
Considering the uncertainties in the ETBE database described below, the total composite
UF is for the derivation of an RfD is 10,000, consisting of four areas of maximum uncertainty.
In the report, A Review of the Reference Dose and Reference Concentration Processes (U.S.
EPA, 2002), the RfD/RfC technical panel concluded that, in cases where maximum uncertainty
exists in four or more areas of uncertainty, or when the total UF is 10,000 or more, it is unlikely
that the database is sufficient to derive a reference value. Because of this uncertainty, an RfD is
not derived and instead an oral minimal data value is presented. The use of the minimal data
value for ETBE is not recommended except in limited circumstances, for example, in screening
level risk assessments or to rank relative risks. Any use of this value should include a discussion
of the uncertainty associated with its derivation.
The minimal data value is based on increased relative kidney weight in F0 generation
male rats exposed to ETBE by gavage as part of a two-generation reproduction and fertility study
(CIT 2004b, unpublished report). This study was chosen as the principal study because it
provides the most sensitive measure of effects of oral exposure to ETBE. Increased relative
kidney weight in F0 generation males was selected as the critical effect because represents the
most sensitive effect and resulted in the lowest BMDL. BMD modeling revealed that the BMDL
associated with the increased relative kidney weight is 143 mg/kg-day. The BMDL provides the
POD for the minimal data value.
A total UF of 10,000 was applied to the POD of 143 mg/kg-day: 10 for interspecies
extrapolation from animals to humans (UFA); 10 for human intraspecies variability (UFH); 10 for
extrapolation from a subchronic to a chronic study (UFS); and 10 to account for database
deficiencies (UFD).
A 10-fold UF was used to account for uncertainties in extrapolating from rats to humans.
The available data do not provide evidence that rats, or any other species, are more sensitive to
ETBE than humans. Consequently, the default UF value of 10 for extrapolating from laboratory
animals to humans was applied.
A 10-fold UF was used to account for variation in susceptibility among members of the
human population (i.e., interindividual variability). Insufficient information is available to
predict potential variability in human susceptibility.
The duration of exposure in the principal study (CIT, 2004b, unpublished report) is 120
days. This length of exposure is greater than the 90 day exposure period commonly utilized in
subchronic studies but falls short of a chronic exposure duration. The only chronic exposure
study in the database for ETBE is a carcinogenicity bioassay that did not report any noncancer
effects other than mortality (Maltoni et al., 1999). Therefore, no data are available to inform the
nature and extent of effects that would be observed with a longer duration of exposure to ETBE.
07/14/2009
C-l
DRAFT - DO NOT CITE OR QUOTE
-------
For these reasons, a 10-fold UF was used to account for the extrapolation from subchronic to
chronic exposure duration.
A 10-fold UF was used to account for deficiencies in the toxicity database on oral
exposure to ETBE. There are no available human occupational or epidemiological studies of
oral exposure to ETBE. There are no standard subchronic or chronic toxicity animal studies
available for oral exposure to ETBE. The toxicity data on oral exposure to ETBE is limited and
largely restricted to a series of unpublished prenatal developmental toxicity and two-generation
reproduction and fertility studies (CIT, 2004a, b, 2003, unpublished reports). Due to the limited
scope and design of the reproductive and development studies, these studies cannot be
considered an adequate assessment of general toxicity from oral exposure to ETBE. In
particular, the lack of systematic histopathological data on the liver and kidney that would be
part of a standard subchronic or chronic toxicity study represents a limitation on the available
data. Note that the database UF is not applied because of a lack of a chronic study per se, but
because of a lack of studies that examined multiple systemic endpoints.
A UF for LOAEL-to-NOAEL extrapolation was not used because the current approach is
to address this factor as one of the considerations in selecting a BMR for BMD modeling. In this
case, a BMR of a change of one SD from the control mean was selected under the assumption
that it represents a minimal biologically significant change.
The oral minimal data value for ETBE was calculated as follows:
Minimal data value = BMDL UF
= 143 mg/kg-day ^ 10,000
= 0.0143 or 1 x 10"2 mg/kg-day
The overall confidence in this chronic oral minimal data value is low. Confidence in the
principal study (CIT, 2004b, unpublished report) is medium. Confidence in the database is low
due to the limited scope and design of the reproductive and development studies that comprise
the available data. These studies cannot be considered an adequate assessment of general
toxicity from oral exposure to ETBE, particularly because the studies lack histopathological data
on the liver and kidney. Reflecting medium confidence in the principal study and low
confidence in the database, confidence in the minimal data value is low.
Figure C-l presents the POD, applied UFs, and derived potential chronic RfVs for
additional endpoints that were modeled using EPA BMDS (version 1.4.1), or for endpoints
where the data were not amenable to BMD modeling as indicated by NOAELs and LOAELs.
This comparison is intended to provide information on additional effects associated with ETBE
exposure.
PODs and potential chronic RfVs that could be derived from the additional effects
identified in Table 5-1 are presented in Figure C-l to allow a comparison with the chosen critical
07/14/2009 C-2 DRAFT - DO NOT CITE OR QUOTE
-------
effect and resulting minimal data value. For reduced body weight gain, and increased kidney
weights in males, the total UF factor applied was 10,000-fold: 10-fold to account for uncertainty
in extrapolating from laboratory animals to humans, 10-fold UF to account for variation in
susceptibility among members of the human population, 10-fold UF to account for subchronic-
to-chronic extrapolation, and 10-fold UF for database deficiencies. For increased liver weight in
male rats of the F1 generation and increased kidney weights in female rats of the F1 generation,
the total UF applied was 1000-fold: 10-fold to account for uncertainty in extrapolating from
laboratory animals to humans, 10-fold UF to account for variation in susceptibility among
members of the human population, and 10-fold UF for database deficiencies. A UF was not used
to account for extrapolating from less than chronic exposure for these endpoints because these
endpoints represent developmental toxicity resulting from a narrow period of exposure.
The increased kidney weight is the effect with the lowest BMDL and the lowest potential
reference value from the data amenable to BMD modeling. Increased kidney weights were
observed in both males and females with effects at lower doses in males. Given the data
suggesting an effect of ETBE on body weight gain (e.g., in males of the FO generation), relative
kidney weights were considered a more appropriate measure than absolute kidney weights
because of the potential effect that body weight changes may have on absolute organ weights.
The dose-response relationship for increased liver weight for oral exposure to ETBE and reduced
body weight gain are also suitable for deriving a chronic reference value, but are associated with
higher BMDLs. An increase in mortality was also considered for the critical effect. Given the
frank effect, and lack of quantitative mortality data, and a consistent dose-response trend, this
endpoint was not considered ideal for the derivation of a chronic reference value and was not
amenable to BMD modeling. Consideration of the available dose-response data to determine an
estimate of oral exposure that is likely to be without an appreciable risk of adverse health effects
over a lifetime has led to the selection of the two-generation reproduction and fertility study in
Sprague-Dawley rats (CIT 2004b, unpublished report) as the principal study and increased
relative kidney weight in FO generation male rats as the critical effect for deriving the oral
minimal data value for ETBE. As discussed above, data suggesting kidney toxicity associated
with oral exposure to ETBE is limited to increased kidney weights in males and females (at
higher doses), with limited histopathological support. Additional evidence of kidney toxicity is
provided by the ETBE inhalation exposure database, which includes effects such as increased
cellular proliferation in the kidney, histological evidence of cellular necrosis (increased incidence
of regenerative foci) in the kidneys, and increased kidney weights (Medinsky et al., 1999; White
et al., 1995).
07/14/2009
C-3
DRAFT - DO NOT CITE OR QUOTE
-------
1000.00
100.00
10.00
a
¦o
£i
a
1.00
0.10
0.01
0.00
NOAEL/
LOAEL
BMD Analyses
LIVER -
KIDNEY -
KIDNEY -
KIDNEY -
KIDNEY -
MORTALITY -
[male SD rats at
56 wksl
LIVER -
KIDNEY -
KIDNEY -
Weight
Weight
(absolute)
(absolute)
(relative)
(absolute)
(relative)
(relative)
(absolute)
(relative)
IFlgen. male
IFOgen. male
IFOgen. male IFlgen. lemale IFlgen. female
IFlgen. male
IFlgen. male
IFlgen. male
SD rats I
SD rats I
SD rats I
SD rats I
SD rats I
SD rats I
SD rats I
SD rats I
B ody W eight B o dy Weight B ody Weight
Change - Dam Change (net Change - F0
[female SD rats] change) - Dam male SD rats
[female SD rats] (days 85-113)
Endpoint [Sex, Strain, Species]
Figure C-l. Potential RfV comparison array for alternative PODs for oral data.
OPOD
flj Animal-to-human
~Human variation
H LOAEL to NOAEL
Q Subchr to Chronic
^Database deficiencies
~ Oral Screening Value
07/14/2009
C-4
DRAFT - DO NOT CITE OR QUOTE
------- |